Spaces:
Running
Running
:/
Browse files
app.py
CHANGED
@@ -3,13 +3,12 @@ import base64
|
|
3 |
import gradio as gr
|
4 |
from mistralai import Mistral
|
5 |
from mistralai.models import OCRResponse
|
6 |
-
from mistralai.exceptions import MistralException
|
7 |
from pathlib import Path
|
8 |
from pydantic import BaseModel
|
9 |
import pycountry
|
10 |
import json
|
11 |
import logging
|
12 |
-
from tenacity import retry, stop_after_attempt, wait_fixed
|
13 |
import tempfile
|
14 |
from typing import Union, Dict, List
|
15 |
from contextlib import contextmanager
|
@@ -32,7 +31,7 @@ class OCRProcessor:
|
|
32 |
self.client = Mistral(api_key=self.api_key)
|
33 |
try:
|
34 |
self.client.models.list() # Validate API key
|
35 |
-
except
|
36 |
raise ValueError(f"Invalid API key: {str(e)}")
|
37 |
|
38 |
@staticmethod
|
@@ -52,19 +51,19 @@ class OCRProcessor:
|
|
52 |
if os.path.exists(temp_file.name):
|
53 |
os.unlink(temp_file.name)
|
54 |
|
55 |
-
@retry(stop=stop_after_attempt(3), wait=wait_fixed(2)
|
56 |
def _call_ocr_api(self, document: Dict) -> OCRResponse:
|
57 |
try:
|
58 |
return self.client.ocr.process(model="mistral-ocr-latest", document=document)
|
59 |
-
except
|
60 |
logger.error(f"OCR API call failed: {str(e)}")
|
61 |
raise
|
62 |
|
63 |
-
@retry(stop=stop_after_attempt(3), wait=wait_fixed(2)
|
64 |
def _call_chat_complete(self, model: str, messages: List[Dict], **kwargs) -> Dict:
|
65 |
try:
|
66 |
return self.client.chat.complete(model=model, messages=messages, **kwargs)
|
67 |
-
except
|
68 |
logger.error(f"Chat complete API call failed: {str(e)}")
|
69 |
raise
|
70 |
|
@@ -176,7 +175,7 @@ class OCRProcessor:
|
|
176 |
def _format_structured_response(file_path: str, content: Dict) -> str:
|
177 |
languages = {lang.alpha_2: lang.name for lang in pycountry.languages if hasattr(lang, 'alpha_2')}
|
178 |
valid_langs = [l for l in content.get("languages", [DEFAULT_LANGUAGE]) if l in languages.values()]
|
179 |
-
|
180 |
response = {
|
181 |
"file_name": Path(file_path).name,
|
182 |
"topics": content.get("topics", []),
|
@@ -195,7 +194,7 @@ def create_interface():
|
|
195 |
placeholder="Enter your Mistral API key here",
|
196 |
type="password"
|
197 |
)
|
198 |
-
|
199 |
def initialize_processor(api_key):
|
200 |
try:
|
201 |
processor = OCRProcessor(api_key)
|
|
|
3 |
import gradio as gr
|
4 |
from mistralai import Mistral
|
5 |
from mistralai.models import OCRResponse
|
|
|
6 |
from pathlib import Path
|
7 |
from pydantic import BaseModel
|
8 |
import pycountry
|
9 |
import json
|
10 |
import logging
|
11 |
+
from tenacity import retry, stop_after_attempt, wait_fixed
|
12 |
import tempfile
|
13 |
from typing import Union, Dict, List
|
14 |
from contextlib import contextmanager
|
|
|
31 |
self.client = Mistral(api_key=self.api_key)
|
32 |
try:
|
33 |
self.client.models.list() # Validate API key
|
34 |
+
except Exception as e:
|
35 |
raise ValueError(f"Invalid API key: {str(e)}")
|
36 |
|
37 |
@staticmethod
|
|
|
51 |
if os.path.exists(temp_file.name):
|
52 |
os.unlink(temp_file.name)
|
53 |
|
54 |
+
@retry(stop=stop_after_attempt(3), wait=wait_fixed(2))
|
55 |
def _call_ocr_api(self, document: Dict) -> OCRResponse:
|
56 |
try:
|
57 |
return self.client.ocr.process(model="mistral-ocr-latest", document=document)
|
58 |
+
except Exception as e:
|
59 |
logger.error(f"OCR API call failed: {str(e)}")
|
60 |
raise
|
61 |
|
62 |
+
@retry(stop=stop_after_attempt(3), wait=wait_fixed(2))
|
63 |
def _call_chat_complete(self, model: str, messages: List[Dict], **kwargs) -> Dict:
|
64 |
try:
|
65 |
return self.client.chat.complete(model=model, messages=messages, **kwargs)
|
66 |
+
except Exception as e:
|
67 |
logger.error(f"Chat complete API call failed: {str(e)}")
|
68 |
raise
|
69 |
|
|
|
175 |
def _format_structured_response(file_path: str, content: Dict) -> str:
|
176 |
languages = {lang.alpha_2: lang.name for lang in pycountry.languages if hasattr(lang, 'alpha_2')}
|
177 |
valid_langs = [l for l in content.get("languages", [DEFAULT_LANGUAGE]) if l in languages.values()]
|
178 |
+
|
179 |
response = {
|
180 |
"file_name": Path(file_path).name,
|
181 |
"topics": content.get("topics", []),
|
|
|
194 |
placeholder="Enter your Mistral API key here",
|
195 |
type="password"
|
196 |
)
|
197 |
+
|
198 |
def initialize_processor(api_key):
|
199 |
try:
|
200 |
processor = OCRProcessor(api_key)
|