added error fixes

app.py

@@ -12,6 +12,7 @@ import time
 from settings import char_remove, gpt_model, RPFAAP2, RPFAAP1, project_id, project_location, processor_id
 from tqdm import tqdm
 import logging
+import google 
 
 logging.basicConfig(filename='app.log', level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
 
@@ -61,8 +62,8 @@ def chat_gpt_document(content, document_type, context):
         document = "RPFAAP2.json"
         desired_format = RPFAAP2
     else: 
-        document = ""
-        desired_format = ""
+        property_info = ["Please Select a Document Type"]
+        return json.dumps(property_info, indent=4)
 
     if context == "": 
         sequence_1 = (document_prefix, content_info, desired_format)
@@ -127,6 +128,7 @@ def detect_image(content, lang):
                 response.error.message))
     
     os.remove(temp_file_path)
+    logging.info(response)
     return(response.full_text_annotation.text)
 
 def detect_document(content):
@@ -182,35 +184,78 @@ def image(content, lang, context):
 def document(content, document_type, context):
     return chat_gpt_document(detect_document(content),document_type,context)
 
-def batch_document(content, document_type, context, progress=gr.Progress()):
-    progress(0, desc="Starting")
-    retries = 5
-    timeout = 5
-    i = 0
-    j = 0 
+unprocessed_documents = []
+global_document_type = None
+global_context = None
+def batch_document(content, document_type, context, progress = gr.Progress()):
+    logging.info(content)
     combined_data = []
-    for x in progress.tqdm(content, desc="Processing"):
-        while True: 
-            k = i+j
-            try: 
-                data = json.loads(chat_gpt_document(detect_document(x),document_type,context))
-                combined_data.append(data)
-                break
-            except openai.error.APIConnectionError:
-                logging.error(f'Retry {k+1} failed: openai.error.APIConnectionError')
-                if i < retries - 1:
-                    logging.error(f'Retrying in {timeout} seconds...')
-                    time.sleep(timeout)
-                    i += 1
-            except openai.error.RateLimitError:
-                logging.error(f'Retry {k+1} failed: openai.error.RateLimitError')
-                if j < retries - 1:
-                    logging.error(f'Retrying in {timeout} seconds...')
-                    time.sleep(timeout)
-                    j += 1
+    global global_document_type
+    global global_context
+    global_document_type = document_type
+    global_context = context
+
+    if progress == "None":
+        for x in content:
+            retries = 1
+            timeout = 1
+            i = 0
+            while True: 
+                try: 
+                    data = json.loads(chat_gpt_document(detect_document(x),document_type,context))
+                    combined_data.append(data)
+                    break
+                except (openai.error.APIConnectionError, openai.error.AuthenticationError, openai.error.RateLimitError, google.api_core.exceptions.RetryError, requests.exceptions.RequestException) as e:
+                    logging.error(f'Retry {i+1} failed: {e}')
+                    if i < retries - 1:
+                        logging.error(f'Retrying in {timeout} seconds...')
+                        time.sleep(timeout)
+                        i += 1
+                    else:
+                        unprocessed_documents.append(x)
+                        break
+
+    else: 
+        progress(0, desc="Starting")
+        for x in progress.tqdm(content, desc="Processing"):
+            retries = 1
+            timeout = 1
+            i = 0
+            while True: 
+                try: 
+                    data = json.loads(chat_gpt_document(detect_document(x),document_type,context))
+                    combined_data.append(data)
+                    break
+                except (openai.error.APIConnectionError, openai.error.AuthenticationError, openai.error.RateLimitError, google.api_core.exceptions.RetryError, requests.exceptions.RequestException) as e:
+                    logging.error(f'Retry {i+1} failed: {e}')
+                    if i < retries - 1:
+                        logging.error(f'Retrying in {timeout} seconds...')
+                        time.sleep(timeout)
+                        i += 1
+                    else:
+                        unprocessed_documents.append(x)
+                        break
+    
     logging.info(combined_data)
+    logging.info(unprocessed_documents)
+    if document_type == "":
+        document_type = "error"
     return save_json(combined_data, document_type)
 
+def retry_unprocessed_documents():
+    # This function will use the documents stored in unprocessed_documents
+    # and call batch_document on them
+    global global_document_type
+    global global_context
+    global unprocessed_documents
+    if unprocessed_documents:
+        output = batch_document(unprocessed_documents, global_document_type, global_context, "None")
+        unprocessed_documents = []
+        return output
+    else:
+        unprocessed_documents = []
+        return save_json("No Unprocessed Documents", "No Unprocessed Documents")
+    
 def save_json(text, filename):
     filename = filename+".json"
     with open(filename, "w", encoding='utf-8') as outfile:
@@ -244,11 +289,13 @@ with gr.Blocks(title="Axon OCR", css=".markdown {text-align: center;}") as app:
                     gr.Textbox(label="Any additional information? (Optional)", placeholder="This is document is an Official Reciept")]
             batch_document_output = gr.File(label="Result")
         batch_document_button = gr.Button("Scan")
+        retry_button = gr.Button("Retry Unprocessed Documents", label="Retry")
          
     
     image_button.click(image, inputs=image_input, outputs=image_output)
     document_button.click(document, inputs=document_input, outputs=document_output)
     batch_document_button.click(batch_document, inputs=batch_document_input, outputs=batch_document_output)
+    retry_button.click(retry_unprocessed_documents, outputs=batch_document_output)
 
 app.queue()
 app.launch(auth=("username", "password"))