jjdZddlmZmZmZddlZddlmZddlm Z m Z m Z m Z ddl mZGdde Zy) z, Mistral OCR transformation implementation. )AnyDictOptionalN)verbose_logger) BaseOCRConfig DocumentTypeOCRRequestData OCRResponse)get_secret_strc eZdZdZdfd ZdedefdZdedededefdZ dd e ded e ed e ed e ede f d Z dd e ededed e edef dZ dededed edef dZdedej$dedefdZxZS)MistralOCRConfigzl Mistral OCR transformation configuration. Reference: https://docs.mistral.ai/api/#tag/ocr returnc"t|yN)super__init__)self __class__s /Users/manta/Documents/Projects/TheRoad-I1/backend/.venv/lib/python3.12/site-packages/litellm/llms/mistral/ocr/transformation.pyrzMistralOCRConfig.__init__s modelc gdS)a Get supported OCR parameters for Mistral OCR. Mistral OCR supports: - pages: List of page numbers to process - include_image_base64: Whether to include base64 encoded images - image_limit: Maximum number of images to return - image_min_size: Minimum size of images to include - bbox_annotation_format: Format for bounding box annotations - document_annotation_format: Format for document annotations )pagesinclude_image_base64 image_limitimage_min_sizebbox_annotation_formatdocument_annotation_format)rrs rget_supported_ocr_paramsz)MistralOCRConfig.get_supported_ocr_paramss   rnon_default_paramsoptional_paramscr|j|}i}|jD]\}}||vs |||<|S)z Map OCR parameters to Mistral-specific format. Mistral accepts these parameters directly, so no transformation needed. Just filter out unsupported params. )r)r items)rr!r"rsupported_params mapped_paramsparamvalues rmap_ocr_paramszMistralOCRConfig.map_ocr_params1sQ 88u8E .446LE5((', e$7rheadersapi_keyapi_baselitellm_paramsc L| td}| tddd|i|}|S)zJ Validate environment and return headers for Mistral OCR. MISTRAL_API_KEYz}Missing Mistral API Key - A call is being made to Mistral but no key is set either in the environment variables or via params AuthorizationzBearer )r ValueError)rr*rr+r,r-kwargss rvalidate_environmentz%MistralOCRConfig.validate_environmentGsU ?01  ?P  wwi0  rc b|d}|jd}|jdr|dS|dS)zt Get complete URL for Mistral OCR endpoint. Returns: https://api.mistral.ai/v1/ocr zhttps://api.mistral.ai/v1/z/v1z/ocrz/v1/ocr)rstripendswith)rr,rr"r-r2s rget_complete_urlz!MistralOCRConfig.get_complete_urlgsI  2H??3'   U #Zt$ $7##rdocumentc tjd|t|tst dt |||d}|j |t|dS)a Transform OCR request to Mistral-specific format. Mistral OCR API accepts: { "model": "mistral-ocr-latest", "document": { "type": "document_url", "document_url": "" }, "pages": [0], # optional "include_image_base64": false, # optional ... } Args: model: Model name (e.g., "mistral-ocr-latest") document: Document dict from user (Mistral format) - already validated in main.py optional_params: Already mapped optional parameters headers: Request headers Returns: OCRRequestData with JSON data z+Mistral OCR transform_ocr_request - model: zExpected document dict, got )rr9N)datafiles)rdebug isinstancedictr1typeupdater )rrr9r"r*r2r;s rtransform_ocr_requestz&MistralOCRConfig.transform_ocr_requestsk@ J5'RS(D);DN;KLM M   O$4t44r raw_response logging_objc b |j}tjd|jt |j dg|j d||j d|j ddS#t $r}tjd||d }~wwxYw) a Return Mistral OCR response in native format. Mistral OCR is the standard format for LiteLLM OCR responses. No transformation needed - return native response. Mistral OCR returns: { "pages": [ { "index": 0, "markdown": "extracted text content", "images": [...], "dimensions": {...} }, ... ], "model": "mistral-ocr-2505-completion", "document_annotation": null, "usage_info": {...} } zMistral OCR response keys: rrdocument_annotation usage_infoocr)rrrFrGobjectz$Error parsing Mistral OCR response: N)jsonrr=keysr get Exceptionerror)rrrCrDr2 response_jsones rtransform_ocr_responsez'MistralOCRConfig.transform_ocr_responses: (--/M  #>}?Q?Q?S>T!U V#''4#''7$1$5$56K$L(,,\:     #Gs!K LG sBB B.B))B.)rN)NNNr)__name__ __module__ __qualname____doc__rstrlistr r?r)rrr3r8rr rBhttpxResponserr rQ __classcell__)rs@rr r sI  c d *    4"&"&)- #  3-  !  J*. $3-$$ $ ! $ $4151515 15  15 15f,,nn, ,  ,rr )rUtypingrrrrXlitellm._loggingr(litellm.llms.base_llm.ocr.transformationrrr r litellm.secret_managers.mainr r rrrr_s5'& + 8N}Nr