igorls / context-builder / 38033688881

87%

Build Type

push

github

Committed by web-flow

Commit Message

fix: detect binary files by NUL bytes and magic numbers (B2) (#27)

* fix: detect binary files by NUL bytes and magic numbers

Windows-1252 accepts every byte, so a PDF header or a single NUL in
the sniff buffer was emitted as text. Treat a NUL as binary unless a
UTF-16/32 BOM is present, recognize common binary signatures, and keep
the existing placeholder plus UTF-16 and Latin-1 transcoding.

Co-authored-by: Igor Lins e Silva <igorls@users.noreply.github.com>

* fix: do not treat a bare MZ prefix as binary

Real PE/DOS images have a NUL in the first 64 bytes, so the NUL rule
already classifies them. A text file that starts with "MZ" stays text.
OggS and fLaC remain signatures, with that trade-off documented.

Co-authored-by: Igor Lins e Silva <igorls@users.noreply.github.com>

* fix: apply binary detection in the auto-diff snapshot path

* refactor: single placeholder path in auto-diff snapshot; test non-UTF-8 text

* test: cover sniff-boundary UTF-8 handling

---------

Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: Igor Lins e Silva <igorls@users.noreply.github.com>

Coverage Stats

37 of 38 new or added lines in 2 files covered. (97.37%)

4 existing lines in 1 file now uncovered.

2702 of 3398 relevant lines covered (79.52%)

4.24 hits per line