Adding logging

app.py

@@ -12,11 +12,13 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain.prompts import PromptTemplate
 from langchain.chains.summarize import load_summarize_chain
 from langchain.callbacks import get_openai_callback
+import logging
 
 chkey = os.environ["API_TOKEN"]
 token = b64encode(f"{chkey}".encode('utf-8')).decode("ascii")
 
 with gr.Blocks(gr.themes.Soft()) as demo:
+    logging.info("*** App Starting ***")
     # Create input Text search box
     input_box = gr.Textbox(label="Input search string for a UK Company Name")
     # State variable to store the Document ID
@@ -38,7 +40,7 @@ with gr.Blocks(gr.themes.Soft()) as demo:
     display_filing_doc_info = gr.Textbox(label="",interactive=False, visible=False)
 
     # OpenAPI Key Input box
-    openapi_key_input = gr.Textbox(label="OpenAI API Key",type='password',interactive=True, visible=False)
+    openapi_key_input = gr.Textbox(label="OpenAI API Key", type='password', interactive=True, visible=False)
 
     # Button to initiate the processing of the Document - OPENAI Call initiated here too
     process_filing_btn = gr.Button("Summarize the Account filing", visible=False)
@@ -54,13 +56,17 @@ with gr.Blocks(gr.themes.Soft()) as demo:
 
     # Function that does the Company search based on a search string. Gets the top 10 results 
     def company_search(text):
+        logging.info("*** New Search Starting ***")
+        logging.info(f'Search term : {text}')
         url = "https://api.company-information.service.gov.uk/advanced-search/companies?company_name_includes=" + text + "&company_status=active&size=10"
+        logging.info(f'Calling Companies House API Advanced search : {url}')
         auth = f'Basic {token}'
         payload={}
         headers = {
             'Authorization': auth
         }
         response = requests.request("GET", url, headers=headers, data=payload)
+        logging.info(f'API Response Code : {response.status_code}')
         select_resp = []
         if response.status_code == 200:
             resp = json.loads(response.text)
@@ -69,6 +75,8 @@ with gr.Blocks(gr.themes.Soft()) as demo:
                 for key, value in comp["registered_office_address"].items():
                     addr.append(value) 
                 select_resp.append(comp["company_number"] + " : " + comp["company_name"] + " : " + ', '.join(addr))
+            resp_joined = (','.join(select_resp))
+            logging.info(f'Response list : {resp_joined}')
             return {output_col: gr.update(visible=True), company_list_box: gr.update(choices=select_resp,interactive=True)}
         else:
             select_resp.append("No matching companies found")
@@ -76,9 +84,11 @@ with gr.Blocks(gr.themes.Soft()) as demo:
     
     # Function to get the Filing information of a selected company
     def company_selected(selected_company, docid):
+        logging.info("* Company selected. Getting Filing History *")
+        logging.info(f'User Selection : {selected_company}')
         regid = selected_company.split(' : ')[0]
         filings_url = "https://api.company-information.service.gov.uk/company/" + regid + "/filing-history?category=accounts&items_per_page=1"
-        print(filings_url)
+        logging.info(f'Calling Companies House API Filings Endpoint : {filings_url}')
         auth = f'Basic {token}'
         payload={}
         headers = {
@@ -86,26 +96,30 @@ with gr.Blocks(gr.themes.Soft()) as demo:
         }
         response = requests.request("GET", filings_url, headers=headers, data=payload)
         resp = json.loads(response.text)
-        #print(resp["items"])
-        if len(resp["items"])>0:
-            resp_value = f'Latest filing done on {resp["items"][0]["date"]}.'
-            if "links" in resp["items"][0]:
-                if "document_metadata" in resp["items"][0]["links"]:
-                    docid = resp["items"][0]["links"]["document_metadata"].rsplit('/',1)[-1]
-                    return {display_filing: gr.Textbox.update(visible=True, value=resp_value), submit_btn: gr.update(visible=True), doc_id : docid}
+        logging.info(f'API Response Code : {response.status_code}')
+        if response.status_code == 200:
+            if len(resp["items"])>0:
+                resp_value = f'Latest filing done on {resp["items"][0]["date"]}.'
+                if "links" in resp["items"][0]:
+                    if "document_metadata" in resp["items"][0]["links"]:
+                        docid = resp["items"][0]["links"]["document_metadata"].rsplit('/',1)[-1]
+                        return {display_filing: gr.Textbox.update(visible=True, value=resp_value), submit_btn: gr.update(visible=True), doc_id : docid}
+                    else:
+                        resp_value += "But Document Metadata is not available."
+                        return {display_filing: gr.Textbox.update(visible=True, value=resp_value), submit_btn: gr.update(visible=False), doc_id : "None"}
                 else:
-                    resp_value += "But Document Metadata is not available."
+                    resp_value += "But Links to the filing not available."
                     return {display_filing: gr.Textbox.update(visible=True, value=resp_value), submit_btn: gr.update(visible=False), doc_id : "None"}
             else:
-                resp_value += "But Links to the filing not available."
-                return {display_filing: gr.Textbox.update(visible=True, value=resp_value), submit_btn: gr.update(visible=False), doc_id : "None"}
+                return {display_filing: gr.Textbox.update(visible=True, value="No record of accounts filed for the company"), submit_btn: gr.update(visible=False), doc_id : "None"}
         else:
             return {display_filing: gr.Textbox.update(visible=True, value="No record of accounts filed for the company"), submit_btn: gr.update(visible=False), doc_id : "None"}
 
     # Function to get the Filing document related to the latest Annual Account filing
     def get_filing(docid):
+        logging.info("* Getting Filing Document for latest filing *")
         doc_url = "https://document-api.company-information.service.gov.uk/document/" + docid + "/content"
-        print(doc_url)
+        logging.info(f'Calling Companies House Documents API : {doc_url}')
         auth = f'Basic {token}'
         payload={}
         headers = {
@@ -113,9 +127,8 @@ with gr.Blocks(gr.themes.Soft()) as demo:
             'Accept': 'application/pdf'
         }
         response = requests.request("GET", doc_url, headers=headers, data=payload)
-        #print(response.text)
+        logging.info(f'API Response Code : {response.status_code}')
         content_type = response.headers['Content-Type']
-        print(content_type)
         resp_value = f'Filing document is of type {content_type}. '
         if content_type == 'application/pdf':
             filename = f'doc_{docid}.pdf'
@@ -124,19 +137,16 @@ with gr.Blocks(gr.themes.Soft()) as demo:
                 f.write(response.content)
             pdf_document = fitz.open(filepath)
             resp_value += f'PDF saved as: {filename}. There are a total of {pdf_document.page_count} pages'
-            print(resp_value)
+            logging.info(resp_value)
             return {display_filing_doc_info: gr.Textbox.update(visible=True, value=resp_value), process_filing_btn: gr.update(visible=True), openapi_key_input: gr.update(visible=True), processed_info: gr.update(visible=True), doc_id : docid}
-        elif content_type == 'application/xhtml+xml':
-            print('Work in progress to process these type of filings')
-            resp_value += 'Work in progress to process these type of filings'
-            return {display_filing_doc_info: gr.Textbox.update(visible=True, value=resp_value), process_filing_btn: gr.update(visible=False), openapi_key_input: gr.update(visible=False), processed_info: gr.update(visible=False), doc_id : "None"}
         else:
-            print('Work in progress to process these type of filings')
             resp_value += 'Work in progress to process these type of filings'
+            logging.info(resp_value)
             return {display_filing_doc_info: gr.Textbox.update(visible=True, value=resp_value), process_filing_btn: gr.update(visible=False), openapi_key_input: gr.update(visible=False), processed_info: gr.update(visible=False), doc_id : "None"}
 
     # Function to initial the Langchain chain with call to OPENAI to Summarize the Annual report
     def langchain_summarize(contents,openai_api_key):
+        logging.info("* Calling Langchain / OPENAI to get the summary *")
         concatenated_content = '`n`n'.join(contents)
         text_splitter = RecursiveCharacterTextSplitter(chunk_size=1200,chunk_overlap=20,length_function=len)
         docs = text_splitter.create_documents([concatenated_content])
@@ -158,6 +168,7 @@ with gr.Blocks(gr.themes.Soft()) as demo:
 
     # Function to extract text from the document and call the LangChain processing function with the text array of pages
     def process_filing(docid, openai_api_key, progress=gr.Progress()):
+        logging.info("* Processing the filing document *")
         progress(0,desc="Starting...")
         filepath = f'./data/doc_{docid}.pdf'
         pdf_document = fitz.open(filepath)
@@ -180,28 +191,29 @@ with gr.Blocks(gr.themes.Soft()) as demo:
             # save list to file
             with open(text_path, 'wb') as f:
                 pickle.dump(contents, f)
-        #print(contents)
+
         resp_value = f'Total of {pdf_document.page_count} pages processed. '
         summary_path = f'./summary/doc_{docid}.txt'
         if os.path.exists(summary_path):
             with open(summary_path, 'r') as f:
                 summary = f.read()
-            print(resp_value)
+            logging.info(resp_value)
             return {processed_info: gr.Textbox.update(visible=True, value=resp_value), summary_text: gr.Textbox.update(visible=True, value=summary)}
         else:
-            #progress(track_tqdm=True,desc="Calling OpenAI to summarize...")
             try:
                 summary, tkn_text = langchain_summarize(contents, openai_api_key)
                 resp_value += tkn_text
+                logging.info(resp_value)
                 with open(summary_path, 'wb') as f:
                     f.write(summary.encode())
                 return {processed_info: gr.Textbox.update(visible=True, value=resp_value), summary_text: gr.Textbox.update(visible=True, value=summary)}
             except Exception as e:
-                print(e)
+                logging.info(e)
                 resp_value += 'LLM Call failed. Please check the OpenAI key again'
+                logging.info(resp_value)
                 return {processed_info: gr.Textbox.update(visible=True, value=resp_value), summary_text: gr.Textbox.update(visible=False)}
             finally:
-                print(resp_value)
+                logging.info(resp_value)
     
     def clear_screen():
         return {output_col: gr.update(visible=False),display_filing: gr.Textbox.update(visible=False),submit_btn: gr.update(visible=False), display_filing_doc_info:gr.update(visible=False), process_filing_btn:gr.update(visible=False),openapi_key_input:gr.update(visible=False),processed_info:gr.update(visible=False),summary_text:gr.update(visible=False)}