Python 3000

Python 3000
Chosen Path
Conversion Script
Status
Example

Python 3.x is a new version of the language, with some backward compatibility issues.

In general, Python 2.x code does not run unmodified under Python 3.x, mostly because the "unicode/string/buffer issue", so some steps are taking place to port this library to the new version.

Chosen Path

There will not be a manual py3k repository; all py3k changes will be back-ported to the trunk ~~in order to enable 2to3 conversion automatically~~ (unified codebase, compatible with both Python 2 and Python 3, without needing the 2to3 converter in the latest revisions).

As this library is pure Python and has no dependencies (beside PIL optionally), there is no need to do a huge re-factor. BTW, initial py3k support took a few changes, see changeset c2f3bae9b379 (just 8 changes). More changes are coming to fix Unicode TTF and image support.

Most string and buffering methods are encapsulated in well-known places (like the _out and sprintf functions), so they should not cause a big impact.

Also, the library is Unicode aware since version 1.7, so also that impact could be mitigated.

NOTE: Until PEP 461 (add % formatting to bytes) is officialy accepted in Python or a suitable alternative is implemented, internal formatting is done using latin1 Unicode strings instead of raw bytes, as this feature is extensively used in the code and allows it to be clean and compact.

Conversion Script

IMPORTANT NOTE: since version 1.7.2 (revision ca2968763421) the codebase has been unified, so there is no need to run any conversion script. (Please skip this section.)

In Windows, you can use py3k.bat, which just calls 2to3.py, install the package and run the basic test:

mkdir fpdf_py3k
c:\python32\tools\Scripts\2to3.py -f all -w -o fpdf_py3k -n fpdf 
c:\Python32\python.exe setup.py install
c:\Python32\python.exe tests\py3k.py

setup.py has the logic to detect the interpreter version and install the correct version of the library.

Status

Currently (version 1.7.2, January 2014) there is early experimental support for Python 3:

most directives are supported (at least add_page, set_font, ln, write, output)
compression is not yet supported (zlib support should be rewritten)
TTF Unicode fonts are not yet supported, use windows-1252 standard fonts (latin1)
image support is working at least for PNG (JPG and GIF are untested, and also depend on PIL)

Example

The following example runs unmodified on Python 2.x and Python 3.x

from fpdf import FPDF

pdf = FPDF()
# compression is not yet supported in py3k version
pdf.compress = False
pdf.add_page()
# Unicode is not yet supported in the py3k version; use windows-1252 standard font
pdf.set_font('Arial', '', 14)  
pdf.ln(10)
pdf.write(5, 'hello world %s áéíóúüñ' % sys.version)
pdf.image("pyfpdf/tutorial/logo.png", 50, 50)
pdf.output('py3k.pdf', 'F')

View the result here: py3k.pdf

It should contain:

hello world 3.2.2 (default, Sep 4 2011, 09:51:08) [MSC v.1500 32 bit (Intel)] áéíóúüñ

(áéíóúüñ is a latin1 test)

And the FPDF logo.