added json block formatting removal

setup.py

@@ -10,7 +10,7 @@ def read_git_requirements(file):
 
 setup(
     name='thepipe_api',
-    version='1.2.5',
+    version='1.2.6',
     author='Emmett McFarlane',
     author_email='[email protected]',
     description='AI-native extractor, powered by multimodal LLMs.',

tests/test_extractor.py

@@ -60,7 +60,7 @@ def test_extract_json_from_response(self):
 
     def test_extract(self):
         results, total_tokens_used = extract(
-            chunks=self.chunks,
+            chunks=self.chunks, # receipt
             schema=self.schema,
         )
 

thepipe/extract.py

@@ -14,14 +14,21 @@
 
 def extract_json_from_response(llm_response: str) -> Union[Dict, List[Dict], None]:
     def clean_response_text(llm_response: str) -> str:
-        return llm_response.encode('utf-8', 'ignore').decode('utf-8')
+        return llm_response.encode('utf-8', 'ignore').decode('utf-8').strip()
 
-    llm_response = llm_response.strip()
+    # try to match inside of code block
     code_block_pattern = r'^```(?:json)?\s*([\s\S]*?)\s*```$'
     match = re.match(code_block_pattern, llm_response, re.MULTILINE | re.DOTALL)
     if match:
-        llm_response = match.group(1).strip()
+        llm_response = match.group(1)
     llm_response = clean_response_text(llm_response)
+
+    # try to remove code block formatting if still present
+    if llm_response.startswith("```json") and llm_response.endswith("```"):
+        llm_response = llm_response[len("```json"):-len("```")]
+    llm_response = clean_response_text(llm_response)
+
+    # parse json by matching curly braces
     try:
         parsed_json = json.loads(llm_response)
         return parsed_json