Wang Qi
ae9f1e7c5b
Fix reasoning not working in multiple model chat ( #18457 )
2026-08-19 09:32:05 +08:00
Jack
b6b6c4de97
fix(deepdoc): score table orientation by per-line detect+recognize ( #18452 )
2026-08-18 22:48:39 +08:00
Jack
b78b559ac5
Remove noocr/SkipOCR dead code path and text-layer-only parity test ( #18443 )
2026-08-18 22:47:31 +08:00
ming1523
6756b48f10
test: add text&code parser-to-chunker parity golden ( #18311 )
2026-08-18 22:45:57 +08:00
Sevenzuo
f1e6b22c4b
fix: honor minimum_should_match in GaussDB search ( #18456 )
...
### Summary
GaussDB DocEngine could return no chunks for conversational queries even
when relevant content was available. `Dealer.search()` supplies
`minimum_should_match` (30%, then 10% on retry), but the GaussDB adapter
discarded it and built a single `plainto_tsquery` from every token. This
effectively required all conversational filler terms to match.
2026-08-18 21:05:49 +08:00
Jack
ebd6be09c7
feat(title-chunker): configurable token-count cap with sentence-boundary re-split ( #18455 )
...
## Summary
The TitleChunker (both `hierarchy` and `group` methods) had no
token-size ceiling, so a long section without sub-headings became one
giant chunk. This adds a configurable `chunk_token_cap` that guarantees
every text chunk stays within a token budget.
2026-08-18 21:03:51 +08:00
Harsh Kashyap
b3d9e02d03
fix(model_meta): Base._get_api_key handles JSON-dict api_key for the model-list verify path ( #18386 )
2026-08-18 20:16:19 +08:00
Harsh Kashyap
5d164085f5
fix(model_meta): LocalAI._get_api_key handles JSON-dict api_key for the verify path ( #18314 )
...
## Summary
Closes #17757 .
2026-08-18 20:15:18 +08:00
qinling0210
2aaa6baf0c
fix(agentic-rag): raise max_parallel_agents for high/ultra to 4, use web search after locate fails repeatedly ( #18430 )
...
### Summary
1. It changes the fallback semantics of the locate phase. When no chunks
are found, the system stays in locate. If the same claim has two
consecutive locate rounds with neither evidence chunks nor newly routed
document scope, web_search is admitted to the candidate tool set on the
next locate round as an external fallback.
2. It makes locate_empty_streak claim-scoped instead of shared in the
global context. This prevents one claim’s empty locate rounds from
affecting sibling claims running in parallel.
3. On the config side, it only raises max_parallel_agents for high /
ultra to 4, without changing max_agent_cycles. This increases parallel
claim execution without deepening per-claim search.
2026-08-18 19:40:07 +08:00
Wang Qi
858f882d21
Fix task time is incorrect ( #18453 )
2026-08-18 19:29:25 +08:00
Wang Qi
b5ad30ed43
Fix pdf cancel at task level not document level, multiple record logs ( #18435 )
2026-08-18 19:29:18 +08:00
Lynn
eb4e1aa2a4
Fix: set enable_thinking to True for qwen3.8-2.4t-a95b ( #18442 )
2026-08-18 19:27:31 +08:00
balibabu
455d34074e
Fix: Initiate a request to check for file changes two seconds after compilation completes. ( #18440 )
2026-08-18 19:27:22 +08:00
Jin Hai
7d4e40ce73
CLI: fix check license ( #18451 )
...
### Summary
Fix license valid can't display correctly.
Signed-off-by: Jin Hai <haijin.chn@gmail.com >
2026-08-18 19:20:42 +08:00
mkaaad
a4ec4f84cf
feat(go-syncer): add GitLab datasource connector with checkpoint resume ( #18425 )
2026-08-18 19:10:51 +08:00
Haruko386
2c8e73ab08
feat[Syncer]: implement Google Cloud Storage data source ( #18420 )
...
### Summary
As title
2026-08-18 18:19:31 +08:00
Haruko386
ab4adeecd2
feat[syncer]: implement notion data source ( #18423 )
...
### Summary
As title
2026-08-18 18:19:02 +08:00
chanx
109cff16c1
fix(web): strip think sections from floating chat widget copy ( #18436 )
2026-08-18 18:00:23 +08:00
euvre
84e4efaccc
fix(web): show introduction for built-in pipelines audio/email/knowledge compiler ( #18411 )
2026-08-18 18:00:06 +08:00
qinling0210
486065d878
Fix: use Infinity connPool.get_conn() in metadata filter push-down ( #18438 )
2026-08-18 17:37:28 +08:00
Wang Qi
b75509c8a6
Fix multiple model chat always use the chat model, not using the passing model ( #18439 )
2026-08-18 17:30:47 +08:00
Jack
c370f629dc
fix(deepdoc): align Go table header detection with Python (parity #4 ) ( #18404 )
2026-08-18 17:24:19 +08:00
Haruko386
d99f461790
document[syncer]: add development document for syncer ( #18431 )
...
### Summary
As title
2026-08-18 17:20:34 +08:00
peewee92
89f3611248
fix(agent): make tag metadata searchable and sortable in agent list ( #18313 )
2026-08-18 17:17:03 +08:00
Lynn
b2baddfe77
Fix: verification error msg ( #18437 )
...
### Summary
As title.
2026-08-18 17:14:52 +08:00
ump45nose
15aad78518
fix: isolate extract html request headers ( #18322 )
2026-08-18 16:49:17 +08:00
balibabu
0158a6bae8
Fix: Edit wiki-style text ( #18419 )
2026-08-18 16:26:38 +08:00
Jack
06463f6a35
fix(deepdoc): preserve cross-page continuation rows on column-count mismatch ( #18416 )
2026-08-18 16:16:02 +08:00
Jack
5aeba40d99
chore(deepdoc): remove dead standalone DLA client, converge on inference.Client ( #18384 )
2026-08-18 16:14:37 +08:00
chanx
c05fe37563
fix(web): surface the validation error when a metadata value save is blocked ( #18418 )
2026-08-18 16:08:48 +08:00
balibabu
cd1bb572b6
Fix: Incorrect translation of "search" on the chunk page. ( #18427 )
2026-08-18 16:08:26 +08:00
balibabu
cd8b336022
Feat: Remove pageindex from the dataset configuration page. ( #18429 )
2026-08-18 16:08:00 +08:00
Wang Qi
2a132d281d
Fix parent chunk with extra newline ( #18426 )
2026-08-18 16:06:30 +08:00
Try everything
4a3b20a889
fix(web): prevent docx preview zoom from skipping a level ( #17323 )
...
### Summary
Fixes a zoom bug in the DOCX preview: the first "zoom in" or "zoom out"
click
skips a level whenever the current zoom is a fit-to-width value.
2026-08-18 15:54:51 +08:00
Muhammad Furqan
c3153ba294
fix(agent/tools): port DeepL to ToolBase so it works as an Agent tool ( #18395 )
...
### Summary
Closes #18394 .
2026-08-18 15:36:11 +08:00
Jack
fb788ee7dc
fix(deepdoc): assign each PDF box to a single table cell matching Python ( #18410 )
2026-08-18 15:13:41 +08:00
mkaaad
94a93748db
Feat: connector go webdav ( #18379 )
...
as title
2026-08-18 15:01:53 +08:00
Wang Qi
86750bd24e
Fix: some pdf with chinese name cannot preview ( #18422 )
2026-08-18 14:54:10 +08:00
Haruko386
8583bc3938
fix(web): remove duplicate tags for locales files ( #18421 )
2026-08-18 14:26:42 +08:00
Lynn
c642e01f4b
Fix: set database version to v0.27.0 ( #18424 )
...
### Summary
As title.
2026-08-18 14:25:01 +08:00
Jack
fa386ddf10
fix(deepdoc): drop per-cell OCR in Go table parser to align with Python ( #18405 )
2026-08-18 13:42:16 +08:00
Wang Qi
75f737de67
Refactor: make page number readable ( #18409 )
2026-08-18 13:27:04 +08:00
Jack
97db1b4c38
chore(table): remove stale manual parity-issues test ( #18403 )
2026-08-18 13:18:57 +08:00
Jack
3a7e275155
fix(deepdoc): add orphan-row cleanup to Go table construction (parity with Python) ( #18402 )
2026-08-18 13:18:25 +08:00
Jack
d3ff25763c
fix(deepdoc): align table orientation scoring with Python recognition confidence ( #18401 )
2026-08-18 13:18:09 +08:00
var4dev
63a454084e
fix(agent): close MCP sessions wrapped in MCPToolBinding ( #18399 )
2026-08-18 13:15:16 +08:00
Jack
5ec11f7b07
fix(pdf): align TSR crop margin with Python (fixed 30px, not 3%) ( #18387 )
2026-08-18 13:11:41 +08:00
Lynn
344bd4d57a
Fix: correct chunk order for non‑PDF documents ( #18412 )
2026-08-18 13:06:13 +08:00
Lynn
2368c54689
Fix: handle title and truncate document preview by sentence ( #18407 )
2026-08-18 11:54:55 +08:00
jay77721
c71991bb7a
feat(ingestion,web): modularize extractor configuration with sub-tabs, independent prompts, and metadata integration ( #18383 )
...
This PR modularizes the **Extractor** component configuration with dedicated feature subtabs, adds independent system prompt configuration, fixes multi-node execution determinism and parameter persistence across save and page refresh, and ensures backward compatibility with legacy flat fields.
2026-08-18 11:38:52 +08:00
