def load_app_features_context(config: AppConfig,
download_manager: DownloadManager):
return AppFeaturesContext(country_lookup=load_lookup_from_config(
config.get('lookup', {}).get('country'),
download_manager=download_manager),
first_name_lookup=load_lookup_from_config(
config.get('lookup', {}).get('first_name'),
download_manager=download_manager),
last_name_lookup=load_lookup_from_config(
config.get('lookup', {}).get('last_name'),
download_manager=download_manager))
def get_ocr_model_for_app_config(
app_config: AppConfig,
enabled: bool = True) -> Optional[OpticalCharacterRecognitionModel]:
ocr_model_config = app_config.get('ocr_models', {}).get('default')
if enabled and ocr_model_config:
return get_lazy_ocr_model_for_config(ocr_model_config)
return None
def get_cv_model_for_app_config(
app_config: AppConfig,
enabled: bool = True) -> Optional[ComputerVisionModel]:
cv_model_config = app_config.get('cv_models', {}).get('default')
if enabled and cv_model_config:
return get_lazy_cv_model_for_config(cv_model_config)
return None
def __init__(self, config: AppConfig):
self.config = config
self.download_manager = DownloadManager(
download_dir=get_download_dir(config))
self.pdfalto_wrapper = PdfAltoWrapper(
self.download_manager.download_if_url(config['pdfalto']['path']))
self.pdfalto_wrapper.ensure_executable()
self.app_context = AppContext(
app_config=config,
download_manager=self.download_manager,
lazy_wapiti_binary_wrapper=LazyWapitiBinaryWrapper(
install_url=config.get('wapiti', {}).get('install_source'),
download_manager=self.download_manager))
self.fulltext_processor_config = FullTextProcessorConfig.from_app_config(
app_config=config)
self.fulltext_models = load_models(
config,
app_context=self.app_context,
fulltext_processor_config=self.fulltext_processor_config)
if config.get('preload_on_startup'):
self.fulltext_models.preload()
self.app_features_context = load_app_features_context(
config, download_manager=self.download_manager)
tei_to_jats_config = config.get('xslt', {}).get('tei_to_jats', {})
self.tei_to_jats_xslt_transformer = XsltTransformerWrapper.from_template_file(
TEI_TO_JATS_XSLT_FILE,
xslt_template_parameters=tei_to_jats_config.get('parameters', {}))
self.doc_to_pdf_enabled = config.get('doc_to_pdf',
{}).get('enabled', True)
self.doc_to_pdf_convert_parameters = config.get('doc_to_pdf',
{}).get('convert', {})
self.doc_converter_wrapper = DocConverterWrapper(
**config.get('doc_to_pdf', {}).get('listener', {}))
def from_app_config(app_config: AppConfig) -> 'FullTextProcessorConfig':
return FullTextProcessorConfig()._replace(
**app_config.get('processors', {}).get('fulltext', {}))