Mostly this involved adding a large number of docstrings to the codecs package, and adjusting the documentation generator and CSS to better accommodate the codecs package (improving the readability of tables, and linking from parent packages to child modules and vice versa for navigation).
Additional changes made at the same time:
encode_jis7_reduced was moved out of dbextra into the expected dbextra_data_7bit, allowing it to be lazy evaluated.
Cp874IncrementalDecoder and Cp874IncrementalEncoder were removed as redundant to Windows874IncrementalDecoder and Windows874IncrementalEncoder, and their labels transferred (the former differed in not mapping the non-repurposed C1 codes, but that hardly warrants being included when the latter already is). This redundancy had passed unnoticed while I was adding single-byte mappings for Python parity.
An additional criterion allowing narrowing down to ISO-2022-JP was added to JapaneseAutodetectIncrementalDecoder (this mainly improves handling of incremental use with small segments per call which will not, or are unlikely to, contain an entire escape sequence in a single segment, but are likely to not switch the decoder state back to ASCII by the end of the segment).
Several instances of code referencing a non-existent local, which would never execute given the current set of error handlers, but would if someone registered an error handler which returned a negative resumption index, were fixed.
A cp042 codec was added to handle code page 42 sometimes used by Windows to map symbol fonts to the Private Use Area (this apparently maps to an entirely different range than x-user-defined, despite being very similar in concept besides not being Extended ASCII).
Mostly this involved adding a large number of docstrings to the codecs package, and adjusting the documentation generator and CSS to better accommodate the codecs package (improving the readability of tables, and linking from parent packages to child modules and vice versa for navigation).
Additional changes made at the same time:
encode_jis7_reducedwas moved out ofdbextrainto the expecteddbextra_data_7bit, allowing it to be lazy evaluated.Cp874IncrementalDecoderandCp874IncrementalEncoderwere removed as redundant toWindows874IncrementalDecoderandWindows874IncrementalEncoder, and their labels transferred (the former differed in not mapping the non-repurposed C1 codes, but that hardly warrants being included when the latter already is). This redundancy had passed unnoticed while I was adding single-byte mappings for Python parity.JapaneseAutodetectIncrementalDecoder(this mainly improves handling of incremental use with small segments per call which will not, or are unlikely to, contain an entire escape sequence in a single segment, but are likely to not switch the decoder state back to ASCII by the end of the segment).cp042codec was added to handle code page 42 sometimes used by Windows to map symbol fonts to the Private Use Area (this apparently maps to an entirely different range thanx-user-defined, despite being very similar in concept besides not being Extended ASCII).