feat: skip OCR (#332)

Author: ciur

papermerge/core/routers/nodes.py

@@ -114,6 +114,7 @@ def create_node(
                 parent_id=pynode.parent_id,
                 size=0,
                 page_count=0,
+                ocr=pynode.ocr,
                 file_name=pynode.title
             )
             if pynode.id:

papermerge/core/schemas/documents.py

@@ -134,6 +134,10 @@ class CreateDocument(BaseModel):
     parent_id: UUID | None
     lang: str | None = None
     file_name: str | None = None
+    # Will OCR be triggered immediately?
+    # True: means, yes, trigger OCR after upload
+    # False: means, skip OCR
+    ocr: bool = True
 
     model_config = {
         "json_schema_extra": {

papermerge/core/signals.py

@@ -212,6 +212,11 @@ def receiver_document_post_upload(
     """
     doc_ver = document_version
     doc = document_version.document
+
+    if not doc.ocr:
+        logger.info(f"Skipping OCR for doc={doc} as doc.ocr=False")
+        return
+
     user = doc.user
 
     logger.debug(

ui/src/components/commander/commander.tsx

@@ -557,7 +557,7 @@ function Commander({
             onEditTagsClick={onEditTagsClick}
             onCreatedNodesByUpload={onCreatedNodesByUpload}
             selected_nodes={selected_nodes}
-            node_id={node_id} />
+            target={nodes!.data!.parent} />
 
             <div className="d-flex align-items-center">
               <SortDropdown

ui/src/components/commander/menu/Menu.tsx

@@ -1,7 +1,7 @@
 import UploadButton from './UploadButton';
 import NewFolder from './NewFolder';
 
-import type { CreatedNodesType, NodeType } from 'types';
+import type { CreatedNodesType, FolderType, NodeType } from 'types';
 import DeleteNodes from './DeleteNodes';
 import RenameNodes from './RenameNode';
 import EditTags from './EditTags';
@@ -14,7 +14,7 @@ type Args = {
   onEditTagsClick: () => void;
   onCreatedNodesByUpload: (created_nodes: CreatedNodesType) => void;
   selected_nodes: Array<string>;
-  node_id: string; // current node id
+  target: FolderType; // current node id
 }
 
 function Menu({
@@ -24,10 +24,10 @@ function Menu({
   onEditTagsClick,
   onCreatedNodesByUpload,
   selected_nodes,
-  node_id
+  target
 }: Args) {
 
-  const upload = <UploadButton node_id={node_id} onCreatedNodesByUpload={onCreatedNodesByUpload} />;
+  const upload = <UploadButton target={target} onCreatedNodesByUpload={onCreatedNodesByUpload} />;
   const new_folder = <NewFolder onClick={onNewFolderClick} />;
   const delete_nodes = <DeleteNodes onClick={onDeleteNodesClick} />;
   const rename_node = <RenameNodes onClick={onRenameClick} />;

ui/src/components/commander/menu/UploadButton.tsx

@@ -1,16 +1,16 @@
 import Button from 'react-bootstrap/Button';
-import { uploader } from 'utils/uploader';
 import Tooltip from 'react-bootstrap/Tooltip';
+import drop_files from '../modals/DropFiles';
 
 import { OverlayTrigger } from 'react-bootstrap';
-import { CreatedNodesType } from 'types';
+import { CreatedNodesType, FolderType } from 'types';
 
 type Args = {
-  node_id: string;
+  target: FolderType;
   onCreatedNodesByUpload: (created_nodes: CreatedNodesType) => void;
 }
 
-function UploadButton({node_id, onCreatedNodesByUpload}: Args) {
+function UploadButton({target, onCreatedNodesByUpload}: Args) {
 
   const onClickProxyUpload = () => {
     let element: HTMLInputElement | null = document.querySelector('input[type=file]');
@@ -30,7 +30,10 @@ function UploadButton({node_id, onCreatedNodesByUpload}: Args) {
       return;
     }
 
-    uploader({files, node_id}).then(
+    drop_files({
+      source_files: files,
+      target: target
+    }).then(
       (created_nodes: CreatedNodesType) => {
         onCreatedNodesByUpload(created_nodes);
       }

ui/src/components/commander/modals/DropFiles.tsx

@@ -1,3 +1,6 @@
+import { useState } from 'react';
+import Form from 'react-bootstrap/Form';
+
 import { MODALS } from 'cconstants';
 import GenericModal from 'components/modals/Generic';
 import { uploader } from 'utils/uploader';
@@ -29,10 +32,19 @@ const DropFilesModal = ({
     files from some folder - it does not make any sense for the upload dialog to
     be open for until all those 200 files get uploaded.
   */
+  const [skipOCR, setSkipOCR] = useState<boolean>(false);
+
+  const onLocalCheck = () => {
+    let new_checked_value = !skipOCR;
+
+    setSkipOCR(new_checked_value);
+  }
+
   const handleSubmit = async () => {
     uploader({
       files: source_files,
-      node_id: target.id
+      node_id: target.id,
+      skip_ocr: skipOCR
     })
     .then(
       (drop_files: CreatedNodesType) => onOK(drop_files)
@@ -56,6 +68,12 @@ const DropFilesModal = ({
         Are you sure you want to upload <span className='text-primary'>
           {source_titles.join(', ')}
         </span> to <span className='text-success'>{target_title}</span>?
+        <p className='pt-3'>
+          <Form.Check
+            onChange={onLocalCheck}
+            type="checkbox"
+            label={`Skip OCR`} />
+        </p>
     </GenericModal>
   );
 }

ui/src/utils/uploader.ts

@@ -5,23 +5,26 @@ import type { CreatedNodesType, NodeType } from 'types';
 type UploaderArgs = {
   files: FileList;
   node_id: string;
+  skip_ocr: boolean;
 }
 
 type CreateDocumentType = {
   title: string;
   parent_id: string;
   ctype: 'document';
+  ocr: boolean;
 }
 
 
-async function uploader({files, node_id}: UploaderArgs): Promise<CreatedNodesType> {
+async function uploader({files, node_id, skip_ocr}: UploaderArgs): Promise<CreatedNodesType> {
   let bulk_create_docs: any = [];
 
   Array.from(files, (file) => {
     let data: CreateDocumentType = {
       title: file.name,
       parent_id: node_id,
-      ctype: 'document'
+      ctype: 'document',
+      ocr: !skip_ocr
     }
 
     bulk_create_docs.push(