Add support for specifying document conversion settings

[cau-git]

• Establish simplified pydantic models to encapsulate conversion settings

Note: Nothing yet functional.

[cau-git]

Some usage examples below.

Case 1: Just enable OCR

import deepsearch as ds
...
api = ds.CpsApi(client)

conv_settings = ConversionSettings.from_defaults(api)
conv_settings.ocr.enabled = True

...
documents = ds.convert_documents(
    api=api,
    proj_key=proj,
    source_path="../data/samples/2206.01062.pdf",
    conversion_settings=conv_settings
)           

Case 2: Enable OCR, and pick a backend

import deepsearch as ds
...
api = ds.CpsApi(client)

conv_settings = ConversionSettings.from_defaults(api)
conv_settings.ocr.enabled = True
conv_settings.ocr.backend = "paddle-ocr" # one of: OCRSettings.get_backends(api)

...
documents = ds.convert_documents(
    api=api,
    proj_key=proj,
    source_path="../data/samples/2206.01062.pdf",
    conversion_settings=conv_settings
)           

Case 3: Just disable table structure model on pipeline

import deepsearch as ds
...
api = ds.CpsApi(client)

conv_settings = ConversionSettings.from_defaults(api)
conv_settings.pipeline.tables = None

...
documents = ds.convert_documents(
    api=api,
    proj_key=proj,
    source_path="../data/samples/2206.01062.pdf",
    conversion_settings=conv_settings
)           

Case 4: Pick project model for table-structure on pipeline

import deepsearch as ds
...
api = ds.CpsApi(client)

conv_settings = ConversionSettings.from_defaults(api)

project_models = ProjectConversionModel.get_models(api, proj_key)

for pm in project_models:
   if pm.name == "my-ts-model-test-1": # basic example: match by name
     conv_settings.pipeline.tables = pm

...
documents = ds.convert_documents(
    api=api,
    proj_key=proj,
    source_path="../data/samples/2206.01062.pdf",
    conversion_settings=conv_settings
)           

[cau-git]

Clean up documentation