reportlab/platypus/paraparser.py
author rgbecker
Sat, 15 Apr 2000 11:11:23 +0000
changeset 134 60e8e0aee073
parent 133 ba7b4eb9be6c
child 162 1cc78f90fe24
permissions -rw-r--r--
Fixed syntax_error handling
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     1
import string
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
     2
import re
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
     3
from types import TupleType
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     4
import sys
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     5
import os
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     6
import copy
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     7
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     8
try:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
     9
	from xml.parsers import xmllib
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    10
	_xmllib_newStyle = 1
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    11
except ImportError:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    12
	import xmllib
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    13
	_xmllib_newStyle = 0
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    14
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    15
from reportlab.lib.colors import stringToColor, white, black, red, Color
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    16
from reportlab.lib.fonts import tt2ps, ps2tt
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    17
from reportlab.lib.enums import TA_LEFT, TA_RIGHT, TA_CENTER, TA_JUSTIFY
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    18
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    19
_re_para = re.compile('^\\s*<\\s*para(\\s+|>)')
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    20
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    21
sizeDelta = 2		# amount to reduce font size by for super and sub script
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    22
subFraction = 0.5	# fraction of font size that a sub script should be lowered
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    23
superFraction = 0.5	# fraction of font size that a super script should be raised
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    24
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
    25
def _num(s):
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    26
	if s[0] in ['+','-']:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    27
		try:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    28
			return ('relative',int(s))
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    29
		except ValueError:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    30
			return ('relative',float(s))
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    31
	else:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    32
		try:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    33
			return int(s)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    34
		except ValueError:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    35
			return float(s)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    36
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    37
def _align(s):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    38
	s = string.lower(s)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    39
	if s=='left': return TA_LEFT
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    40
	elif s=='right': return TA_RIGHT
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    41
	elif s=='justify': return TA_JUSTIFY
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    42
	elif s in ('centre','center'): return TA_CENTER
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    43
	else: raise ValueError
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    44
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    45
_paraAttrMap = {'font': ('fontName', None),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    46
				'fontsize': ('fontSize', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    47
				'leading': ('leading', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    48
				'lindent': ('leftIndent', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    49
				'rindent': ('rightIndent', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    50
				'findent': ('firstLineIndent', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    51
				'align': ('alignment', _align),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    52
				'spaceb': ('spaceBefore', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    53
				'spacea': ('spaceAfter', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    54
				'bfont': ('bulletFontName', None),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    55
				'bfontsize': ('bulletFontIndent',_num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    56
				'bindent': ('bulletFontIndent',_num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    57
				'color':('textColor',stringToColor),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    58
				'fg': ('textColor',stringToColor)}
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    59
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    60
#things which are valid font attributes
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    61
_fontAttrMap = {'size': ('fontSize', _num),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    62
				'name': ('fontName', None),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    63
				'fg': 	('textColor', stringToColor),
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    64
				'color':('textColor', stringToColor)}
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    65
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    66
def _addAttributeNames(m):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    67
	K = m.keys()
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    68
	for k in K:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    69
		n = string.lower(m[k][0])
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    70
		if not m.has_key(n):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    71
			m[n] = m[k]
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    72
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    73
_addAttributeNames(_paraAttrMap)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    74
_addAttributeNames(_fontAttrMap)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    75
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    76
def _applyAttributes(obj, attr):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    77
	for k, v in attr.items():
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    78
		if type(v) is TupleType and v[0]=='relative':
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    79
			v = v[1]+getattr(obj,k,0)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
    80
		setattr(obj,k,v)
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
    81
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    82
#characters not supported: epsi, Gammad, gammad, kappav, rhov, Upsi, upsi
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    83
greeks = {
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    84
	'alpha':'a',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    85
	'beta':'b',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    86
	'chi':'c',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    87
	'Delta':'D',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    88
	'delta':'d',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    89
	'epsiv':'e',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    90
	'eta':'h',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    91
	'Gamma':'G',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    92
	'gamma':'g',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    93
	'iota':'i',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    94
	'kappa':'k',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    95
	'Lambda':'L',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    96
	'lambda':'l',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    97
	'mu':'m',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    98
	'nu':'n',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
    99
	'Omega':'W',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   100
	'omega':'w',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   101
	'omicron':'x',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   102
	'Phi':'F',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   103
	'phi':'f',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   104
	'phiv':'j',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   105
	'Pi':'P',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   106
	'pi':'p',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   107
	'piv':'v',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   108
	'Psi':'Y',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   109
	'psi':'y',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   110
	'rho':'r',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   111
	'Sigma':'S',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   112
	'sigma':'s',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   113
	'sigmav':'V',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   114
	'tau':'t',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   115
	'Theta':'Q',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   116
	'theta':'q',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   117
	'thetav':'j',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   118
	'Xi':'X',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   119
	'xi':'x',
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   120
	'zeta':'z'
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   121
}
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   122
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   123
#------------------------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   124
class ParaFrag:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   125
	"""class ParaFrag contains the intermediate representation of string
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   126
	segments as they are being parsed by the XMLParser.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   127
	"""
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   128
	def __init__(self,**attr):
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   129
		for k,v in attr.items():
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   130
			setattr(self,k,v)
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   131
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   132
	def clone(self,**attr):
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   133
		n = apply(ParaFrag,(),self.__dict__)
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   134
		if attr != {}: apply(ParaFrag.__init__,(n,),attr)
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   135
		return n
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   136
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   137
#------------------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   138
# The ParaFormatter will be able to format the following xml
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   139
# tags:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   140
#	   < b > < /b > - bold
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   141
#	   < i > < /i > - italics
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   142
#	   < u > < /u > - underline
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   143
#	   < super > < /super > - superscript
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   144
#	   < sub > < /sub > - subscript
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   145
#	   <font name=fontfamily/fontname color=colorname size=float>
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   146
#
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   147
#		The whole may be surrounded by <para> </para> tags
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   148
#
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   149
# It will also be able to handle any MathML specified Greek characters.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   150
#------------------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   151
class ParaParser(xmllib.XMLParser):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   152
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   153
	#----------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   154
	# First we will define all of the xml tag handler functions.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   155
	#
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   156
	# start_<tag>(attributes)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   157
	# end_<tag>()
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   158
	#
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   159
	# While parsing the xml ParaFormatter will call these
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   160
	# functions to handle the string formatting tags.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   161
	# At the start of each tag the corresponding field will
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   162
	# be set to 1 and at the end tag the corresponding field will
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   163
	# be set to 0.	Then when handle_data is called the options
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   164
	# for that data will be aparent by the current settings.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   165
	#----------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   166
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   167
	#### bold
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   168
	def start_b( self, attributes ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   169
		self._push(bold=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   170
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   171
	def end_b( self ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   172
		self._pop(bold=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   173
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   174
	#### italics
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   175
	def start_i( self, attributes ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   176
		self._push(italic=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   177
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   178
	def end_i( self ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   179
		self._pop(italic=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   180
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   181
	#### underline
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   182
	def start_u( self, attributes ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   183
		self._push(underline=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   184
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   185
	def end_u( self ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   186
		self._pop(underline=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   187
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   188
	#### super script
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   189
	def start_super( self, attributes ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   190
		self._push(super=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   191
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   192
	def end_super( self ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   193
		self._pop(super=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   194
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   195
	#### sub script
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   196
	def start_sub( self, attributes ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   197
		self._push(sub=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   198
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   199
	def end_sub( self ):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   200
		self._pop(sub=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   201
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   202
	#### greek script
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   203
	if _xmllib_newStyle:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   204
		def handle_entityref(self,name):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   205
			if greeks.has_key(name):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   206
				self._push(greek=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   207
				self.handle_data(greeks[name])
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   208
				self._pop(greek=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   209
			else:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   210
				xmllib.XMLParser.handle_entityref(self,name)
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   211
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   212
		def syntax_error(self,lineno,message):
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   213
			self._syntax_error(message)
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   214
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   215
	else:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   216
		def start_greekLetter(self, attributes,letter):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   217
			self._push(greek=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   218
			self.handle_data(letter)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   219
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   220
		def syntax_error(self,message):
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   221
			self._syntax_error(message)
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   222
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   223
	def _syntax_error(self,message):
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   224
		if message[:10]=="attribute " and message[-17:]==" value not quoted": return
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   225
		self.errors.append(message)
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   226
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   227
	def start_greek(self, attributes):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   228
		self._push(greek=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   229
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   230
	def end_greek(self):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   231
		self._pop(greek=1)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   232
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   233
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   234
	def start_font(self,attr):
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   235
		A = self.getAttributes(attr,_fontAttrMap)
113
2c20ecb9015e Fixed start_font
rgbecker
parents: 112
diff changeset
   236
		apply(self._push,(),A)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   237
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   238
	def end_font(self):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   239
		self._pop()
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   240
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   241
	def start_para(self,attr):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   242
		style = self._style
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   243
		if attr!={}:
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   244
			style = copy.deepcopy(style)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   245
			_applyAttributes(style,self.getAttributes(attr,_paraAttrMap))
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   246
			self._style = style
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   247
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   248
		# initialize semantic values
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   249
		frag = ParaFrag()
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   250
		frag.sub = 0
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   251
		frag.super = 0
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   252
		frag.rise = 0
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   253
		frag.fontName, frag.bold, frag.italic = ps2tt(style.fontName)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   254
		frag.fontSize = style.fontSize
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   255
		frag.underline = 0
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   256
		frag.textColor = style.textColor
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   257
		frag.greek = 0
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   258
		self._stack = [frag]
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   259
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   260
	def end_para(self):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   261
		self._pop()
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   262
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   263
	def _push(self,**attr):
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   264
		frag = copy.copy(self._stack[-1])
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   265
		_applyAttributes(frag,attr)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   266
		self._stack.append(frag)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   267
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   268
	def _pop(self,**kw):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   269
		frag = self._stack[-1]
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   270
		del self._stack[-1]
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   271
		for k, v in kw.items():
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   272
			assert getattr(frag,k)==v
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   273
		return frag
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   274
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   275
	def getAttributes(self,attr,attrMap):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   276
		A = {}
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   277
		for k, v in attr.items():
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   278
			k = string.lower(k)
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   279
			if k in attrMap.keys():
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   280
				j = attrMap[k]
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   281
				func = j[1]
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   282
				try:
123
daa339ac3104 Fixes to attribute handling and test script
rgbecker
parents: 119
diff changeset
   283
					A[j[0]] = (func is None) and v or apply(func,(v,))
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   284
				except:
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   285
					self._syntax_error('%s: invalid value %s'%(k,v))
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   286
			else:
134
60e8e0aee073 Fixed syntax_error handling
rgbecker
parents: 133
diff changeset
   287
				self._syntax_error('invalid attribute name %s'%k)
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   288
		return A
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   289
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   290
	#----------------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   291
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   292
	def __init__(self,verbose=0):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   293
		if _xmllib_newStyle:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   294
			xmllib.XMLParser.__init__(self,verbose=verbose)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   295
		else:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   296
			xmllib.XMLParser.__init__(self)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   297
			# set up handlers for various tags
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   298
			self.elements = {	'b': (self.start_b, self.end_b),
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   299
							'u': (self.start_u, self.end_u),
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   300
							'i': (self.start_i, self.end_i),
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   301
							'super': (self.start_super, self.end_super),
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   302
							'sub': (self.start_sub, self.end_sub),
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   303
							'font': (self.start_font, self.end_font),
132
8a9576d734b9 Fix for 1.5.2 xmllib + <para>
rgbecker
parents: 129
diff changeset
   304
							'greek': (self.start_greek, self.end_greek),
8a9576d734b9 Fix for 1.5.2 xmllib + <para>
rgbecker
parents: 129
diff changeset
   305
							'para': (self.start_para, self.end_para)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   306
							}
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   307
132
8a9576d734b9 Fix for 1.5.2 xmllib + <para>
rgbecker
parents: 129
diff changeset
   308
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   309
			# automatically add handlers for all of the greek characters
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   310
			for item in greeks.keys():
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   311
				self.elements[item] = (lambda attr,self=self,letter=greeks[item]:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   312
					self.start_greekLetter(attr,letter), self.end_greek)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   313
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   314
			# set up dictionary for greek characters, this is a class variable
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   315
			self.entitydefs = copy.copy(self.entitydefs)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   316
			for item in greeks.keys():
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   317
				self.entitydefs[item] = '<%s/>' % item
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   318
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   319
	def _reset(self, style):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   320
		'''reset the parser'''
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   321
		xmllib.XMLParser.reset(self)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   322
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   323
		# initialize list of string segments to empty
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   324
		self.errors = []
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   325
		self.fragList = []
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   326
		self._style = style
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   327
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   328
	#----------------------------------------------------------------
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   329
	def handle_data(self,data):
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   330
		"Creates an intermediate representation of string segments."
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   331
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   332
		frag = copy.copy(self._stack[-1])
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   333
		#save our data
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   334
		frag.text = data
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   335
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   336
		# if sub and super are both one they will cancel each other out
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   337
		if frag.sub == 1 and frag.super == 1:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   338
			frag.sub = 0
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   339
			frag.super = 0
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   340
112
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
   341
		if frag.sub:
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
   342
			frag.rise = -frag.fontSize*subFraction
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
   343
			frag.fontSize = max(frag.fontSize-sizeDelta,3)
115
49b2fae9ea60 Fixes to superscript stuff
rgbecker
parents: 113
diff changeset
   344
		elif frag.super:
112
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
   345
			frag.rise = frag.fontSize*superFraction
115
49b2fae9ea60 Fixes to superscript stuff
rgbecker
parents: 113
diff changeset
   346
			frag.fontSize = max(frag.fontSize-sizeDelta,3)
112
1d4892961fdb Added rise attribute
rgbecker
parents: 102
diff changeset
   347
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   348
		if frag.greek: frag.fontName = 'symbol'
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   349
		# bold, italic, and underline
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   350
		frag.fontName = tt2ps(frag.fontName,frag.bold,frag.italic)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   351
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   352
		self.fragList.append(frag)
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   353
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   354
	#----------------------------------------------------------------
102
1818e7fa3738 Added clone method to ParaFrag
rgbecker
parents: 96
diff changeset
   355
	def parse(self, text, style):
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   356
		"""Given a formatted string will return a list of
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   357
		ParaFrag objects with their calculated widths.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   358
		If errors occur None will be returned and the
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   359
		self.errors holds a list of the error messages.
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   360
		"""
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   361
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   362
		# the xmlparser requires that all text be surrounded by xml
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   363
		# tags, therefore we must throw some unused flags around the
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   364
		# given string
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   365
		self._reset(style)	# reinitialise the parser
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   366
		if not(len(text)>=6 and text[0]=='<' and _re_para.match(text)):
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   367
			text = "<para>"+text+"</para>"
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   368
		self.feed(text)
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   369
		self.close()	# force parsing to complete
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   370
		style = self._style
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   371
		del self._style
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   372
		if len(self.errors)==0:
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   373
			fragList = self.fragList
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   374
			self.fragList = []
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   375
			return style, fragList
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   376
		else:
119
b4dc589c8364 <para> tag added in layout.py paraparser.py
rgbecker
parents: 115
diff changeset
   377
			return style, None
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   378
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   379
if __name__=='__main__':
129
623baa767856 Splitting layout.py
rgbecker
parents: 128
diff changeset
   380
	from reportlab.platypus.paragraph import cleanBlockQuotedText
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   381
	_parser=ParaParser()
133
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   382
	def check_text(text,p=_parser):
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   383
		print '##########'
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   384
		text = cleanBlockQuotedText(text)
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   385
		l,rv = p.parse(text,style)
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   386
		if rv is None:
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   387
			for l in _parser.errors:
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   388
				print l
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   389
		else:
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   390
			print 'ParaStyle', l.fontName,l.fontSize,l.textColor
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   391
			for l in rv:
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   392
				print l.fontName,l.fontSize,l.textColor,l.bold, l.rise, l.text[:25]
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   393
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   394
	style=ParaFrag()
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   395
	style.fontName='Times-Roman'
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   396
	style.fontSize = 12
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   397
	style.textColor = black
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   398
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   399
	text='''
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   400
	<b><i><greek>a</greek>D</i></b>&beta;
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   401
	<font name="helvetica" size="15" color=green>
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   402
	Tell me, O muse, of that ingenious hero who travelled far and wide
113
2c20ecb9015e Fixed start_font
rgbecker
parents: 112
diff changeset
   403
	after</font> he had sacked the famous town of Troy. Many cities did he visit,
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   404
	and many were the nations with whose manners and customs he was acquainted;
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   405
	moreover he suffered much by sea while trying to save his own life
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   406
	and bring his men safely home; but do what he might he could not save
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   407
	his men, for they perished through their own sheer folly in eating
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   408
	the cattle of the Sun-god Hyperion; so the god prevented them from
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   409
	ever reaching home. Tell me, too, about all these things, O daughter
115
49b2fae9ea60 Fixes to superscript stuff
rgbecker
parents: 113
diff changeset
   410
	of Jove, from whatsoever source you<super>1</super> may know them.
96
2a9cca4c5cf0 Beginnings of a paragraph parser
rgbecker
parents:
diff changeset
   411
	'''
133
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   412
	check_text(text)
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   413
	check_text('<para> </para>')
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   414
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   415
	check_text('''
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   416
	<font color=red>&tau;</font>Tell me, O muse, of that ingenious hero who travelled far and wide
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   417
	after he had sacked the famous town of Troy. Many cities did he visit,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   418
	and many were the nations with whose manners and customs he was acquainted;
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   419
	moreover he suffered much by sea while trying to save his own life
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   420
	and bring his men safely home; but do what he might he could not save
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   421
	his men, for they perished through their own sheer folly in eating
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   422
	the cattle of the Sun-god Hyperion; so the god prevented them from
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   423
	ever reaching home. Tell me, too, about all these things, O daughter
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   424
	of Jove, from whatsoever source you may know them.''')
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   425
	check_text('''
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   426
	Telemachus took this speech as of good omen and rose at once, for
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   427
	he was bursting with what he had to say. He stood in the middle of
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   428
	the assembly and the good herald Pisenor brought him his staff. Then,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   429
	turning to Aegyptius, "Sir," said he, "it is I, as you will shortly
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   430
	learn, who have convened you, for it is I who am the most aggrieved.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   431
	I have not got wind of any host approaching about which I would warn
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   432
	you, nor is there any matter of public moment on which I would speak.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   433
	My grieveance is purely personal, and turns on two great misfortunes
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   434
	which have fallen upon my house. The first of these is the loss of
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   435
	my excellent father, who was chief among all you here present, and
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   436
	was like a father to every one of you; the second is much more serious,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   437
	and ere long will be the utter ruin of my estate. The sons of all
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   438
	the chief men among you are pestering my mother to marry them against
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   439
	her will. They are afraid to go to her father Icarius, asking him
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   440
	to choose the one he likes best, and to provide marriage gifts for
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   441
	his daughter, but day by day they keep hanging about my father's house,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   442
	sacrificing our oxen, sheep, and fat goats for their banquets, and
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   443
	never giving so much as a thought to the quantity of wine they drink.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   444
	No estate can stand such recklessness; we have now no Ulysses to ward
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   445
	off harm from our doors, and I cannot hold my own against them. I
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   446
	shall never all my days be as good a man as he was, still I would
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   447
	indeed defend myself if I had power to do so, for I cannot stand such
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   448
	treatment any longer; my house is being disgraced and ruined. Have
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   449
	respect, therefore, to your own consciences and to public opinion.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   450
	Fear, too, the wrath of heaven, lest the gods should be displeased
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   451
	and turn upon you. I pray you by Jove and Themis, who is the beginning
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   452
	and the end of councils, [do not] hold back, my friends, and leave
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   453
	me singlehanded- unless it be that my brave father Ulysses did some
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   454
	wrong to the Achaeans which you would now avenge on me, by aiding
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   455
	and abetting these suitors. Moreover, if I am to be eaten out of house
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   456
	and home at all, I had rather you did the eating yourselves, for I
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   457
	could then take action against you to some purpose, and serve you
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   458
	with notices from house to house till I got paid in full, whereas
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   459
	now I have no remedy."''')
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   460
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   461
	check_text('''
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   462
But as the sun was rising from the fair sea into the firmament of
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   463
heaven to shed light on mortals and immortals, they reached Pylos
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   464
the city of Neleus. Now the people of Pylos were gathered on the sea
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   465
shore to offer sacrifice of black bulls to Neptune lord of the Earthquake.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   466
There were nine guilds with five hundred men in each, and there were
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   467
nine bulls to each guild. As they were eating the inward meats and
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   468
burning the thigh bones [on the embers] in the name of Neptune, Telemachus
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   469
and his crew arrived, furled their sails, brought their ship to anchor,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   470
and went ashore. ''')
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   471
	check_text('''
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   472
So the neighbours and kinsmen of Menelaus were feasting and making
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   473
merry in his house. There was a bard also to sing to them and play
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   474
his lyre, while two tumblers went about performing in the midst of
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   475
them when the man struck up with his tune.]''')
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   476
	check_text('''
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   477
"When we had passed the [Wandering] rocks, with Scylla and terrible
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   478
Charybdis, we reached the noble island of the sun-god, where were
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   479
the goodly cattle and sheep belonging to the sun Hyperion. While still
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   480
at sea in my ship I could bear the cattle lowing as they came home
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   481
to the yards, and the sheep bleating. Then I remembered what the blind
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   482
Theban prophet Teiresias had told me, and how carefully Aeaean Circe
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   483
had warned me to shun the island of the blessed sun-god. So being
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   484
much troubled I said to the men, 'My men, I know you are hard pressed,
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   485
but listen while I tell you the prophecy that Teiresias made me, and
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   486
how carefully Aeaean Circe warned me to shun the island of the blessed
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   487
sun-god, for it was here, she said, that our worst danger would lie.
ba7b4eb9be6c Debugging xml changes
rgbecker
parents: 132
diff changeset
   488
Head the ship, therefore, away from the island.''')