updates for OCRWorker (#565)

Author: ciur

.dockerignore

@@ -17,3 +17,4 @@ ui2/.vscode
 ui2/.pnp.cjs
 ui2/.pnp.loader.mjs
 ui2/.env.development.local
+ui2/public/papermerge-runtime-config.js

.gitignore

@@ -38,3 +38,4 @@ simplest.yml
 ui/public/runtime/config.js
 .ruff_cache/
 .pytest_cache/
+.enwardrc

docker/standard/Dockerfile

@@ -14,11 +14,14 @@ ENV CORE_APP=/core_app
 ENV PAPERMERGE__DATABASE__URL=sqlite:////db/db.sqlite3
 ENV PAPERMERGE__AUTH__USERNAME=admin
 ENV PAPERMERGE__AUTH__EMAIL=admin@example.com
-ENV PAPERMERGE__OCR__DEFAULT_LANGUAGE=deu
 ENV PAPERMERGE__MAIN__API_PREFIX=""
+ENV PAPERMERGE__OCR__LANG_CODES="deu,eng,ron"
+ENV PAPERMERGE__OCR__DEFAULT_LANG_CODE="deu"
+ENV PAPERMERGE__OCR__AUTOMATIC="false"
 
 RUN apk update && apk add linux-headers python3-dev \
     gcc \
+    curl \
     libc-dev \
     supervisor \
     imagemagick \
@@ -27,6 +30,7 @@ RUN apk update && apk add linux-headers python3-dev \
     poppler-utils
 
 RUN pip install --upgrade poetry roco==0.4.2
+RUN curl -L -o /bin/env2js https://github.com/papermerge/env2js/releases/download/0.2/env2js.amd64
 
 COPY poetry.lock pyproject.toml README.md LICENSE ${CORE_APP}/
 
@@ -37,6 +41,7 @@ COPY docker/standard/entrypoint.sh /entrypoint.sh
 COPY docker/standard/bundles/supervisor/* /etc/papermerge/
 COPY docker/standard/bundles/nginx/* /etc/nginx/
 COPY docker/standard/logging.yaml /etc/papermerge/
+COPY docker/standard/core.js.tmpl /${CORE_APP}/core.js.tmpl
 COPY ./papermerge ${CORE_APP}/papermerge/
 COPY alembic.ini ${CORE_APP}/
 

docker/standard/core.js.tmpl

@@ -0,0 +1,5 @@
+window.__PAPERMERGE_RUNTIME_CONFIG__ = {
+  ocr__lang_codes: "{{ .PAPERMERGE__OCR__LANG_CODES }}",
+  ocr__default_lang_code: "{{ .PAPERMERGE__OCR__DEFAULT_LANG_CODE }}",
+  ocr__automatic: {{ .PAPERMERGE__OCR__AUTOMATIC }}
+}

docker/standard/entrypoint.sh

@@ -55,8 +55,15 @@ case $CMD in
     ;;
   server)
     exec_init
+    # TODO: replace roco with env2js
     roco > /usr/share/nginx/html/auth_server/papermerge-runtime-config.js
-    roco > /usr/share/nginx/html/ui/papermerge-runtime-config.js
+    /bin/env2js -f /core_app/core.js.tmpl > /usr/share/nginx/html/ui/papermerge-runtime-config.js
+    exec /usr/bin/supervisord -c /etc/papermerge/supervisord.conf
+    ;;
+  server_without_init)
+    # TODO: replace roco with env2js
+    roco > /usr/share/nginx/html/auth_server/papermerge-runtime-config.js
+    /bin/env2js -f /core_app/core.js.tmpl > /usr/share/nginx/html/ui/papermerge-runtime-config.js
     exec /usr/bin/supervisord -c /etc/papermerge/supervisord.conf
     ;;
   create_token.sh)

papermerge/core/config.py

@@ -19,7 +19,15 @@ class Settings(BaseSettings):
     papermerge__main__cf_domain: str | None = None
     papermerge__database__url: str = "sqlite:////db/db.sqlite3"
     papermerge__redis__url: str | None = None
-    papermerge__ocr__default_language: str = 'deu'
+    papermerge__ocr__default_lang_code: str = 'deu'
+    # When is OCR triggered ?
+    # `ocr__automatic` = True means that OCR will be performed without
+    #   end user intervention i.e. via background scheduler like celery scheduler
+    # `ocr__automatic` = False means that OCR will be performed only
+    #   if requested by end user. In this case user can choose to
+    #   start schedule OCR on upload; also in this case use can choose to
+    #   scheduler OCR later on any document.
+    papermerge__ocr__automatic: bool = False
     papermerge__search__url: str | None = None
 
 

papermerge/core/features/document/cli/cli.py

@@ -0,0 +1,32 @@
+import uuid
+import typer
+
+from papermerge.core.tasks import send_task
+from papermerge.core.db.engine import Session
+
+from papermerge.core import dbapi, constants, types
+
+
+app = typer.Typer(help="OCR tasks")
+
+
+@app.command()
+def schedule_ocr(node_id: uuid.UUID, force: bool = False, lang: str | None = None):
+    """Schedules OCR for given node ID"""
+    with Session() as db_session:
+        node_type: types.CType = dbapi.get_node_type(db_session, node_id)
+
+        if node_type == "document":
+            if lang is None:
+                lang = dbapi.get_document_lang(db_session, node_id)
+            send_task(
+                constants.WORKER_OCR_DOCUMENT,
+                kwargs={
+                    "document_id": str(node_id),
+                    "lang": lang,
+                },
+                route_name="ocr",
+            )
+        else:
+            # get all descendants of node_id
+            pass

papermerge/core/features/document/db/api.py

@@ -27,10 +27,11 @@
 )
 from papermerge.core.features.document.schema import DocumentCFVRow
 from papermerge.core.features.document.ordered_document_cfv import OrderedDocumentCFV
+from papermerge.core import config
 
 from .selectors import select_doc_cfv, select_docs_by_type
 
-
+settings = config.get_settings()
 
 logger = logging.getLogger(__name__)
 
@@ -634,6 +635,17 @@ def upload(
         route_name="s3",
     )
 
+    if not settings.papermerge__ocr__automatic:
+        if doc.ocr is True:
+            # user chose "schedule OCR" when uploading document
+            tasks.send_task(
+                constants.WORKER_OCR_DOCUMENT,
+                kwargs={
+                    "document_id": str(doc.id),
+                    "lang": doc.lang,
+                },
+                route_name="ocr",
+            )
 
     return validated_model, None
 

ui2/index.html

@@ -9,6 +9,7 @@
   <body>
     <div id="root"></div>
     <div id="modals"></div>
+    <script type="module" src="/papermerge-runtime-config.js"></script>
     <script type="module" src="/src/main.tsx"></script>
   </body>
 </html>

ui2/papermerge-runtime-config.js

@@ -0,0 +1,5 @@
+window.__PAPERMERGE_RUNTIME_CONFIG__ = {
+  ocr__lang_codes: "deu,eng,ron, spa, ita, fra",
+  ocr__default_lang_code: "eng",
+  ocr__automatic: false
+}