CHANGES

=======

This is a summary of changes made to the reportlab source code for each release.

Please refer to subversion backlogs (using the release dates) for more details

or for releases which we have not provide a higher level changes list for.

The contributors lists are in no order and apologies to those accidentally not

mentioned. If we missed you, please let us know!

RELEASE 3.2  01/06/2015

   * Steven Jacobs

   * Philip Semanchuk

   * Marius Gedminas

   * masklinn

   * Kale Franz

   * Albertas Agejavas

   • Anders Hammarquist

   * jvanzuela @ bitbucket

   * Glen Lindermann

   * Greg Jones

   * James Bynd

RELEASE 3.1  22/04/2014

If you are running ReportLab 3.0.x, the changes are minor.

ReportLab 3.0 now supports Python 2.7, 3.3 and higher.  

There has been a substantial internal rewrite to ensure consistent use of unicode strings for

  natural-language text, and of bytes for all file format internals.  The intent

  is to make as few API changes as possible so that there should be little or no

  impact on users and their applications.  Changes are too numerous but can be

  seen on Bitbucket.

### Python 3.x compatibility

  * Python 3.x compatibility.  A single line of code should run on 2.7 and 3.3

  * __init__.py restricts to 2.7 or >=3.3

  * __init__.py allow the import of on optional reportlab.local_rl_mods to allow monkey patching etc.

  * rl_config now imports rl_settings & optionally local_rl_settings

  * xmllib is gone, alongside the paraparser stuff that caused issues in favour of HTMLParser.

  * some obsolete C extensions (sgmlop and pyHnj) are gone

  * Improved support for multi-threaded systems to the _rl_accel extension module.

### New features

  * highlight annotation (contributed by Ben Echols)

### Other

  * numerous very minor fixes, visible through BitBucket.

###Charts / graphics enhancements

  * Added SimpleTimeSeriesPlot

  * added _computeMaxSpace

  * added in lineStyle (for bars)

  * improved SVG rendering

###Charts/graphics bug fixes:

  * piecharts.py: fix Pie3d __init__ to call its superclass

  * linecharts.py: fix swatch creation

###PDF

  * Fixes to testshapes & pdfform resetting

  * colors.py

  * various minor fixes

###Platypus

  * Defined a small bullet rather than a big circle as the default for unordered lists

  * fixed attribute spelling bug

  * fixed CJK + endDots

###Acknowledgements

This is a minor release focusing mainly on improved documentation.  There are a 

number of minor enhancements, and a larger number of previous-undocumented

enhancements which we have documented better.

   * Manuals have been reformatted with more pleasing code snippets and tables of 

     contents, and reviewed and expanded

   * Added support for HTML-style list objects

   * Added flexible mechanism for drawing bullets

   * Allowed XPreformatted objects to use Asian line wrapping

     can now set up a 'chapter first page template' which will always be followed

     by a 'continuation template' on the next page break, saving the programmer from

     having to issue control flow commands in the story.

   * added a TopPadder flowable, which will 'wrap' another Flowable and move it 

     to the bottom of the current page.  

   * More helpful error messages when large tables cannot be rendered

   * Trailing dots for use on contents pages

   * Support for UPCA bar codes

   * We now have a semi-intelligent system for labelling pie charts with 

     callout lines.  Thanks to James Martin-Collar, a maths student at Warwick 

     University, who did this as his summer internship.

   * Axes - added startOffset and endOffset properties; allowed for axis 

     background annotations.

   * Bar charts - allow more control of z Index (i.e. drawing order of axes and

     lines)

   * Pie charts - fixed bugs in 3d appearance

   * SVG output back end has seen some bugs fixed and now outputs resizeable SVG

   * Alex Buck

   * Felix Labrecque <felixl@densi.com>

   * Peter Johnson <johnson.peter@gmail.com>

   * James Martin-Collar

   * Guillaume Francois

Many new features have been added and numerous bugs have been fixed.

Thanks to everybody who has contributed to the open-source toolkit in

the run-up to the 2.5 release, whether by reporting bugs, sending patches,

or contributing to the reportlab-users mailing list.

Major contributors are credited in the user documentation.

   * Support for colour separated PDF output and other optimisations and

     features for high-quality printing, including enforcement of colour

     models for CMYK, RGB, and "spot colours"

   * Long table optimisations are now turned on by default.  Previously,

     documents with very long tables spanning many pages could take a long

     time to create because we considered the whole table to work out row

     and column sizes.  A patch was submitted some time ago to fix this

     controlled by a flag in the rl_config file, but this was set 'off'

     for compatibility.  Users are often not aware of this and we haven't

     found any real-world cases where the new layout technique works badly,

     so we are turning this behaviour on.

   * New support for QR barcodes - [try our demo!](https://www.reportlab.com/demos/qr/)

###PDF

   * Colour separation and other enhancements for high-end print

   * Python 2.7 support

###Charts

   * reportlab.graphics.charts.axes

       * ValueAxis

           * avoidBoundSpace - Space to allow above and below

           * abf_ignore_zero - Set to True to make the avoidBoundFrac calculations treat zero as non-special

           * keepTickLabelsInside - Ensure tick labels do not project beyond bounds of axis if true

       * NormalDateXValueAxis

           * specialTickClear - clear rather than delete close ticks when forced first/end dates

       * AdjYValueAxis

           * labelVOffset - add this to the labels

   * reportlab.graphics.charts.barcharts

       * BarChart

           * categoryLabelBarSize - width to leave for a category label to go between categories

           * categoryLabelBarOrder - where any label bar should appear first/last

           * barRecord (advanced) - callable(bar,label=labelText,value=value,**kwds) to record bar information

   * reportlab.graphics.charts.legends

       * SubColProperty

           * dx - x offset from default position

           * dy - y offset from default position

       * Legend

           * swdx - x position adjustment for the swatch

           * swdy - y position adjustment for the swatch

   * reportlab.graphics.charts.piecharts

       * Pie

           * wedgeRecord (advanced) - callable(wedge,*args,**kwds)

   * reportlab.graphics.charts.utils

       * DrawTimeCollector - generic mechanism for collecting information about nodes at the time they are about to be drawn

And many bugfixes...

This is the Big One, and the reason some apps may break. You must now pass in 

text either in UTF-8 or as unicode string objects. The library will handle 

everything to do with output encoding. There is more information on this below.

Since this is the biggest change, we'll start by reviewing how it worked in the 

past. In ReportLab 1.x, any string input you passed to our APIs was supposed to 

be in the same encoding as the font you selected for output. If using the 

default fonts in Acrobat Reader (Helvetica/Times/Courier), you would have 

implicitly used WinAnsi encoding, which is almost exactly the same as Latin-1. 

However, if using TrueType fonts, you would have been using UTF-8. For Asian 

fonts, you had a wide choice of encodings but had to specify which one (e.g 

Shift-JIS or EUC for Japanese). This state of affairs meant that you had to 

make sure that every piece of text input was in the same encoding as the font 

used to display it.

With ReportLab 2, none of that necessary. Instead:

Here is what's different now:

  Any text you pass to a canvas API (drawString etc.), Paragraph or other 

  flowable constructor, into a table cell, or as an attribute of a graphic (e.g. 

  chart.title.text), is supposed to be unicode. If you use a traditional Python 

  string, it is assumed to be UTF-8. If you pass a Unicode object, we know it's 

  unicode. 

  Fonts still work in different ways, and the built-in ones will still use 

  WinAnsi or MacRoman internally while TrueType will use UTF-8. However, the 

  library hides this from you; it converts as it writes out the PDF file. As 

  before, it's still your job to make sure the font you use has the characters 

  you need, or you may get either a traceback or a visible error character. 

  Asian CID fonts

  You no longer need to specify the encoding for the built-in Asian fonts, 

  just the face name. ReportLab knows about the standard fonts in Adobe's Asian 

  Language Packs. 

  The standard Truetype fonts differ slightly for Asian languages (e.g 

  msmincho.ttc). These can now be read and used, albeit somewhat inefficiently. 

  Asian word wrapping

  Previously we could display strings in Asian languages, but could not 

  properly wrap paragraphs as there are no gaps between the words. We now have a 

  basic word wrapping algorithm.

  A convenience tag, <unichar/> has also been added. You can now do <unichar 

  code="0xfc"/> or <unichar name='LATIN SMALL LETTER U WITH DIAERESIS'/> and get 

  a lowercase u umlaut. Names should be those in the Unicode Character Database.

  Accents, Greeks and symbols

  The correct way to refer to all non-ASCII characters is to use their 

  unicode representation. This can be literal Unicode or UTF-8. Special symbols 

  and Greek letters (collectively, "greeks") inserted in paragraphs using the 

  greek tag (e.g. <greek>lambda</greek>) or using the entity references (e.g. 

  λ) are now processed in a different way than in version 1. Previously, 

  these were always rendered using the Zapf Dingbats font. Now they are always 

  output in the font you specified, unless that font does not support that 

  character. If the font does not support the character, and the font you 

  specified was an Adobe Type 1 font, Zapf Dingbats is used as a fallback. 

  However, at present there is no fallback in the case of TTF fonts. Note that 

  this means that documents that contain greeks and specify a TTF font may need 

  changing to explicitly specify the font to use for the greek character, or you 

  will see a black square in place of that character when you view your PDF 

  output in Acrobat Reader.

    FreeTextAnnotation allows showing and hiding of an arbitrary PDF "form" 

    (reusable chunk of PDF content) depending on whether the document is printed or 

    viewed on-screen, or depending on whether the mouse is hovered over the 

    content, etc.

    ReportLab now supports using TTF fonts packaged in .TTC files

    You no longer need to specify the encoding for the built-in Asian