src/reportlab/platypus/paraparser.py
author robin
Wed, 26 Mar 2014 12:32:02 +0000
changeset 4067 712e1822ca31
parent 3947 826044f2a532
parent 4059 9b48d0d9f137
child 4077 ac3fcb7cc6f4
permissions -rw-r--r--
merge py33 to default
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
3617
ae5744e97c42 reportlab: copyright date changes
robin
parents: 3552
diff changeset
     1
#Copyright ReportLab Europe Ltd. 2000-2012
494
54257447cfe9 Changed to indirect copyright
rgbecker
parents: 433
diff changeset
     2
#see license.txt for license details
2332
2a7ab4405e18 Remove $Header:, fix CopyRight & history
rgbecker
parents: 2321
diff changeset
     3
#history http://www.reportlab.co.uk/cgi-bin/viewcvs.cgi/public/reportlab/trunk/reportlab/platypus/paraparser.py
2321
3454f5b41760 Unicode and UTF8 support changes
andy
parents: 2200
diff changeset
     4
__version__=''' $Id$ '''
3032
22224b1b4d24 New docstrings mainly for module titles
damian
parents: 2964
diff changeset
     5
__doc__='''The parser used to process markup within paragraphs'''
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     6
import string
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
     7
import re
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     8
import sys
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     9
import os
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    10
import copy
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
    11
import base64
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
    12
from pprint import pprint as pp
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
    13
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
    14
try:
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
    15
    import pickle as pickle
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
    16
except:
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
    17
    import pickle
2693
3c61a57aecd1 missing import statement
andy
parents: 2670
diff changeset
    18
import unicodedata
279
e7d8b3631d5c Global sequencer put in the 'story builder'.
andy_robinson
parents: 267
diff changeset
    19
import reportlab.lib.sequencer
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
    20
518
5be3fcb26c78 Semantic Name changes
rgbecker
parents: 514
diff changeset
    21
from reportlab.lib.abag import ABag
4004
ebd460ef8d2f rename utils.py UniChr --> uniChr
robin
parents: 3994
diff changeset
    22
from reportlab.lib.utils import ImageReader, isPy3, annotateException, encode_label, asUnicode, asBytes, uniChr
248
c103b7a55e79 Color fixes; thanks to J Alet
rgbecker
parents: 238
diff changeset
    23
from reportlab.lib.colors import toColor, white, black, red, Color
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    24
from reportlab.lib.fonts import tt2ps, ps2tt
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    25
from reportlab.lib.enums import TA_LEFT, TA_RIGHT, TA_CENTER, TA_JUSTIFY
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    26
from reportlab.lib.units import inch,mm,cm,pica
3955
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    27
if isPy3:
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    28
    from html.parser import HTMLParser
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    29
    from html.entities import name2codepoint
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    30
else:
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    31
    from HTMLParser import HTMLParser
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    32
    from htmlentitydefs import name2codepoint
3dbd805c708c paraparser.py: fix HTMLParser imports
robin
parents: 3954
diff changeset
    33
2410
f505ed647678 reportlab: add fix for <para/> and new test
rgbecker
parents: 2376
diff changeset
    34
_re_para = re.compile(r'^\s*<\s*para(?:\s+|>|/>)')
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    35
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
    36
sizeDelta = 2       # amount to reduce font size by for super and sub script
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
    37
subFraction = 0.5   # fraction of font size that a sub script should be lowered
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
    38
superFraction = 0.5 # fraction of font size that a super script should be raised
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    39
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
    40
DEFAULT_INDEX_NAME='_indexAdd'
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
    41
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    42
def _convnum(s, unit=1, allowRelative=True):
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    43
    if s[0] in ('+','-') and allowRelative:
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    44
        try:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    45
            return ('relative',int(s)*unit)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    46
        except ValueError:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    47
            return ('relative',float(s)*unit)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    48
    else:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    49
        try:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    50
            return int(s)*unit
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    51
        except ValueError:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    52
            return float(s)*unit
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    53
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    54
def _num(s, unit=1, allowRelative=True):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    55
    """Convert a string like '10cm' to an int or float (in points).
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    56
       The default unit is point, but optionally you can use other
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    57
       default units like mm.
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    58
    """
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    59
    if s.endswith('cm'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    60
        unit=cm
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    61
        s = s[:-2]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    62
    if s.endswith('in'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    63
        unit=inch
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    64
        s = s[:-2]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    65
    if s.endswith('pt'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    66
        unit=1
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    67
        s = s[:-2]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    68
    if s.endswith('i'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    69
        unit=inch
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    70
        s = s[:-1]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    71
    if s.endswith('mm'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    72
        unit=mm
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    73
        s = s[:-2]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    74
    if s.endswith('pica'):
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    75
        unit=pica
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
    76
        s = s[:-4]
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    77
    return _convnum(s,unit,allowRelative)
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    78
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    79
def _numpct(s,unit=1,allowRelative=False):
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    80
    if s.endswith('%'):
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    81
        return _PCT(_convnum(s[:-1],allowRelative=allowRelative))
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    82
    else:
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    83
        return _num(s,unit,allowRelative)
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    84
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    85
class _PCT:
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    86
    def __init__(self,v):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    87
        self._value = v*0.01
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    88
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    89
    def normalizedValue(self,normalizer):
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
    90
        normalizer = normalizer or getattr(self,'_normalizer')
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    91
        return normalizer*self._value
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    92
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    93
def _valignpc(s):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    94
    s = s.lower()
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    95
    if s in ('baseline','sub','super','top','text-top','middle','bottom','text-bottom'):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    96
        return s
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    97
    if s.endswith('%'):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    98
        n = _convnum(s[:-1])
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
    99
        if isinstance(n,tuple):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   100
            n = n[1]
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   101
        return _PCT(n)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   102
    n = _num(s)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   103
    if isinstance(n,tuple):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   104
        n = n[1]
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   105
    return n
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   106
2836
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   107
def _autoLeading(x):
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   108
    x = x.lower()
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   109
    if x in ('','min','max','off'):
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   110
        return x
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   111
    raise ValueError('Invalid autoLeading=%r' % x )
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   112
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   113
def _align(s):
3731
b233dd0577ff another round of changes mostly type related
rptlab
parents: 3723
diff changeset
   114
    s = s.lower()
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   115
    if s=='left': return TA_LEFT
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   116
    elif s=='right': return TA_RIGHT
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   117
    elif s=='justify': return TA_JUSTIFY
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   118
    elif s in ('centre','center'): return TA_CENTER
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   119
    else: raise ValueError
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   120
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   121
_paraAttrMap = {'font': ('fontName', None),
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   122
                'face': ('fontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   123
                'fontsize': ('fontSize', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   124
                'size': ('fontSize', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   125
                'leading': ('leading', _num),
2836
66fb84201abe platypus: add support for autoLeading
rgbecker
parents: 2748
diff changeset
   126
                'autoleading': ('autoLeading', _autoLeading),
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   127
                'lindent': ('leftIndent', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   128
                'rindent': ('rightIndent', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   129
                'findent': ('firstLineIndent', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   130
                'align': ('alignment', _align),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   131
                'spaceb': ('spaceBefore', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   132
                'spacea': ('spaceAfter', _num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   133
                'bfont': ('bulletFontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   134
                'bfontsize': ('bulletFontSize',_num),
2860
3f14d66194c2 platypus: added bulletOffsetY inspired by haraldarminmassa@gmail.com
rgbecker
parents: 2857
diff changeset
   135
                'boffsety': ('bulletOffsetY',_num),
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   136
                'bindent': ('bulletIndent',_num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   137
                'bcolor': ('bulletColor',toColor),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   138
                'color':('textColor',toColor),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   139
                'backcolor':('backColor',toColor),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   140
                'bgcolor':('backColor',toColor),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   141
                'bg':('backColor',toColor),
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   142
                'fg': ('textColor',toColor),
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   143
                }
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   144
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   145
_bulletAttrMap = {
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   146
                'font': ('bulletFontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   147
                'face': ('bulletFontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   148
                'size': ('bulletFontSize',_num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   149
                'fontsize': ('bulletFontSize',_num),
2860
3f14d66194c2 platypus: added bulletOffsetY inspired by haraldarminmassa@gmail.com
rgbecker
parents: 2857
diff changeset
   150
                'offsety': ('bulletOffsetY',_num),
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   151
                'indent': ('bulletIndent',_num),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   152
                'color': ('bulletColor',toColor),
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   153
                'fg': ('bulletColor',toColor),
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   154
                }
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   155
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   156
#things which are valid font attributes
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   157
_fontAttrMap = {'size': ('fontSize', _num),
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   158
                'face': ('fontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   159
                'name': ('fontName', None),
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   160
                'fg':   ('textColor', toColor),
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   161
                'color':('textColor', toColor),
2446
6b9268ab33c3 allow solid para background
andy
parents: 2410
diff changeset
   162
                'backcolor':('backColor',toColor),
6b9268ab33c3 allow solid para background
andy
parents: 2410
diff changeset
   163
                'bgcolor':('backColor',toColor),
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   164
                }
3552
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   165
#things which are valid span attributes
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   166
_spanAttrMap = {'size': ('fontSize', _num),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   167
                'face': ('fontName', None),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   168
                'name': ('fontName', None),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   169
                'fg':   ('textColor', toColor),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   170
                'color':('textColor', toColor),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   171
                'backcolor':('backColor',toColor),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   172
                'bgcolor':('backColor',toColor),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   173
                'style': ('style',None),
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   174
                }
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   175
#things which are valid font attributes
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   176
_linkAttrMap = {'size': ('fontSize', _num),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   177
                'face': ('fontName', None),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   178
                'name': ('fontName', None),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   179
                'fg':   ('textColor', toColor),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   180
                'color':('textColor', toColor),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   181
                'backcolor':('backColor',toColor),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   182
                'bgcolor':('backColor',toColor),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   183
                'dest': ('link', None),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   184
                'destination': ('link', None),
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   185
                'target': ('link', None),
2594
746800f5caf9 reportlab: fix up links in paragraphs
rgbecker
parents: 2585
diff changeset
   186
                'href': ('link', None),
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   187
                }
2744
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   188
_anchorAttrMap = {'fontSize': ('fontSize', _num),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   189
                'fontName': ('fontName', None),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   190
                'name': ('name', None),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   191
                'fg':   ('textColor', toColor),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   192
                'color':('textColor', toColor),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   193
                'backcolor':('backColor',toColor),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   194
                'bgcolor':('backColor',toColor),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   195
                'href': ('href', None),
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   196
                }
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   197
_imgAttrMap = {
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   198
                'src': ('src', None),
3434
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
   199
                'width': ('width',_numpct),
3c14212cc997 platypus: preliminary working version of % height/width for <img> tag
rgbecker
parents: 3368
diff changeset
   200
                'height':('height',_numpct),
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   201
                'valign':('valign',_valignpc),
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   202
                }
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   203
_indexAttrMap = {
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   204
                'name': ('name',None),
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   205
                'item': ('item',None),
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   206
                'offset': ('offset',None),
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   207
                'format': ('format',None),
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   208
                }
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   209
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   210
def _addAttributeNames(m):
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   211
    K = list(m.keys())
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   212
    for k in K:
1944
a50f8e3f93f8 laissez faire case
rgbecker
parents: 1940
diff changeset
   213
        n = m[k][0]
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   214
        if n not in m: m[n] = m[k]
3731
b233dd0577ff another round of changes mostly type related
rptlab
parents: 3723
diff changeset
   215
        n = n.lower()
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   216
        if n not in m: m[n] = m[k]
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   217
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   218
_addAttributeNames(_paraAttrMap)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   219
_addAttributeNames(_fontAttrMap)
3552
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   220
_addAttributeNames(_spanAttrMap)
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   221
_addAttributeNames(_bulletAttrMap)
2747
46005202d9d0 paraparser: fix link/anchor attributes
rgbecker
parents: 2745
diff changeset
   222
_addAttributeNames(_anchorAttrMap)
46005202d9d0 paraparser: fix link/anchor attributes
rgbecker
parents: 2745
diff changeset
   223
_addAttributeNames(_linkAttrMap)
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   224
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   225
def _applyAttributes(obj, attr):
3723
99aa837b6703 second stage of port to Python 3.3; working hello world
rptlab
parents: 3721
diff changeset
   226
    for k, v in attr.items():
3787
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
   227
        if isinstance(v,(list,tuple)) and v[0]=='relative':
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   228
            if hasattr(obj, k):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   229
                v = v[1]+getattr(obj,k)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   230
            else:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   231
                v = v[1]
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   232
        setattr(obj,k,v)
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   233
1931
784fce255e2d Added in more special entities as suggested by Christoph Zwerschke
rgbecker
parents: 1736
diff changeset
   234
#Named character entities intended to be supported from the special font
2200
be0cfccc662a Fixed up tabs and whitespace in all source files
andy_robinson
parents: 2053
diff changeset
   235
#with additions suggested by Christoph Zwerschke who also suggested the
1931
784fce255e2d Added in more special entities as suggested by Christoph Zwerschke
rgbecker
parents: 1736
diff changeset
   236
#numeric entity names that follow.
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   237
greeks = {
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   238
    'Aacute': u'\xc1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   239
    'aacute': u'\xe1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   240
    'Acirc': u'\xc2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   241
    'acirc': u'\xe2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   242
    'acute': u'\xb4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   243
    'AElig': u'\xc6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   244
    'aelig': u'\xe6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   245
    'Agrave': u'\xc0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   246
    'agrave': u'\xe0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   247
    'alefsym': u'\u2135',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   248
    'Alpha': u'\u0391',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   249
    'alpha': u'\u03b1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   250
    'and': u'\u2227',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   251
    'ang': u'\u2220',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   252
    'Aring': u'\xc5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   253
    'aring': u'\xe5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   254
    'asymp': u'\u2248',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   255
    'Atilde': u'\xc3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   256
    'atilde': u'\xe3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   257
    'Auml': u'\xc4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   258
    'auml': u'\xe4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   259
    'bdquo': u'\u201e',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   260
    'Beta': u'\u0392',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   261
    'beta': u'\u03b2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   262
    'brvbar': u'\xa6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   263
    'bull': u'\u2022',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   264
    'cap': u'\u2229',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   265
    'Ccedil': u'\xc7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   266
    'ccedil': u'\xe7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   267
    'cedil': u'\xb8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   268
    'cent': u'\xa2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   269
    'Chi': u'\u03a7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   270
    'chi': u'\u03c7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   271
    'circ': u'\u02c6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   272
    'clubs': u'\u2663',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   273
    'cong': u'\u2245',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   274
    'copy': u'\xa9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   275
    'crarr': u'\u21b5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   276
    'cup': u'\u222a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   277
    'curren': u'\xa4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   278
    'dagger': u'\u2020',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   279
    'Dagger': u'\u2021',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   280
    'darr': u'\u2193',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   281
    'dArr': u'\u21d3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   282
    'deg': u'\xb0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   283
    'delta': u'\u03b4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   284
    'Delta': u'\u2206',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   285
    'diams': u'\u2666',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   286
    'divide': u'\xf7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   287
    'Eacute': u'\xc9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   288
    'eacute': u'\xe9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   289
    'Ecirc': u'\xca',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   290
    'ecirc': u'\xea',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   291
    'Egrave': u'\xc8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   292
    'egrave': u'\xe8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   293
    'empty': u'\u2205',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   294
    'emsp': u'\u2003',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   295
    'ensp': u'\u2002',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   296
    'Epsilon': u'\u0395',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   297
    'epsilon': u'\u03b5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   298
    'epsiv': u'\u03b5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   299
    'equiv': u'\u2261',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   300
    'Eta': u'\u0397',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   301
    'eta': u'\u03b7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   302
    'ETH': u'\xd0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   303
    'eth': u'\xf0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   304
    'Euml': u'\xcb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   305
    'euml': u'\xeb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   306
    'euro': u'\u20ac',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   307
    'exist': u'\u2203',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   308
    'fnof': u'\u0192',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   309
    'forall': u'\u2200',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   310
    'frac12': u'\xbd',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   311
    'frac14': u'\xbc',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   312
    'frac34': u'\xbe',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   313
    'frasl': u'\u2044',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   314
    'Gamma': u'\u0393',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   315
    'gamma': u'\u03b3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   316
    'ge': u'\u2265',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   317
    'harr': u'\u2194',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   318
    'hArr': u'\u21d4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   319
    'hearts': u'\u2665',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   320
    'hellip': u'\u2026',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   321
    'Iacute': u'\xcd',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   322
    'iacute': u'\xed',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   323
    'Icirc': u'\xce',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   324
    'icirc': u'\xee',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   325
    'iexcl': u'\xa1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   326
    'Igrave': u'\xcc',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   327
    'igrave': u'\xec',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   328
    'image': u'\u2111',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   329
    'infin': u'\u221e',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   330
    'int': u'\u222b',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   331
    'Iota': u'\u0399',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   332
    'iota': u'\u03b9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   333
    'iquest': u'\xbf',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   334
    'isin': u'\u2208',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   335
    'Iuml': u'\xcf',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   336
    'iuml': u'\xef',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   337
    'Kappa': u'\u039a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   338
    'kappa': u'\u03ba',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   339
    'Lambda': u'\u039b',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   340
    'lambda': u'\u03bb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   341
    'lang': u'\u2329',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   342
    'laquo': u'\xab',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   343
    'larr': u'\u2190',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   344
    'lArr': u'\u21d0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   345
    'lceil': u'\uf8ee',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   346
    'ldquo': u'\u201c',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   347
    'le': u'\u2264',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   348
    'lfloor': u'\uf8f0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   349
    'lowast': u'\u2217',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   350
    'loz': u'\u25ca',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   351
    'lrm': u'\u200e',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   352
    'lsaquo': u'\u2039',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   353
    'lsquo': u'\u2018',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   354
    'macr': u'\xaf',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   355
    'mdash': u'\u2014',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   356
    'micro': u'\xb5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   357
    'middot': u'\xb7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   358
    'minus': u'\u2212',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   359
    'mu': u'\xb5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   360
    'Mu': u'\u039c',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   361
    'nabla': u'\u2207',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   362
    'nbsp': u'\xa0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   363
    'ndash': u'\u2013',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   364
    'ne': u'\u2260',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   365
    'ni': u'\u220b',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   366
    'notin': u'\u2209',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   367
    'not': u'\xac',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   368
    'nsub': u'\u2284',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   369
    'Ntilde': u'\xd1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   370
    'ntilde': u'\xf1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   371
    'Nu': u'\u039d',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   372
    'nu': u'\u03bd',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   373
    'Oacute': u'\xd3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   374
    'oacute': u'\xf3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   375
    'Ocirc': u'\xd4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   376
    'ocirc': u'\xf4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   377
    'OElig': u'\u0152',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   378
    'oelig': u'\u0153',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   379
    'Ograve': u'\xd2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   380
    'ograve': u'\xf2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   381
    'oline': u'\uf8e5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   382
    'omega': u'\u03c9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   383
    'Omega': u'\u2126',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   384
    'Omicron': u'\u039f',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   385
    'omicron': u'\u03bf',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   386
    'oplus': u'\u2295',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   387
    'ordf': u'\xaa',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   388
    'ordm': u'\xba',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   389
    'or': u'\u2228',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   390
    'Oslash': u'\xd8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   391
    'oslash': u'\xf8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   392
    'Otilde': u'\xd5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   393
    'otilde': u'\xf5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   394
    'otimes': u'\u2297',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   395
    'Ouml': u'\xd6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   396
    'ouml': u'\xf6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   397
    'para': u'\xb6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   398
    'part': u'\u2202',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   399
    'permil': u'\u2030',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   400
    'perp': u'\u22a5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   401
    'phis': u'\u03c6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   402
    'Phi': u'\u03a6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   403
    'phi': u'\u03d5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   404
    'piv': u'\u03d6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   405
    'Pi': u'\u03a0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   406
    'pi': u'\u03c0',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   407
    'plusmn': u'\xb1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   408
    'pound': u'\xa3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   409
    'prime': u'\u2032',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   410
    'Prime': u'\u2033',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   411
    'prod': u'\u220f',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   412
    'prop': u'\u221d',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   413
    'Psi': u'\u03a8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   414
    'psi': u'\u03c8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   415
    'radic': u'\u221a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   416
    'rang': u'\u232a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   417
    'raquo': u'\xbb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   418
    'rarr': u'\u2192',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   419
    'rArr': u'\u21d2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   420
    'rceil': u'\uf8f9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   421
    'rdquo': u'\u201d',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   422
    'real': u'\u211c',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   423
    'reg': u'\xae',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   424
    'rfloor': u'\uf8fb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   425
    'Rho': u'\u03a1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   426
    'rho': u'\u03c1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   427
    'rlm': u'\u200f',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   428
    'rsaquo': u'\u203a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   429
    'rsquo': u'\u2019',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   430
    'sbquo': u'\u201a',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   431
    'Scaron': u'\u0160',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   432
    'scaron': u'\u0161',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   433
    'sdot': u'\u22c5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   434
    'sect': u'\xa7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   435
    'shy': u'\xad',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   436
    'sigmaf': u'\u03c2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   437
    'sigmav': u'\u03c2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   438
    'Sigma': u'\u03a3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   439
    'sigma': u'\u03c3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   440
    'sim': u'\u223c',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   441
    'spades': u'\u2660',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   442
    'sube': u'\u2286',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   443
    'sub': u'\u2282',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   444
    'sum': u'\u2211',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   445
    'sup1': u'\xb9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   446
    'sup2': u'\xb2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   447
    'sup3': u'\xb3',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   448
    'supe': u'\u2287',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   449
    'sup': u'\u2283',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   450
    'szlig': u'\xdf',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   451
    'Tau': u'\u03a4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   452
    'tau': u'\u03c4',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   453
    'there4': u'\u2234',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   454
    'thetasym': u'\u03d1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   455
    'thetav': u'\u03d1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   456
    'Theta': u'\u0398',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   457
    'theta': u'\u03b8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   458
    'thinsp': u'\u2009',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   459
    'THORN': u'\xde',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   460
    'thorn': u'\xfe',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   461
    'tilde': u'\u02dc',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   462
    'times': u'\xd7',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   463
    'trade': u'\uf8ea',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   464
    'Uacute': u'\xda',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   465
    'uacute': u'\xfa',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   466
    'uarr': u'\u2191',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   467
    'uArr': u'\u21d1',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   468
    'Ucirc': u'\xdb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   469
    'ucirc': u'\xfb',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   470
    'Ugrave': u'\xd9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   471
    'ugrave': u'\xf9',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   472
    'uml': u'\xa8',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   473
    'upsih': u'\u03d2',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   474
    'Upsilon': u'\u03a5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   475
    'upsilon': u'\u03c5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   476
    'Uuml': u'\xdc',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   477
    'uuml': u'\xfc',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   478
    'weierp': u'\u2118',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   479
    'Xi': u'\u039e',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   480
    'xi': u'\u03be',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   481
    'Yacute': u'\xdd',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   482
    'yacute': u'\xfd',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   483
    'yen': u'\xa5',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   484
    'yuml': u'\xff',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   485
    'Yuml': u'\u0178',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   486
    'Zeta': u'\u0396',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   487
    'zeta': u'\u03b6',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   488
    'zwj': u'\u200d',
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   489
    'zwnj': u'\u200c',
1931
784fce255e2d Added in more special entities as suggested by Christoph Zwerschke
rgbecker
parents: 1736
diff changeset
   490
    }
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   491
4004
ebd460ef8d2f rename utils.py UniChr --> uniChr
robin
parents: 3994
diff changeset
   492
known_entities = dict([(k,uniChr(v)) for k,v in name2codepoint.items()])
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   493
for k in greeks:
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   494
    if k not in known_entities:
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   495
        known_entities[k] = greeks[k]
3994
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
   496
f = isPy3 and asBytes or asUnicode
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
   497
K = list(known_entities.keys())
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
   498
for k in K:
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
   499
    known_entities[f(k)] = known_entities[k]
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
   500
del k, f, K
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
   501
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   502
#------------------------------------------------------------------------
518
5be3fcb26c78 Semantic Name changes
rgbecker
parents: 514
diff changeset
   503
class ParaFrag(ABag):
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   504
    """class ParaFrag contains the intermediate representation of string
3961
8c5743bc6489 paraparser.py: remove reference to XMLParser
robin
parents: 3957
diff changeset
   505
    segments as they are being parsed by the ParaParser.
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   506
    fontname, fontSize, rise, textColor, cbDefn
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   507
    """
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   508
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   509
_greek2Utf8=None
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   510
def _greekConvert(data):
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   511
    global _greek2Utf8
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   512
    if not _greek2Utf8:
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   513
        from reportlab.pdfbase.rl_codecs import RL_Codecs
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   514
        import codecs
3812
22a7f2e7ad1f paraparser.py: remove unused local var
robin
parents: 3809
diff changeset
   515
        #our decoding map
22a7f2e7ad1f paraparser.py: remove unused local var
robin
parents: 3809
diff changeset
   516
        dm = codecs.make_identity_dict(range(32,256))
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   517
        for k in range(0,32):
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   518
            dm[k] = None
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   519
        dm.update(RL_Codecs._RL_Codecs__rl_codecs_data['symbol'][0])
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   520
        _greek2Utf8 = {}
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   521
        for k,v in dm.items():
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   522
            if not v:
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   523
                u = '\0'
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   524
            else:
3787
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
   525
                if isPy3:
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
   526
                    u = chr(v)
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
   527
                else:
3900
056aff5b1a58 paraparser.py: fix _greekConvert for 2.7
robin
parents: 3896
diff changeset
   528
                    u = unichr(v).encode('utf8')
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   529
            _greek2Utf8[chr(k)] = u
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   530
    return ''.join(map(_greek2Utf8.__getitem__,data))
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   531
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   532
#------------------------------------------------------------------
267
52a348f6c4c3 noted replication of XML markup comment between paraparser.py and paragraph.py
aaron_watters
parents: 266
diff changeset
   533
# !!! NOTE !!! THIS TEXT IS NOW REPLICATED IN PARAGRAPH.PY !!!
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   534
# The ParaFormatter will be able to format the following
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   535
# tags:
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   536
#       < /b > - bold
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   537
#       < /i > - italics
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   538
#       < u > < /u > - underline
2644
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   539
#       < strike > < /strike > - strike through
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   540
#       < super > < /super > - superscript
1736
dafc17db33d2 Attempt to use sup as well as super
rgbecker
parents: 1683
diff changeset
   541
#       < sup > < /sup > - superscript
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   542
#       < sub > < /sub > - subscript
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   543
#       <font name=fontfamily/fontname color=colorname size=float>
3552
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   544
#        <span name=fontfamily/fontname color=colorname backcolor=colorname size=float style=stylename>
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   545
#       < bullet > </bullet> - bullet text (at head of para only)
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   546
#       <onDraw name=callable label="a label"/>
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   547
#       <index [name="callablecanvasattribute"] label="a label"/>
2670
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   548
#       <link>link text</link>
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   549
#           attributes of links 
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   550
#               size/fontSize=num
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   551
#               name/face/fontName=name
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   552
#               fg/textColor/color=color
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   553
#               backcolor/backColor/bgcolor=color
3fdd642a7b76 minor cosmetic changes
rgbecker
parents: 2664
diff changeset
   554
#               dest/destination/target/href/link=target
2745
0b44535fa1a5 paraparser/paragraph.py: fix comments
rgbecker
parents: 2744
diff changeset
   555
#       <a>anchor text</a>
2744
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   556
#           attributes of anchors 
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   557
#               fontSize=num
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   558
#               fontName=name
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   559
#               fg/textColor/color=color
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   560
#               backcolor/backColor/bgcolor=color
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   561
#               href=href
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   562
#       <a name="anchorpoint"/>
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   563
#       <unichar name="unicode character name"/>
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   564
#       <unichar value="unicode code point"/>
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   565
#       <img src="path" width="1in" height="1in" valign="bottom"/>
3440
739ddbe7feab paaraparser/paragraph.py: add info re percentage in <img> for idea contributed by Roberto Alsina <ralsina@netmanagers.com.ar>
rgbecker
parents: 3434
diff changeset
   566
#               width="w%" --> fontSize*w/100   idea from Roberto Alsina
739ddbe7feab paaraparser/paragraph.py: add info re percentage in <img> for idea contributed by Roberto Alsina <ralsina@netmanagers.com.ar>
rgbecker
parents: 3434
diff changeset
   567
#               height="h%" --> linewidth*h/100 <ralsina@netmanagers.com.ar>
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   568
#       <greek> - </greek>
1683
7fa753e4420a Removed all trailing whitespace
andy_robinson
parents: 1677
diff changeset
   569
#
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   570
#       The whole may be surrounded by <para> </para> tags
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   571
#
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   572
# It will also be able to handle any MathML specified Greek characters.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   573
#------------------------------------------------------------------
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
   574
class ParaParser(HTMLParser):
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   575
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   576
    #----------------------------------------------------------
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   577
    # First we will define all of the xml tag handler functions.
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   578
    #
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   579
    # start_<tag>(attributes)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   580
    # end_<tag>()
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   581
    #
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   582
    # While parsing the xml ParaFormatter will call these
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   583
    # functions to handle the string formatting tags.
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   584
    # At the start of each tag the corresponding field will
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   585
    # be set to 1 and at the end tag the corresponding field will
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   586
    # be set to 0.  Then when handle_data is called the options
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   587
    # for that data will be aparent by the current settings.
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   588
    #----------------------------------------------------------
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   589
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   590
    def __getattr__( self, attrName ):
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   591
        """This way we can handle <TAG> the same way as <tag> (ignoring case)."""
2369
f3cc620c14ed paraparser.py: minor speedup
rgbecker
parents: 2368
diff changeset
   592
        if attrName!=attrName.lower() and attrName!="caseSensitive" and not self.caseSensitive and \
f3cc620c14ed paraparser.py: minor speedup
rgbecker
parents: 2368
diff changeset
   593
            (attrName.startswith("start_") or attrName.startswith("end_")):
f3cc620c14ed paraparser.py: minor speedup
rgbecker
parents: 2368
diff changeset
   594
                return getattr(self,attrName.lower())
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   595
        raise AttributeError(attrName)
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   596
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   597
    #### bold
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   598
    def start_b( self, attributes ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   599
        self._push(bold=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   600
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   601
    def end_b( self ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   602
        self._pop(bold=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   603
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   604
    def start_strong( self, attributes ):
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   605
        self._push(bold=1)
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   606
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   607
    def end_strong( self ):
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   608
        self._pop(bold=1)
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   609
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   610
    #### italics
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   611
    def start_i( self, attributes ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   612
        self._push(italic=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   613
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   614
    def end_i( self ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   615
        self._pop(italic=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   616
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   617
    def start_em( self, attributes ):
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   618
        self._push(italic=1)
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   619
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   620
    def end_em( self ):
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   621
        self._pop(italic=1)
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   622
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   623
    #### underline
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   624
    def start_u( self, attributes ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   625
        self._push(underline=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   626
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   627
    def end_u( self ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   628
        self._pop(underline=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   629
2644
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   630
    #### strike
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   631
    def start_strike( self, attributes ):
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   632
        self._push(strike=1)
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   633
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   634
    def end_strike( self ):
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   635
        self._pop(strike=1)
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   636
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   637
    #### link
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   638
    def start_link(self, attributes):
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   639
        self._push(**self.getAttributes(attributes,_linkAttrMap))
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   640
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   641
    def end_link(self):
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   642
        frag = self._stack[-1]
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   643
        del self._stack[-1]
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   644
        assert frag.link!=None
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   645
2744
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   646
    #### anchor
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   647
    def start_a(self, attributes):
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   648
        A = self.getAttributes(attributes,_anchorAttrMap)
2893
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   649
        name = A.get('name',None)
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   650
        if name is not None:
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   651
            name = name.strip()
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   652
            if not name:
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   653
                self._syntax_error('<a name="..."/> anchor variant requires non-blank name')
2744
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   654
            if len(A)>1:
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   655
                self._syntax_error('<a name="..."/> anchor variant only allows name attribute')
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   656
                A = dict(name=A['name'])
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   657
            A['_selfClosingTag'] = 'anchor'
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   658
        else:
2893
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   659
            href = A.get('href','').strip()
7432e06445ba paraparser.py: improved checking on <a> tag attributes
rgbecker
parents: 2861
diff changeset
   660
            A['link'] = href    #convert to our link form
3931
ceee76b69e7c fix href bug reported by bitbucket.kmouts
robin
parents: 3900
diff changeset
   661
            A.pop('href',None)
2744
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   662
        self._push(**A)
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   663
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   664
    def end_a(self):
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   665
        frag = self._stack[-1]
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   666
        sct = getattr(frag,'_selfClosingTag','')
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   667
        if sct:
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   668
            assert sct=='anchor' and frag.name,'Parser failure in <a/>'
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   669
            defn = frag.cbDefn = ABag()
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   670
            defn.label = defn.kind = 'anchor'
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   671
            defn.name = frag.name
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   672
            del frag.name, frag._selfClosingTag
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   673
            self.handle_data('')
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   674
            self._pop()
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   675
        else:
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   676
            del self._stack[-1]
9472eedb9702 reportlab/platypus: add two way <a> tag
rgbecker
parents: 2742
diff changeset
   677
            assert frag.link!=None
2742
8edd54153201 paraparser: allow <a> as alias for <link>
rgbecker
parents: 2694
diff changeset
   678
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   679
    def start_img(self,attributes):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   680
        A = self.getAttributes(attributes,_imgAttrMap)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   681
        if not A.get('src'):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   682
            self._syntax_error('<img> needs src attribute')
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   683
        A['_selfClosingTag'] = 'img'
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   684
        self._push(**A)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   685
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   686
    def end_img(self):
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   687
        frag = self._stack[-1]
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   688
        assert getattr(frag,'_selfClosingTag',''),'Parser failure in <img/>'
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   689
        defn = frag.cbDefn = ABag()
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   690
        defn.kind = 'img'
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   691
        defn.src = getattr(frag,'src',None)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   692
        defn.image = ImageReader(defn.src)
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   693
        size = defn.image.getSize()
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   694
        defn.width = getattr(frag,'width',size[0])
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   695
        defn.height = getattr(frag,'height',size[1])
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   696
        defn.valign = getattr(frag,'valign','bottom')
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   697
        del frag._selfClosingTag
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   698
        self.handle_data('')
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   699
        self._pop()
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
   700
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   701
    #### super script
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   702
    def start_super( self, attributes ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   703
        self._push(super=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   704
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   705
    def end_super( self ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   706
        self._pop(super=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   707
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   708
    start_sup = start_super
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   709
    end_sup = end_super
1736
dafc17db33d2 Attempt to use sup as well as super
rgbecker
parents: 1683
diff changeset
   710
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   711
    #### sub script
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   712
    def start_sub( self, attributes ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   713
        self._push(sub=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   714
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   715
    def end_sub( self ):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   716
        self._pop(sub=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   717
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   718
    #### greek script
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   719
    #### add symbol encoding
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   720
    def handle_charref(self, name):
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   721
        try:
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   722
            if name[0]=='x':
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   723
                n = int(name[1:],16)
1931
784fce255e2d Added in more special entities as suggested by Christoph Zwerschke
rgbecker
parents: 1736
diff changeset
   724
            else:
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   725
                n = int(name)
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   726
        except ValueError:
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   727
            self.unknown_charref(name)
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   728
            return
4004
ebd460ef8d2f rename utils.py UniChr --> uniChr
robin
parents: 3994
diff changeset
   729
        self.handle_data(uniChr(n))   #.encode('utf8'))
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   730
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   731
    def syntax_error(self,lineno,message):
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
   732
        self._syntax_error(message)
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   733
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   734
    def _syntax_error(self,message):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   735
        if message[:10]=="attribute " and message[-17:]==" value not quoted": return
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   736
        self.errors.append(message)
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   737
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   738
    def start_greek(self, attr):
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   739
        self._push(greek=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   740
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   741
    def end_greek(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   742
        self._pop(greek=1)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   743
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   744
    def start_unichar(self, attr):
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   745
        if 'name' in attr:
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   746
            if 'code' in attr:
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   747
                self._syntax_error('<unichar/> invalid with both name and code attributes')
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   748
            try:
3809
cbd390e6d557 paraparser.py: fix unichar
robin
parents: 3787
diff changeset
   749
                v = unicodedata.lookup(attr['name'])
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   750
            except KeyError:
3809
cbd390e6d557 paraparser.py: fix unichar
robin
parents: 3787
diff changeset
   751
                self._syntax_error('<unichar/> invalid name attribute\n"%s"' % ascii(name))
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   752
                v = '\0'
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   753
        elif 'code' in attr:
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   754
            try:
3937
f685e0de4282 paraparser.py:fix start_unichar for puython2
robin
parents: 3931
diff changeset
   755
                v = int(eval(attr['code']))
f685e0de4282 paraparser.py:fix start_unichar for puython2
robin
parents: 3931
diff changeset
   756
                v = chr(v) if isPy3 else unichr(v)
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   757
            except:
3809
cbd390e6d557 paraparser.py: fix unichar
robin
parents: 3787
diff changeset
   758
                self._syntax_error('<unichar/> invalid code attribute %s' % ascii(attr['code']))
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   759
                v = '\0'
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   760
        else:
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   761
            v = None
2664
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   762
            if attr:
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   763
                self._syntax_error('<unichar/> invalid attribute %s' % list(attr.keys())[0])
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   764
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   765
        if v is not None:
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   766
            self.handle_data(v)
2585
ee08fea4505b reportlab: added pound sign
rgbecker
parents: 2584
diff changeset
   767
        self._push(_selfClosingTag='unichar')
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   768
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   769
    def end_unichar(self):
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   770
        self._pop()
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
   771
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   772
    def start_font(self,attr):
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   773
        self._push(**self.getAttributes(attr,_fontAttrMap))
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   774
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   775
    def end_font(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   776
        self._pop()
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   777
3552
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   778
    def start_span(self,attr):
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   779
        A = self.getAttributes(attr,_spanAttrMap)
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   780
        if 'style' in A:
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   781
            style = self.findSpanStyle(A.pop('style'))
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   782
            D = {}
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   783
            for k in 'fontName fontSize textColor backColor'.split():
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   784
                v = getattr(style,k,self)
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   785
                if v is self: continue
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   786
                D[k] = v
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   787
            D.update(A)
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   788
            A = D
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   789
        self._push(**A)
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   790
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   791
    end_span = end_font
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
   792
2663
927cc273c5a5 <br/> work in progress
andy
parents: 2646
diff changeset
   793
    def start_br(self, attr):
2664
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   794
        self._push(_selfClosingTag='br',lineBreak=True,text='')
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
   795
        
2663
927cc273c5a5 <br/> work in progress
andy
parents: 2646
diff changeset
   796
    def end_br(self):
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
   797
        #print('\nend_br called, %d frags in list' % len(self.fragList))
2664
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   798
        frag = self._stack[-1]
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   799
        assert frag._selfClosingTag=='br' and frag.lineBreak,'Parser failure in <br/>'
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   800
        del frag._selfClosingTag
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   801
        self.handle_data('')
c9faa3a99e93 reportlab/platypus: <br/> tags now working
rgbecker
parents: 2663
diff changeset
   802
        self._pop()
2663
927cc273c5a5 <br/> work in progress
andy
parents: 2646
diff changeset
   803
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   804
    def _initial_frag(self,attr,attrMap,bullet=0):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   805
        style = self._style
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   806
        if attr!={}:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   807
            style = copy.deepcopy(style)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   808
            _applyAttributes(style,self.getAttributes(attr,attrMap))
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   809
            self._style = style
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   810
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   811
        # initialize semantic values
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   812
        frag = ParaFrag()
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   813
        frag.sub = 0
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   814
        frag.super = 0
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   815
        frag.rise = 0
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   816
        frag.underline = 0
2644
e762ad1c8909 reportlab: add support for strike through
rgbecker
parents: 2594
diff changeset
   817
        frag.strike = 0
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   818
        frag.greek = 0
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
   819
        frag.link = None
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   820
        if bullet:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   821
            frag.fontName, frag.bold, frag.italic = ps2tt(style.bulletFontName)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   822
            frag.fontSize = style.bulletFontSize
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   823
            frag.textColor = hasattr(style,'bulletColor') and style.bulletColor or style.textColor
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   824
        else:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   825
            frag.fontName, frag.bold, frag.italic = ps2tt(style.fontName)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   826
            frag.fontSize = style.fontSize
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   827
            frag.textColor = style.textColor
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   828
        return frag
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   829
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   830
    def start_para(self,attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   831
        self._stack = [self._initial_frag(attr,_paraAttrMap)]
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   832
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   833
    def end_para(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   834
        self._pop()
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   835
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   836
    def start_bullet(self,attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   837
        if hasattr(self,'bFragList'):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   838
            self._syntax_error('only one <bullet> tag allowed')
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   839
        self.bFragList = []
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   840
        frag = self._initial_frag(attr,_bulletAttrMap,1)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   841
        frag.isBullet = 1
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   842
        self._stack.append(frag)
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   843
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   844
    def end_bullet(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   845
        self._pop()
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
   846
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   847
    #---------------------------------------------------------------
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   848
    def start_seqdefault(self, attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   849
        try:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   850
            default = attr['id']
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   851
        except KeyError:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   852
            default = None
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   853
        self._seq.setDefaultCounter(default)
266
081154da1a78 Added Sequencer and associated XML tags
andy_robinson
parents: 253
diff changeset
   854
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   855
    def end_seqdefault(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   856
        pass
1683
7fa753e4420a Removed all trailing whitespace
andy_robinson
parents: 1677
diff changeset
   857
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   858
    def start_seqreset(self, attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   859
        try:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   860
            id = attr['id']
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   861
        except KeyError:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   862
            id = None
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   863
        try:
2368
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   864
            base = int(attr['base'])
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   865
        except:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   866
            base=0
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   867
        self._seq.reset(id, base)
266
081154da1a78 Added Sequencer and associated XML tags
andy_robinson
parents: 253
diff changeset
   868
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   869
    def end_seqreset(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   870
        pass
744
2abd99baf95b Accepts seqdefault/seqDefault and seqreset/seqReset
andy_robinson
parents: 677
diff changeset
   871
2368
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   872
    def start_seqchain(self, attr):
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   873
        try:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   874
            order = attr['order']
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   875
        except KeyError:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   876
            order = ''
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   877
        order = order.split()
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   878
        seq = self._seq
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   879
        for p,c in zip(order[:-1],order[1:]):
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   880
            seq.chain(p, c)
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   881
    end_seqchain = end_seqreset
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   882
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   883
    def start_seqformat(self, attr):
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   884
        try:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   885
            id = attr['id']
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   886
        except KeyError:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   887
            id = None
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   888
        try:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   889
            value = attr['value']
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   890
        except KeyError:
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   891
            value = '1'
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   892
        self._seq.setFormat(id,value)
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   893
    end_seqformat = end_seqreset
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   894
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   895
    # AR hacking in aliases to allow the proper casing for RML.
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   896
    # the above ones should be deprecated over time. 2001-03-22
2368
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   897
    start_seqDefault = start_seqdefault
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   898
    end_seqDefault = end_seqdefault
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   899
    start_seqReset = start_seqreset
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   900
    end_seqReset = end_seqreset
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   901
    start_seqChain = start_seqchain
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   902
    end_seqChain = end_seqchain
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   903
    start_seqFormat = start_seqformat
791a362e9cae added seqchain/format tags
rgbecker
parents: 2341
diff changeset
   904
    end_seqFormat = end_seqformat
1683
7fa753e4420a Removed all trailing whitespace
andy_robinson
parents: 1677
diff changeset
   905
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   906
    def start_seq(self, attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   907
        #if it has a template, use that; otherwise try for id;
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   908
        #otherwise take default sequence
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   909
        if 'template' in attr:
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   910
            templ = attr['template']
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   911
            self.handle_data(templ % self._seq)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   912
            return
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   913
        elif 'id' in attr:
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   914
            id = attr['id']
1683
7fa753e4420a Removed all trailing whitespace
andy_robinson
parents: 1677
diff changeset
   915
        else:
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   916
            id = None
2694
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   917
        increment = attr.get('inc', None)
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   918
        if not increment:
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   919
            output = self._seq.nextf(id)
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   920
        else:
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   921
            #accepts "no" for do not increment, or an integer.
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   922
            #thus, 0 and 1 increment by the right amounts.
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   923
            if increment.lower() == 'no':
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   924
                output = self._seq.thisf(id)
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   925
            else:
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   926
                incr = int(increment)
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   927
                output = self._seq.thisf(id)
dd0ea6474ea0 fixes to crashing PTO trailer when empty, and numbering
andy
parents: 2693
diff changeset
   928
                self._seq.reset(id, self._seq._this() + incr)
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   929
        self.handle_data(output)
1683
7fa753e4420a Removed all trailing whitespace
andy_robinson
parents: 1677
diff changeset
   930
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   931
    def end_seq(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   932
        pass
266
081154da1a78 Added Sequencer and associated XML tags
andy_robinson
parents: 253
diff changeset
   933
4059
9b48d0d9f137 paraparse.py: add ondraw=onDraw aliases
robin
parents: 4004
diff changeset
   934
    def start_ondraw(self,attr):
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   935
        defn = ABag()
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   936
        if 'name' in attr: defn.name = attr['name']
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   937
        else: self._syntax_error('<onDraw> needs at least a name attribute')
506
68bd275f16e2 Added onDraw tag to paragraphs
rgbecker
parents: 494
diff changeset
   938
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   939
        if 'label' in attr: defn.label = attr['label']
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   940
        defn.kind='onDraw'
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   941
        self._push(cbDefn=defn)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   942
        self.handle_data('')
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   943
        self._pop()
4059
9b48d0d9f137 paraparse.py: add ondraw=onDraw aliases
robin
parents: 4004
diff changeset
   944
    start_onDraw=start_ondraw 
9b48d0d9f137 paraparse.py: add ondraw=onDraw aliases
robin
parents: 4004
diff changeset
   945
    end_onDraw=end_ondraw=end_seq
3111
86a3158c50bd reportlab: improved support for onDraw and SimpleIndex
rgbecker
parents: 3032
diff changeset
   946
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   947
    def start_index(self,attr):
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   948
        attr=self.getAttributes(attr,_indexAttrMap)
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   949
        defn = ABag()
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   950
        if 'item' in attr:
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   951
            label = attr['item']
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   952
        else:
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   953
            self._syntax_error('<index> needs at least an item attribute')
3326
ce725978d11c Initial Python3 compatibility fixes
damian
parents: 3188
diff changeset
   954
        if 'name' in attr:
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   955
            name = attr['name']
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   956
        else:
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   957
            name = DEFAULT_INDEX_NAME
3187
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   958
        format = attr.get('format',None)
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   959
        if format is not None and format not in ('123','I','i','ABC','abc'):
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   960
            raise ValueError('index tag format is %r not valid 123 I i ABC or abc' % offset)
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   961
        offset = attr.get('offset',None)
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   962
        if offset is not None:
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   963
            try:
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   964
                offset = int(offset)
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   965
            except:
2d5a6655556e tableofcontents/paraparser: allow for format and offset parameters
rgbecker
parents: 3165
diff changeset
   966
                raise ValueError('index tag offset is %r not an int' % offset)
3856
5316e8678dd9 paraparser.py: fix label encoding
robin
parents: 3826
diff changeset
   967
        defn.label = encode_label((label,format,offset))
3165
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   968
        defn.name = name
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   969
        defn.kind='index'
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   970
        self._push(cbDefn=defn)
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   971
        self.handle_data('')
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   972
        self._pop()
cbda9e7d0ee3 reportlab: new index support
rgbecker
parents: 3137
diff changeset
   973
    end_index=end_seq
2663
927cc273c5a5 <br/> work in progress
andy
parents: 2646
diff changeset
   974
3826
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
   975
    def start_unknown(self,attr):
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
   976
        pass
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
   977
    end_unknown=end_seq
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
   978
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   979
    #---------------------------------------------------------------
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   980
    def _push(self,**attr):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   981
        frag = copy.copy(self._stack[-1])
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   982
        _applyAttributes(frag,attr)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   983
        self._stack.append(frag)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   984
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   985
    def _pop(self,**kw):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   986
        frag = self._stack[-1]
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   987
        del self._stack[-1]
3723
99aa837b6703 second stage of port to Python 3.3; working hello world
rptlab
parents: 3721
diff changeset
   988
        for k, v in kw.items():
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   989
            assert getattr(frag,k)==v
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   990
        return frag
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   991
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   992
    def getAttributes(self,attr,attrMap):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   993
        A = {}
3723
99aa837b6703 second stage of port to Python 3.3; working hello world
rptlab
parents: 3721
diff changeset
   994
        for k, v in attr.items():
1940
baa0abc136c4 Henning von Bargen's caseSensitive flag
rgbecker
parents: 1932
diff changeset
   995
            if not self.caseSensitive:
3731
b233dd0577ff another round of changes mostly type related
rptlab
parents: 3723
diff changeset
   996
                k = k.lower()
3721
0c93dd8ff567 initial changes from 2to3-3.3
rptlab
parents: 3656
diff changeset
   997
            if k in list(attrMap.keys()):
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   998
                j = attrMap[k]
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
   999
                func = j[1]
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1000
                try:
3931
ceee76b69e7c fix href bug reported by bitbucket.kmouts
robin
parents: 3900
diff changeset
  1001
                    A[j[0]] = v if func is None else func(v)
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1002
                except:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1003
                    self._syntax_error('%s: invalid value %s'%(k,v))
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1004
            else:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1005
                self._syntax_error('invalid attribute name %s'%k)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1006
        return A
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
  1007
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1008
    #----------------------------------------------------------------
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1009
3826
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1010
    def __init__(self,verbose=0, caseSensitive=0, ignoreUnknownTags=1):
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1011
        HTMLParser.__init__(self)
3826
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1012
        self.verbose = verbose
4059
9b48d0d9f137 paraparse.py: add ondraw=onDraw aliases
robin
parents: 4004
diff changeset
  1013
        #HTMLParser is cas insenstive anyway, but the rml interface still needs this
9b48d0d9f137 paraparse.py: add ondraw=onDraw aliases
robin
parents: 4004
diff changeset
  1014
        #all start/end_ methods should have a lower case version for HMTMParser
3826
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1015
        self.caseSensitive = caseSensitive
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1016
        self.ignoreUnknownTags = ignoreUnknownTags
266
081154da1a78 Added Sequencer and associated XML tags
andy_robinson
parents: 253
diff changeset
  1017
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1018
    def _iReset(self):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1019
        self.fragList = []
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1020
        if hasattr(self, 'bFragList'): delattr(self,'bFragList')
250
a1bcf9c6c21e <bullet> xml tag added
rgbecker
parents: 248
diff changeset
  1021
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1022
    def _reset(self, style):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1023
        '''reset the parser'''
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1024
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1025
        HTMLParser.reset(self)
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1026
        # initialize list of string segments to empty
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1027
        self.errors = []
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1028
        self._style = style
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1029
        self._iReset()
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1030
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1031
    #----------------------------------------------------------------
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1032
    def handle_data(self,data):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1033
        "Creates an intermediate representation of string segments."
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1034
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1035
        #The old parser would only 'see' a string after all entities had
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1036
        #been processed.  Thus, 'Hello &trade; World' would emerge as one
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1037
        #fragment.    HTMLParser processes these separately.  We want to ensure
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1038
        #that successive calls like this are concatenated, to prevent too many
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1039
        #fragments being created.
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1040
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1041
        frag = copy.copy(self._stack[-1])
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1042
        if hasattr(frag,'cbDefn'):
2857
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
  1043
            kind = frag.cbDefn.kind
487dc2450eec reprotlab: inline images horizontal positioning OK
rgbecker
parents: 2836
diff changeset
  1044
            if data: self._syntax_error('Only empty <%s> tag allowed' % kind)
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
  1045
        elif hasattr(frag,'_selfClosingTag'):
2663
927cc273c5a5 <br/> work in progress
andy
parents: 2646
diff changeset
  1046
            if data!='': self._syntax_error('No content allowed in %s tag' % frag._selfClosingTag)
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
  1047
            return
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1048
        else:
1736
dafc17db33d2 Attempt to use sup as well as super
rgbecker
parents: 1683
diff changeset
  1049
            # if sub and super are both on they will cancel each other out
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1050
            if frag.sub == 1 and frag.super == 1:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1051
                frag.sub = 0
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1052
                frag.super = 0
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1053
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1054
            if frag.sub:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1055
                frag.rise = -frag.fontSize*subFraction
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1056
                frag.fontSize = max(frag.fontSize-sizeDelta,3)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1057
            elif frag.super:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1058
                frag.rise = frag.fontSize*superFraction
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1059
                frag.fontSize = max(frag.fontSize-sizeDelta,3)
112
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
  1060
2584
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
  1061
            if frag.greek:
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
  1062
                frag.fontName = 'symbol'
0fed2bd8ef90 reportlab: fixed <greek> added <unichar [name=..|code=../> to paragraph
rgbecker
parents: 2575
diff changeset
  1063
                data = _greekConvert(data)
514
3784fe357a72 Slight optimisation in handle_data for cbdefn frags
rgbecker
parents: 508
diff changeset
  1064
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1065
        # bold, italic, and underline
2861
2096955de8cf platypus: autoLeading vertical pos improvement
rgbecker
parents: 2860
diff changeset
  1066
        frag.fontName = tt2ps(frag.fontName,frag.bold,frag.italic)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1067
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1068
        #save our data
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1069
        frag.text = data
514
3784fe357a72 Slight optimisation in handle_data for cbdefn frags
rgbecker
parents: 508
diff changeset
  1070
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1071
        if hasattr(frag,'isBullet'):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1072
            delattr(frag,'isBullet')
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1073
            self.bFragList.append(frag)
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1074
        else:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1075
            self.fragList.append(frag)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1076
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1077
    def handle_cdata(self,data):
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1078
        self.handle_data(data)
211
52541f1643b6 CDATA handler added
rgbecker
parents: 209
diff changeset
  1079
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1080
    def _setup_for_parse(self,style):
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1081
        self._seq = reportlab.lib.sequencer.getSequencer()
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1082
        self._reset(style)  # reinitialise the parser
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1083
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1084
    def _complete_parse(self):
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1085
        "Reset after parsing, to be ready for next paragraph"
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1086
        del self._seq
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1087
        style = self._style
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1088
        del self._style
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1089
        if len(self.errors)==0:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1090
            fragList = self.fragList
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1091
            bFragList = hasattr(self,'bFragList') and self.bFragList or None
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1092
            self._iReset()
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1093
        else:
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1094
            fragList = bFragList = None
2575
0cba68b93555 reportlab-utf8 moved to trunk
rgbecker
parents: 2446
diff changeset
  1095
1677
1450177dd19e Exterminated all tab characters and added a test to make sure
andy_robinson
parents: 1160
diff changeset
  1096
        return style, fragList, bFragList
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
  1097
3787
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1098
    def _tt_handle(self,tt):
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1099
        "Iterate through a pre-parsed tuple tree (e.g. from pyRXP)"
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1100
        #import pprint
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1101
        #pprint.pprint(tt)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1102
        #find the corresponding start_tagname and end_tagname methods.
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1103
        #These must be defined.
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1104
        tag = tt[0]
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1105
        try:
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1106
            start = getattr(self,'start_'+tag)
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1107
            end = getattr(self,'end_'+tag)
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1108
        except AttributeError:
3826
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1109
            if not self.ignoreUnknownTags:
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1110
                raise ValueError('Invalid tag "%s"' % tag)
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1111
            start = self.start_unknown
02f216b6e38e paraparser.py: support old behaviours eg ignore unknown tags etc etc
robin
parents: 3812
diff changeset
  1112
            end = self.end_unknown
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1113
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1114
        #call the start_tagname method
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1115
        start(tt[1] or {})
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1116
        #if tree node has any children, they will either be further nodes,
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1117
        #or text.  Accordingly, call either this function, or handle_data.
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1118
        C = tt[2]
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1119
        if C:
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1120
            M = self._tt_handlers
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1121
            for c in C:
3731
b233dd0577ff another round of changes mostly type related
rptlab
parents: 3723
diff changeset
  1122
                M[isinstance(c,(list,tuple))](c)
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1123
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1124
        #call the end_tagname method
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1125
        end()
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1126
3787
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1127
    def _tt_start(self,tt):
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1128
        self._tt_handlers = self.handle_data,self._tt_handle
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1129
        self._tt_handle(tt)
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1130
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1131
    def tt_parse(self,tt,style):
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1132
        '''parse from tupletree form'''
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1133
        self._setup_for_parse(style)
3787
8f9be6d6f75c convert paraparser to use pyRXP directly (or any TT producer)
robin
parents: 3731
diff changeset
  1134
        self._tt_start(tt)
2376
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1135
        return self._complete_parse()
7e70411a7236 flowables.py: minor change to PTOContainer
rgbecker
parents: 2369
diff changeset
  1136
3552
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
  1137
    def findSpanStyle(self,style):
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
  1138
        raise ValueError('findSpanStyle not implemented in this parser')
20ecbcc53c15 paraparser.py add support for <span style=stylename>
rgbecker
parents: 3440
diff changeset
  1139
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1140
    #HTMLParser interface
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1141
    def parse(self, text, style):
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1142
        "attempt replacement for parse"
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1143
        self._setup_for_parse(style)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1144
        text = asUnicode(text)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1145
        if not(len(text)>=6 and text[0]=='<' and _re_para.match(text)):
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1146
            text = u"<para>"+text+u"</para>"
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1147
        try:
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1148
            self.feed(text)
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1149
        except:
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1150
            annotateException('paragraph text %s caused exception' % ascii(text))
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1151
        return self._complete_parse()
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1152
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1153
    def handle_starttag(self, tag, attrs):
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1154
        "Called by HTMLParser when a tag starts"
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1155
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1156
        #tuple tree parser used to expect a dict.  HTML parser
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1157
        #gives list of two-element tuples
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1158
        if isinstance(attrs, list):
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1159
            d = {}
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1160
            for (k,  v) in attrs:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1161
                d[k] = v
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1162
            attrs = d
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1163
        if not self.caseSensitive: tag = tag.lower()
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1164
        try:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1165
            start = getattr(self,'start_'+tag)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1166
        except AttributeError:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1167
            if not self.ignoreUnknownTags:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1168
                raise ValueError('Invalid tag "%s"' % tag)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1169
            start = self.start_unknown
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1170
        #call it
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1171
        start(attrs or {})
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1172
        
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1173
    def handle_endtag(self, tag):
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1174
        "Called by HTMLParser when a tag ends"
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1175
        #find the existing end_tagname method
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1176
        if not self.caseSensitive: tag = tag.lower()
3954
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1177
        try:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1178
            end = getattr(self,'end_'+tag)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1179
        except AttributeError:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1180
            if not self.ignoreUnknownTags:
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1181
                raise ValueError('Invalid tag "%s"' % tag)
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1182
            end = self.end_unknown
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1183
        #call it
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1184
        end()
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1185
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1186
    def handle_entityref(self, name):
44dbe56eb858 first attempt at HTMLParser-based paraparser
Andy Robinson <andy@reportlab.com>
parents: 3937
diff changeset
  1187
        "Handles a named entity.  "
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1188
        try:
3994
9c38aba205e1 paraparser.py: fix entityref handling
robin
parents: 3961
diff changeset
  1189
            v = known_entities[name]
3957
c7cedb1dde29 paraparser.py: remove usage of pyRXPU, remove attempted optimization of texts
robin
parents: 3956
diff changeset
  1190
        except:
3994