The frontend was wired against /api/documents/{id}/chunks/* (canonical
doc-centric chunkset) but the backend never exposed those routes — the
chunk tab in the doc workspace 404'd. The domain entities (Chunk,
ChunkEdit, ChunkPush) and persistence repos already existed since #205;
what was missing was the service + API layer that connects them.
ChunkService owns all canonical chunkset invariants (sequence ordering,
soft-delete + audit log atomicity) and shares the chunker port with
AnalysisService so chunking strategy stays a single implementation.
AnalysisService grew a duck-typed promoter hook that copies the chunks
of the first successful analysis into the canonical chunkset. The hook
is idempotent so subsequent ad-hoc analyses (Studio / OCR Debug) never
overwrite hand-edited state.
Routes added (all additive, /api/documents prefix):
GET /{id}/chunks
POST /{id}/chunks
PATCH /{id}/chunks/{chunkId}
DELETE /{id}/chunks/{chunkId}
POST /{id}/chunks/{chunkId}/split
POST /{id}/chunks/merge
POST /{id}/rechunk
GET /{id}/tree
GET /{id}/diff?store=...
POST /{id}/chunks/push
- domain : StoreKind.NEO4J ajouté à l'enum
- service : validation config par kind étendue (Neo4j requires non-empty index_name)
- frontend : Neo4jConfigForm (index_name + database optionnelle), dispatcher StoreConfigForm, dropdown kind, i18n FR/EN
- tests service : 2 cas Neo4j (create OK + missing index_name 422)
Auth (URI/user/password) reste pilotée par les variables d'environnement
globales (cf. infra/neo4j.py) — la config par store ne porte que les
paramètres routables (index, database).
Couche service entre l'API et les repos SQLite :
- list_stores avec enrichissement document_count via document_store_links
- get/create/update/delete avec validations
- Slug normalisé lowercase, motif kebab-case strict
- Unicité name + slug (409 sur collision)
- Single-default invariant via clear_default_except
- Validation config par kind (OpenSearch require index_name) — extensible
- Delete refusé sur le store seedé "default" et sur tout store avec liens
- Erreurs typées (Validation/NotFound/Conflict) avec hint http_status
Étend le repo avec les méthodes manquantes pour le CRUD :
- find_by_name : valider unicité name au create/update
- update : remplace les champs mutables (id et created_at intacts)
- clear_default_except : promotion mono-default lors d'un changement de default
- delete : retourne True/False selon suppression effective
Tests dédiés sur chacune des nouvelles méthodes.
Backend
- HealthResponse exposes inspectModeEnabled / chunksModeEnabled /
askModeEnabled (additive; defaults true). main.py /api/health
populates them from settings.
- infra/settings.py: three new env-var-driven booleans (defaults true)
parsed in from_env() like the existing reasoning_enabled flag.
- tests/test_api_endpoints.py: extra assertion that /api/health
surfaces the three new fields with their defaults.
Frontend — flag store
- features/feature-flags/store.ts: FeatureFlag union extended with
inspectMode / chunksMode / askMode. New entries in featureRegistry
are gated on context fields populated from health. Missing fields
fall back to true so a frontend pointed at an older backend keeps
every mode visible.
- store gains a modeFlags() helper returning Record<DocMode, boolean>
so the routing guard does not need to know the FeatureFlag union.
Frontend — routing
- shared/routing/resolveMode.ts: pure resolver. If the requested mode
is enabled, return it; else first enabled in priority ask > chunks
> inspect; else null.
- app/router/index.ts: beforeEach guard scoped to ROUTES.DOC_WORKSPACE.
Disabled mode → rewrite ?mode= to the first enabled one. All three
off → redirect to /docs?reason=no-mode-enabled.
Frontend — flash
- pages/DocsLibraryPage.vue: shows a banner when ?reason=no-mode-enabled
is set. #211 will move this into the proper library page banner.
- i18n flags.allModesDisabled added in fr + en.
Tests
- shared/routing/resolveMode.test.ts (6 cases): every (requested,
enabled) combination including all-disabled, priority order,
missing requested.
- features/feature-flags/store.test.ts: three new cases covering the
new fields in /api/health, fall-back-to-true on missing fields, and
modeFlags() shape.
Refs #210
CLI script to migrate existing tenants to the data model introduced
by #202-205. Idempotent and resumable — re-running is a no-op.
tools/migrate_06.py
- Step 1: backfill_lifecycle — infers documents.lifecycle_state from
analysis_jobs.status + chunks_json presence
Failed analyses, no completed -> Failed
Completed + chunks_json -> Chunked
Completed, no chunks_json -> Parsed
Otherwise -> Uploaded
- Step 2: materialize_chunks — promotes the latest chunks_json blob
for each doc into rows in the new chunks table. Stable id derivation
via uuid5(NAMESPACE_OID, '<doc>|<seq>|<text>') so re-running lands
on the same ids
- Step 3: backfill_links — for any doc that has chunks materialized,
creates a (doc, default-store) link in Ingested state with a freshly
computed chunkset_hash. Treats 'chunks exist' as proxy for 'doc was
ingested into the legacy single index' — sufficient for the typical
pre-0.6.0 deployment, with OpenSearch reindex documented separately
- Step 4: reaggregate — applies #203's aggregate_lifecycle rule so
doc-chunked transitions to Ingested
Persistence
- migration_progress table for resumability + per-step idempotency
CLI
python -m tools.migrate_06 # full migration
python -m tools.migrate_06 --dry-run # plan only, no writes
python -m tools.migrate_06 --only-step <name> # rerun a phase
Tests
- 9 tests: inference rule per (completed, failed, chunks_json) tuple,
end-to-end migration on a hand-built three-doc snapshot, idempotency
(second run = no writes), --dry-run writes nothing, deterministic
chunk ids across reruns
Refs #206
The data and domain layers for the chunks editor (#219-224 in 0.6.0).
Chunks were previously stored as a JSON blob in analysis_jobs.chunks_json;
this commit makes them first-class persisted entities with stable IDs,
soft-delete, and an immutable audit log.
Domain
- Chunk: persistent entity with id, document_id, sequence, text,
headings, source_page, bboxes, doc_items, token_count, timestamps,
deleted_at (soft delete)
- ChunkEdit: immutable audit row (action, actor, at, before, after,
parents, children, reason)
- ChunkPush: snapshot of which chunk_ids landed in which store at push
- ChunkEditAction enum: insert/update/delete/merge/split
- domain/chunk_editing.py: pure operations on a chunkset (insert,
update, delete, merge, split). Each returns a new chunkset and the
affected chunk(s); errors raise ChunkEditingError.
Persistence
- Three new tables: chunks, chunk_edits, chunk_pushes (FK + indexes)
- SqliteChunkRepository (insert, insert_many, update, soft_delete,
find_for_document, find_by_id; respects deleted_at)
- SqliteChunkEditRepository (append-only audit log; paginated reads
ordered newest-first; per-chunk history)
- SqliteChunkPushRepository (per-(doc, store) latest snapshot)
Ports
- ChunkRepository, ChunkEditRepository, ChunkPushRepository protocols
added to domain/ports.py
Tests
- 17 tests for the pure chunk-editing operations covering insert /
update / delete / merge / split, sequence shifts, lineage, error
paths (out-of-range, missing id, deleted target, cross-document)
- 11 tests for the three repositories: round-trips, soft-delete
filtering, history ordering, lineage round-trip, cascade-delete with
document, find_latest semantics
Service orchestration (ChunkEditingService — atomic chunk + audit
write) and the API endpoints land in a follow-up commit on the same
feature branch / next release. The data + domain foundation here is
what unblocks #219-224.
Refs #205
Adds the pure-domain hash function that summarises a chunkset for
stale-detection purposes. Recorded on each DocumentStoreLink at push
time (#203 ships the column slot); compared against the recomputed
current hash to flip a link to Stale when the source has drifted.
domain/hashing.py
- chunkset_hash(chunks: Iterable[ChunkResult]) -> str
- SHA-256 over (text, source_page, headings) per chunk
- Excludes bboxes / doc_items / token_count by design
- 0x1F separator between chunks defends against the join-attack
(split A+B vs concat AB)
Tests
- 9 tests: determinism, sensitivity per included field, exclusion of
rendering-only fields, join-attack resistance, order sensitivity,
empty-input invariant
- Locked fixture: a hand-built 3-chunk input has a fixed expected hash;
CI fails loud if anyone changes the canonical inputs without updating
the fixture deliberately (and the release notes)
Service integration (recompute on chunk write, set on push) lands with
#205 once chunks are first-class — direct integration on the legacy
chunks_json path is deliberately deferred to keep #204 focused.
Refs #204
Introduces the data layer for multi-store ingestion. Documents can now
live in multiple stores, each with its own Ingested/Stale/Failed state.
The doc-level lifecycle (#202) becomes the aggregate over all per-store
links, computed by a pure domain function.
Domain
- Store entity (name, slug, kind, embedder, config, is_default)
- DocumentStoreLink entity with mark_ingested / mark_stale / mark_failed
helpers
- StoreKind and DocumentStoreLinkState enums
- aggregate_lifecycle(): pure function — Failed > Stale > Ingested
> fallback (the doc's pre-link Uploaded/Parsed/Chunked state)
Persistence
- New tables 'stores' and 'document_store_links' with the right indexes
(doc_id, store_id, state) and a UNIQUE (doc, store) on the link
- Default 'opensearch' store seeded idempotently in init_db, embedder
pulled from DEFAULT_EMBEDDER (fallback bge-m3)
- SqliteStoreRepository (find_by_slug, find_by_id, get_default, …)
- SqliteDocumentStoreLinkRepository with ON CONFLICT … DO UPDATE upsert
Ports
- StoreRepository and DocumentStoreLinkRepository protocols added
Tests
- 14 new tests: seed idempotency, insert/find round-trips, UNIQUE
constraint, cascade delete with the document, every link state
round-trips, aggregation rule with all branches
Refs #203
Adds a first-class lifecycle state to every document, distinct from
AnalysisJob.status. The lifecycle describes the document as a whole and
is the foundation for the doc-centric pivot in 0.6.0.
Domain
- DocumentLifecycleState enum (Uploaded/Parsed/Chunked/Ingested/Stale/Failed)
- Document.lifecycle_state and lifecycle_state_at fields
- Document.transition_to() validates against a transition table
(domain/lifecycle.py) and returns a DocumentLifecycleChanged event
- InvalidLifecycleTransitionError on disallowed transitions
Persistence
- ALTER TABLE documents to add the two columns (default 'Uploaded')
- New index idx_documents_lifecycle_state for filter perf
- _COLUMN_MIGRATIONS refactored to support multiple tables
- _POST_MIGRATION_DDL list for indexes on freshly-added columns
- SqliteDocumentRepository.update_lifecycle()
Services
- AnalysisService drives transitions on parse / chunk / re-chunk / fail
via _transition_document(); idempotent and resilient (logs WARN and
continues if a stale state is somehow encountered)
API
- DocumentResponse exposes lifecycleState + lifecycleStateAt
(additive — existing 'status' field kept for backwards compat)
Frontend
- Document type extended with lifecycleState and lifecycleStateAt
- DocumentLifecycleState union literal mirroring the backend enum
Tests
- 24 new tests in test_lifecycle.py covering transitions, idempotency,
invariant preservation, and event emission
- test_repos.py: round-trip + every-enum-value check + update_lifecycle
- test_chunking.py: rechunk path now mocks document_repo correctly
Refs #202
Two patterns in Docling's serialization were mirrored 1:1 by the graph
projection and produced node explosions on real documents:
- An InlineGroup (paragraph of mixed style runs) emits one `groups[]`
entry plus N `texts[]` runs. Naive iteration created one Paragraph
node per run.
- A Picture's `children` carry internal text labels extracted by the
layout model (flowchart boxes, chart axis labels, diagram callouts).
Each child became its own Paragraph node, drowning the figure.
`build_collapse_index` (in the shared `infra.docling_tree` helper) now
returns the `skip_refs` set + `inline_meta` overrides for both cases.
The Neo4j `tree_writer` and the in-memory `docling_graph` consume the
same index, so both projections stay in sync.
InlineGroups are projected as a single :Paragraph carrying the
concatenated text and the union of children's provs (re-indexed).
Pictures keep their :Figure node and prov; their descendants are
dropped. Captions live in the picture's separate `captions` field, not
in `children`, so they are unaffected.
* docs: rename Clean Architecture → Hexagonal Architecture (ports & adapters)
Le backend suit le pattern ports & adapters (ports dans domain/ports.py,
adaptateurs dans infra/), pas Clean Architecture au sens Uncle Bob.
Aligne la terminologie dans README, docs/architecture.md, ADR guide,
audit master, fiche audit 01, et la nav mkdocs.
Les noms de fichiers et la commande /audit:clean-architecture restent
stables pour preserver les liens croises et les skills existants.
* feat(settings): add paste-image size/type limits surfaced via /api/health
Introduces MAX_PASTE_IMAGE_SIZE_MB (default 10) and
PASTE_ALLOWED_IMAGE_TYPES (default image/png,image/jpeg,image/webp)
env vars so the upcoming Verify-mode clipboard-paste handler can
validate client-side against the same limits the backend enforces.
Follows the existing MAX_FILE_SIZE_MB pattern. Ships the accepted
design doc at docs/design/195-copy-paste-image-verify-mode.md.
Refs #195
Backend — live runner
- New `POST /api/documents/:id/rag` endpoint. Loads `document_json` from
SQLite, reconstructs the DoclingDocument, wraps the model id in
`ModelIdentifier(ollama_name=...)`, and calls `agent._rag_loop`
off-thread (blocking sync call). Returns a `RAGResult` in the shape
the existing v1 import path already consumes, so the frontend overlay
is fully reused.
- `_rag_loop` is private upstream; we call it because `run()` wraps the
answer in a synthetic DoclingDocument and drops the iteration trace.
- Settings: `RAG_ENABLED`, `OLLAMA_HOST`, `RAG_MODEL_ID`. Router mounts
unconditionally; handler 503s when the flag is off or deps aren't
installed. `rag_available` surfaced in `/api/health`.
- Maps known docling-agent bugs to readable HTTP errors: 502 with
"the model couldn't produce a parseable answer" when `_rag_loop`
raises `IndexError` from `find_json_dicts([])[0]` after 3 + 3
rejection-sampling retries (model-dependent).
- Tests: 11 cases (flag off, query empty, no analysis, happy path,
model_id wrap, Ollama env, IndexError → 502, other errors → 500,
deps missing → 503).
Backend — bug fix
- Default `BATCH_PAGE_SIZE` flipped from `10` to `0` to match the
dataclass default. The old default silently dropped `document_json`
(see `domain/services.merge_results`) for any doc > 10 pages, which
broke the reasoning tunnel. Set `BATCH_PAGE_SIZE>0` explicitly on
memory-constrained deploys if batching is wanted.
Frontend — runner UX
- `features/reasoning/api.ts:runReasoning()` — POST wrapper.
- `RunReasoningDialog.vue` — query textarea + optional model_id
override. Blocks close while running, 20-40s loading state,
synthesises a sidecar-shaped envelope so the panel surfaces query +
model the same way an imported trace would.
- `ReasoningWorkspace.vue` — primary "Run reasoning" button; "Import
trace" relegated to ghost secondary.
- Store: `runDialogOpen`, `running`, `setRunning`.
Frontend — answer polish
- Answer rendered through `marked` + DOMPurify (models emit markdown
lists; `pre-wrap` rendered them as plain "1. …" strings).
- Dedicated answer block with orange border, "ANSWER" label, "Copy"
button (clipboard + "Copied ✓" feedback).
- IterationCard: drop the duplicate `response` block (the main answer
is authoritative); style reasons equal to `"fallback"` (docling-agent
`select_from_failure` placeholder) as italic muted "— no structured
rationale".
Frontend — node details contents
- Clicking a SectionHeader (or any node with compound children) lists
its contained elements in `NodeDetailsPanel` under a new "Contents"
block. Children come from the same `parentMap` used for Cytoscape
compound parenting (explicit PARENT_OF + synthetic section scope),
inverted once and cached as a computed.
- Click a child row → pan the viewport to it + swap the selection.
Housekeeping
- `cytoscape-navigator` removed from `package-lock.json` (follow-up
from the minimap removal in the previous commit).
Adds the `docling-agent` reasoning-trace viewer as a Studio tunnel, per
`docs/design/reasoning-trace.md`. Users pick an analyzed document, import
a RAGResult JSON, and the iterations are overlaid on the document graph.
Graph source is decoupled from Neo4j: a new pure builder
(`infra/docling_graph.build_graph_payload`) reads `document_json` from
SQLite and emits the same Cytoscape-shaped payload that `fetch_graph`
returns from Neo4j. Neo4j stays exclusive to the Maintain ingestion
pipeline. Shared DoclingDocument helpers live in `infra/docling_tree.py`
so TreeWriter and the builder can't drift on label taxonomy or tree walks.
Also removes the Cytoscape minimap (cytoscape-navigator) from GraphView:
second render instance hurt perf on large documents for no UX win.
Backend
- new `GET /api/documents/:id/reasoning-graph` (SQLite-only)
- new `infra/docling_tree.py`, `infra/docling_graph.py`
- `analysis_repo.find_latest_completed_by_document`
- tests: `test_docling_graph.py` (builder), `test_graph_api.py` (endpoint)
Frontend
- `features/reasoning/` — store, overlay, types, panel, import dialog,
workspace, doc picker
- new `ReasoningPage` + `/reasoning` and `/reasoning/:docId` routes
- `GraphView` gains a `fetcher` prop so reasoning can inject the
SQLite-backed fetcher while Maintain keeps using the Neo4j one
- drops minimap (nav container, dep, CSS)
- legend filters + section parenting extracted for reuse
- i18n base strings (FR + EN)
The `doc_items` field was added to `ChunkResult` earlier in the
feature branch (used by ChunkWriter for DERIVED_FROM edges), but the
test fixture was never updated. CI caught it now that the PR is open.
Fixes: tests/test_chunking.py::TestChunkResult::test_serializable
ChunkWriter mirrors chunks into Neo4j after OpenSearch indexing, creating
HAS_CHUNK edges and DERIVED_FROM back-references to the source Elements
(via doc_items propagated from the local chunker).
Graph API: GET /api/documents/{id}/graph returns a cytoscape-shaped
payload with nodes + edges for Document / Element / Page / Chunk.
Hard cap at 200 pages returns HTTP 413 per design §8.4.
Frontend: new Graph tab in Studio results, rendered with Cytoscape.js +
dagre layout (lazy-loaded, ~175 KB gz). Legend, node styling per element
label, directional edges styled per edge type.
README gains a Neo4j section with the schema, three demo Cypher
queries, and env vars. Backend tests skip cleanly when the neo4j python
package is not installed locally.
Refs #186
Serialize a DoclingDocument to a Neo4j graph: Document + Page + Element
nodes with dynamic specific labels (SectionHeader, Paragraph, Table,
Figure, …), plus HAS_ROOT / PARENT_OF / NEXT / ON_PAGE edges. Replace-on-
write for idempotent re-ingestion.
The reader returns the verbatim document_json stored on the Document
node — reconstruction from graph nodes is deferred to v0.6.
Wired into AnalysisService._finalize_analysis: runs after conversion,
degrades gracefully by default, fails fast when neo4j_required is set.
Refs #186
Add Neo4j as an optional graph-native storage layer (ingestion profile).
Introduces infra/neo4j with a singleton async driver wrapper and an
idempotent bootstrap of constraints + indexes, wired into the FastAPI
lifespan. Integration tests skip when no live Neo4j is reachable.
Refs #186
Hybrid approach: reuse LocalChunker to chunk the DoclingDocument JSON
returned by Serve, so chunking works identically in both local and
remote modes without calling Serve's chunk endpoint.
Backend:
- _build_chunker() always returns LocalChunker (remove engine guard)
- Use docling-core[chunking] extra for required dependencies
- Skip client-side batching in remote mode (Serve manages its own
resources, and batching discards document_json needed for chunking)
- Fix Serve form fields: remove generate_page_images (not a Serve
field), use repeated form keys for to_formats and page_range
- Log Serve error response body on 4xx/5xx for diagnosis
- Fix FastAPI 204 DELETE routes missing response_model=None
Frontend:
- Update chunking feature flag to enable Prepare UI in remote mode
Closes#51
- Move DEFAULT_PAGE_WIDTH/HEIGHT to domain/value_objects.py and import in both converters
- Add opensearch_default_limit to Settings (configurable via OPENSEARCH_DEFAULT_LIMIT env var)
- Pass settings.conversion_timeout to ServeConverter, removing independent _DEFAULT_TIMEOUT
- Update OpenSearchStore to accept default_limit from Settings via constructor
- Conditionally mount ingestion router only when OpenSearch + embedding are configured
- Add `ingestionAvailable` field to /api/health response
- Add `ingestion` feature flag to frontend (hides Search nav, Ingest button,
OpenSearch badge, indexed badges/filters when disabled)
- Skip ingestion polling when flag is off
- Make OpenSearch + embedding optional in docker-compose via profiles
- Add docker-compose.ingestion.yml override for full-stack ingestion
- Set BATCH_PAGE_SIZE=5 default in Docker local image
- Lead Quick Start with one-liner docker run command
- Document ingestion as opt-in with dedicated section
- Add BATCH_PAGE_SIZE, MAX_FILE_SIZE_MB, MAX_PAGE_COUNT, RATE_LIMIT_RPM to config tables
- Update test counts (380 backend, 159 frontend)
- Date CHANGELOG 0.4.0, bump frontend version to 0.4.0
- Sync CONTRIBUTING.md with E2E Karate test sections
Closes#180
Backend: GET /api/ingestion/search?q=…&doc_id=… endpoint with
SearchResponse schema. Frontend: search bar in Documents page, results
with filename, page, chunk index, relevance score. 3 new API tests.
- local_converter.py: remove redundant `_default_converter = None` in
except block of `_ensure_default_converter` (variable was already None,
re-raised immediately — dead store)
- test_analysis_service.py: replace bare `await task` with
`await asyncio.gather(task)` to satisfy static analysis
- Replace French mode strings (configurer/verifier/preparer) with English
equivalents (configure/verify/prepare) in StudioPage.vue and tests
- Extract _build_conversion_options, _run_conversion, _finalize_analysis
from _run_analysis_inner to respect Single Responsibility Principle
- Rename _get_default_converter to _ensure_default_converter to reflect
its lazy-init side effect
Closes#136, closes#137, closes#138
Add state-machine guard clauses to AnalysisJob transition methods
(mark_running, mark_completed, mark_failed, update_progress) to prevent
invalid status transitions. Make all domain value objects immutable with
frozen=True. Add 11 tests covering guard clause behavior.
Closes#132, closes#133
- Add DocumentRepository and AnalysisRepository protocols in domain/ports.py (#128)
- Refactor persistence repos from module functions to SqliteDocumentRepository
and SqliteAnalysisRepository classes
- Inject repos into AnalysisService and new DocumentService class via
constructor, removing direct imports of persistence and infra.settings (#129)
- Move _merge_results, _classify_error, _extract_html_body to domain/services.py (#130)
- Update main.py composition root to build and wire all dependencies
- Switch api/documents.py to Depends pattern matching api/analyses.py
- Update all tests to use injected mocks instead of module-level patches
Closes#128, closes#129, closes#130
Re-read the job from DB before mark_completed so that
progress_current/progress_total written during batched conversion
are not overwritten by the stale in-memory object.
Add regression unit test and e2e assertion on final progress values.
Replace hardcoded 5 MB upload limit with a configurable setting.
Backend exposes the value via /api/health, frontend reads it
dynamically for validation and UI messages.
Closes#48
Use Docling's native page_range parameter to split large PDFs into
sequential batches, preventing memory exhaustion and timeouts.
Progress is reported via existing polling mechanism.
Closes#56
Use `import infra.local_converter as lc_mod` consistently instead of
mixing `import` and `from ... import` for the same module.
Addresses CodeQL review comment on PR #58.
Docling's native document_timeout is the only mechanism that can
interrupt processing inside a blocked thread (OCR, table extraction).
Without it, asyncio.wait_for cannot stop a frozen conversion.
Configurable via DOCUMENT_TIMEOUT env var (default: 120s).
Closes#57 (C1)
