2497 Commits
Author SHA1 Message Date
Sebastian Husch Lee bc4729fd53 feat: add tool result offload compactor (#12487) 2026-10-09 13:47:35 +02:00
Ootso Dharchowdhuryandanakin87 ca03bbf89d fix: unwrap Annotated in _unwrap_optional for State parameter detection (#13169)
Co-authored-by: anakin87 <[email protected]>
2026-10-08 14:45:52 +00:00
Sebastian Husch Lee 58f38e5b4a fix: correct LLMDocumentContentExtractor failure messages and docstring (#13175) 2026-10-08 10:55:36 +00:00
Sebastian Husch Lee ec4c02868f fix: don't return a snapshot path when saving the pipeline snapshot fails (#13174) 2026-10-08 10:54:03 +00:00
9b166e6507 fix: always close PDF in _convert_pdf_to_images; narrow bare except in SentenceSplitter (#12942)
Co-authored-by: Timur Rakhmatullin <[email protected]>
Co-authored-by: Claude Opus 4.6 <[email protected]>
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: David S. Batista <[email protected]>
2026-10-08 10:45:14 +01:00
Sebastian Husch Lee acdf073b99 test: update expected OpenAI token count in test_count_complex (#13176) 2026-10-08 10:42:05 +01:00
Rachit Mittal d9d5686b31 fix: preserve cross-socket connections in Pipeline.connect (#13149) 2026-10-08 10:51:11 +02:00
David S. Batista 517881afa6 feat: collecting agents tool in telemetry events (#13125) 2026-10-07 18:19:42 +02:00
Stefano Fiorucci 56c342d52e feat!: improve HTMLToDocument encoding detection (#13153) 2026-10-07 15:58:21 +02:00
bogdankosticandDavid S. Batista f8aaedbcd8 fix: map raw reasoning text to reasoning_text field in OpenAIResponse components (#13124)
Co-authored-by: David S. Batista <[email protected]>
2026-10-07 14:23:32 +02:00
Aditya17-bot 3807058535 fix: write EvaluationRunResult CSV reports as UTF-8 (#13143) 2026-10-07 13:04:50 +02:00
Zeke Z. fc4522f7ab fix: keep both filters when FilterPolicy.MERGE combines OR/NOT filters (#12816)
Signed-off-by: Zhang.ZJ <[email protected]>
2026-10-07 09:59:41 +02:00
Lukas 0383da04c4 fix: read text boxes in DOCXToDocument (#12769) 2026-10-07 09:37:50 +02:00
Sebastian Husch Lee ef6eee2daf chore: use OpenAIResponsesChatGenerator as the default in LLMRanker and QueryExpander (#13129) 2026-10-07 09:23:30 +02:00
5d298af165 fix!: parse list-of-parts content in ChatMessage.from_openai_dict_format (#12446)
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
Co-authored-by: Julian Risch <[email protected]>
2026-10-06 20:18:35 +02:00
Sebastian Husch Lee a9ed2f9370 chore: clarify QueryExpander fallback warnings (#13131) 2026-10-06 11:54:53 +02:00
Sebastian Husch Lee 4a59e5023f docs: clarify that LLMRanker fallback documents are unranked (#13128) 2026-10-06 09:59:53 +02:00
Stefano Fiorucci 9b8c3ba911 feat!: add lifecycle management for Tools/Toolsets (#13024) 2026-10-05 12:13:43 +02:00
Christopher Krah da212420c6 fix: parse empty tool call arguments in the non-streaming OpenAI generators (#13047) 2026-10-05 08:33:01 +02:00
0e4b7cd171 fix: preserve row_number columns in CSVToDocument (#12788)
Co-authored-by: LimbC-C <[email protected]>
Co-authored-by: anakin87 <[email protected]>
2026-10-02 13:46:30 +00:00
Eddy ZHANG 643d1b9af7 fix: raise a clear error when JsonSchemaValidator gets no messages (#13078) 2026-10-02 15:30:58 +02:00
414e6c7665 fix: sort SentenceWindowRetriever context before merging text (#12976)
Co-authored-by: Claude Opus 5.5 <[email protected]>
Co-authored-by: bogdankostic <[email protected]>
2026-10-02 13:47:47 +02:00
07b22cb339 feat(embedders)!: default to text-embedding-3-small (#11742)
Co-authored-by: bogdankostic <[email protected]>
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
2026-10-02 12:55:57 +02:00
6f2fee053e fix: raise ValueError for reversed page ranges in expand_page_range (#11492)
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Sonnet 5 <[email protected]>
2026-10-02 11:42:47 +02:00
bogdankosticandSebastian Husch Lee 472da278db feat!: align page_number metadata across splitters and fix drift in overlapping splits (#13006)
Co-authored-by: Sebastian Husch Lee <[email protected]>
2026-10-01 15:36:11 +02:00
Marcuswang0824 c26422883f fix: validate non-textual inputs in LostInTheMiddleRanker (#13037) 2026-10-01 06:53:16 +00:00
Julian RischandClaude 3bcf7e7e58 fix: read abbreviation lists as UTF-8 regardless of locale (#13052)
Co-authored-by: Claude <[email protected]>
2026-10-01 08:48:24 +02:00
Grapette.Landanakin87 2ae9acdf31 fix: render every content part of ChatPromptBuilder message templates (#12907)
Co-authored-by: anakin87 <[email protected]>
2026-09-30 09:14:55 +00:00
bb5b39f42e fix: stop ConditionalRouter and BranchJoiner from_dict from mutating the caller's data (#12935)
Co-authored-by: David S. Batista <[email protected]>
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
2026-09-29 10:47:50 +00:00
Harsh Kashyap 1ee5dc721c fix: treat None group values as missing in MetaFieldGroupingRanker (#12892) 2026-09-29 12:35:33 +02:00
5909e70c53 fix: skip BM25L/BM25Plus lower bound for query terms missing from a document (#12895)
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
2026-09-29 12:05:19 +02:00
Stefano Fiorucci bf5ec569eb test: make test_link_content_fetcher_multiple_html_streams more robust (#12927) 2026-09-29 11:41:48 +02:00
Ambuj UpadhyayandClaude Opus 5.5 a4d46c0380 docs: exit_reason's tool result is not always last_message (#12891)
Co-authored-by: Claude Opus 5.5 <[email protected]>
2026-09-29 07:57:02 +02:00
Christopher Krahandanakin87 11a0dd260a fix: keep an XLSX cell from breaking the markdown table (#12972)
Co-authored-by: anakin87 <[email protected]>
2026-09-28 16:09:01 +00:00
IronMurphyandanakin87 c2e809e1a2 fix(MetaFieldRanker): apply missing_meta when no document has the field (#12963)
Co-authored-by: anakin87 <[email protected]>
2026-09-28 13:09:19 +00:00
Chaimaa Abi 436fc1c7c8 fix: handle scalar content values in JSONConverter (#12983) 2026-09-28 13:36:58 +02:00
Julian RischandClaude Opus 5.5 43b12cce8e perf: query the Document Store once in SentenceWindowRetriever (#12997)
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
2026-09-28 13:21:43 +02:00
NanHan Qingandanakin87 693ec05945 fix: release the token counter when CompactionHook is closed (#12985)
Co-authored-by: anakin87 <[email protected]>
2026-09-28 09:46:17 +00:00
0edc59c2d4 fix: ignore dimension for custom mock embeddings (#12980)
Co-authored-by: carey-bk <[email protected]>
Co-authored-by: anakin87 <[email protected]>
2026-09-28 09:22:58 +00:00
JayanthandClaude Opus 5.5 3153560b06 fix: keep hyperlink addresses in DOCXToDocument table cells (#12988)
Signed-off-by: Jayanth <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
2026-09-28 10:24:50 +02:00
54f7e25ac6 fix: accept dict tool arguments in ChatMessage.from_openai_dict_format (#12790)
Co-authored-by: Sebastian Husch Lee <[email protected]>
Co-authored-by: Sebastian Husch Lee <[email protected]>
2026-09-28 09:05:55 +02:00
Rain. 6704f3903c fix: default BM25 tokenizer splits CJK characters for bare-term retrieval (#12837) 2026-09-28 08:37:37 +02:00
Eddy ZHANGandbogdankostic bf679e3979 test: enable mypy typing checks for the whole test folder (#12930)
Co-authored-by: bogdankostic <[email protected]>
2026-09-25 11:38:35 +02:00
Madan Kumarandbogdankostic 3d159a47cd fix: keep inter-sentence whitespace when a sentence ends with a closing quote (#12431)
Co-authored-by: bogdankostic <[email protected]>
2026-09-25 10:16:01 +02:00
PeterSmith0127-lcmandDavid S. Batista b3d1fbdca2 fix: validate top_k in MultiRetriever (#12719)
Co-authored-by: David S. Batista <[email protected]>
2026-09-24 16:51:43 +00:00
Lucienandbogdankostic 9a0af6afe1 fix: accept dictionary-valued flat pipeline inputs (#12884)
Co-authored-by: bogdankostic <[email protected]>
2026-09-24 15:32:02 +00:00
d8dd4778be fix: raise TypeError for non-string TextCleaner inputs (#12792)
Co-authored-by: Cursor <[email protected]>
Co-authored-by: anakin87 <[email protected]>
2026-09-24 14:47:06 +00:00
Serhii Zghama 1ca716f480 fix: don't treat a generator's own "error" field as failure in LLMMetadataExtractor (#12852) 2026-09-24 16:26:59 +02:00
847ec4dd60 fix: do not mutate caller-provided client_kwargs in LinkContentFetcher (#12829)
Signed-off-by: Ewuji John <[email protected]>
Co-authored-by: Julian Risch <[email protected]>
Co-authored-by: Claude Opus 5.5 (1M context) <[email protected]>
2026-09-24 12:44:38 +00:00
Grapette.L cba3db2f39 fix(core): compare Ellipsis callable parameters by return type (#12905) 2026-09-24 13:31:25 +02:00