PeculiarVentures / pdf / 41426318606

77%

push

github

refactor(deps): replace pvtsutils and pvutils with @peculiar/utils (#150)

Use @peculiar/utils ^4.1.0 root exports (`hex`, `binary`, `utf8`,
`utf16`, `base64`, `base64url`, `bytes`, `toArrayBuffer`,
`toUint8Array`, `isBufferSource`) instead of pvtsutils' `Convert` and
`BufferSourceConverter` and pvutils' `isEqualBuffer`; no `/legacy`
imports. Call sites keep their ArrayBuffer results. Public signatures
use @peculiar/utils' `BufferSource` type (ArrayBuffer, SharedArrayBuffer
or any ArrayBufferView), which replaces pvtsutils' type of the same
name.

Cross-reference stream fields are the one place that decoded
odd-length hex, which pvtsutils padded with a leading zero; it now asks
`hex.decode` for that explicitly. New specs pin the bytes that reach the
PDF: hex strings (white space, odd digit count, lowercase output,
invalid digits), UTF-16BE text strings with the BOM and surrogate
pairs, one-byte-per-character string output, cross-reference stream
fields and the zero-padded signature /Contents. They give the same
results with pvtsutils.

Remaining differences on invalid input: UTF-8 decoding replaces
malformed sequences with U+FFFD instead of throwing URIError, and
UTF-8 encoding of lone surrogates (e.g. in an encryption password)
produces U+FFFD instead of throwing.

2721 of 4032 branches covered (67.49%)

Branch coverage included in aggregate %.

142 of 165 new or added lines in 49 files covered. (86.06%)

7190 of 8909 relevant lines covered (80.7%)

1299.24 hits per line