Merge pull request #351 from edenai/SD2-1394-fix-mistral-ocr-async

[Fix] bounding box optional for ocr async

Author: Daggx

edenai_apis/apis/mistral/mistral_api.py

@@ -343,12 +343,12 @@ def ocr__ocr_async__get_job_result(
         pages = []
         for page in response_data["pages"]:
             raw_text += page["markdown"]
-            # markdown_lines = page["markdown"].split("\n")
-            # lines = []
-            # for line_text in markdown_lines:
-            #     line = Line(text=line_text, bounding_box=BoundingBox())
-            #     lines.append(line)
-            # pages.append(Page(lines=lines))
+            markdown_lines = page["markdown"].split("\n")
+            lines = []
+            for line_text in markdown_lines:
+                line = Line(text=line_text, confidence=100)
+                lines.append(line)
+            pages.append(Page(lines=lines))
 
         return AsyncResponseType(
             original_response=response_data,

edenai_apis/features/ocr/ocr_async/ocr_async_dataclass.py

@@ -187,8 +187,8 @@ class Line(BaseModel):
 
     text: str = Field(description="Text detected in the line")
     words: Sequence[Word] = Field(default_factory=list, description="List of words")
-    bounding_box: BoundingBox = Field(
-        description="Bounding boxes of the words in the line"
+    bounding_box: Optional[BoundingBox] = Field(
+        default=None, description="Bounding box of the line, can be None"
     )
     confidence: Optional[float] = Field(..., description="Confidence of the line")