Add support for uploading documents via REST API

sbrunner · Aug 14, 2023 · 2e0be8c · 2e0be8c
1 parent e9d23fd
commit 2e0be8c
Show file tree

Hide file tree

Showing 10 changed files with 283 additions and 15 deletions.
diff --git a/config.md b/config.md
@@ -99,7 +99,7 @@
   - **`auto_cut`** _(object)_: The auto mask configuration, the mask is used to definitively mask the source image. Default: `{"enabled": false}`.
     - **`enabled`** _(boolean)_: Enable the auto cut. Default: `true`.
     - **`auto_mask`**: Refer to _[#/definitions/auto_mask](#definitions/auto_mask)_.
-  - **`no_remove_to_continue`** _(boolean)_: Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the pdf. Default: `false`.
+  - **`no_remove_to_continue`** _(boolean)_: Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the PDF. Default: `false`.
   - **`deskew`** _(object)_: The deskew configuration.
     - **`min_angle`** _(number)_: The minimum angle to detect the image skew [degree]. Default: `-45`.
     - **`max_angle`** _(number)_: The maximum angle to detect the image skew [degree]. Default: `45`.
@@ -124,6 +124,12 @@
     - **`graduation_text_font_color`** _(array)_: Default: `[0, 0, 0]`.
       - **Items** _(integer)_
     - **`graduation_text_margin`** _(integer)_: Default: `6`.
+  - **`rest_upload`** _(object)_: Upload the final PDF via Paperless REST API.
+    - **`enabled`** _(boolean)_: Enable the upload of the PDF via REST API. Default: `false`.
+    - **`api_url`** _(string, required)_: The URL address of the REST API, usually http://server.name/api.
+    - **`api_token`** _(string, required)_: The API token.
+  - **`consume_folder`** _(object)_: Send the final PDF to Paperless using the consume folder.
+    - **`enabled`** _(boolean)_: Enable using the consume folder. Default: `true`.
 - <a id="definitions/contour"></a>**`contour`** _(object)_: The configuration used to find the contour.
   - **`min_box_size`** _(number)_: The minimum box size to find the content [mm]. Default: `{"crop": 3, "empty": 10, "limit": 10}`.
   - **`min_box_black`** _(number)_: The minimum black in a box on content find [%]. Default: `2`.

diff --git a/poetry.lock b/poetry.lock
diff --git a/process.md b/process.md
@@ -119,7 +119,7 @@
   - **`auto_cut`** _(object)_: The auto mask configuration, the mask is used to definitively mask the source image. Default: `{"enabled": false}`.
     - **`enabled`** _(boolean)_: Enable the auto cut. Default: `true`.
     - **`auto_mask`**: Refer to _[#/definitions/auto_mask](#definitions/auto_mask)_.
-  - **`no_remove_to_continue`** _(boolean)_: Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the pdf. Default: `false`.
+  - **`no_remove_to_continue`** _(boolean)_: Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the PDF. Default: `false`.
   - **`deskew`** _(object)_: The deskew configuration.
     - **`min_angle`** _(number)_: The minimum angle to detect the image skew [degree]. Default: `-45`.
     - **`max_angle`** _(number)_: The maximum angle to detect the image skew [degree]. Default: `45`.
@@ -144,3 +144,9 @@
     - **`graduation_text_font_color`** _(array)_: Default: `[0, 0, 0]`.
       - **Items** _(integer)_
     - **`graduation_text_margin`** _(integer)_: Default: `6`.
+  - **`rest_upload`** _(object)_: Upload the final PDF via Paperless REST API.
+    - **`enabled`** _(boolean)_: Enable the upload of the PDF via REST API. Default: `false`.
+    - **`api_url`** _(string, required)_: The URL address of the REST API, usually http://server.name/api.
+    - **`api_token`** _(string, required)_: The API token.
+  - **`consume_folder`** _(object)_: Send the final PDF to Paperless using the consume folder.
+    - **`enabled`** _(boolean)_: Enable using the consume folder. Default: `true`.
diff --git a/pyproject.toml b/pyproject.toml
@@ -64,6 +64,7 @@ typing-extensions = { version = "4.7.1", optional = true }
 Jinja2 = { version = "3.1.2", optional = true }
 natsort = { version = "8.4.0", optional = true }
 nbformat = { version = "5.9.2", optional = true }
+requests = { version = "2.31.0", optional = true }
 
 [tool.poetry.extras]
 process = [
@@ -82,7 +83,8 @@ process = [
     "typing-extensions",
     "Jinja2",
     "natsort",
-    "nbformat"
+    "nbformat",
+    "requests"
 ]
 
 [tool.poetry.group.dev.dependencies]
@@ -94,6 +96,7 @@ pytest-rerunfailures = "12.0"
 pyroma = "4.2"
 typing-extensions = "4.7.1"
 c2cwsgiutils = { version = "6.0.0.dev142", extras = ["test_images"] }
+types-requests = "2.31.0.2"
 
 [build-system]
 requires = ["poetry-core>=1.0.0", "poetry-dynamic-versioning", "poetry-plugin-tweak-dependencies-version"]

diff --git a/scan_to_paperless/config.py b/scan_to_paperless/config.py
@@ -4,6 +4,8 @@
 
 from typing import TypedDict, Union
 
+from typing_extensions import Required
+
 APPEND_CREDIT_CARD_DEFAULT = False
 """ Default value of the field path 'Arguments append_credit_card' """
 
@@ -163,13 +165,15 @@ class Arguments(TypedDict, total=False):
     """
     No REMOVE_TO_CONTINUE.
 
-    Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the pdf.
+    Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the PDF.
 
     default: False
     """
 
     deskew: "_ArgumentsDeskew"
     rule: "Rule"
+    rest_upload: "RestUpload"
+    consume_folder: "ConsumeFolder"
 
 
 class AutoCut(TypedDict, total=False):
@@ -334,6 +338,10 @@ class AutoRotate(TypedDict, total=False):
 """ Default value of the field path 'Arguments colors' """
 
 
+CONSUME_FOLDER_ENABLED_DEFAULT = True
+""" Default value of the field path 'Consume folder enabled' """
+
+
 CONTOUR_KERNEL_SIZE_DEFAULT = 1.5
 """ Default value of the field path 'Contour contour_kernel_size' """
 
@@ -424,6 +432,23 @@ class Configuration(TypedDict, total=False):
     """
 
 
+class ConsumeFolder(TypedDict, total=False):
+    """
+    Consume folder.
+
+    Send the final PDF to Paperless using the consume folder
+    """
+
+    enabled: bool
+    """
+    Consume folder enabled.
+
+    Enable using the consume folder
+
+    default: True
+    """
+
+
 class Contour(TypedDict, total=False):
     """
     Contour.
@@ -1019,6 +1044,10 @@ class Ps2Pdf(TypedDict, total=False):
     """
 
 
+REST_UPLOAD_ENABLED_DEFAULT = False
+""" Default value of the field path 'REST upload enabled' """
+
+
 ROTATE_EVEN_DEFAULT = False
 """ Default value of the field path 'Mode rotate_even' """
 
@@ -1075,6 +1104,41 @@ class Ps2Pdf(TypedDict, total=False):
 """ Default value of the field path 'Rule minor_graduation_space' """
 
 
+class RestUpload(TypedDict, total=False):
+    """
+    REST upload.
+
+    Upload the final PDF via Paperless REST API
+    """
+
+    enabled: bool
+    """
+    REST upload enabled.
+
+    Enable the upload of the PDF via REST API
+
+    default: False
+    """
+
+    api_url: Required[str]
+    """
+    REST upload API url.
+
+    The URL address of the REST API, usually http://server.name/api
+
+    Required property
+    """
+
+    api_token: Required[str]
+    """
+    REST upload API token.
+
+    The API token
+
+    Required property
+    """
+
+
 class Rule(TypedDict, total=False):
     """
     Rule.

diff --git a/scan_to_paperless/config_schema.json b/scan_to_paperless/config_schema.json
@@ -443,7 +443,7 @@
         "no_remove_to_continue": {
           "type": "boolean",
           "default": false,
-          "description": "Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the pdf.",
+          "description": "Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the PDF.",
           "title": "No REMOVE_TO_CONTINUE"
         },
         "deskew": {
@@ -488,7 +488,6 @@
             }
           }
         },
-
         "rule": {
           "type": "object",
           "title": "Rule",
@@ -569,6 +568,43 @@
               "default": 6
             }
           }
+        },
+        "rest_upload": {
+          "type": "object",
+          "title": "REST upload",
+          "description": "Upload the final PDF via Paperless REST API",
+          "properties": {
+            "enabled": {
+              "type": "boolean",
+              "default": false,
+              "description": "Enable the upload of the PDF via REST API",
+              "title": "REST upload enabled"
+            },
+            "api_url": {
+              "type": "string",
+              "description": "The URL address of the REST API, usually http://server.name/api",
+              "title": "REST upload API url"
+            },
+            "api_token": {
+              "type": "string",
+              "description": "The API token",
+              "title": "REST upload API token"
+            }
+          },
+          "required": ["api_url", "api_token"]
+        },
+        "consume_folder": {
+          "type": "object",
+          "title": "Consume folder",
+          "description": "Send the final PDF to Paperless using the consume folder",
+          "properties": {
+            "enabled": {
+              "type": "boolean",
+              "default": true,
+              "description": "Enable using the consume folder",
+              "title": "Consume folder enabled"
+            }
+          }
         }
       }
     },

diff --git a/scan_to_paperless/process.py b/scan_to_paperless/process.py
@@ -23,6 +23,7 @@
 import nbformat
 import numpy as np
 import pikepdf
+import requests
 import ruamel.yaml.compat
 from deskew import determine_skew_debug_images
 from PIL import Image, ImageDraw, ImageFont
@@ -2374,7 +2375,34 @@ def finalize(
         if progress:
             call(["cp", temporary_pdf.name, os.path.join(root_folder, f"{count}-pikepdf.pdf")])
             count += 1
-        call(["cp", temporary_pdf.name, destination])
+
+        if (
+            config["args"]
+            .setdefault("consume_folder", {})
+            .setdefault("enabled", schema.CONSUME_FOLDER_ENABLED_DEFAULT)
+        ):
+            call(["cp", temporary_pdf.name, destination])
+        if (
+            config["args"]
+            .setdefault("rest_upload", cast(schema.RestUpload, {}))
+            .setdefault("enabled", schema.REST_UPLOAD_ENABLED_DEFAULT)
+        ):
+            token = config["args"]["rest_upload"]["api_token"]
+            url = config["args"]["rest_upload"]["api_url"]
+            url = f"{url}/documents/post_document/"
+            headers = {"authorization": f"Token {token}"}
+
+            with open(temporary_pdf.name, "rb") as document_file:
+                files = {"document": document_file}
+                title = os.path.basename(root_folder)
+                data = {"title": title}
+                response = requests.post(url, headers=headers, data=data, files=files, timeout=120)
+                if not response.ok:
+                    raise ScanToPaperlessException(
+                        f"Failed ({response.status_code}) upload to "
+                        f"'{url}' with token '{token}'\n{response.text}"
+                    )
+                print(f"Uploaded {temporary_pdf.name} with title {title}")
 
 
 def _process_code(name: str) -> None:
@@ -2519,7 +2547,7 @@ def _process(
         trace = traceback.format_exc()
 
         out = {"error": str(exception), "traceback": trace.split("\n")}
-        for attribute in ("returncode", "cmd"):
+        for attribute in ("returncode", "cmd", "description", "error_text"):
             if hasattr(exception, attribute):
                 out[attribute] = getattr(exception, attribute)
         for attribute in ("output", "stdout", "stderr"):

diff --git a/scan_to_paperless/process_schema.json b/scan_to_paperless/process_schema.json
@@ -485,7 +485,7 @@
         "no_remove_to_continue": {
           "type": "boolean",
           "default": false,
-          "description": "Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the pdf.",
+          "description": "Don't wait for the deletion of the REMOVE_TO_CONTINUE file before exporting the PDF.",
           "title": "No REMOVE_TO_CONTINUE"
         },
         "deskew": {
@@ -610,6 +610,43 @@
               "default": 6
             }
           }
+        },
+        "rest_upload": {
+          "type": "object",
+          "title": "REST upload",
+          "description": "Upload the final PDF via Paperless REST API",
+          "properties": {
+            "enabled": {
+              "type": "boolean",
+              "default": false,
+              "description": "Enable the upload of the PDF via REST API",
+              "title": "REST upload enabled"
+            },
+            "api_url": {
+              "type": "string",
+              "description": "The URL address of the REST API, usually http://server.name/api",
+              "title": "REST upload API url"
+            },
+            "api_token": {
+              "type": "string",
+              "description": "The API token",
+              "title": "REST upload API token"
+            }
+          },
+          "required": ["api_url", "api_token"]
+        },
+        "consume_folder": {
+          "type": "object",
+          "title": "Consume folder",
+          "description": "Send the final PDF to Paperless using the consume folder",
+          "properties": {
+            "enabled": {
+              "type": "boolean",
+              "default": true,
+              "description": "Enable using the consume folder",
+              "title": "Consume folder enabled"
+            }
+          }
         }
       }
     }