Changes

This chapter describes the changes coming with each new version of pygount.

Version 1.2.5, 2021-05-16

  • Removed support for Python 3.5. Probably it still works but there is no easy way to test this anymore because 3.5 has reached end of life a while ago.

Version 1.2.4, 2020-08-11

  • Fixed scanning of “.” (for current folder), which was skipped entirely (issue #56).

Version 1.2.3, 2020-07-05

  • Improved detection of text files by trying to guess a lexer for *.txt before assuming it is text. This basically fixes the detection of CMakelists.txt as CMake file #53). However, it will only work with some files due to multiple issues with the regular expression Pygments uses in versions up to 2.6.1 to detect CMake headers. This should be fixed once pull request #1491 is applied.

Version 1.2.2, 2020-06-24

  • Changed preprocessor statements to count as code, unlike Pygments which treats them as special comments (contributed by nkr0, issue #51).

Version 1.2.1, 2020-04-02

  • Fixed broken links in README on PyPI by moving the documentation to ReadTheDocs.

  • Improved API:

    • Changed factory functions to methods and added deprecation warnings:

      • source_analysis()SourceAnalysis.from_file()

      • pseudo_source_analysis()SourceAnalysis.from_state()

    • Changed attributes in SourceAnalysis to read-only properties.

    • Renamed properties holding counts from xxx to xxx_count.

    • Added API reference to documentation.

    • Added a couple of type hints and assertions.

Version 1.2.0, 2020-03-30

  • Added file count to summary.

  • Changed installation to fail when attempting to install on Python earlier than 3.5.

  • Improved API:

    • Changed SourceAnalysis.state to be a proper enum instead of a string.

    • Added ProjectSummary to summarize multiple files.

  • Cleaned up project:

    • Changed continuous integration from Travis CI to Github actions in the hope that the CI build does not automatically break after a while because things constantly change in the CI backend.

    • Changed README format from reStructuredText to Markdown.

    • Improved badges in README: added a badge for supported Python versions and unified the layout by using <https://shields.io>.

    • Removed obsolete development files (for ant, tox etc).

Version 1.1.0, 2020-03-10

  • Fixed --folders-to-skip and --names-to-skip which simply were ignored (contributed by pclausen, issue #17).

  • Added option --format=summary to get a language overview and sum total (based on a contribution by Yuriy Petrovskiy, issue #16).

  • Added Python 3.7 and 3.8 to the list of supported versions.

  • Dropped support for Python 3.3 and 3.4, mostly because it became hard to test without going through major hoops.

Version 1.0.0, 2017-07-04

  • Fixed confusing warning about XML file <unknown> caused by SAX parser. As a workaround, <unknown> is now replaced by the actual path of the XML file that cannot be parsed.

  • Added Python 3.6 to the list of supported versions (issue #14).

Version 0.9, 2017-05-04

  • Fixed AssertionError when option --encoding=chardet was specified.

  • Changed warning message “no fallback encoding specified, using](encoding>” to a debug message because it did not add any interesting information as the encoding actually used is visible in the info message for each file.

  • Added detection of binary files and excluded them from the analysis. In particular Django model objects (*.mo) are not considered Modelica source code anymore (issue #11).

  • Added detection of DocBook XML by DTD (issue #10).

  • Added support for suffices to indicate PL/SQL files according to Oracle FAQ entry on file extensions (issue #12).

  • Added possibility to specify a fallback encoding for encoding ‘chardet’. Use e.g. --encoding=chardet;cp1252.

Version 0.8, 2016-10-07

  • Fixed option --verbose. Now each analyzed source code results in at least one informational message in the log.

  • Added detection of duplicates using size and then MD5 code as criteria (issue #2). Use the option --duplicates to still count duplicate source code.

  • Improved detetion of programming language, which is now more consistent and yields the same language between Python invocations.

Version 0.7, 2016-09-28

  • Fixed that option --generated was ignored.

  • Added support for a couple of languages not supported by pygments yet:

    • m4, VBScript and WebFOCUS use minimalistic lexers that can distinguish between comments and code.

    • OMG IDL repurposes the existing Java lexer.

  • Added detection of certain XML dialects as separate language (issue #8).

Version 0.6, 2016-09-26

  • Fixed that source files could end up as __error__ if the first non ASCII characters showed up only after 16 kilobyte and the encoding was not UTF-8. Now pygount attempts to read the whole file as UTF-8 before assuming it actually is UTF-8.

  • Changed lines in plain text files to count as comments (issue #9). Before pygments treated them as ResourceBundle.

  • Changed that empty files have __empty__ as language (issue #7).

  • Extended workaround for pygments issue #1284 to replace any lexer *+Evoque by *.

Version 0.5, 2016-09-22

  • Added that generated source code is excluded from analysis (issue #1). Use option --generated to specify patterns that indicate generated code.

  • Added workaround for pygments sometimes detecting the same XML file as XML and other times as XML+Evoque (probably depending on the hash seed). Now XML+Evoque is always changed to XML.

  • Added __pycache__ as default --folders-to-skip.

  • Added notes on pseudo languages for source code that cannot be analyzed.

Version 0.4, 2016-09-11

  • Fixed LookupError on broken encoding in magic comment (issue #4).

  • Added options --folders-to-skip and --names-to-skip to specify which files should be excluded from analysis.

  • Added comma (,) and colon (:) to list of “white characters” that do not count as code if there is nothing else in the line.

  • Improved pattern matching: for all options that according to --help take PATTERNS you can now specify that the patterns are regular expressions instead of shell patterns (using [regex]) and that they should extend the default patterns (using [...]).

  • Improved documentation: added notes on how code is counted and how pygount compares to other similar tools.

Version 0.3, 2016-08-20

  • Fixed @rem comments in DOS batch files (issue #3).

  • Cleaned up code.

Version 0.2, 2016-07-10

  • Fixed that files starting with underscore (e.g. __init__.py) were excluded from analysis.

  • Changed chardet package to be optional.

  • Added possibility to specify single files and glob patterns to analyze.

  • Added that lines containing only certain characters are treated as white space instead of code. Currently this concerns brackets (()[]{}) and semicolon (;).

  • Added that Python’s pass statement is treated as white space instead of code.

  • Cleaned up and (slightly) optimized code.

Version 0.1, 2016-07-05

  • Initial public release.