Files
ragflow/internal/deepdoc/parser/pdf/pdfsync/pdfsync.go

44 lines
1.7 KiB
Go
Raw Normal View History

feat(agent): Go ingestion pipeline progress mirroring and DeepDOC parser hardening (#16795) feat(ingestion): mirror Go pipeline progress into the document table; harden resume guards - pipeline: bind the owning document via WithDocumentID; after each TrackProgress event aggregate ingestion_task_log progress and mirror progress/run/progress_msg back into the document table, so GET /api/v1/datasets/{dataset_id}/documents reflects live Go pipeline progress without a bespoke endpoint. - canvas: extend the S3 resume guard to reject legacy no-op nodes (e.g. ExitLoop) so component_total equals the count of progress-reporting components and the aggregate percent can reach 100%. - runtime/canvas: route progress through TrackProgress; add interrupt test coverage (r3_interrupt_test.go). - dao/entity: add IngestionTask.DocumentID column and AggregateProgress support used by the mirror; IngestionTaskLog keeps a Checkpoint column alongside the progress fields. feat(deepdoc): cache DocAnalyzer inference results in Redis (1h TTL) - Redis-backed DocAnalyzerCache decorator over inference.Client; cache key = "ddoc:cache:<method>:" + sha256 of the JPEG-encoded image bytes (deterministic). - TTL = 1h; hits skip the inner HTTP call and return cached JSON; inner errors are not cached. refactor(deepdoc): align figure cropping with Python cropout + bounded page caches - CropSectionByDLA mirrors Python cropout: best-overlap DLA figure/equation region, fallback to section bbox per page, vertical concat on gray background. - sliding-window page-image cache bounds peak memory to the recent window instead of the whole PDF. - rename DLADebug -> DLARegions across parser/chunker/tests. refactor(parser): drop lib_type selector; align NewXxxParser with NewPDFParser - remove config["lib_type"] lookup and the libType param/field/switch from all nine constructors; surface the CGO-required error at ParseWithResult time instead of construction time; drop resolveLibType, its test, and the four lib_type constants. feat(utility): add a reusable workerpool for bounded concurrent execution - internal/utility/workerpool.go (+ tests). refactor: translate Chinese prose comments to English in non-harness Go files. chore: upgrade github.com/cloudwego/eino from v0.9.9 to v0.9.12.
2026-07-10 10:36:10 +08:00
//go:build cgo
// Package pdfsync provides a single process-wide mutex that serializes ALL
// access to the PDFium C library, regardless of which Go binding issues the
// call.
//
// Both the cgo pdfium binding (package pdfium) and the Rust pdf_oxide binding
// (package pdfoxide) are linked into the same final binary against the same
// PDFium static library. build.sh collapses the duplicate FPDF_* symbols with
// --allow-multiple-definition, so the process contains exactly ONE PDFium
// instance. PDFium is documented as NOT thread-safe for any call — concurrent
// calls corrupt the global heap and crash with SIGSEGV even when the calls
// operate on different documents.
//
// Therefore every native entry point in either binding must hold this one
// shared lock. Two independent mutexes would give the illusion of safety
// while still allowing a cgo pdfium call and a Rust pdf_oxide call to
// interleave onto the same PDFium instance. A single shared mutex is correct
// in both the one-instance and the (hypothetical, build-breaking) two-instance
// cases; two mutexes are only correct in the two-instance case.
package pdfsync
import "sync"
// Mu serializes all PDFium C API calls across every binding linked into the
// process. Acquire it (via With / WithErr) around every FPDF_* call and every
// pdf_oxide Rust call that re-enters the same PDFium instance.
var Mu sync.Mutex
// With runs f while holding the shared PDFium mutex.
func With(f func()) {
Mu.Lock()
defer Mu.Unlock()
f()
}
// WithErr runs f while holding the shared PDFium mutex and propagates its
// error unchanged.
func WithErr(f func() error) error {
Mu.Lock()
defer Mu.Unlock()
return f()
}