1

backend/app/services/providers/deepgram_client.py

@@ -0,0 +1,52 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+import requests
+
+from backend.app.config import settings
+
+
+class DeepgramClient:
+    endpoint = "https://api.deepgram.com/v1/listen?model=nova-3&smart_format=true&language=es"
+
+    def transcribe_file(self, path: Path) -> dict:
+        if not settings.deepgram_api_key:
+            raise RuntimeError("DEEPGRAM_API_KEY no esta configurada")
+
+        headers = {
+            "Authorization": f"Token {settings.deepgram_api_key}",
+            "Content-Type": self._content_type(path),
+        }
+        with path.open("rb") as audio_file:
+            response = requests.post(self.endpoint, headers=headers, data=audio_file, timeout=600)
+
+        if response.status_code == 401:
+            raise RuntimeError("API Key de Deepgram invalida")
+        if not response.ok:
+            try:
+                error = response.json()
+            except ValueError:
+                error = response.text
+            raise RuntimeError(f"Deepgram error {response.status_code}: {error}")
+        return response.json()
+
+    def transcript_from_response(self, data: dict) -> tuple[str, float | None]:
+        alternative = data.get("results", {}).get("channels", [{}])[0].get("alternatives", [{}])[0]
+        transcript = alternative.get("paragraphs", {}).get("transcript") or alternative.get("transcript") or ""
+        confidence = alternative.get("confidence")
+        if not transcript.strip():
+            raise RuntimeError("Deepgram no devolvio transcripcion")
+        return transcript.strip(), confidence
+
+    def _content_type(self, path: Path) -> str:
+        ext = path.suffix.lower()
+        return {
+            ".mp3": "audio/mpeg",
+            ".wav": "audio/wav",
+            ".m4a": "audio/mp4",
+            ".mp4": "audio/mp4",
+            ".webm": "audio/webm",
+            ".ogg": "audio/ogg",
+            ".flac": "audio/flac",
+        }.get(ext, "application/octet-stream")

backend/app/services/providers/mistral_client.py

@@ -0,0 +1,52 @@
+from __future__ import annotations
+
+import base64
+from pathlib import Path
+
+import requests
+
+from backend.app.config import settings
+
+
+class MistralClient:
+    endpoint = "https://api.mistral.ai/v1/chat/completions"
+    model = "pixtral-12b-2409"
+
+    def ocr_image(self, image_path: Path) -> str:
+        if not settings.mistral_api_key:
+            raise RuntimeError("MISTRAL_API_KEY no esta configurada")
+        data_url = self._image_data_url(image_path)
+        response = requests.post(
+            self.endpoint,
+            headers={
+                "Content-Type": "application/json",
+                "Authorization": f"Bearer {settings.mistral_api_key}",
+            },
+            json={
+                "model": self.model,
+                "messages": [
+                    {
+                        "role": "user",
+                        "content": [
+                            {
+                                "type": "text",
+                                "text": "Extract all text from this image and format it as Markdown. If there are tables, represent them using Markdown table syntax. Do not include commentary.",
+                            },
+                            {"type": "image_url", "image_url": data_url},
+                        ],
+                    }
+                ],
+                "max_tokens": 3000,
+            },
+            timeout=180,
+        )
+        if response.status_code == 401:
+            raise RuntimeError("API Key de Mistral invalida")
+        if not response.ok:
+            raise RuntimeError(f"Mistral error {response.status_code}: {response.text}")
+        data = response.json()
+        return data["choices"][0]["message"]["content"]
+
+    def _image_data_url(self, image_path: Path) -> str:
+        encoded = base64.b64encode(image_path.read_bytes()).decode("ascii")
+        return f"data:image/jpeg;base64,{encoded}"