feat: skip ocr (#36)

CHANGELOG.md

@@ -9,6 +9,12 @@ can be found in [changelog.d folder](https://github.com/papermerge/papermerge-cl
 
 <!-- towncrier release notes start -->
 
+## 0.8.0 - 2024-02-22
+
+## Added
+
+- `--skip-ocr` flag. Works only with Papermerge REST API >= 3.1
+
 ## 0.7.1 - 2024-02-20
 
 ### Fixed

README.md

@@ -116,6 +116,14 @@ successful import - add `--delete` flag:
 PLEASE BE CAREFUL WITH `--delete` FLAG AS IT WILL IRREVERSIBLE DELETE THE LOCAL
 COPY OF THE UPLOADED DOCUMENT!
 
+Choose to skip OCR of imported documents with `--skip-ocr` flag:
+
+    $ papermerge-cli import --skip-ocr /path/to/folder/
+
+Skip OCR flag can be used with folders (will apply to all docs in the folder)
+or with individual documents.
+`--skip-ocr` flag will work only with Papermerge REST API >= v3.1
+
 ### search
 
 Search for node (document or folder) by text or by tags:
@@ -169,3 +177,13 @@ or in case of uuid is a folder:
 You can also specify the format/type of the downloaded archive (e.g. in case node is either a folder):
 
      $ papermerge-cli download --uuid <folder-uuid>  -f /path/to/file-system/folder.targz -t targz
+
+
+## Version Compatiblity
+
+
+| CLI Version | REST API version | Remarks|
+|-------------|------------------|--------|
+| 0.7.0       | 3.0.x            ||
+| 0.7.1       | 3.0.x            ||
+| 0.8.0       | 3.1.x            | Skip OCR feature introduced|

papermerge_cli/lib/importer.py

@@ -14,7 +14,8 @@ def upload_file_or_folder(
     token: str,
     file_or_folder: Path,
     parent_id=None,
-    delete: bool = False
+    delete: bool = False,
+    skip_ocr: bool = False,
 ) -> None:
     user: User = get_me(host=host, token=token)
 
@@ -27,6 +28,7 @@ def upload_file_or_folder(
             host=host,
             token=token,
             file_path=file_or_folder,
+            skip_ocr=skip_ocr,
             parent_id=parent_id
         )
         if delete:
@@ -40,6 +42,7 @@ def upload_file_or_folder(
                 host=host,
                 token=token,
                 file_path=Path(entry.path),
+                skip_ocr=skip_ocr,
                 parent_id=parent_id
             )
 

papermerge_cli/main.py

@@ -92,11 +92,19 @@
         help='Delete local(s) file after successful upload.'
     )
 ]
+SkipOCR = Annotated[
+    bool,
+    typer.Option(
+        is_flag=True,
+        help='Skip OCR i.e. do not trigger OCR operation on upload.'
+             ' Works only with REST API >= 3.1'
+    )
+]
 TargetNodeID = Annotated[
     uuid.UUID,
     typer.Option(
-        help="UUID of the target/destination folder. "
-             "Default value is user's Inbox folder's UUID."
+        is_flag=True,
+        help="Trigger OCR"
     )
 ]
 OrderBy = Annotated[
@@ -138,6 +146,7 @@ def import_command(
     ctx: typer.Context,
     file_or_folder: FileOrFolderPath,
     delete: DeleteAfterImport = False,
+    skip_ocr: SkipOCR = False,
     target_id: TargetNodeID | None = None
 ):
     """Import recursively folders and documents from local filesystem
@@ -150,6 +159,7 @@ def import_command(
             host=ctx.obj['HOST'],
             token=ctx.obj['TOKEN'],
             file_or_folder=Path(file_or_folder),
+            skip_ocr=skip_ocr,
             parent_id=target_id,
             delete=delete
         )

papermerge_cli/rest/documents.py

@@ -9,14 +9,16 @@ def upload(
     host: str,
     token: str,
     file_path: Path,
-    parent_id: UUID
+    parent_id: UUID,
+    skip_ocr: bool = False,
 ) -> Document:
     api_client = ApiClient[Document](token=token, host=host)
 
     doc_to_create = CreateDocument(
         title=file_path.name,
         file_name=file_path.name,
-        parent_id=parent_id
+        parent_id=parent_id,
+        ocr=not skip_ocr
     )
 
     response_doc: Document = api_client.post(

papermerge_cli/schema/documents.py

@@ -13,6 +13,9 @@ class CreateDocument(BaseModel):
     parent_id: UUID | None
     lang: str | None = None
     file_name: str | None = None
+    # if true then OCR the document
+    # if false then skip OCR part
+    ocr: bool = True
 
 
 class Page(BaseModel):

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "papermerge-cli"
-version = "0.7.1"
+version = "0.8.0"
 description = "Command line utility for your Papermerge DMS instance"
 authors = ["Eugen Ciur <[email protected]>"]
 license = "Apache 2.0"