Commit Graph

74 Commits (4d4acec418b5fa67d457d1bf85846476d1e92e73)

Author SHA1 Message Date
lyingbug 8fae51c640
style(frontend): polish knowledge base settings hints and sub-sections (#4193)
- Replace heavy warning alerts for locked states (embedding model, indexing
  strategy, storage engine) with subtle lock-icon hints
- Redesign the AI-generated description card with empty state and footer actions
- Replace the graph-database-disabled alert with an inline empty-state card
- Drop thick left-border panels in advanced and chunking settings in favor of
  rounded secondary containers; widen descriptions for switch rows
2026-10-12 18:04:35 +08:00
lyingbug 0c5d28abb7
feat: index and recall images with multimodal embedding models (#3683)
* feat(models): expose embedding input modalities

Whether an embedding model maps images into the same space as text is
what will decide if a knowledge base can store image vectors. The
catalog already had the field, but most multimodal embedding models
were missing it and embedding rows never reported it.

- Catalog: declare input modalities for the Doubao and Tongyi vision
  embeddings, the Aliyun multimodal / qwen-vl embedding families and
  jina-embeddings-v4, as listed in each vendor's documentation. The
  migration baseline changes for exactly these nine entries.
- Model responses resolve capabilities for embedding rows too, so
  capabilities.input reflects the catalog merged with spec.input.
- Model editor: an "Image input" switch for embedding models. It follows
  the catalog and writes spec.input only when the row differs from it;
  picking another catalog model or switching vendor drops the
  declaration. Model cards show the image badge for such models.

* feat(embedding): embed images with multimodal embedding models

Adds the image side of an embedder, for models that map images into the
same space as their text vectors so a text query can find an image.

- api: EmbedImage (bytes + MIME type, sent inline as base64 because the
  object store is usually unreachable from the vendor) and an
  ImageEmbedder implemented by all four protocols, each in the image
  shape its reference documents: Ark image_url, DashScope contents[].image,
  Gemini inlineData parts, and on the OpenAI shape an object keyed by the
  vendor's image_field (Jina: "image"). A plain OpenAI-compatible
  endpoint names no image field and takes no images.
- Settings: image_field, max_image_batch_size (default one per request),
  max_image_bytes and image_mime_types, filled from vendor docs for the
  DashScope multimodal models, gemini-embedding-2 and Jina.
- embedding: an optional ImageEmbedder interface reached through
  AsImageEmbedder, so existing Embedder implementations and fakes are
  untouched. An embedder takes images only when the catalog (or the
  row's spec.input) lists image AND the endpoint can carry one. Images
  are checked against the documented limits before any request, then
  split at the documented images per request. The concurrency, debug
  and Langfuse decorators forward it; logs and traces carry the MIME
  type and size, never the bytes.

The migration baseline changes only for the entries whose resolved
embedding settings now carry image limits.

* feat(knowledge): index images by their own vector and recall them

When the knowledge base's embedding model embeds images, each image now
gets a vector of its own next to its caption and OCR chunks, so a query
can find an image by what it shows even where the caption left it out.

Ingestion
- After the caption/OCR chunks are indexed, the multimodal task embeds
  the image (embedding.PrepareImage converts and shrinks it to the
  vendor's documented limits) and stores it as a new image_vector chunk:
  Content is the caption (OCR text when there is none) for reranking and
  the answer context; its index row carries the image's vector with
  SourceType ImageSourceType. Scanned PDF pages are skipped, their OCR
  text serves better. A failure here is logged and never fails the task.
- The image vector gets its own chunk rather than a second index row on
  the caption chunk: OpenSearch and Weaviate key documents by ChunkID,
  so a second row would overwrite the caption's text vector, and copy,
  move and enable/disable all follow chunk IDs.
- CompositeRetrieveEngine.BatchIndexVectors writes precomputed vectors
  through each engine's normal BatchIndex, keeping row order (the
  composite's de-duplication does not).
- Paths that rebuild vectors from text leave image_vector alone:
  syncChunkIndex only syncs its enabled flag, updateChunkVector skips
  it, and a caption edit updates its Content only.
- The multimodal task's caption/OCR index rows now set IsEnabled; only
  backends with a column default were storing them as enabled.

Retrieval
- A text query scores lower against an image than against text, so when
  the model embeds images the document vector search is widened by half
  (capped at 500) and run at min(VectorThreshold, 0.1); each store
  group's hits are then held to their own threshold before score
  normalization: text at VectorThreshold, images at min(VectorThreshold,
  0.1). Keyword hits on image rows are dropped, their Content being the
  already-indexed caption.
- image_vector hits are searchable and merge like caption hits: parent
  context is attached, and the caption is not repeated from ImageInfo.

Adds golang.org/x/image for scaling and WebP/BMP decoding.

* fix(models): report no thinking levels for embedding, rerank and ASR models

Embedding rows now carry capabilities, and the end-to-end run showed them
listing off/auto/minimal/low/medium/high: these references resolve no chat
protocol, and an empty level map reads as "every level supported". A model
without a chat protocol does not think, so its capabilities say so.

* fix(compaction): never read a cancelled turn's stream as a summary

streamSummary selects on the turn's context and the provider stream at
once. When the turn is already cancelled both are ready (providers close
the stream once their context ends), and select picks at random, so a
stopped turn could come back as a completed, partial summary with a nil
error. TestSummarizationIsNotRetriedOnceTheTurnIsCancelled caught it
intermittently in CI and reproduces under -race with several -cpu values.

Check the context before each receive and again when the stream closes.

* feat(knowledge): put image vectors behind a knowledge-base switch, off by default

Image vector indexing and recall used to turn on whenever the knowledge
base's embedding model declared image input. A model that takes images is
often chosen for text alone, so an upgrade would have added an embedding
call per image, widened vector recall 1.5x and lowered the image threshold
to 0.1 for those knowledge bases without anyone asking.

ImageProcessingConfig gains image_vector_enabled (default false; stored
JSON without the key reads as off, so existing knowledge bases stay off
with no migration). Both sides now need the switch and the model:

- indexImageVector returns before touching the model unless the KB opted
  in (KnowledgeBase.IsImageVectorEnabled, which also needs vector indexing).
- applyImageRecall resolves the embedding model only when some KB in the
  search opted in, and records the opted-in KBs per store group
  (storeGroup.ImageKBIDs). Only those groups widen the pool and lower the
  threshold; image hits of any other KB, including ones indexed before the
  switch was turned off, are dropped.

The KB editor shows the switch in the image processing section, with a
warning when the selected embedding model does not declare image input;
docs describe the switch, its cost and what turning it off does.

* fix(knowledge): leave image vectors behind when cloning into a KB that is off

CloneChunk copied every chunk of the source document and then CopyIndices
for all of them, so copying a document from a knowledge base with image
vectors on into one with the switch off (same model and store) still
created image_vector chunks and their index rows, bypassing the switch
that indexImageVector honours.

Image vector chunks are now dropped from the copy when the target has not
opted in. Every engine's CopyIndices copies only the rows of chunks in the
mapping, so their index rows stay behind too.

* fix(search): refill the vector pool past stale image vectors

Image hits of a knowledge base whose switch is off were dropped only after
the engine had ranked and cut its results at TopK. Image vectors indexed
before the switch was turned off could fill that pool and push out valid
text hits: with MatchCount 5 the pool is 50, and 50 stale image rows
ranked above the text left the search empty.

No engine can filter rows by source type (Postgres and SQLite do not even
take ExcludeChunkIDs), so the exclusion cannot move into the query. When
the document vector pool comes back full and holds stale image rows, that
search alone runs again on a pool twice as large, until it holds TopK rows
that are not stale, the index runs out, or the pool reaches
maxRetrievalPoolSize (at most four extra calls from a pool of 50). The pool
is then cut back to TopK such rows, matching an engine-side exclusion
unless more than maxRetrievalPoolSize-TopK stale rows rank above the text.

* feat(i18n): translate the image vector switch and image input keys into zh-TW

The zh-TW locale arrived from main after this branch added five keys: the
knowledge base's image vector switch (label, description, unsupported-model
warning) and the embedding model's image input flag (label, description).

* fix(knowledge): leave image vectors behind when moving into a KB that is off

A reuse_vectors move relocates a document's chunks and index rows in place:
MoveKnowledgeIndices rebinds the rows by document ID and
MoveChunksByKnowledgeID rebinds the chunks. Moving a document from a
knowledge base with image vectors on into one with the switch off therefore
carried its image_vector chunks and their vectors along, the same bypass the
clone path had.

When the target has not opted in, the image_vector chunks are now deleted
during the move, their index rows first and then the chunks, so a retry
finds the chunks again rather than leaving orphaned rows. A target that
opted in still receives them unchanged.

* fix(retriever): keep image vectors out of keyword-only engines

BatchIndexVectors handed image rows to every engine of the composite,
including one that serves keyword retrieval alone (Elasticsearch v7 next to
a vector store). There the row holds nothing but the image's caption, which
the caption's own chunk already indexes, and retrieval discards every
keyword hit on it.

Image rows now go only to engines that serve vector retrieval. An engine
that serves both keeps one row for both kinds, so its keyword search still
meets them; the search side handles that.

* fix(search): refill the keyword pool past image rows

Every engine that serves both kinds of retrieval stores one row per index
entry and runs its keyword search over that row's content, so an image
vector row, whose content is its caption, is a keyword hit wherever the
caption's own chunk is. filterImageHits drops all of them, but only after
the engine has cut its ranking at TopK, so they crowded out text and caption
hits; a keyword ranking that put the image rows first came back empty.

The keyword pool is now refilled the way the vector pool is refilled past
stale image vectors: when it came back full and held image rows, the
keyword search runs again on a pool twice as large, up to
maxRetrievalPoolSize, and is cut back to TopK hits that are kept.

Leaving the rows out of the keyword index instead would mean writing them
without searchable content (Milvus and Tencent VectorDB derive a BM25
sparse vector from it, and how they take an empty one is unverified) or a
source-type filter in every engine's keyword query (ten engines, and
filterable fields that existing collections lack).

* revert: let image vectors move with the document into a KB that is off

This reverts commit e259977fe915f7d0bc30c4b02bc7af04da1c8dd4.

Deleting the image vector index rows by chunk ID before the move is not
reliable: the Weaviate and Elasticsearch v8 delete wrappers ignore partial
failures, and OpenSearch rejects more than 1000 IDs with ErrBatchTooLarge.
A partial delete followed by the chunk delete leaves index rows whose chunks
are gone, which nothing ever cleans up.

Without it, the image_vector chunks and their index rows move together
with the document. While the target's switch is off they are filtered at
query time (the pool refill keeps them from crowding out text), they are
removed with the document on delete, reparse or knowledge base delete, both
chunks and index rows by knowledge ID, and they become active again if the
target turns the switch on. The clone path keeps dropping them, since there
the rows are copied rather than moved.

* fix(search): keep the image row refill from losing hits it already had

refillPool replaced a pool with the answer to each larger search. A refill
that answered only in part (RetrieveResult.Error, swallowed into a warning
by retainPartialResults, as when some collections of the store did not
answer) could hold fewer text rows than the complete smaller pool, or none:
40 stale image rows above 10 text came back from a partial refill as the
40 image rows alone, and the search lost all 10 text hits. A refill that
failed outright failed the whole store group, discarding a first round that
had succeeded.

A refill is now best effort. A failed or incomplete refill search stops
the refill and leaves the last complete pool in place, cut back to TopK
kept rows as before, so a refill can no longer leave a search worse off
than it was without it. Merging the partial answer into the pool was not
done: its rows would need deduplicating and its ranking is cut at a
different TopK over a different set of collections, while keeping the
previous pool is exact.

* docs(knowledge): say what happens to image vectors in a KB that is off and what the refill costs

The image vector section now says that with the switch off, image vectors
that were indexed before it was turned off, or that came along when the
document was moved in from a KB that has it on, are kept but filtered at
query time, are removed when the document is deleted or reparsed or the
KB is deleted, and are recalled again once the switch is turned on; a copy
into such a KB leaves them behind.

It also states the cost of keeping dropped image rows from crowding text
out of the pools: the vector and keyword pools are searched again with
TopK doubled up to the 500 candidate cap, which adds queries for store
groups that hold such rows, and text can still be missed when more than
about 500 candidates are image rows. A failed or partial refill keeps the
previous complete pool.

* fix(search): log unexpected refill result sets, and describe the refill stop condition precisely
2026-10-08 18:19:53 +08:00
tayadwx 68fe9599cd
feat(gallery): add an image gallery tab with schema-driven filters (#3694)
* feat(image): observe image attributes and let the policy decide OCR

Describe each image with the VLM as before, but ask it to also observe two
attributes about the image in the same round — how much body text it carries
(contain.text) and whether it presents data visually (contain.data_visual) —
then let a small pure function decide whether that image is worth a second,
separate OCR round.

Per-class OCR was the first shape of this idea; a class (chart / photo / logo)
is only an indirect proxy for 'is there text worth reading here', and the class
table has to grow whenever a new kind of image shows up. The decision actually
needs the observed attributes, so it reads them directly, and the attribute set
is a backend registry row rather than a new class.

The observation and the caption share one VLM call; OCR stays a separate call
(prompt isolation, see #3291). An attribute the model does not answer for is
left absent rather than filled with a conservative default, so the trace never
claims an observation that did not happen, and an incomplete observation falls
back to the on_unobserved policy (OCR still runs). The registry also carries the
panel's display text, so a new attribute needs no frontend change.

Off by default: image_attrs_enabled=false keeps today's behaviour (one caption
and one OCR per image). KB config: image_processing_config.image_attrs_enabled /
image_actions; per-upload override via process_config. Registry:
GET /api/v1/image-attrs/schema.

* feat(gallery): add image gallery tab with schema-driven filters

Adds a 4th knowledge-base tab parallel to Documents/Wiki/Graph that
browses every image asset in a KB.

Backend:
- GET /api/v1/knowledge-bases/:id/images lists image assets projected
  from chunks.image_info, with keyword (caption/OCR), multi-value
  attribute (attr_<name>=<value>), is_enabled and sort filters applied
  in memory, plus pagination.
- Resolves the source knowledge item title (SourceName) via a batched
  GetKnowledgeBatch so the viewer can show a human-readable source.

Frontend:
- ImageGallery.vue: thumbnail grid + search/sort + schema-driven filter
  panel (from GET /image-attrs/schema) + single-image viewer with
  prev/next, attributes/caption/OCR/source.
- Wired as the gallery tab in KnowledgeBase.vue; i18n keys in all 5 locales.

* feat(gallery): drive filters/search/sort from a configurable attribute contract

The gallery UI is no longer hardcoded: it renders from a self-describing
contract served by GET /knowledge-bases/:id/gallery-config (live attribute
sources, every attribute's definition with usage flags merged across the
config tiers, and the caller's search activation state).

- Attribute sources form an open registry (types.RegisterGalleryAttrSource);
  a builtin source models caption/ocr_text/timestamps/enabled so the gallery
  has zero special cases.
- Usage resolution: user > KB > system > source, field by field; the system
  tier lives in system_settings (gallery.policy, JSON, no rebuild), the KB
  tier slot is wired in the engine but not stored yet.
- Search activation is the user's own record: mode (all|custom) plus
  per-field on/off in users.preferences.gallery; source declarations never
  set defaults (unrecorded = off in custom mode).
- image_info is read through a gallery-local struct, removing the compile
  dependency on the attribute-pipeline types; the gallery builds and runs
  with or without that feature deployed.
- ListImages validates search_in / sort_by / attr_filters against the
  contract; ineligible fields are dropped, not served.

* fix(frontend): inline the shared graph-drawer stylesheet (@import instead of <style src>)

plugin-vue 6.0.6 keys unscoped src-style descriptors by the imported file
path, so AgentStreamDisplay.vue and WikiBrowser.vue — both importing
components/css/wiki-graph-drawer.less — overwrite each other during the
build and the second one crashes with "reading 'scoped'". Adding the
gallery changed the module order enough to make the collision hit every
time. Inlining the same stylesheet keeps the emitted CSS identical and
removes the shared descriptor key.

* feat(gallery): render attributes from short labels with descriptions in tooltips

A source could only say one thing about a value, so the filter panel had to put
a full sentence inside a checkbox. Attributes and their values now carry two
pieces of text: a short label for the space-constrained controls (checkbox,
search field, sort option) and a description for wherever there is room
(tooltip, detail panel).

The contract's value list becomes an array of {value, label, description}, and
the display helpers resolve each field separately: the gallery stamps the short
name on a control and moves the sentence into a tooltip. Translation overlays
are looked up per field, so a source that ships no gallery-specific text still
picks up the attribute pipeline's own wording.

* fix(gallery): let locale text win over the contract's wording

The filter panel read the backend's own label before asking the locale,
so a translated value never got a chance: the contract ships its own
default-language wording, and it shadowed the overlay for every locale.
Built keys were also off by a segment — imageAttr.<name> has to be
imageAttr.<name>.label, and the gallery's own keys live under
knowledgeEditor.wikiBrowser.gallery.attr — so builtin and system
attributes fell through to English on a Chinese UI.

* feat(gallery): top bar layout, pinned panels and per-value verdicts

The toolbar moves onto its own row above the grid, so the controls no
longer eat into the space the images are there for: search box, the two
scope switches, sort field and direction, and the result count all sit in
one flex row. The settings behind them are no longer inline either — each
opens as a popover under its control and can be pinned to a right-hand
rail, where the two panels take turns as tabs instead of both claiming
width at once.

Filtering gains a third state per attribute value. Where a value could
only be picked or not picked, it can now be "off" (hide those images),
"on" (show them whatever else says) or neutral (leave them alone), which
is what lets one attribute's rules talk about an image another attribute
tried to hide without the pair of them cancelling each other out: a
forced display outranks a forced hide, and anything no rule speaks about
stays visible.

The switch that used to only exist inside the panel now has a toolbar
home as well. "筛选:全显示" genuinely drops the attribute constraint
instead of imposing it against a panel the user cannot see.

Filtering used to empty the list the moment it was switched on. Any
image whose attribute was never observed — every image ingested before
the attribute pipeline existed — read as a mismatch for every filter,
which is not narrowing a list but losing it. Absence now stays silence:
an attribute no image carries places no constraint on it, until the
pipeline catches up and fills the value in.

Co-authored-by: WorkBuddy

* feat(gallery): expose the image attribute pipeline as a gallery source

The gallery's contract resolved to a single source, so the filter panel
could only offer "Enabled" while the attributes the observe-and-describe
pipeline fills in — contain.text, contain.data_visual — never reached the
UI. The registry is a gallery attribute source like any other: every
attribute it declares is offered to the gallery, filterable by default,
with the config tiers able to override the usage flags at runtime.

The registration lives in the attribute pipeline's own commit; this branch
carried an earlier, squapshot of that work which predated the split of an
attribute value into a short label and a description, so the registry's
value shape and this file have been brought in line with it. Both PRs now
hold byte-identical copies of the source hook, which is what lets either
one land first and the other merge cleanly afterwards.

Co-authored-by: WorkBuddy

* fix(gallery): viewer race, pin icon, panel anchoring and scope wording

Opening the viewer rendered the img element with the previous or empty
source while the proxy fetch was still producing a blob, and the error
that empty source fired locked the viewer in its failure state even
after the real image arrived. The resolved source now lands before the
flag is consulted, and a token keeps a slow fetch of one image from
overwriting a faster one that navigated past it.

The pin button drew nothing: TDesign has no "pushpin" glyph, so the
button rendered as an empty shell. It now uses pin / pin-filled, keeps
its place beside the tab bar, and a pinned rail assigns its columns
explicitly — the panel precedes the main section in the DOM, so leaving
the grid to source order handed the wide column to the panel and
squeezed the images into the rail.

The floating panel no longer appears at a hardcoded offset: it anchors
just below the arrow that summoned it, the arrow flips up while its
panel is out, a second press retracts, and any click elsewhere in the
toolbar closes it. When a rail is pinned the arrows turn to the
requested tab instead of spawning a second surface.

The scope words (全部 / 全显示 / 自定义) become plain status text rather
than a second dropdown — they name the current mode, and only the arrow
beside them opens the panel that changes it.

Co-authored-by: WorkBuddy

* fix(web): align the OCR-on-unobserved switch with the other setting rows

The switch sat bare inside its setting-row while every other switch on
the page is wrapped in setting-control, which is what reserves the right
column and centers the control vertically. With the row laid out from
the top, the bare switch hugged the row's upper edge and crowded the
description text — visibly out of place next to its neighbors. Wrapping
it in setting-control gives it the same right-aligned, centered slot as
the rest.

Co-authored-by: WorkBuddy

* fix(web): move the OCR-on-unobserved row out of the vertical schema row

The row lived inside setting-row-vertical, whose descendant rule stretches
every setting-info and setting-control beneath it to 100% width. Two
full-width flex children side by side cannot fit, so the control overflowed
the container and the switch landed past the right edge — visibly further
out than the switch on the row above. The row and its note now sit beside
the vertical block instead of inside it, under their own condition, so the
switch shares the ordinary row layout and its right edge lines up with the
"image attribute observation" switch above.

Co-authored-by: WorkBuddy

* fix(gallery): applying a verdict now activates the filter scope

Setting an off/on verdict while the panel's master switch was off
silently ignored it: buildParams only attaches attr_rules in custom
scope, and setVerdict never flipped the scope. A verdict the user just
expressed now switches the scope to custom; clearing the last meaningful
verdict returns to all, since custom with no rules constrains nothing.

* fix(gallery): search checkboxes stay visible and idle panels read as dimmed

The search panel hid its field checkboxes while the master switch was
on, so the only way to reach them was to flip the switch first. They now
sit next to the switch like the filter panel's verdicts do: dimmed in
all-field mode (where they carry no weight) yet clickable, and touching
one drops the mode to custom. The filter panel's verdict rows dim the
same way while the filter switch is off.

* fix(gallery): bulk verdict row, neutral default mark, panel polish

The clear-filters button never matched what the panel is for, so it is
now a bulk verdict row: one set of -/off/on buttons that applies the
chosen position to every value of every value-typed attribute, with the
default mark turned into a language-neutral dash. The panel head drops
its second divider (the tab bar already draws one) and the tabs follow
the toolbar's order: search left, filter right.

* fix(gallery): drop the all-fields caption and center the order arrow

The caption above the search checkboxes showed and hid with the master
switch, shoving the checkboxes down on every flip; it said nothing the
dimmed state doesn't, so it is gone. The order button's arrow is an
inline svg, which sits on the text baseline and reads as riding high;
it now centers against the label.

* fix(gallery): center the order arrow for real

vertical-align never applied: the icon is a flex item inside the
button's inline-flex text wrapper, and the wrapper's default stretch
pins the explicitly-sized 16px icon to the top of the 22px line box.
align-items: center on the wrapper is what actually centers it.

* fix(gallery): shorten panel tab labels in en and ru

The two-word labels overflowed the 300px panel and forced the tab bar
into scroll arrows; single-word labels fit in every locale that was
affected.

* docs: drop the image pipeline rewrite from this branch

The rewritten 'Go 侧接手后' paragraph belongs to the image attribute observation
work (#3635), which this branch carries as a prerequisite and which already
documents the caption_ocr / observation_driven split. Keeping a competing
rewrite of the same paragraph on the gallery branch makes it collide with
upstream, which extended that paragraph with the custom_instructions and inline
HTML table notes. Restoring the shared wording keeps the gallery branch free of
this file so upstream's edits merge cleanly.

* style: gofmt the gallery attribute types

Struct field tags in the resolved gallery config were one column narrow, so
gofmt rebroke the alignment block. Whitespace only.

* style(gallery): use design tokens in the gallery styles

The style guard ratchet fails on any new violation. ImageGallery.vue shipped
with 47 TDesign token fallbacks, 21 literal font sizes, 4 literal radii and a
0.15s transition; all of them now read the --app-* / --td-* tokens, which is
what the guard asks for and what dark mode depends on.

* lint(gallery): fix the golangci-lint findings in the gallery code

lll: the three over-long swagger @Param lines and one registry row were
shortened. ineffassign: ctx was assigned before the access check and then
reassigned, so the first value was dead; it is now declared once, after the
check. errcheck: the three unchecked c.Error calls in ListImages. revive: the
AttrSpec / AttrValue doc comment block was attached to the wrong type and split
into one block each. staticcheck S1016: the gallery registration converts the
registry value instead of restating its fields.

* lint(gallery): wrap the image query select and check the last c.Error

* style: reformat the wrapped image select in chunk.go

* fix(gallery): address review findings on ordering, KB switch and blob lifetime

- sortImageAssets: swap operands for desc instead of negating, keeping a
  strict weak ordering so ties page stably
- ImageGallery: reload on knowledgeBaseId change, drop stale list
  responses, fetch thumbnails in parallel with one cached blob per URL,
  revoke blobs on unmount, default sort to builtin:created_at
- KnowledgeBase: non-wiki KBs with a stale wiki/graph tab fall back to
  documents again
- handler: move ListImages godoc out of ListKnowledgeFolders' block and
  document attr_filters / attr_rules
- DecideOCR: a payload with no OCR conditions uses the built-in policy
  instead of silently skipping OCR

* docs(parsing): describe the caption_ocr and observation_driven image pipelines

Carried over from #3635, which this PR supersedes.

* fix(gallery): load local:// and other storage handles through the KB file proxy

The gallery's own scheme check only caught resource://, provider:// and
storage://, so local:// (the default storage), minio://, cos:// and s3://
images went straight into <img src> and never rendered. Resolve through
isProviderFileURL, the scheme list chat and wiki already use, and lock the
routing with a test.

* fix(kb-editor): keep API-customised OCR conditions on save and upload

The KB editor always wrote the registry default into image_actions.ocr.on,
so any unrelated edit replaced a knowledge base's custom OCR conditions
with the default table; the upload dialog did the same for the per-document
override. Keep a stored custom list and fall back to the registry default
only when there is none, and show the conditions actually in effect instead
of the default table in both panels.

* perf(gallery): de-duplicate, filter, sort and page images in SQL

Every gallery request used to read every image-bearing chunk of the KB,
OCR text included, decode it in Go and page in memory. A single query now
expands the image_info arrays, keeps one row per image URL, applies the
keyword / attribute / verdict / enabled filters and the sort, and returns
only the page with the filtered total.

The expansion carries narrow scalars only; filters and JSON sort keys are
evaluated on the winning entry of each image, and the page's JSON is read
back for its rows alone. Postgres runs the query with JIT off, since the
fixed 100-rows-per-call estimate of jsonb_array_elements otherwise prices
it ~100x high and compilation dominates.

On 20k images with 2 KB of OCR text each (40k chunks), a page drops from
~750 ms to ~100-200 ms on postgres and ~200 ms on sqlite; keyword search
over OCR text stays a full scan (~400-600 ms) until the text is indexed.
The filter semantics are pinned against both databases.

* perf(gallery): serve the gallery from a trigger-maintained chunk_images table

Reading images out of chunks.image_info meant parsing every image chunk of
the knowledge base on each request, so a page still cost time in
proportion to the KB. chunk_images (versioned/000113, sqlite/000032) keeps
one row per image_info entry with caption, OCR text and attributes as
columns; triggers on chunks keep it in step with every write path, and the
migration backfills existing chunks.

Copies of one image stay separate rows and the query lists the copy on the
most recently updated chunk through an indexed anti-join, so writes decide
nothing and concurrent writers need no locking. On postgres, caption and
OCR text get pg_trgm indexes when the extension is available.

On 20k images with 2 KB of OCR text each (40k chunks):
  postgres  first page 10 ms, page 500 41 ms, keyword 2 ms, filters 23 ms
  sqlite    first page  3 ms, page 500 39 ms, keyword 147 ms, filters 51 ms
down from ~100-600 ms. Writing those 40k image chunks costs 7.0 s instead
of 1.3 s on postgres (1.9 s instead of 1.0 s on sqlite), mostly the trigram
indexes; status and enable-flag updates patch rows in place.

---------

Co-authored-by: tayadwx <tayadwx@users.noreply.github.com>
Co-authored-by: wizardchen <wizardchen@tencent.com>
2026-09-25 12:33:59 +08:00
lyingbug b4aa5bbe8d
refactor(frontend): style consistency pass — shared shells, tokens, unified confirmations (#3376)
* refactor(frontend): batch 1 of style consistency cleanup

- theme.css: alias --td-font-family-mono to --app-font-family-mono so the
  29 monospace call sites follow the user's code-font setting
- replace 115 hardcoded rgba(7,192,95,x) brand tints with
  color-mix(in srgb, var(--td-brand-color) N%, transparent) so they track
  the dark-mode brand color
- strip 358 dead var(--td-*, fallback) fallbacks (tokens are guaranteed by
  theme.css); keep the 3 undefined tokens untouched
- menu.vue: drop 9 unreferenced style blocks; remove TencentSans font file
  and its global import
- AgentStreamDisplay: drop 6 unused keyframes; move the teleported wiki
  drawer styles into components/css/wiki-graph-drawer.less imported by
  both chat and WikiBrowser so styling no longer depends on load order
- creatChat: remove four !important fixed-width media queries; chat root no
  longer forces font-size: 20px
- list pages: create buttons use native theme="primary" (drops the purple
  gradient on AgentList); header-action-btn no longer needs !important
- add styleGuard.test.mjs ratchet test (brand rgba, legacy blue, token
  fallbacks, z-index !important, raster icons)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): unify destructive confirmations on useConfirmDelete

- useConfirmDelete now uses the danger dialog theme with a red confirm
  button; it is the single entry point for delete/leave confirmations
- AgentList, KnowledgeBaseList and OrganizationList drop their hand-rolled
  del-*-dialog (span-as-button, circle.png) in favour of useConfirmDelete;
  circle.png asset removed
- FAQ single-entry delete gains a confirmation (it deleted immediately);
  new knowledgeEditor.faq.confirmDelete key in all five locales
- VectorStore/StorageBackend delete dialogs and the generated-question
  popconfirm use the danger theme like the rest of the app
- style guard raster-icon baseline lowered to 10

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* feat(frontend): shared modal shell with Esc, overlay close and unsaved guard

Add composables/useModalShell and wire it into the four full-screen
settings-style modals (Settings, AgentEditorModal,
OrganizationSettingsModal, KnowledgeBaseEditorModal):

- Esc closes the modal everywhere (previously only Settings); it defers
  to any open TDesign dialog/drawer/popup and to the agent editor's own
  placeholder popups
- overlay click closes everywhere (Settings gains it)
- editors snapshot their form after load/save; closing with unsaved
  changes now asks "有未保存的更改 / 放弃更改 / 继续编辑" instead of
  silently discarding
- new common.unsavedChanges.* keys in all five locales

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* fix(frontend): mark instant-effect sections in the knowledge base editor

Share, data source and activity sections apply changes immediately, but
the editor footer still showed Cancel / Save, so users either re-saved
or assumed Cancel would undo. Those sections now show an "applies
immediately" note with a single Close button; other sections keep the
Cancel / Save footer. New knowledgeEditor.footer.instantEffect key in all
five locales.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): consistent success wording and confirm labels (zh-CN)

- toasts use the "已X" form everywhere: 25 zh-CN values such as
  保存成功 / 删除成功 / 智能体创建成功 become 已保存 / 已删除 / 已创建智能体;
  common.copySuccess now reads the same as common.copied
- confirm buttons in dialogs and popconfirms use the bare verb
  (删除 / 重置 / 移除 / 应用 ...) instead of 确认X; dialog titles keep 确认删除
- GeneralSettings no longer pops a bare "成功" toast after theme / font /
  font-size changes, which are self-evident and applied instantly

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): extract SettingsModalShell for the four settings-style modals

Settings, AgentEditorModal, OrganizationSettingsModal and
KnowledgeBaseEditorModal each carried a copy of the same overlay / panel /
sidebar-nav / footer markup and ~20 CSS blocks, already drifting
(1000/1080/1100px widths, three shadows, different nav paddings).

components/SettingsModalShell.vue now owns that chrome: one 1080x780
panel, grouped nav with badges, optional loading overlay, footer with a
note slot and an actions slot. Consumers keep only their content and
wire `@close` to their useModalShell.requestClose. Settings keeps its
sub-menu nav through the `nav` slot; OrganizationSettingsModal keeps its
image icon through the `nav-icon` slot.

Net: -1025 lines across the four modals.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): shared resource-card mixins and EmptyState for list pages

KnowledgeBaseList, AgentList and OrganizationList each re-declared the
same 136px card (border, radius, shadow, hover, header/title/description/
bottom, more-button reveal, feature badges) with small drifts, plus a
copy of the empty-state block using the upload illustration for every
kind of emptiness.

- components/css/resource-card.less: .resource-card(),
  .resource-card-grid() and .resource-feature-badge() mixins; the pages
  keep only their own variants (type gradients, favourite star,
  decoration). More-buttons stay visible on touch devices.
- components/EmptyState.vue: icon / title / description / actions; the
  11 empty states now use a matching TDesign icon instead of upload.svg
- badge accent colours move from hard-coded old-blue / material orange /
  pink / slate rgba to color-mix over theme tokens (new
  --app-accent-purple in theme.css); style guard legacy-blue baseline
  drops from 40 to 18

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): shared settings-section mixins for header, row, info, control

25 settings/system/integration pages each re-declared .section-header,
.setting-row, .setting-info and .setting-control with five different
header margins (18-32px), three description sizes and assorted paddings.
components/css/settings-section.less now provides
.settings-section-header() / .setting-row() / .setting-info() /
.setting-control() / .setting-save-state(); pages keep only their
deliberate deviations (e.g. Ollama's column control, TenantInfo's fixed
label column). PlatformAPIKeys' style block switches to less so it can
use the mixins. Net -570 lines.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): provider-card mixins, dark-safe vendor badges, one spin keyframe

- components/css/provider-card.less: .provider-card() /
  -interactive() / -add() / -badge() / -badge-color() / -body() /
  -header() / -title() / -subtitle() / -more() shared by the model,
  vector store, web search, storage, parser and sandbox settings cards;
  MCP and skill cards reuse the 26px badge variant
- vendor badge colours are now `.provider-card-badge-color(#hex)`: the
  foreground is mixed 75/25 with the text colour so it stays readable in
  dark mode, the background is the same colour at 12% alpha
- 18 copies of a rotate-360 keyframe (steer-spin, chat-header-spin,
  sandbox-terminal-spin, ...) collapse into a global `wk-spin` in
  theme.css; 23 files updated, one source test adjusted

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): use TDesign password toggle for API keys; align chat toolbar buttons

- ModelEditorDialog: the API key field is a plain t-input type="password"
  and relies on TDesign's built-in show/hide eye instead of a hand-rolled
  suffix toggle, matching every other secret field in the app
- chat answer toolbar buttons are 28px with 6px radius, the same size as
  the user-message actions and the input control bar

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* refactor(frontend): app design tokens, tokenized literals, global TDesign overrides

theme.css gains an app-level token set derived from the de-facto scale
already used in the codebase, so values do not change:
  --app-radius-{xs,sm,md,lg,xl,pill} (4/6/8/10/12/999px)
  --app-text-{2xs..4xl}              (10..24px)
  --app-space-{1..10}                (4px grid)
  --app-motion-{instant,fast,base,slow} (120/150/200/300ms)
  --z-{sticky,dropdown,modal,drawer,popover,toast}

A codemod mapped 675 border-radius, 1904 font-size and 551 transition
duration literals in 191 files onto those tokens (value-preserving; five
source-text tests now accept either form).

assets/theme/tdesign-overrides.less holds global-intent TDesign
overrides; its `.wk-popover` skin replaces six per-file copies of the
blurred popover panel (TenantMembers x2, UserProfile,
OrganizationSettingsModal, systemAdminDialog.less,
share-to-space-panel.overlay.less) without !important.

styleGuard ratchets now also cover radius / font-size / motion literals,
`transition: all`, total `!important` and z-index !important.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

* fix(frontend): keep symmetric side insets for the chat column on narrow viewports

.chat had padding 0 0 20px 20px so the scrollbar could sit on the content
edge, but that also left the message list and input box flush against the
right edge whenever the viewport was narrower than 960px + sidebar. The
inset now lives on .msg_list and .input-container (--chat-content-inset,
20px each side, max-width 960px + insets); the scroll container still
spans the full column so the scrollbar stays on the edge. Wide layouts are
unchanged (960px content, centered).

* chore(frontend): strip trailing whitespace flagged by pre-push

* test(frontend): accept tokenized font-size and shell-rendered loading overlay

Two source-text assertions only ran in CI (Node 24 discovers .test.ts,
local Node 20 does not):
- ChatQuestionMinimap: the 13px preview font is now var(--app-text-md)
- modelUsageDetails: the editors' loading overlay is rendered by
  SettingsModalShell via :loading instead of an inline v-if
Both regexes now accept either form.

---------

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
2026-09-17 23:11:08 +08:00
lyingbug 5574ebc011
feat(knowledge): derive knowledge base descriptions from document profiles (#3370)
* feat(knowledge): derive knowledge base descriptions from document profiles

Knowledge bases only had a hand-written description, which most users
leave empty, so agents bound to several knowledge bases had nothing to
route on and searched all of them. This adds a generated description
that is derived from per-document data instead of maintained by a model.

Document profile: the summary call now returns JSON with a 2-3 sentence
summary plus a gist, 3-5 topic keywords, a document type and one
typical question (knowledges.profile). Input is capped at 8k chars and
output at 1024 tokens; plain-text replies from custom templates still
work and simply carry no profile. The summary chunk embeds the gist and
short summary.

Knowledge base aggregate: a database-only fold over those rows (counts
per file type, tag, normalized topic and doc type; stratified samples of
titles and questions; a hash of the inputs). Deleting, moving or
re-parsing a document changes nothing but the rows, so the next rebuild
is exact without any model judgement.

Description text: one small model call over the aggregate yields a
gist, merged topics and typical questions (knowledge_bases.
generated_profile). The call is skipped when the aggregate hash is
unchanged; failures keep the previous text and record the error; an
empty knowledge base clears it. The manual description is never
touched, and the agent runtime context renders both so the model can
pass knowledge_base_ids instead of fanning out.

Refreshes are requested from summary task exit, post-process completion,
deletion, cross-KB moves and enabling the option, debounced per
knowledge base in 30 s windows via kb:profile on the summary queue.
POST /knowledge-bases/:id/profile/generate regenerates synchronously.
profile_config (enabled, model_id, custom_instructions) opts a document
knowledge base in; the settings dialog shows the generated card with
regenerate and adopt-as-description actions in five locales. Search
results now log hits per chunk type so the value of indexing summary
chunks can be measured.

Migrations 000101 (PostgreSQL) and 000021 (SQLite) add the columns; the
SQLite schema test expects the new version and columns.

* fix(knowledge): keep generated descriptions out of duplicates and refresh clones

DuplicateKnowledgeBase copies the source row through JSON, which carried
generated_profile onto a knowledge base that has no documents yet, so
an empty duplicate advertised the source's topics. The duplicate now
starts without a generated description and keeps profile_config so it
derives its own once documents arrive.

CopyKnowledgeBase builds the target field by field and skipped
profile_config, and the clone task never asked for a description after
copying the documents, so a clone stayed undescribed until its next
upload. The target now carries a copy of profile_config and both clone
paths request a refresh on completion.

* fix(knowledge): refresh the loaded clone target and carry document profiles

ProcessKBClone requested the description refresh with the reservation
stub it builds before CopyKnowledgeBase runs. For the main create path
that stub has neither a type nor profile_config, so the refresh helper
rejected it and nothing was enqueued. The refresh now uses the loaded
target returned by CopyKnowledgeBase.

cloneKnowledge copied the summary text but not the structured profile,
so even a scheduled refresh would have aggregated titles without topics
until every clone was re-summarized. Cloned documents now carry an
independent copy of the profile.
2026-09-17 22:21:55 +08:00
wizardchen 4c0d42c4cb feat(agent): keep long sandbox turns from truncating, overflowing, or rewriting files
Refuse truncated tool calls, compact-and-retry once on overflow, and let
sandbox files be appended, paged, and batch-edited instead of rewritten.
Models can set context_window (default 200K) so compaction sizes itself
from the real window; agent max_iterations of -1 means unlimited.
2026-09-02 22:09:09 +08:00
leon4z 988cbb0330
fix(models): show usage details when deletion is blocked (#2975)
* fix(models): show usage details when deletion is blocked

* fix(models): open the matching agent section and cap usage lists

Agent "open configuration" now lands on the tab that owns the binding
(VLM/ASR vs follow-ups), and delete failures keep the server message when
structured details cannot be parsed. Listed KB/agent rows are capped at 50
while totals still drive the deletion guard and dialog headings.

---------

Co-authored-by: Leon <leon_hfut@foxmail.com>
Co-authored-by: wizardchen <wizardchen@tencent.com>
2026-09-02 17:20:37 +08:00
asmh1989 60d2ffa4fb fix(ui): 统一剪贴板复制为公共方法,HTTP 环境回退 execCommand 2026-08-12 15:38:01 +08:00
MidoriKurage 1431e9414e feat(agent): prefill default rerank model on agent creation
Prefill rerank_model_id in the agent create dialog with the same
is_default-first, first-active fallback as chat

Extract the selection into frontend/src/utils/modelDefaults.ts and reuse
it in the KB dialog.
2026-08-10 17:20:01 +08:00
wizardchen f1ef364820 fix(knowledge): harden auto-tagging and let it respect manual tags
Automatic tagging shipped with a few paths that quietly disabled it and a
prompt that spent most of its budget on data the model never needed.

Robustness:
- Treat an omitted confidence as a deliberate selection instead of zero.
  Models frequently drop the field, and the 0.75 floor then filtered out
  every match, silently turning the feature off. Percentage-scale values
  are rescaled rather than accepted as out-of-range certainty.
- Classify the first 500 tags when a knowledge base has more, logging a
  warning, instead of dropping the task entirely. Large knowledge bases
  previously got no auto-tagging at all with no user-visible signal.
- Reuse common.ParseLLMJsonResponse for the model reply, and give that
  helper a balanced-scan fallback. The local extraction cut the payload at
  the last "}" or "]", so a trailing citation such as "see [1]" produced
  invalid JSON and burned the task's retries.

Prompt:
- Address candidates by ordinal rather than UUID. Tag names are short while
  a UUID is 36 characters, so this cuts candidate tokens roughly tenfold,
  keeps the model from having to reproduce (and corrupt) an id, and is what
  makes the raised 500 ceiling affordable. Ordinals are range-checked and
  mapped back to ids server-side.
- Drop the rationale field. Nothing read it, and at max_tags=10 it competed
  with the 1024-token output budget and risked truncating the JSON.
- Fence document text in <document> so file content cannot be read as
  instructions.

Behaviour:
- Add auto_tag_config.skip_if_tagged, defaulting to true, so documents the
  user already tagged at upload time keep their classification. Existing
  tags are now loaded before the model call, so a skipped document costs no
  LLM round-trip. Configs stored without the field adopt the default.

Also promote AddKnowledgeTagRelations to KnowledgeRepository instead of
reaching for it through a type assertion that failed at runtime, and route
the max_tags bounds through AutoTagConfig.Normalize so the prompt, the match
filter and the persisted config cannot drift apart.

Adds handler-level coverage for scope, cancellation, idempotency, ordinal
validation and the new toggle, plus SDK, Swagger and API doc updates.
2026-08-07 15:35:58 +08:00
lkx 9426ee1470 feat: 增加自动添加标签的功能 2026-08-07 15:01:01 +08:00
wizardchen a1a07a52eb fix(frontend): satisfy KB editor type-check after create save
Use a narrowed createdKbId local when reloading data and emitting success
so vue-tsc accepts the post-create flow.
2026-07-31 11:54:20 +08:00
wizardchen 23ae5e179c fix(frontend): close editor on save and guide post-create setup
Align agent and knowledge base editors so edit saves close the modal,
first-time create stays open with contextual footer hints, and follow-up
actions point users to sharing/integration or data-source configuration.
2026-07-31 11:54:20 +08:00
wizardchen a3a0308b62 feat(knowledge): add scoped activity audit trail 2026-07-23 10:50:32 +08:00
wizardchen f8b3e5f148 feat(storage): add multi-instance storage backends per workspace
Introduce tenant-scoped storage backend registry so workspaces can register
multiple storage instances and bind each knowledge base to a specific backend,
with migration from legacy tenant-level storage config and new settings UI.
2026-07-15 14:54:39 +08:00
wizardchen 139ec1e6b8 fix(storage): align default provider fallback with STORAGE_ALLOW_LIST
Extract storage allowlist into internal/storageallowlist so KB creation
uses FirstAllowed when tenant default_provider is empty or disallowed.
Share pickUsableStorageProvider via editorResources for consistent UI
selection, and add regression tests for OBS status and allow-list defaults.
2026-07-15 11:05:00 +08:00
ShadowSmallBaby b3bb60f03c fix: 修复/api/v1/tenants/kv/storage-engine-config接口不返回华为云obs配置导致创建知识库弹窗存储引擎华为云OBS未正确显示不可用标识
fix: 修复租户default_provider为空时降级至local,而local未被允许时导致创建知识库报错: Storage provider is not allowed by STORAGE_ALLOW_LIST
2026-07-15 10:54:14 +08:00
wizardchen 41975abcc5 refactor(terminology): rename user-facing tenant labels to workspace
Align product copy, API docs, Swagger descriptions, and i18n strings
with workspace terminology while keeping internal tenant identifiers
and headers unchanged for API compatibility.
2026-07-14 16:14:43 +08:00
wizardchen c1cddb4d4d Add instruction fields to knowledge base configuration 2026-07-13 12:16:19 +08:00
wizardchen e40ad802d7 feat(datasource): add RSS feed connector with full-text ingestion
Enable private RSS feeds as a knowledge base data source with custom
headers, article body fetching, and Markdown output for incremental sync.
2026-06-23 17:30:38 +08:00
wizardchen 10e63b2227 feat(frontend): add MCP config section and unify modal navigation styles
- Add dedicated MCP service configuration section in AgentEditorModal (agent mode only)
- Unify sidebar nav hover/active states across settings and editor modals
- Refine tool card and tag chip colors to use design tokens and color-mix
- Tighten suggested questions button padding and border-radius
2026-06-22 14:33:54 +08:00
wizardchen 20b01a6cd5 feat(embed): session sidebar grouping and embed theme alignment (#1657)
Group embed sessions in the sidebar by channel, replace session source filter
with grouping buckets, and align shared chat components with embed theme colors.
2026-06-12 23:10:39 +08:00
wizardchen d8a26f7c23 refactor: remove enable_multimodal and adjust parent-child configuration
- Removed the `enable_multimodal` field from the `ChunkingConfigOverride` interface and related components to streamline configuration.
- Updated the logic in `UploadConfirmDialog.vue` to default `enableParentChild` to false when initializing from knowledge base info.
- Modified the `mergeChunkingConfig` function to ensure that `EnableParentChild` is authoritative, allowing explicit control over its state.
- Added a new test case in `knowledge_process_config_test.go` to verify the behavior of parent-child toggles in process configuration resolution.
2026-06-09 20:38:24 +08:00
wizardchen 815c72c964 refactor(frontend): streamline knowledge base and agent handling with chat resources store
- Updated components to utilize the new `useChatResourcesStore` for managing knowledge bases and agents, improving data fetching and caching mechanisms.
- Refactored `AgentSelector`, `IMChannelPanel`, `Input-field`, and `menu` components to replace direct API calls with store methods, enhancing maintainability and performance.
- Adjusted knowledge base loading logic to ensure consistency across components and reduce redundant API requests.
- Improved session state handling in `settings` store to prevent overwriting user selections during state restoration.
2026-06-09 18:05:33 +08:00
wizardchen f4af9cca97 feat(contextualGuides): add multiple contextual guide components for agent and knowledge base creation
- Introduced new components: AgentCreateContextualGuide, KbCreateContextualGuide, TenantModelsGuide, and SpotlightGuide to enhance user onboarding.
- Implemented dynamic step configurations for each guide, allowing tailored user experiences based on context.
- Enhanced the existing NewUserGuide component to utilize the new SpotlightGuide for improved visual guidance.
- Updated localization files to include new strings for contextual guides, ensuring a comprehensive user experience across languages.
- Refactored existing components to integrate with the new guide system, improving maintainability and user interaction.
2026-06-03 14:42:46 +08:00
wizardchen 48381fbaf5 feat: implement tenant default storage provider handling in Knowledge Base creation
This commit introduces functionality to utilize the tenant's default storage provider when creating a Knowledge Base. It includes updates to the frontend to load the default provider from settings and apply it during Knowledge Base initialization. Additionally, the backend has been enhanced to ensure that the storage provider is set correctly based on tenant configuration, improving consistency across the application. Tests have been added to verify the correct application of the default storage provider in various scenarios.
2026-05-29 13:47:32 +08:00
wizardchen 83808cc5b7 feat(frontend): improve agent and KB editor ID display and intent prompts UX
Expose copyable resource IDs in edit modals and replace the intent prompt
dropdown with independent toggle buttons so multi-intent selection wraps cleanly.
2026-05-28 21:11:49 +08:00
ochan.kwon 8b7d00b260 feat: expose KB ↔ vector store binding in list, editor, and detail UI
Backend
-------
ListKnowledgeBases now enriches each row with the resolved vector_store
metadata (name / source / engine_type / status) via a new
buildKBListResponse helper. Store views are batch-resolved once per
request through BatchResolveStoreView so an N-KB list costs one
vector-store service call rather than N — closing the N+1 limitation
flagged in #1372's known-limitations section. Cross-tenant shared KBs
continue to render via SharedStoreDisplay so the owning tenant's store
inventory cannot be correlated across rows; the underlying vector_store_id
UUID is stripped from those responses.

Resolver failures degrade gracefully: bound KBs render as unavailable
instead of breaking the list. Test coverage pins the env / bound /
shared distinction, the batch-call-count invariant, and the
graceful-failure path.

Frontend
--------
KB editor modal gains a new "Vector Store" section. Create mode shows a
dropdown that combines the system default (env store) and the tenant's
configured user stores, fetched once at mount via the existing
listVectorStores API. Edit mode shows the bound store read-only via a
new VectorStoreBadge component with an explicit immutability hint —
matching the backend's `<-:create` GORM tag and the service-layer
UpdateKnowledgeBaseRequest DTO that already omit the field.

KB list cards surface a small engine-type badge for own-tenant bound
KBs, and a warning badge when the bound store is unavailable. Env-bound
and shared KBs render no badge (visual noise control). KB detail header
shows the bound store via the same VectorStoreBadge component; shared
KBs fall through to the badge's internal "shared" branch with no name /
engine / id rendered.

The KB editor's create-time error handler translates the typed
ErrVectorStoreBindingInvalid (2200) and ErrVectorStoreUnavailable
(2201) into localized messages and jumps the user back to the
VectorStore section so they can pick a different store or fall back to
the system default.

The KB row type gains five optional fields (vector_store_id / name /
engine_type / source / status). i18n: 18 new keys added to en-US,
ko-KR, zh-CN; ru-RU receives English placeholders pending translation
(consistent with prior PRs in this locale).

Part of #993 (Phase 2: Per-KB VectorStore Binding).
Phase 2 roadmap item: PR 5 (KB binding UI + list-response enrichment).
Depends on #994, #1310, #1372, #1386 (all backend in the Phase 2 series).
2026-05-22 17:40:10 +08:00
wizardchen c6a2f6c8cc refactor(permissions): enhance role-based access controls across knowledge base and agent views
- Updated visibility conditions for various UI elements based on user roles, ensuring that only users with appropriate permissions (Admin, Contributor, or KB creator) can access certain actions.
- Introduced computed properties to manage permissions for sharing knowledge bases, managing data sources, and executing graph extraction.
- Adjusted the AgentList and KnowledgeBase components to reflect these changes, improving user experience by preventing unauthorized actions from being displayed.
- Ensured that the backend routes are aligned with the updated frontend role checks, enhancing security and clarity in user interactions.
2026-05-18 17:28:58 +08:00
wizardchen f58139a945 fix(frontend): persist graph extract toggle state on KB save (#1297)
The "启用实体关系提取" switch in GraphSettings was silently dropped on
save: buildSubmitData only attached extract_config to the request when
both indexingStrategy.graphEnabled AND nodeExtractConfig.enabled were
true. When the graph indexing strategy was off (the default), the
toggle change never reached the backend and reloading the KB always
showed the switch as off.

Always emit extract_config based on the user's actual toggle state so
the value round-trips correctly through updateKBConfig regardless of
the indexing strategy selection.
2026-05-13 00:40:58 +08:00
wizardchen cc6ac6ac50 fix(frontend): unify radio button theme and disabled selection
- Add global t-radio-button outline styles in theme.css (brand fill when checked).

- Align ModelEditor, share KB, datasource, agent/MCP settings with t-radio-button.

- Remove redundant :deep overrides in AgentEditorModal, KB editor, FAQEntryManager, etc.

- Use brand disabled tone for disabled+checked radio buttons so selection stays visible.
2026-05-07 21:51:22 +08:00
Claude 49adf0d73c docs(chunking): align overlap default + sharpen UI/source/repo docs
Three documentation passes around the adaptive chunking work:

UI
- Frontend ChunkOverlap default consolidated to 80 (was 100), matching
  chunker.DefaultChunkOverlap on the backend. Both DEFAULT_CHUNKING_PRESET
  and initFormData updated. The KB-load fallback also uses 80 when a
  loaded KB has no chunk_overlap stored.
- All four locales (en-US, zh-CN, ko-KR, ru-RU) get rewritten chunking
  setting descriptions: each now states the validated range, the default,
  and the situations where you'd deviate (FAQ vs narrative, embedder
  token limits, language-specific corpora).

Source code
- splitter.go: DefaultChunkSize / DefaultChunkOverlap constants get a
  longer block-comment explaining the per-language token math and the
  use-case sweet spots, plus the migration note on what the old
  inconsistent defaults were.
- KBChunkingSettings.vue: new comment block above ChunkingConfig
  documents the slider min/max for each setting, why those bounds
  exist, and the recommended TokenLimit values per embedding model.

Repo docs
- New docs/CHUNKING.md: end-to-end guide covering why chunking matters,
  the adaptive 3-tier architecture, per-setting reference with ranges
  and sweet spots, parent-child explanation, the token-limit table per
  embedder (OpenAI / Voyage / Cohere / BGE / MiniLM / Jina), 7 use-case
  presets, the debug panel workflow, the API surface, and known
  trade-offs (recursive strategy hidden from UI, no auto-reindex on
  strategy switch, OCR limitations).
- CHANGELOG.md gets a new [Unreleased] section consolidating all the
  adaptive-chunking work shipped on this branch: 5 features, 8
  improvements, 6 fixes, 1 docs entry. The entry references
  docs/CHUNKING.md for deeper explanation.

https://claude.ai/code/session_01XADhx6mtu2ZYW3DE9Lun6k
2026-05-06 17:17:07 +08:00
Claude beb7d005e9 fix(chunker): QA audit — pointer DTOs, profile reuse, stats correctness
Resolves the issues surfaced by the post-Phase-3 self-review.

High severity
- Strategy / TokenLimit / Languages can now be reset. The KB-config
  handler uses pointer DTOs so the absence of a field in the payload
  means "no change" while an explicit empty string / 0 / [] clears
  the value. The frontend now sends the fields unconditionally so
  the user can revert to defaults via the clear icon — previously
  these fields were write-once.
- DocProfile is no longer computed twice in the preview endpoint.
  Diagnostics carries the profile that drove tier selection (when
  auto-strategy ran); the handler only re-profiles for explicit
  strategies that bypass the profiler.
- Chunk-size stats (avg / min / max / stddev) are now computed over
  the FULL chunk set before the response is truncated to
  previewMaxChunks. Avoids systematically misleading metrics for
  large documents.
- SplitWithDiagnostics defaults SelectedTier to TierLegacy so an
  empty-text request never leaves the diag with a blank tier string.

Medium severity
- Split() reverts to its own loop instead of delegating to
  SplitWithDiagnostics. Diagnostics struct allocation is reserved
  for callers that actually want the trace (preview endpoint).
  Matters in SplitParentChild where Split runs once per parent.
- Preview endpoint runs the splitter on a goroutine wrapped in a
  5s context.WithTimeout. CJK input at the 256k-rune ceiling
  could otherwise tie up the worker for several seconds.

Low severity
- KBChunkingSettings emits arrays via spread copies so parent-side
  mutation cannot drift the form's reactive state.
- KBChunkingDebug font-family includes a monospace fallback so the
  chunk content stays monospaced even when the TDesign CSS variable
  is missing.

Documentation
- DocProfile JSON shape is now annotated as part of the public
  preview API; field renames would be breaking.

https://claude.ai/code/session_01XADhx6mtu2ZYW3DE9Lun6k
2026-05-06 17:17:07 +08:00
Claude 46d9c396c6 feat(frontend): expose chunking strategy in KB editor
Surfaces the adaptive 3-tier chunking controls in the existing
"Chunking" section of the KB editor modal. Users can now pick a
strategy from a dropdown without crafting JSON for the API:

- Automatic (recommended)  → backend profiler picks the tier
- Markdown-optimized       → forces heading-aware splitting
- Smart structure detection → forces heuristic boundary detection
- Classic                  → original recursive splitter (legacy)

The fifth backend value `recursive` is intentionally hidden from the
UI — it's functionally near `legacy` and would dilute the choice.

Advanced options (collapsible, disabled when strategy=legacy):
- TokenLimit (0–8192 approx tokens; 0 = char-based budget)
- Languages multi-select (DE/EN/ZH; empty = auto-detect)

Behaviour:
- New KBs default to strategy="auto"
- Existing KBs without the field render with empty strategy and keep
  classic behavior on save unless the user actively picks one
- Overlap warning when overlap >= chunkSize/2

Wired through:
- KBChunkingSettings.vue: strategy select + advanced collapsible +
  ChunkingConfig interface extended
- KnowledgeBaseEditorModal.vue: form-state defaults, kb-load mapping,
  buildSubmitData (chunking_config) + updateKBConfig (documentSplitting)
  pass strategy/token_limit/languages
- api/initialization/index.ts: types extended
- i18n: en-US / zh-CN / ko-KR / ru-RU all carry strategy + advanced
  keys (6 keys each)

https://claude.ai/code/session_01XADhx6mtu2ZYW3DE9Lun6k
2026-05-06 17:17:07 +08:00
Claude f140c12518 feat(frontend): unify font rendering and add font picker in settings
The UI used hardcoded "PingFang SC" in 50+ components, while the global
font stack in App.vue used a different fallback chain. Outside of mainland
China this caused visibly different fonts in the sidebar, breadcrumbs,
search and agent views compared to settings and document panes.

- Introduce --app-font-family / --app-font-family-mono CSS variables in
  theme.css and route TDesign's --td-font-family through them
- Add a useFont composable that persists the user's choice in localStorage
  and applies it via document.documentElement.style before mount
- Replace hardcoded font-family declarations across the app with the new
  variables (TencentSans logo and the SVG attribute list are preserved)
- Add UI font and monospace font selectors in General Settings with
  i18n strings for en-US, zh-CN, ko-KR and ru-RU

https://claude.ai/code/session_01S1FVPf4jRfWXkTSnpMmMp7
2026-05-05 00:48:40 +08:00
wizardchen 7dd6ed47c1 feat: Enhance chunking configuration for wiki indexing strategy
- Introduced chunkingDirty state to track manual adjustments to chunking settings, preventing automatic overrides during strategy changes.
- Added WIKI_ONLY_CHUNKING_PRESET for specific configurations when using the wiki-only indexing strategy, ensuring optimal settings in creation mode.
- Implemented logic to apply or revert chunking presets based on the current indexing strategy, maintaining user-defined settings in edit mode.
- Updated resetState function to clear chunkingDirty state upon modal reset.
2026-04-23 11:21:28 +08:00
wizardchen a8a3d9e694 feat: Enhance wiki extraction and configuration options
- Introduced extraction granularity settings in the wiki configuration, allowing users to control the level of detail in entity and concept extraction (focused, standard, exhaustive).
- Updated the UI to reflect these new options, including tooltips for better user guidance.
- Refactored the knowledge base editor to support the new extraction granularity feature, ensuring a seamless user experience.
- Improved the backend logic for wiki ingest tasks to utilize the new granularity settings, enhancing the accuracy of extracted content.
- Added tests to validate the new extraction granularity functionality and ensure robust performance across different configurations.
2026-04-22 21:18:32 +08:00
wizardchen 3f97a7a090 feat: Add locked indexing strategy warning and UI enhancements in KnowledgeBaseEditorModal
- Introduced a new warning message in English and Chinese to inform users that the indexing strategy cannot be changed once the knowledge base contains content.
- Updated the UI to visually indicate when indexing options are locked, preventing user interaction with disabled checkboxes.
- Enhanced the styling of the indexing strategy section to improve user experience and clarity.

These changes ensure users are aware of the restrictions on indexing strategy modifications, enhancing the overall usability of the knowledge base editor.
2026-04-22 21:18:20 +08:00
wizardchen 86c227dd02 feat: Revamp indexing strategy UI in KnowledgeBaseEditorModal
- Enhanced the indexing strategy section by adding descriptions and a new layout for better user interaction.
- Introduced toggle functionality for vector indexing and improved the visual presentation of indexing options with new styles and badges.
- Updated the component's CSS to create a more responsive and user-friendly design for indexing check items.

These changes significantly improve the usability and clarity of the indexing strategy settings in the knowledge base editor.
2026-04-22 21:18:12 +08:00
wizardchen 28deb8282f feat: Add comprehensive documentation for WeKnora knowledge base and wiki features
- Introduced multiple new documentation files, including a complete documentation index, quick reference guides, and comprehensive analyses of the knowledge base and wiki settings.
- The new files provide detailed insights into the architecture, data models, API endpoints, and implementation strategies for the knowledge base and wiki functionalities.
- Enhanced navigation and usability for developers and team members by summarizing key components, configurations, and data flows related to the knowledge base system.

These additions significantly improve the clarity and accessibility of documentation, supporting better onboarding and reference for current and future team members.
2026-04-22 21:18:10 +08:00
wizardchen d4d005422d feat: Implement wiki_config updates in knowledge base management
Added functionality to enable updates for the `wiki_config` field in the `KnowledgeBaseConfig` struct. This includes:

1. **Type Definition Update**: Added `WikiConfig` field to `KnowledgeBaseConfig` to allow JSON unmarshaling during update requests.
2. **Service Layer Enhancement**: Updated the `UpdateKnowledgeBase` method to persist changes to `wiki_config` when provided in the request.
3. **Frontend TypeScript Adjustments**: Enhanced type definitions in the frontend to support `wiki_config`, improving IDE autocomplete and type checking.

This fix addresses the critical issue where `wiki_config` was not being updated due to its absence in the `KnowledgeBaseConfig` struct, ensuring proper handling of wiki configurations during knowledge base updates.
2026-04-22 21:17:21 +08:00
wizardchen e147209e77 feat: Implement comprehensive wiki feature for knowledge bases
This commit introduces the complete wiki feature for WeKnora, enabling AI-powered wiki page generation and management. The implementation includes:

**Backend Changes:**
- Wiki data model: WikiPage type with support for multiple page types (Summary, Entity, Concept, Index, Log)
- Database schema: wiki_pages table with full migration support
- WikiPageService: CRUD operations and page management
- WikiPageRepository: GORM-based persistence layer
- Wiki ingest pipeline: Automated generation of wiki pages from knowledge documents
  * Summary page generation using LLM
  * Entity and concept extraction in a single LLM call
  * Synthesis opportunity detection
  * Index page rebuilding
  * Log page maintenance
- Wiki boost feature: Enhance chat retrieval with wiki context
- Wiki linting: Maintenance and validation utilities
- Agent wiki tools: Enable agents to query and interact with wiki pages
- Wiki prompts: Comprehensive LLM prompt templates for all wiki generation tasks
- Language support: Reuse existing middleware language infrastructure for LLM prompts

**Frontend Changes:**
- Wiki browser UI: View all wiki pages with filtering and search
- Wiki API client: Knowledge base wiki management endpoints
- Knowledge base editor: Configure wiki settings (language, auto-ingest, synthesis model)
- i18n updates: Support for English, Korean, Russian, and Chinese interfaces

**Configuration:**
- Container DI: Wire up all wiki services
- Router: Register wiki API endpoints
- Task handling: Support async wiki ingest tasks

**Testing:**
- Unit tests for wiki page types
- Service layer tests
- Endpoint tests for wiki operations
- Integration tests with LLM mocking

**Documentation:**
- Language refactoring analysis and guides
- Implementation completion reports
- Quick reference guides for developers

**Key Features:**
✅ LLM-powered wiki page generation from documents
✅ Multi-language support (9+ languages)
✅ Automatic extraction of entities and concepts
✅ Synthesis opportunity detection
✅ Index and log page maintenance
✅ Progressive wiki building across multiple documents
✅ Agent-based wiki interaction
✅ Chat retrieval enhancement with wiki context
✅ Full frontend UI for wiki browsing

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-22 21:17:18 +08:00
wizardchen 0fcab3a413 feat(video): update ASR terminology and file type labels for improved clarity
- Changed 'Audio Processing' to 'Audio & Video' in English, Korean, Russian, and Chinese translations to reflect expanded functionality.
- Updated ASR configuration titles and descriptions to include video processing capabilities, enhancing user understanding of the feature.
- Renamed file type labels from 'Audio Files' to 'Audio & Video' across multiple languages for consistency and clarity.
- Enhanced document upload handling in the KnowledgeBase to filter unsupported file types, providing user feedback for filtered images, videos, and audio files.

This update improves the internationalization of the application and clarifies the capabilities of the ASR feature.
2026-04-14 22:59:14 +08:00
wizardchen 1a4b7b390d feat: enhance document summary functionality with expandable/collapsible feature and overflow detection, improving user interaction 2026-04-11 18:51:16 +08:00
wizardchen 155fc690a7 feat(audio): add audio file preview support and enhance UI components
- Integrated audio file preview functionality in document content and preview components.
- Updated localization files to include audio-related messages.
- Enhanced the UI to display an audio player for supported audio formats.
- Implemented audio loading state and error handling for unsupported audio playback.
2026-04-03 13:41:16 +08:00
wizardchen c93f4e94ae feat(asr): add ASR model connection check and update related components
- Implemented a new function to check ASR model connectivity via the /v1/audio/transcriptions endpoint.
- Updated ModelEditorDialog.vue to integrate ASR model checks and adjust UI elements accordingly.
- Enhanced localization files to reflect changes in ASR terminology and descriptions.
- Added ASR configuration options to the knowledge base and updated related migration scripts.
- Refactored model handling to support ASR as a distinct model type in the application.
2026-04-02 21:27:27 +08:00
wizardchen 624e24ba49 feat(asr): integrate Automatic Speech Recognition (ASR) support
- Added ASR configuration options to the knowledge base and model settings.
- Implemented ASR model selection and transcription capabilities in the knowledge processing pipeline.
- Enhanced file upload support to include audio formats for ASR processing.
- Updated localization files to include ASR-related labels and descriptions.
- Refactored model handling to accommodate ASR as a new model type.
2026-04-02 21:27:27 +08:00
ochan.kwon 25379f5df5 fix(frontend): use storage_provider_config instead of legacy storage_config
The backend introduced storage_provider_config as the authoritative
field in commit dbab7b0, but the frontend was never updated to use it.
This caused the UI to always display "local" regardless of the actual
storage provider, and saving KB settings could overwrite the correct
storage_provider_config with an incorrect value.

Changes:
- KnowledgeBaseEditorModal.vue: read from storage_provider_config with
  storage_config fallback, dual-write both fields on save
- KnowledgeBaseList.vue: update feature badge logic and add type def
  for storage_provider_config, exclude "local" from badge display
- api/knowledge-base/index.ts: add storage_provider_config to API type
- ko-KR.ts, ru-RU.ts: add missing i18n keys — multimodal section
  (sidebar.multimodal, multimodal.title/description), channel labels
  (channelLabel, channelWeb, channelApi, etc.), and agent.think
2026-04-01 20:44:13 +08:00
wizardchen d5ec5dc0ea feat(i18n): update localization files for new UI elements and actions
- Added translations for "Refresh logs" in English, Korean, Russian, and Chinese.
- Enhanced the knowledge base editor modal to include a badge for data source count.
- Improved data source settings UI with a confirmation dialog for deletion and a refresh button in sync logs.
- Updated the data source sync logs component to fetch logs on visibility change and added a refresh option in the header.
2026-03-30 20:46:07 +08:00
wizardchen adabf8e089 feat(datasource): implement data source management functionality
- Added a new API for managing data sources, including CRUD operations and connection validation.
- Introduced UI components for data source settings, sync logs, and editor dialog, enhancing user experience in managing external data sources.
- Updated localization files to support new data source features in multiple languages.
- Implemented cron expression humanization for better scheduling visibility.
- Enhanced knowledge base editor to include data source management options.
2026-03-30 20:46:07 +08:00