1794 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
davebrokit	f3e37eafae	Introduce IBlock and ILettersBlock interfaces (Round 2) (#1263) ... * Code review changes - Keep the Bounds property on the image classes so this isn't a major breaking API change - Don't expose letters collection * Minor fix * Switch to using BoundingBox in the library --------- Co-authored-by: davmarksman <david@brokit.co.uk> nightly-latest	2026-02-28 16:25:51 +00:00
Eliot Jones	a4047247a8	Additional digital corpora testing (#1261) ... * add additional testing pairs 0010-0011 for integration some of these files required the skip missing fonts flag set to true - propagate use lenient parsing for dictionaries inside arrays, handles a corrupt file 0012710 not in this test set * add pair 0012-0013	2026-02-22 21:01:03 +00:00
BobLd	34db05f735	Fix Benchmarks solution and add BruteForceBenchmarks	2026-02-22 20:05:57 +00:00
Eliot Jones	9c0d6893e0	revert flate decode handling to more lenient processing (#1254) ... * revert flate decode handling to more lenient processing the change to use zlib/a adler checksum verification flow meant that invalid flate streams would not be decoded correctly. this caused issues for files that included invalid/missing checksums. this reverts the processing to the old approach for files like #1235 * fix object stream offset handling and track circular refs * update tests * normalize line endings for mac runner * fixes for mac clownery * add next pair to common crawl action * add a test case for the root cause of the int overflow	2026-02-22 15:49:50 +00:00
BobLd	adb5713621	Make LinkAnnotation public	2026-02-18 14:09:20 +00:00
BobLd	665623561a	Add UglyToad.PdfPig.Benchmarks and misc performance improvements	2026-02-17 11:53:09 +00:00
EliotJones	7e370a0eab	confirm tag is on master before running publish	2026-02-15 18:39:45 +00:00
EliotJones	b9aa53166d	replace release flow single job with pr process ... since actions do not have permissions to push directly to master and bot accounts to achieve the same are hard to manage we will change the release flow to work as follows. 1. manually invoke `prepare_release_pr.yml` action, this creates a new branch with the version of all project files updated and creates a pull request for that version. this pr then should be merged using rebase merge 2. `tag_release.yml` checks if the newest commit name starts with the text "Release " and also verifies if it changed the version of the package csproj. if both those conditions are met it will create and push a new tag, e.g. `v0.1.17` to master 3. `publish_nuget.yml` listens for new `v*` tags on master and triggers the nuget deployment this is all chat gpt code so who knows if it will work	2026-02-15 18:39:45 +00:00
BobLd	0a2b1e076f	Improve HasFormXObjectCircularReference and fix #1250	2026-02-15 18:06:35 +00:00
EliotJones	f732718852	both Tj and TJ operators should increment text sequence #1241 ... the text sequence was added to group letters added by the same content stream operation together for layout analysis. TJ allows for individual letter positioning in the same operation which modifies the text matrix while writing whereas Tj just writes the text at the current location. both should be semantically equivalent for this sequence number. this was originally added in #61	2026-02-15 17:50:07 +00:00
Alexander Vinnikov	d69e3fa6a9	make ext gstates stacked	2026-02-11 18:39:33 +00:00
BobLd	d6e86b057e	Handle empty encoding in Type1FontSimple and fix #1248	2026-02-10 06:14:39 +00:00
BobLd	c27f1b6553	Only throw if ArrayToken length is less than 4 in ToRectangle() and fix #1238	2026-02-01 16:10:24 +00:00
BobLd	629891e80e	Make LinkAnnotation internal to fix unit tests	2026-02-01 15:50:21 +00:00
David Ochsner	13825b9424	Add Links to Pdf Generation	2026-02-01 15:13:55 +00:00
BobLd	7c4f5e2424	Introduce StackDepthGuard class to check for stack depth in CoreTokenScanner and fix #1217	2025-12-23 16:24:04 +01:00
BobLd	bd573b2a1c	Increment version to 0.1.14	2025-12-23 12:00:19 +01:00
BobLd	baeac0d0c3	Do not return glyph bbox and path in Type1Font if character name is '.notdef' 0.1.13	2025-12-23 11:09:00 +01:00
BobLd	ee0cb1dc4a	Use file header offset when doing brute force find and fix #1223	2025-12-07 13:43:22 +00:00
BobLd	c70b343caa	Minor Type1FontParser optimisations	2025-12-02 17:29:55 +00:00
Carlo Kok	ce563db133	Clarify handling of optional content group names ... Updated comments to clarify handling of optional content group names in PDFs.	2025-11-23 10:52:21 +00:00
BobLd	37a5dffcaa	Skip in ReadCompositeGlyph when glyphIndex is out of range and fix #1213	2025-11-21 21:31:18 +00:00
EliotJones	aef0a78ee6	update release logic to check out master before commit	2025-11-14 09:50:07 +00:00
BobLd	005e52783e	Ensure no key end up missing in ResolveInternal and fix #1209	2025-11-13 14:53:28 +00:00
BobLd	f4e7db5b5b	Simply order by offset also when not doing brute force to fix #1208	2025-11-13 14:26:35 +00:00
BobLd	52ecef0e28	Increment version to 0.1.13	2025-11-13 09:34:17 +00:00
BobLd	599ce04bef	Add test that fails before revert of e11dc6b v0.1.12	2025-11-07 22:31:43 +00:00
BobLd	2a6ee918b7	Revert "Avoid a lot of seeks by making most tokenizers no longer read to far by using seek." ... This reverts commit e11dc6bf40.	2025-11-07 22:31:43 +00:00
BobLd	9d3cd0a429	Check for array size before slice in ColorSpaceDetailsByteConverter.Convert()	2025-11-07 20:02:39 +00:00
BobLd	b49e5aa697	Check for index out of range in GlyphDataTable.ReadFlags() and fix #1199	2025-11-07 19:39:18 +00:00
Alexander Vinnikov	181fa9d837	make transform stack consistent	2025-11-03 20:06:40 +00:00
BobLd	6ce6986d78	Update test run command to use Release configuration	2025-10-29 20:29:42 +00:00
BobLd	37c9fef20b	Do not slice the stream to the length breaks decoding in FlateDecode	2025-10-29 20:27:40 +00:00
Bert Huijben	e11dc6bf40	Avoid a lot of seeks by making most tokenizers no longer read to far by using seek. ... Optimize the FirstPassParser to just fetch a final chunk before doing things char-by-char backwards.	2025-10-28 06:48:41 +00:00
BobLd	40bcc22ea1	Add CMap caching at document level and add MurmurHash3 hashing function	2025-10-26 16:20:27 +00:00
Bert Huijben	94d515061e	Update NameToUnicodeConvertAglSpecification to test what was intended	2025-10-25 19:47:56 +01:00
BobLd	3555521634	Fix regression introduced in 3592fc8 where slicing the stream to the length breaks decoding	2025-10-23 19:08:05 +01:00
Bert Huijben	6fba565d66	Avoid doing a true file seek for simple peeking the next char in the token parser	2025-10-20 06:33:34 +01:00
Bert Huijben	3592fc8438	Use zlib information to verify compressed content before using it	2025-10-15 18:46:36 +01:00
BobLd	c9034f991c	Only apply RemoveStridePadding() when bytes per pixel is one and fix #1183	2025-10-15 12:57:25 +01:00
BobLd	255e70f0a7	Set Type 3 font ascent to Top instead of Height, see #1164	2025-10-14 12:18:52 +01:00
BobLd	2216ade1f2	Trim excess in long lived font collections	2025-10-14 10:26:41 +01:00
BobLd	cf0c33b1e0	Improve DfsIterative() performance	2025-10-13 12:35:21 +01:00
BobLd	ffba176060	Improve GroupIndexes() performance with #1178	2025-10-13 12:35:21 +01:00
BobLd	b14f45f59f	Add more tests to NearestNeighbourWordExtractorTests	2025-10-12 19:51:04 +01:00
ricflams	c28d114b79	Guard against circular references in XRef tables/streams ... - Detect and prevent an xref table/stream at a certain offset from being read twice; malformed xref tables with circular references could otherwise cause the table-reading to loop forever. - Another approach could be to prevent TryReadTableAtOffset from changing the bytes' CurrentOffset to the lastObjPosition in its attempt to read a table (eg restore CurrentOffset after the attempt to read a Table) so the outer bytes-loop could continue its search through the entire bytes unaffected.	2025-10-01 06:32:38 +01:00
Richard Flamsholt	d7d01f842e	Update test Issue874: No longer missing a font ... Including the stream-xref means that the formerly missing font is no longer missing, so simply run the two test-cases under the (stricter) assumption of SkipMissingFonts=false.	2025-09-30 18:35:45 +01:00
Richard Flamsholt	33a8d829ee	Update test Issue874: Also more text on page 2 ... Page two has had four more characters added, which is now delected by this xref-stream fix	2025-09-30 18:35:45 +01:00
Richard Flamsholt	57921c7e9b	Update test Issue874: Now finds more text on page 1 ... With the fix for including associated streams, this test now finds more text on the first page. I've verified using Aspose.PDF and by viewing the ErcotFacts.pdf file being tested that yes, it was indeed missing part of the text before.	2025-09-30 18:35:45 +01:00
ricflams	5a6b3970f0	Add table-xref's associated stream-xrefs ... - If an XrefTable has an associated stream, as indicated via the XrefStm-property, then read and add that XrefStream - Any table can have 0 or 1 such associated streams - A caveat: such an associated stream might also theoretically be part of the Parts-sequence in which case it would be encountered both by looping through all those parts along with all the regular tables and now also by association to any of those tables. It doesn't seem harmful since the offsets are flattened eventually anyway and stored by their offset-key into a mapping-table.	2025-09-30 18:35:45 +01:00