perf(wiki): move ingest log to event table and index to on-demand API

Fixes O(n²) write amplification during wiki ingest on large KBs. Previously
every ingest/retract op re-wrote the single `slug='log'` row end-to-end and
every batch re-wrote the entire `slug='index'` directory markdown. On a 40k-
doc KB the log row grew to tens of MB and the index row to several MB, so
each batch triggered giant TOAST updates that dominated ingest wall time.

Log: new `wiki_log_entries` event table (`id DESC` indexed per KB) replaces
the single TEXT row. Batch ingest now collects entries and flushes them
once per batch via `AppendBatch`. Each entry stores `pages_affected` as
JSONB `[{slug,title}]` so the UI can render real titles; custom Scan falls
back to legacy `[]string` so older rows still deserialize.

Index: `wiki_pages[slug=index].content` keeps only the LLM-generated intro
(a few KB). The directory is now served by a structured paginated API
(`GetIndexView`) that reads `slug/title/summary` per type with cursor
pagination, so the agent and the frontend only pull the slice they need.
`RebuildIndexPage` degrades to a no-op; agent `wiki_read_page('index')`
synthesizes a small top-K overview and points callers at `wiki_search`.

Ingest resilience: LLM calls wrap with 3-attempt exponential backoff on
transient errors (5xx/408/429, transport resets/timeouts). Summary/extract
failures now bubble up so the batch's failed-op requeue path runs instead
of silently dropping the doc.

Frontend: sidebar Index/Log entries switch to dedicated views. Index view
streams intro → Summary → Entity → Concept → Synthesis → Comparison via
IntersectionObserver (with a nextTick re-check so small KBs still load
every section). Log view uses cursor pagination; pages_affected renders
titles with slug tooltip.

- new migration: migrations/versioned/000040_wiki_log_entries.{up,down}.sql
- tests: log repo pagination + legacy Scan, ListByTypeLight windowing,
  renderIndexOverviewForAgent output, isTransientLLMError classifier

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
wizardchen
2026-05-08 21:43:53 +08:00
committed by lyingbug
parent bf12868923
commit d0144f3586
25 changed files with 2353 additions and 272 deletions
+56 -17
View File
@@ -19,9 +19,10 @@ import (
// WikiPageHandler handles HTTP requests for wiki page operations
type WikiPageHandler struct {
wikiService interfaces.WikiPageService
kbService interfaces.KnowledgeBaseService
lintService *service.WikiLintService
wikiService interfaces.WikiPageService
kbService interfaces.KnowledgeBaseService
lintService *service.WikiLintService
logEntryService interfaces.WikiLogEntryService
}
// NewWikiPageHandler creates a new wiki page handler
@@ -29,11 +30,13 @@ func NewWikiPageHandler(
wikiService interfaces.WikiPageService,
kbService interfaces.KnowledgeBaseService,
lintService *service.WikiLintService,
logEntryService interfaces.WikiLogEntryService,
) *WikiPageHandler {
return &WikiPageHandler{
wikiService: wikiService,
kbService: kbService,
lintService: lintService,
wikiService: wikiService,
kbService: kbService,
lintService: lintService,
logEntryService: logEntryService,
}
}
@@ -274,12 +277,19 @@ func (h *WikiPageHandler) DeletePage(c *gin.Context) {
}
// GetIndex godoc
// @Summary Get wiki index page
// @Description Returns the wiki index page (creates default if not exists)
// @Summary Get wiki index view
// @Description Returns the wiki index as intro text plus per-type paginated
// @Description directory groups. The heavy directory markdown that used to
// @Description live in wiki_pages.content was replaced with this structured
// @Description response so a KB with tens of thousands of pages no longer
// @Description materializes megabytes of TEXT on every index open.
// @Tags Wiki
// @Produce json
// @Param kb_id path string true "Knowledge base ID"
// @Success 200 {object} types.WikiPage
// @Param kb_id path string true "Knowledge base ID"
// @Param types query string false "Comma-separated page types (default: all content types)"
// @Param limit query int false "Per-group window size, 1-200 (default 50)"
// @Param cursor query string false "Opaque offset cursor from previous response"
// @Success 200 {object} types.WikiIndexResponse
// @Security Bearer
// @Router /api/v1/knowledgebase/{kb_id}/wiki/index [get]
func (h *WikiPageHandler) GetIndex(c *gin.Context) {
@@ -289,22 +299,43 @@ func (h *WikiPageHandler) GetIndex(c *gin.Context) {
return
}
page, err := h.wikiService.GetIndex(c.Request.Context(), kbID)
var pageTypes []string
if raw := c.Query("types"); raw != "" {
for _, t := range strings.Split(raw, ",") {
t = strings.TrimSpace(t)
if t != "" {
pageTypes = append(pageTypes, t)
}
}
}
limit := 50
if raw := c.Query("limit"); raw != "" {
if v, convErr := strconv.Atoi(raw); convErr == nil && v > 0 {
limit = v
}
}
resp, err := h.wikiService.GetIndexView(c.Request.Context(), kbID, pageTypes, limit, c.Query("cursor"))
if err != nil {
c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
return
}
c.JSON(http.StatusOK, page)
c.JSON(http.StatusOK, resp)
}
// GetLog godoc
// @Summary Get wiki operation log
// @Description Returns the wiki operation log page (creates default if not exists)
// @Description Returns a paginated feed of wiki operation events (ingest, retract, ...)
// @Description newest-first. Pagination is cursor-based: pass `next_cursor` from the
// @Description previous response back as `cursor` to fetch the next page.
// @Tags Wiki
// @Produce json
// @Param kb_id path string true "Knowledge base ID"
// @Success 200 {object} types.WikiPage
// @Param kb_id path string true "Knowledge base ID"
// @Param cursor query string false "Opaque cursor from the previous page (empty = newest)"
// @Param limit query int false "Page size, 1-200 (default 50)"
// @Success 200 {object} types.WikiLogEntryListResponse
// @Security Bearer
// @Router /api/v1/knowledgebase/{kb_id}/wiki/log [get]
func (h *WikiPageHandler) GetLog(c *gin.Context) {
@@ -314,13 +345,21 @@ func (h *WikiPageHandler) GetLog(c *gin.Context) {
return
}
page, err := h.wikiService.GetLog(c.Request.Context(), kbID)
cursor := c.Query("cursor")
limit := 50
if raw := c.Query("limit"); raw != "" {
if v, convErr := strconv.Atoi(raw); convErr == nil && v > 0 {
limit = v
}
}
resp, err := h.logEntryService.List(c.Request.Context(), kbID, cursor, limit)
if err != nil {
c.JSON(http.StatusInternalServerError, gin.H{"error": err.Error()})
return
}
c.JSON(http.StatusOK, page)
c.JSON(http.StatusOK, resp)
}
// Graph query parameter bounds. The defaults cap an `overview` request at