Merge pull request #13 from scub-france/refact-fixes

Mke some fixes and refacto

document-parser/api/documents.py

@@ -87,6 +87,6 @@ async def preview(
         return Response(content=png_bytes, media_type="image/png")
     except ValueError as e:
         raise HTTPException(status_code=404, detail=str(e))
-    except Exception as e:
+    except Exception:
         logger.exception("Failed to generate preview")
-        raise HTTPException(status_code=422, detail=f"Failed to generate preview: {e}")
+        raise HTTPException(status_code=422, detail="Failed to generate preview")

document-parser/api/schemas.py

@@ -8,7 +8,7 @@ from __future__ import annotations
 
 from datetime import datetime
 
-from pydantic import BaseModel, ConfigDict
+from pydantic import BaseModel, ConfigDict, field_validator
 
 
 def _to_camel(name: str) -> str:
@@ -61,6 +61,20 @@ class PipelineOptionsRequest(BaseModel):
     generate_page_images: bool = False
     images_scale: float = 1.0
 
+    @field_validator("table_mode")
+    @classmethod
+    def validate_table_mode(cls, v: str) -> str:
+        if v not in ("accurate", "fast"):
+            raise ValueError('table_mode must be "accurate" or "fast"')
+        return v
+
+    @field_validator("images_scale")
+    @classmethod
+    def validate_images_scale(cls, v: float) -> float:
+        if v <= 0 or v > 10:
+            raise ValueError("images_scale must be between 0 (exclusive) and 10")
+        return v
+
 
 class CreateAnalysisRequest(BaseModel):
     documentId: str  # camelCase to match existing frontend contract

document-parser/domain/parsing.py

@@ -255,23 +255,23 @@ def convert_document(
         and images_scale == 1.0
     )
 
-    if is_default:
-        conv = get_default_converter()
-    else:
-        conv = build_converter(
-            do_ocr=do_ocr,
-            do_table_structure=do_table_structure,
-            table_mode=table_mode,
-            do_code_enrichment=do_code_enrichment,
-            do_formula_enrichment=do_formula_enrichment,
-            do_picture_classification=do_picture_classification,
-            do_picture_description=do_picture_description,
-            generate_picture_images=generate_picture_images,
-            generate_page_images=generate_page_images,
-            images_scale=images_scale,
-        )
-
     with _converter_lock:
+        if is_default:
+            conv = get_default_converter()
+        else:
+            conv = build_converter(
+                do_ocr=do_ocr,
+                do_table_structure=do_table_structure,
+                table_mode=table_mode,
+                do_code_enrichment=do_code_enrichment,
+                do_formula_enrichment=do_formula_enrichment,
+                do_picture_classification=do_picture_classification,
+                do_picture_description=do_picture_description,
+                generate_picture_images=generate_picture_images,
+                generate_page_images=generate_page_images,
+                images_scale=images_scale,
+            )
+
         result = conv.convert(file_path)
 
     doc = result.document

document-parser/persistence/analysis_repo.py

@@ -3,7 +3,7 @@
 from __future__ import annotations
 
 from domain.models import AnalysisJob, AnalysisStatus
-from persistence.database import get_db
+from persistence.database import get_connection
 
 
 def _row_to_job(row) -> AnalysisJob:
@@ -30,45 +30,36 @@ _SELECT_WITH_DOC = """
 
 
 async def insert(job: AnalysisJob) -> None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         await db.execute(
             """INSERT INTO analysis_jobs (id, document_id, status, created_at)
                VALUES (?, ?, ?, ?)""",
             (job.id, job.document_id, job.status.value, str(job.created_at)),
         )
         await db.commit()
-    finally:
-        await db.close()
 
 
-async def find_all() -> list[AnalysisJob]:
-    db = await get_db()
-    try:
+async def find_all(*, limit: int = 200, offset: int = 0) -> list[AnalysisJob]:
+    async with get_connection() as db:
         cursor = await db.execute(
-            f"{_SELECT_WITH_DOC} ORDER BY aj.created_at DESC"
+            f"{_SELECT_WITH_DOC} ORDER BY aj.created_at DESC LIMIT ? OFFSET ?",
+            (limit, offset),
         )
         rows = await cursor.fetchall()
         return [_row_to_job(r) for r in rows]
-    finally:
-        await db.close()
 
 
 async def find_by_id(job_id: str) -> AnalysisJob | None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         cursor = await db.execute(
             f"{_SELECT_WITH_DOC} WHERE aj.id = ?", (job_id,)
         )
         row = await cursor.fetchone()
         return _row_to_job(row) if row else None
-    finally:
-        await db.close()
 
 
 async def update_status(job: AnalysisJob) -> None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         await db.execute(
             """UPDATE analysis_jobs
                SET status = ?, content_markdown = ?, content_html = ?,
@@ -81,28 +72,20 @@ async def update_status(job: AnalysisJob) -> None:
              job.id),
         )
         await db.commit()
-    finally:
-        await db.close()
 
 
 async def delete(job_id: str) -> bool:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         cursor = await db.execute("DELETE FROM analysis_jobs WHERE id = ?", (job_id,))
         await db.commit()
         return cursor.rowcount > 0
-    finally:
-        await db.close()
 
 
 async def delete_by_document(document_id: str) -> int:
     """Delete all analysis jobs for a given document. Returns count deleted."""
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         cursor = await db.execute(
             "DELETE FROM analysis_jobs WHERE document_id = ?", (document_id,)
         )
         await db.commit()
         return cursor.rowcount
-    finally:
-        await db.close()

document-parser/persistence/database.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 import logging
 import os
+from contextlib import asynccontextmanager
 
 import aiosqlite
 
@@ -24,7 +25,7 @@ CREATE TABLE IF NOT EXISTS documents (
 
 CREATE TABLE IF NOT EXISTS analysis_jobs (
     id                TEXT PRIMARY KEY,
-    document_id       TEXT NOT NULL REFERENCES documents(id),
+    document_id       TEXT NOT NULL REFERENCES documents(id) ON DELETE CASCADE,
     status            TEXT NOT NULL DEFAULT 'PENDING',
     content_markdown  TEXT,
     content_html      TEXT,
@@ -52,3 +53,13 @@ async def get_db() -> aiosqlite.Connection:
     db.row_factory = aiosqlite.Row
     await db.execute("PRAGMA foreign_keys = ON")
     return db
+
+
+@asynccontextmanager
+async def get_connection():
+    """Context manager that opens and auto-closes a database connection."""
+    db = await get_db()
+    try:
+        yield db
+    finally:
+        await db.close()

document-parser/persistence/document_repo.py

@@ -3,7 +3,7 @@
 from __future__ import annotations
 
 from domain.models import Document
-from persistence.database import get_db
+from persistence.database import get_connection
 
 
 def _row_to_document(row) -> Document:
@@ -19,8 +19,7 @@ def _row_to_document(row) -> Document:
 
 
 async def insert(doc: Document) -> None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         await db.execute(
             """INSERT INTO documents (id, filename, content_type, file_size, page_count, storage_path, created_at)
                VALUES (?, ?, ?, ?, ?, ?, ?)""",
@@ -28,49 +27,36 @@ async def insert(doc: Document) -> None:
              doc.page_count, doc.storage_path, str(doc.created_at)),
         )
         await db.commit()
-    finally:
-        await db.close()
 
 
-async def find_all() -> list[Document]:
-    db = await get_db()
-    try:
+async def find_all(*, limit: int = 200, offset: int = 0) -> list[Document]:
+    async with get_connection() as db:
         cursor = await db.execute(
-            "SELECT * FROM documents ORDER BY created_at DESC"
+            "SELECT * FROM documents ORDER BY created_at DESC LIMIT ? OFFSET ?",
+            (limit, offset),
         )
         rows = await cursor.fetchall()
         return [_row_to_document(r) for r in rows]
-    finally:
-        await db.close()
 
 
 async def find_by_id(doc_id: str) -> Document | None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         cursor = await db.execute("SELECT * FROM documents WHERE id = ?", (doc_id,))
         row = await cursor.fetchone()
         return _row_to_document(row) if row else None
-    finally:
-        await db.close()
 
 
 async def update_page_count(doc_id: str, page_count: int) -> None:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         await db.execute(
             "UPDATE documents SET page_count = ? WHERE id = ?",
             (page_count, doc_id),
         )
         await db.commit()
-    finally:
-        await db.close()
 
 
 async def delete(doc_id: str) -> bool:
-    db = await get_db()
-    try:
+    async with get_connection() as db:
         cursor = await db.execute("DELETE FROM documents WHERE id = ?", (doc_id,))
         await db.commit()
         return cursor.rowcount > 0
-    finally:
-        await db.close()

document-parser/services/analysis_service.py

@@ -13,6 +13,9 @@ from persistence import analysis_repo, document_repo
 
 logger = logging.getLogger(__name__)
 
+# Maximum time (seconds) allowed for a single document conversion.
+CONVERSION_TIMEOUT = int(__import__("os").environ.get("CONVERSION_TIMEOUT", "600"))
+
 
 async def create(document_id: str, *, pipeline_options: dict | None = None) -> AnalysisJob:
     """Create a new analysis job and launch background processing."""
@@ -24,12 +27,25 @@ async def create(document_id: str, *, pipeline_options: dict | None = None) -> A
     job.document_filename = doc.filename
     await analysis_repo.insert(job)
 
-    # Fire-and-forget background task
-    asyncio.create_task(_run_analysis(job.id, doc.storage_path, doc.filename, pipeline_options))
+    # Fire background task with error logging callback
+    task = asyncio.create_task(
+        _run_analysis(job.id, doc.storage_path, doc.filename, pipeline_options)
+    )
+    task.add_done_callback(_on_task_done)
 
     return job
 
 
+def _on_task_done(task: asyncio.Task) -> None:
+    """Log unhandled exceptions from background analysis tasks."""
+    if task.cancelled():
+        logger.warning("Analysis task was cancelled")
+        return
+    exc = task.exception()
+    if exc:
+        logger.error("Unhandled exception in analysis task: %s", exc, exc_info=exc)
+
+
 async def find_all() -> list[AnalysisJob]:
     return await analysis_repo.find_all()
 
@@ -46,22 +62,23 @@ async def _run_analysis(
     job_id: str, file_path: str, filename: str, pipeline_options: dict | None = None,
 ) -> None:
     """Background task: run Docling conversion and update job status."""
-    job = await analysis_repo.find_by_id(job_id)
-    if not job:
-        logger.error("Analysis job %s not found", job_id)
-        return
-
-    job.mark_running()
-    await analysis_repo.update_status(job)
-    logger.info("Analysis started: %s (file: %s)", job_id, filename)
-
     try:
+        job = await analysis_repo.find_by_id(job_id)
+        if not job:
+            logger.error("Analysis job %s not found", job_id)
+            return
+
+        job.mark_running()
+        await analysis_repo.update_status(job)
+        logger.info("Analysis started: %s (file: %s)", job_id, filename)
+
         # Build kwargs from pipeline options
         convert_kwargs = pipeline_options or {}
 
-        # Run blocking Docling conversion in a thread
-        result: ConversionResult = await asyncio.to_thread(
-            convert_document, file_path, **convert_kwargs,
+        # Run blocking Docling conversion in a thread with timeout
+        result: ConversionResult = await asyncio.wait_for(
+            asyncio.to_thread(convert_document, file_path, **convert_kwargs),
+            timeout=CONVERSION_TIMEOUT,
         )
 
         pages_json = json.dumps([asdict(p) for p in result.pages])
@@ -79,7 +96,21 @@ async def _run_analysis(
 
         logger.info("Analysis completed: %s (%d pages)", job_id, result.page_count)
 
+    except asyncio.TimeoutError:
+        logger.error("Analysis timed out after %ds: %s", CONVERSION_TIMEOUT, job_id)
+        await _mark_failed(job_id, f"Conversion timed out after {CONVERSION_TIMEOUT}s")
+
     except Exception as e:
         logger.exception("Analysis failed: %s", job_id)
-        job.mark_failed(str(e))
-        await analysis_repo.update_status(job)
+        await _mark_failed(job_id, str(e))
+
+
+async def _mark_failed(job_id: str, error: str) -> None:
+    """Safely mark a job as failed, handling DB errors gracefully."""
+    try:
+        job = await analysis_repo.find_by_id(job_id)
+        if job:
+            job.mark_failed(error)
+            await analysis_repo.update_status(job)
+    except Exception:
+        logger.exception("Could not mark job %s as failed", job_id)

document-parser/services/document_service.py

@@ -18,11 +18,18 @@ UPLOAD_DIR = os.environ.get("UPLOAD_DIR", "./uploads")
 MAX_FILE_SIZE = 50 * 1024 * 1024  # 50 MB
 
 
+# PDF magic bytes: %PDF
+_PDF_MAGIC = b"%PDF"
+
+
 async def upload(filename: str, content_type: str, file_content: bytes) -> Document:
     """Save uploaded file to disk and persist metadata."""
     if len(file_content) > MAX_FILE_SIZE:
         raise ValueError("File too large (max 50 MB)")
 
+    if not file_content[:4].startswith(_PDF_MAGIC):
+        raise ValueError("Invalid file: not a PDF document")
+
     os.makedirs(UPLOAD_DIR, exist_ok=True)
 
     safe_name = f"{uuid.uuid4()}_{filename}"

document-parser/tests/test_schemas.py

@@ -1,7 +1,9 @@
-"""Tests for API schemas — camelCase serialization."""
+"""Tests for API schemas — camelCase serialization and validation."""
 
 from datetime import datetime
 
+import pytest
+
 from api.schemas import (
     AnalysisResponse,
     CreateAnalysisRequest,
@@ -109,6 +111,28 @@ class TestPipelineOptionsRequest:
         assert data["do_ocr"] is False
         assert data["do_table_structure"] is True  # default preserved
 
+    def test_invalid_table_mode_rejected(self):
+        with pytest.raises(ValueError, match='table_mode must be "accurate" or "fast"'):
+            PipelineOptionsRequest(table_mode="invalid")
+
+    def test_negative_images_scale_rejected(self):
+        with pytest.raises(ValueError, match="images_scale must be between"):
+            PipelineOptionsRequest(images_scale=-1.0)
+
+    def test_zero_images_scale_rejected(self):
+        with pytest.raises(ValueError, match="images_scale must be between"):
+            PipelineOptionsRequest(images_scale=0)
+
+    def test_excessive_images_scale_rejected(self):
+        with pytest.raises(ValueError, match="images_scale must be between"):
+            PipelineOptionsRequest(images_scale=11.0)
+
+    def test_boundary_images_scale_accepted(self):
+        opts = PipelineOptionsRequest(images_scale=0.1)
+        assert opts.images_scale == 0.1
+        opts2 = PipelineOptionsRequest(images_scale=10.0)
+        assert opts2.images_scale == 10.0
+
 
 class TestCreateAnalysisRequest:
     def test_parses_document_id(self):

frontend/package-lock.json

@@ -8,6 +8,7 @@
       "name": "docling-studio",
       "version": "0.1.0",
       "dependencies": {
+        "dompurify": "^3.3.3",
         "marked": "^17.0.4",
         "pinia": "^2.3.0",
         "vue": "^3.4.0",
@@ -765,6 +766,12 @@
       "integrity": "sha512-dWHzHa2WqEXI/O1E9OjrocMTKJl2mSrEolh1Iomrv6U+JuNwaHXsXx9bLu5gG7BUWFIN0skIQJQ/L1rIex4X6w==",
       "dev": true
     },
+    "node_modules/@types/trusted-types": {
+      "version": "2.0.7",
+      "resolved": "https://registry.npmjs.org/@types/trusted-types/-/trusted-types-2.0.7.tgz",
+      "integrity": "sha512-ScaPdn1dQczgbl0QFTeTOmVHFULt394XJgOQNoyVhZ6r2vLnMLJfBPd53SB52T/3G36VI1/g2MZaX0cwDuXsfw==",
+      "optional": true
+    },
     "node_modules/@vitejs/plugin-vue": {
       "version": "5.2.4",
       "resolved": "https://registry.npmjs.org/@vitejs/plugin-vue/-/plugin-vue-5.2.4.tgz",
@@ -1064,6 +1071,14 @@
         "node": ">=6"
       }
     },
+    "node_modules/dompurify": {
+      "version": "3.3.3",
+      "resolved": "https://registry.npmjs.org/dompurify/-/dompurify-3.3.3.tgz",
+      "integrity": "sha512-Oj6pzI2+RqBfFG+qOaOLbFXLQ90ARpcGG6UePL82bJLtdsa6CYJD7nmiU8MW9nQNOtCHV3lZ/Bzq1X0QYbBZCA==",
+      "optionalDependencies": {
+        "@types/trusted-types": "^2.0.7"
+      }
+    },
     "node_modules/entities": {
       "version": "7.0.1",
       "resolved": "https://registry.npmjs.org/entities/-/entities-7.0.1.tgz",

frontend/package.json

@@ -11,10 +11,11 @@
     "test:run": "vitest run"
   },
   "dependencies": {
-    "vue": "^3.4.0",
-    "vue-router": "^4.6.4",
+    "dompurify": "^3.3.3",
+    "marked": "^17.0.4",
     "pinia": "^2.3.0",
-    "marked": "^17.0.4"
+    "vue": "^3.4.0",
+    "vue-router": "^4.6.4"
   },
   "devDependencies": {
     "@vitejs/plugin-vue": "^5.2.0",

frontend/src/app/router/index.js

@@ -15,6 +15,11 @@ const routes = [
     path: '/settings',
     name: 'settings',
     component: () => import('../../pages/SettingsPage.vue')
+  },
+  {
+    path: '/:pathMatch(.*)*',
+    name: 'not-found',
+    redirect: '/'
   }
 ]
 

frontend/src/features/analysis/ui/ImageGallery.vue

@@ -38,7 +38,7 @@ const props = defineProps({
 const images = computed(() => {
   const result = []
   for (const page of props.pages) {
-    for (const el of page.elements) {
+    for (const el of (page.elements || [])) {
       if (el.type === 'picture') {
         result.push({ ...el, page: page.page_number })
       }
@@ -86,7 +86,7 @@ const images = computed(() => {
 .card-type {
   font-size: 12px;
   font-weight: 600;
-  color: #22C55E;
+  color: var(--success);
   text-transform: uppercase;
 }
 

frontend/src/features/analysis/ui/MarkdownViewer.vue

@@ -5,6 +5,7 @@
 <script setup>
 import { computed } from 'vue'
 import { marked } from 'marked'
+import DOMPurify from 'dompurify'
 import { useI18n } from '../../../shared/i18n.js'
 
 const props = defineProps({ content: String })
@@ -12,7 +13,7 @@ const { t } = useI18n()
 
 const rendered = computed(() => {
   if (!props.content) return `<p class="empty">${t('results.noMarkdown')}</p>`
-  return marked.parse(props.content)
+  return DOMPurify.sanitize(marked.parse(props.content))
 })
 </script>
 

frontend/src/features/analysis/ui/ResultTabs.vue

@@ -77,7 +77,7 @@ const pageMarkdown = computed(() => {
   const page = currentPageData.value
   if (!page) return ''
 
-  return page.elements
+  return (page.elements || [])
     .map(el => formatElement(el))
     .filter(Boolean)
     .join('\n\n')

frontend/src/features/document/ui/DocumentList.vue

@@ -25,13 +25,8 @@
 
 <script setup>
 import { useDocumentStore } from '../store.js'
+import { formatSize } from '../../../shared/format.js'
 const store = useDocumentStore()
-
-function formatSize(bytes) {
-  if (!bytes) return ''
-  const mb = bytes / (1024 * 1024)
-  return mb >= 1 ? `${mb.toFixed(1)} MB` : `${(bytes / 1024).toFixed(0)} KB`
-}
 </script>
 
 <style scoped>

frontend/src/pages/HistoryPage.vue

@@ -55,18 +55,13 @@ import { ref, onMounted } from 'vue'
 import { HistoryList, useHistoryStore } from '../features/history/index.js'
 import { useDocumentStore } from '../features/document/store.js'
 import { useI18n } from '../shared/i18n.js'
+import { formatSize } from '../shared/format.js'
 
 const historyStore = useHistoryStore()
 const docStore = useDocumentStore()
 const { t } = useI18n()
 const tab = ref('analyses')
 
-function formatSize(bytes) {
-  if (!bytes) return ''
-  const mb = bytes / (1024 * 1024)
-  return mb >= 1 ? `${mb.toFixed(1)} MB` : `${(bytes / 1024).toFixed(0)} KB`
-}
-
 function formatDate(iso) {
   if (!iso) return ''
   return new Date(iso).toLocaleString()

frontend/src/pages/StudioPage.vue

@@ -268,7 +268,7 @@
 </template>
 
 <script setup>
-import { ref, computed, watch, nextTick, onMounted, reactive } from 'vue'
+import { ref, computed, watch, nextTick, onMounted, onBeforeUnmount, reactive } from 'vue'
 import { useRoute, useRouter } from 'vue-router'
 import { useDocumentStore } from '../features/document/store.js'
 import { useAnalysisStore } from '../features/analysis/store.js'
@@ -368,6 +368,10 @@ onMounted(async () => {
     router.replace({ query: {} })
   }
 })
+
+onBeforeUnmount(() => {
+  analysisStore.stopPolling()
+})
 </script>
 
 <style scoped>

frontend/src/shared/format.js

@@ -0,0 +1,10 @@
+/**
+ * Format a byte count as a human-readable size string.
+ * @param {number|null|undefined} bytes
+ * @returns {string}
+ */
+export function formatSize(bytes) {
+  if (!bytes) return ''
+  const mb = bytes / (1024 * 1024)
+  return mb >= 1 ? `${mb.toFixed(1)} MB` : `${(bytes / 1024).toFixed(0)} KB`
+}