Merge pull request #778 from sbrunner/jsonschema-default

Use the default from the jsonschema

README.md

@@ -151,7 +151,7 @@ the images.
 
 ### Add a mask
 
-If your scanner add some margin around the scanned image it will relay case some issue the deskew and the
+If your scanner add some margin around the scanned image it will relay case some issue the skew and the
 content detection.
 
 To solve that you can add a black and white image named `mask.png` in the root folder and draw in black the

config.md

@@ -19,12 +19,12 @@
 - **`extension`** _(string)_: The extension of generate image (png or tiff). Default: `png`.
 - **`default_args`**: Refer to _#/definitions/args_.
 - **`viewer`** _(string)_: The command used to start the viewer. Default: `eog`.
-- **`modes`** _(object)_: Customize the modes. Can contain additional properties.
+- **`modes`** _(object)_: Customize the modes. Can contain additional properties. Default: `{'adf': {'scanimage_arguments': ['--source=ADF']}, 'double': {'scanimage_arguments': ['--source=ADF'], 'auto_bash': True, 'rotate_even': True}, 'multi': {'scanimage_arguments': ['--batch-prompt']}, 'one': {'scanimage_arguments': ['--batch-count=1']}}`.
   - **Additional Properties** _(object)_
     - **`scanimage_arguments`** _(array)_: Additional scanimage arguments.
       - **Items** _(string)_
-    - **`auto_bash`** _(boolean)_: Run the ADF in tow step odd and even, needed for scanner that don't support double face.
-    - **`rotate_even`** _(boolean)_: Rotate the even pages, to use in conjunction with auto_bash.
+    - **`auto_bash`** _(boolean)_: Run the ADF in tow step odd and even, needed for scanner that don't support double face. Default: `False`.
+    - **`rotate_even`** _(boolean)_: Rotate the even pages, to use in conjunction with auto_bash. Default: `False`.
 
 ## Definitions
 
@@ -35,16 +35,16 @@
     - **Items** _(integer)_
   - **`de_noise_morphology`** _(boolean)_: Apply a morphology operation to remove noise. Default: `True`.
   - **`inverse_mask`** _(boolean)_: Inverse the mask. Default: `False`.
-  - **`de_noise_size`** _(integer)_: The size of the artifact that will be de noise. Default: `20`.
+  - **`de_noise_size`** _(integer)_: The size of the artifact that will be de noise. Default: `1000`.
   - **`de_noise_level`** _(integer)_: The threshold level used in de noise on the blurry image. Default: `220`.
-  - **`buffer_size`** _(integer)_: The size of the buffer add on the mask. Default: `50 an case of mask, 20 in case of cut`.
+  - **`buffer_size`** _(integer)_: The size of the buffer add on the mask. Default: `20`.
   - **`buffer_level`** _(integer)_: The threshold level used in buffer on the blurry image. Default: `20`.
   - **`additional_filename`** _(string)_: An image file used to add on the mask.
 - **`args`** _(object)_: Cannot contain additional properties.
-  - **`level`** _(['boolean', 'integer'])_: true: => do level on 15% - 85% (under 15 % will be black above 85% will be white), false: => 0% - 100%, <number>: => (0 + <number>)% - (100 - number)%.
+  - **`level`** _(['boolean', 'integer'])_: true: => do level on 15% - 85% (under 15 % will be black above 85% will be white), false: => 0% - 100%, <number>: => (0 + <number>)% - (100 - number)%. Default: `False`.
   - **`auto_level`** _(boolean)_: If no level specified, do auto level. Default: `False`.
   - **`min_level`** _(number)_: Min level if no level end no auto-level. Default: `15`.
-  - **`max_level`** _(number)_: Max level if no level end no auto-level. Default: `15`.
+  - **`max_level`** _(number)_: Max level if no level end no auto-level. Default: `85`.
   - **`cut_white`** _(number)_: Set the near white pixels on the image to white. Default: `250`.
   - **`cut_black`** _(number)_: Set the near black pixels on the image to black. Default: `0`.
   - **`no_crop`** _(boolean)_: Don't do any crop. Default: `False`.
@@ -53,7 +53,7 @@
   - **`dpi`** _(number)_: The DPI used to convert the mm to pixel. Default: `300`.
   - **`sharpen`** _(boolean)_: Do the sharpen. Default: `False`.
   - **`dither`** _(boolean)_: Do the dither. Default: `False`.
-  - **`tesseract`** _(boolean)_: Use tesseract to to an OCR on the document. Default: `False`.
+  - **`tesseract`** _(boolean)_: Use tesseract to to an OCR on the document. Default: `True`.
   - **`tesseract_lang`** _(string)_: The used language for tesseract. Default: `fra+eng`.
   - **`append_credit_card`** _(boolean)_: Do an assisted split. Default: `False`.
   - **`assisted_split`** _(boolean)_: Do an assisted split. Default: `False`.
@@ -67,7 +67,7 @@
   - **`contour_kernel_size_empty`** _(number)_: The block size used in a box on content find if the page is empty [mm]. Default: `1.5`.
   - **`threshold_block_size_empty`** _(number)_: The block size used in a box on threshold for content find if the page is empty [mm]. Default: `1.5`.
   - **`threshold_value_c_empty`** _(number)_: A variable used on threshold, should be low on low contrast image, used in a box on content find if the page is empty. Default: `70`.
-  - **`min_box_size_limit`** _(number)_: The minimum box size to find the limits based on content [mm]. Default: `3`.
+  - **`min_box_size_limit`** _(number)_: The minimum box size to find the limits based on content [mm]. Default: `10`.
   - **`min_box_black_limit`** _(number)_: The minimum black in a box on content find the limits based on content [%]. Default: `2`.
   - **`contour_kernel_size_limit`** _(number)_: The block size used in a box on content find the limits based on content [mm]. Default: `1.5`.
   - **`threshold_block_size_limit`** _(number)_: The block size used in a box on threshold for content find the limits based on content [mm]. Default: `1.5`.
@@ -83,7 +83,7 @@
   - **`jpeg_quality`** _(integer)_: The JPEG quality. Default: `90`.
   - **`background_color`** _(array)_: The background color. Default: `[255, 255, 255]`.
     - **Items** _(integer)_
-  - **`auto_mask`**: The auto mask configuration, the mask is used to mask the image on crop and deskew calculation. Refer to _#/definitions/auto_mask_.
+  - **`auto_mask`**: The auto mask configuration, the mask is used to mask the image on crop and skew calculation. Refer to _#/definitions/auto_mask_.
   - **`auto_cut`**: The auto mask configuration, the mask is used to definitively mask the source image. Refer to _#/definitions/auto_mask_.
   - **`deskew_min_angle`** _(number)_: The minimum angle to detect the image skew [degree]. Default: `-10`.
   - **`deskew_max_angle`** _(number)_: The maximum angle to detect the image skew [degree]. Default: `10`.

process.md

@@ -53,16 +53,16 @@
     - **Items** _(integer)_
   - **`de_noise_morphology`** _(boolean)_: Apply a morphology operation to remove noise. Default: `True`.
   - **`inverse_mask`** _(boolean)_: Inverse the mask. Default: `False`.
-  - **`de_noise_size`** _(integer)_: The size of the artifact that will be de noise. Default: `20`.
+  - **`de_noise_size`** _(integer)_: The size of the artifact that will be de noise. Default: `1000`.
   - **`de_noise_level`** _(integer)_: The threshold level used in de noise on the blurry image. Default: `220`.
-  - **`buffer_size`** _(integer)_: The size of the buffer add on the mask. Default: `50 an case of mask, 20 in case of cut`.
+  - **`buffer_size`** _(integer)_: The size of the buffer add on the mask. Default: `20`.
   - **`buffer_level`** _(integer)_: The threshold level used in buffer on the blurry image. Default: `20`.
   - **`additional_filename`** _(string)_: An image file used to add on the mask.
 - **`args`** _(object)_
-  - **`level`** _(['boolean', 'integer'])_: true: => do level on 15% - 85% (under 15 % will be black above 85% will be white), false: => 0% - 100%, <number>: => (0 + <number>)% - (100 - number)%.
+  - **`level`** _(['boolean', 'integer'])_: true: => do level on 15% - 85% (under 15 % will be black above 85% will be white), false: => 0% - 100%, <number>: => (0 + <number>)% - (100 - number)%. Default: `False`.
   - **`auto_level`** _(boolean)_: If no level specified, do auto level. Default: `False`.
   - **`min_level`** _(number)_: Min level if no level end no auto-level. Default: `15`.
-  - **`max_level`** _(number)_: Max level if no level end no auto-level. Default: `15`.
+  - **`max_level`** _(number)_: Max level if no level end no auto-level. Default: `85`.
   - **`cut_white`** _(number)_: Set the near white pixels on the image to white. Default: `250`.
   - **`cut_black`** _(number)_: Set the near black pixels on the image to black. Default: `0`.
   - **`no_crop`** _(boolean)_: Don't do any crop. Default: `False`.
@@ -71,7 +71,7 @@
   - **`dpi`** _(number)_: The DPI used to convert the mm to pixel. Default: `300`.
   - **`sharpen`** _(boolean)_: Do the sharpen. Default: `False`.
   - **`dither`** _(boolean)_: Do the dither. Default: `False`.
-  - **`tesseract`** _(boolean)_: Use tesseract to to an OCR on the document. Default: `False`.
+  - **`tesseract`** _(boolean)_: Use tesseract to to an OCR on the document. Default: `True`.
   - **`tesseract_lang`** _(string)_: The used language for tesseract. Default: `fra+eng`.
   - **`append_credit_card`** _(boolean)_: Do an assisted split. Default: `False`.
   - **`assisted_split`** _(boolean)_: Do an assisted split. Default: `False`.
@@ -85,7 +85,7 @@
   - **`contour_kernel_size_empty`** _(number)_: The block size used in a box on content find if the page is empty [mm]. Default: `1.5`.
   - **`threshold_block_size_empty`** _(number)_: The block size used in a box on threshold for content find if the page is empty [mm]. Default: `1.5`.
   - **`threshold_value_c_empty`** _(number)_: A variable used on threshold, should be low on low contrast image, used in a box on content find if the page is empty. Default: `70`.
-  - **`min_box_size_limit`** _(number)_: The minimum box size to find the limits based on content [mm]. Default: `3`.
+  - **`min_box_size_limit`** _(number)_: The minimum box size to find the limits based on content [mm]. Default: `10`.
   - **`min_box_black_limit`** _(number)_: The minimum black in a box on content find the limits based on content [%]. Default: `2`.
   - **`contour_kernel_size_limit`** _(number)_: The block size used in a box on content find the limits based on content [mm]. Default: `1.5`.
   - **`threshold_block_size_limit`** _(number)_: The block size used in a box on threshold for content find the limits based on content [mm]. Default: `1.5`.
@@ -101,7 +101,7 @@
   - **`jpeg_quality`** _(integer)_: The JPEG quality. Default: `90`.
   - **`background_color`** _(array)_: The background color. Default: `[255, 255, 255]`.
     - **Items** _(integer)_
-  - **`auto_mask`**: The auto mask configuration, the mask is used to mask the image on crop and deskew calculation. Refer to _#/definitions/auto_mask_.
+  - **`auto_mask`**: The auto mask configuration, the mask is used to mask the image on crop and skew calculation. Refer to _#/definitions/auto_mask_.
   - **`auto_cut`**: The auto mask configuration, the mask is used to definitively mask the source image. Refer to _#/definitions/auto_mask_.
   - **`deskew_min_angle`** _(number)_: The minimum angle to detect the image skew [degree]. Default: `-10`.
   - **`deskew_max_angle`** _(number)_: The maximum angle to detect the image skew [degree]. Default: `10`.

scan_to_paperless/__init__.py

@@ -7,9 +7,9 @@
 from ruamel.yaml.main import YAML
 
 if sys.version_info.minor >= 8:
-    from scan_to_paperless import config as stp_config
+    from scan_to_paperless import config as schema
 else:
-    from scan_to_paperless import config_old as stp_config  # type: ignore
+    from scan_to_paperless import config_old as schema  # type: ignore
 
 CONFIG_FILENAME = "scan-to-paperless.yaml"
 
@@ -23,13 +23,13 @@
 CONFIG_PATH = os.path.join(CONFIG_FOLDER, CONFIG_FILENAME)
 
 
-def get_config(config_filename: str) -> stp_config.Configuration:
+def get_config(config_filename: str) -> schema.Configuration:
     """Get the configuration."""
     if os.path.exists(config_filename):
         yaml = YAML()
         yaml.default_flow_style = False
         with open(config_filename, encoding="utf-8") as config_file:
-            config = cast(stp_config.Configuration, yaml.load(config_file))
+            config = cast(schema.Configuration, yaml.load(config_file))
             if "exteds" in config:
 
                 base_config = get_config(
@@ -38,7 +38,7 @@ def get_config(config_filename: str) -> stp_config.Configuration:
                     )
                 )
 
-                strategies_config = cast(stp_config.MergeStrategies, config.get("strategies", {}))
+                strategies_config = cast(schema.MergeStrategies, config.get("strategies", {}))
                 merger = Merger(
                     [
                         (list, strategies_config.get("list", ["override"])),
@@ -47,7 +47,7 @@ def get_config(config_filename: str) -> stp_config.Configuration:
                     strategies_config.get("fallback", ["merge"]),
                     strategies_config.get("type_conflict", ["merge"]),
                 )
-                config = cast(stp_config.Configuration, merger.merge(base_config, config))
+                config = cast(schema.Configuration, merger.merge(base_config, config))
             return config
     print(f"Missing config file: {config_filename}")
     return {}