mirror of
https://github.com/storytold/craft-libs.git
synced 2026-11-13 22:12:12 +00:00
Dylan M. Taylor
d516b44691
Add craft-doc-binary: WordCraft's .doc reader on craft-wordmodel and craft-cfb * craft-doc-binary: WordCraft's crates/docbin as it is at wordcraft 07f97737 The source, unchanged, so the next commit shows only what the move changes. It has no manifest yet, so it is not built. * craft-doc-binary: on craft-wordmodel, craft-cfb and craft-codepage - The compound file is read with craft-cfb (bounded directory, streams refused above 1 GiB before anything is allocated), in place of the cfb crate and WordCraft's cfb_guard precheck; an oversized stream is DocbinError::Limit. - 8-bit text uses craft-codepage in place of wordcraft_doc::encoding. - wordcraft-doc is craft-wordmodel; wordcraft-geom's helpers are its geom. The tests keep building their files with the cfb crate (a dev-dependency), so craft-cfb's reader is checked against another implementation's writer. The repeating-DIFAT regression is craft-cfb's own; the guard asserts in the valid-files test go with the guard.
1.1 KiB
1.1 KiB
craft-doc-binary
Reader for the Word 97-2003 binary file format (.doc, [MS-DOC]) into the craft-wordmodel
document model. It is WordCraft's reader, on craft-cfb for the compound file and
craft-codepage for 8-bit text.
// Not a compound file: an error, never a panic.
assert!(craft_doc_binary::read(b"not a .doc").is_err());
assert!(!craft_doc_binary::is_encrypted(b"not a .doc"));
read(bytes),read_with_password(bytes, Option<&str>),is_encrypted(bytes).- Text in pieces (8-bit and UTF-16), paragraph and character formatting, styles, lists, tables, sections, headers and footers, notes, comments, fields, bookmarks, pictures and OfficeArt shapes, document properties.
- XOR obfuscation, RC4 and RC4 CryptoAPI password protection.
There is no writer.
Limits
Streams (1 GiB each, refused by craft-cfb before anything is allocated), paragraphs, pieces,
property counts, table nesting and drawing records are capped; the constants are in the source
next to what they bound, and each has a test. Hostile input gives an error or a best-effort
document, never a panic.