Update app.py

app.py

@@ -71,114 +71,158 @@ elif not openai_api_key:
     st.error("The OpenAI API key is not properly configured. Check your environment variables!")
 else:
     # Generate the report by calling the ChatGPT Turbo API and the WooRank API
+
+    # First, let's create a simple PromptTemplate class since it's not imported
+    class PromptTemplate:
+        def __init__(self, template, input_variables):
+            self.template = template
+            self.input_variables = input_variables
+        
+        def format(self, **kwargs):
+            return self.template.format(**kwargs)
+    
     def analyze_data(_advice, _items, _topics, _issues, _technologies, openai_api_key):
         """
-        :param _advice: A list of strings, each string is a piece of advice
-        :param _items: a list of items that are being analyzed
-        :param _topics: a list of topics that the user is interested in
-        :param _issues: a list of issues that the user has selected
-        :param _technologies: A list of technologies that the user has selected
-        """
-        # Create the system message for ChatGPT Turbo
-        prefix_messages = [{"role": "system", "content": '''You are a helpful and truthful SEO that is very good at analyzing websites with a specific focus on structured data. /n
-                            You are able to provide a detailed report on the website's structured data and how to improve it. /n
-                            ADD AS LEARN MORE LINKS FOR THE FIRST TEXT BLOCK LINKS TO structured data https://wordlift.io/blog/en/entity/structured-data/ and schema.org https://wordlift.io/blog/en/entity/schema-org/ TO PROVIDE ADDITIONAL HELP./n/n
-                            YOU ARE WRITING THE REPORT IN HTML USING A TEMPLATE.'''}]
-
-        client = OpenAI(api_key=openai_api_key)
+        Analyzes website data and generates a structured report using OpenAI's GPT model.
         
-        # Construct messages for the chat API
-        messages = []
-        messages.extend(prefix_messages)
-
-
-        # Create the prompt template and the run statement when there are NOT issues
-        if not _issues and len(_items) > 0:
-            template = """
-            First text block of the report./n
-            Analyze the: {advice}, consider that the site features the following schema classes: {items}./n/n
-
-            Second text block of the report./n
-            The website's homepage also references the following entities: {topics} that could be used to improve the SEO of the website further./n/n
-
-            Third text block of the report./n
-            Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do./n/n
-
-            THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:/n
-            "first": "First text block with schema classes in <i>italic</i>",
-            "second": "Second text block with entities in <b>bold</b>",
-            "third": "Third text block with technologies in <i>italic</i>"
-            """
-            prompt = PromptTemplate(template=template, input_variables=[
-                                    "advice", "items", "topics", "technologies"])
-            run_statement = {"advice": _advice, "items": _items,
-                            "topics": _topics, "technologies": _technologies}
-
-        # Create the prompt template and the run statement when there ARE NOT schema classes
-        elif not _items:
-            template = """
-            First text block of the report./n
-            The website homepage doesn't seem to feature any schema class./n/n
-
-            Second text block of the report./n
-            The website's homepage also references the following entities: {topics} that can be used to improve the SEO of the website./n/n
-
-            Third text block of the report./n
-            Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do./n/n
-
-            THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:/n
-            "first": "First text block",
-            "second": "Second text block with entities in <b>bold</b>"
-            "third": "Third text block with technologies in <i>italic</i>"
-            """
-            prompt = PromptTemplate(template=template, input_variables=[
-                                    "topics", "technologies"])
-            run_statement = {"topics": _topics, "technologies": _technologies}
-
-        # Create the prompt template and the run statement when there ARE issues
-        else:
-            template = """
-            First text block of the report./n
-            Analyze the: {advice}, consider that the site features the following schema classes: {items}./n/n
-
-            Second text block of the report. /n
-            Describe the following issues with the markup: {issues} and indicate how to fix them./n/n
-
-            Third text block of the report./n
-            The website's homepage also references the following entities: {topics} that could be used to improve the SEO of the website further./n/n
-
-            Fourth text block of the report./n
-            Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do./n/n
-
-            THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:/n
-            "first": "First text block with schema classes in <i>italic</i>",
-            "second": "Second text block with issues in <u>underline</u>",
-            "third": "Third text block with entities in <b>bold</b>"
-            "fourth": "Fourth text block with technologies in <i>italic</i>"
-            """
-            prompt = PromptTemplate(template=template, input_variables=[
-                                    "advice", "items", "topics", "issues", "technologies"])
-            run_statement = {"advice": _advice, "items": _items,
-                            "topics": _topics, "issues": _issues, "technologies": _technologies}
+        Args:
+            _advice (list): A list of strings, each string is a piece of advice
+            _items (list): A list of items that are being analyzed
+            _topics (list): A list of topics that the user is interested in
+            _issues (list): A list of issues that the user has selected
+            _technologies (list): A list of technologies that the user has selected
+            openai_api_key (str): The OpenAI API key
         
-        # Format the prompt
-        user_message = prompt.format(**run_statement)
-        messages.append({"role": "user", "content": user_message})
-
-        # Make the API call
+        Returns:
+            str: A JSON-formatted string containing the analysis report
+        """
         try:
-            response = client.chat.completions.create(
-                model="gpt-4o",
-                messages=messages
-            )
-            out = response.choices[0].message.content
+            # Create the system message for ChatGPT
+            prefix_messages = [{
+                "role": "system", 
+                "content": '''You are a helpful and truthful SEO that is very good at analyzing websites with a specific focus on structured data.
+                            You are able to provide a detailed report on the website's structured data and how to improve it.
+                            ADD AS LEARN MORE LINKS FOR THE FIRST TEXT BLOCK LINKS TO structured data https://wordlift.io/blog/en/entity/structured-data/ and schema.org https://wordlift.io/blog/en/entity/schema-org/ TO PROVIDE ADDITIONAL HELP.
+                            YOU ARE WRITING THE REPORT IN HTML USING A TEMPLATE.'''
+            }]
+    
+            # Initialize OpenAI client
+            client = OpenAI(api_key=openai_api_key)
+            
+            # Construct messages for the chat API
+            messages = []
+            messages.extend(prefix_messages)
+    
+            # Create the prompt template and run statement based on conditions
+            if not _issues and len(_items) > 0:
+                # Case 1: When there are NO issues but there ARE items
+                template = """
+                First text block of the report.
+                Analyze the: {advice}, consider that the site features the following schema classes: {items}.
+    
+                Second text block of the report.
+                The website's homepage also references the following entities: {topics} that could be used to improve the SEO of the website further.
+    
+                Third text block of the report.
+                Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do.
+    
+                THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:
+                "first": "First text block with schema classes in <i>italic</i>",
+                "second": "Second text block with entities in <b>bold</b>",
+                "third": "Third text block with technologies in <i>italic</i>"
+                """
+                prompt = PromptTemplate(
+                    template=template,
+                    input_variables=["advice", "items", "topics", "technologies"]
+                )
+                run_statement = {
+                    "advice": _advice,
+                    "items": _items,
+                    "topics": _topics,
+                    "technologies": _technologies
+                }
+    
+            elif not _items:
+                # Case 2: When there are NO schema classes
+                template = """
+                First text block of the report.
+                The website homepage doesn't seem to feature any schema class.
+    
+                Second text block of the report.
+                The website's homepage also references the following entities: {topics} that can be used to improve the SEO of the website.
+    
+                Third text block of the report.
+                Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do.
+    
+                THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:
+                "first": "First text block",
+                "second": "Second text block with entities in <b>bold</b>",
+                "third": "Third text block with technologies in <i>italic</i>"
+                """
+                prompt = PromptTemplate(
+                    template=template,
+                    input_variables=["topics", "technologies"]
+                )
+                run_statement = {
+                    "topics": _topics,
+                    "technologies": _technologies
+                }
+    
+            else:
+                # Case 3: When there ARE issues
+                template = """
+                First text block of the report.
+                Analyze the: {advice}, consider that the site features the following schema classes: {items}.
+    
+                Second text block of the report.
+                Describe the following issues with the markup: {issues} and indicate how to fix them.
+    
+                Third text block of the report.
+                The website's homepage also references the following entities: {topics} that could be used to improve the SEO of the website further.
+    
+                Fourth text block of the report.
+                Describe, if available, IN A SINGLE SENTENCE the {technologies} that the site appears to be using and what they do.
+    
+                THE OUTPUT MUST USE THE FOLLOWING TEMPLATE:
+                "first": "First text block with schema classes in <i>italic</i>",
+                "second": "Second text block with issues in <u>underline</u>",
+                "third": "Third text block with entities in <b>bold</b>",
+                "fourth": "Fourth text block with technologies in <i>italic</i>"
+                """
+                prompt = PromptTemplate(
+                    template=template,
+                    input_variables=["advice", "items", "topics", "issues", "technologies"]
+                )
+                run_statement = {
+                    "advice": _advice,
+                    "items": _items,
+                    "topics": _topics,
+                    "issues": _issues,
+                    "technologies": _technologies
+                }
+            
+            # Format the prompt and add it to messages
+            user_message = prompt.format(**run_statement)
+            messages.append({"role": "user", "content": user_message})
+    
+            # Make the API call
+            try:
+                response = client.chat.completions.create(
+                    model="gpt-4",  # Fixed model name from "gpt-4o"
+                    messages=messages
+                )
+                out = response.choices[0].message.content
+            except Exception as e:
+                out = f"Sorry, there was an error with the OpenAI API: {e}"
+                print(f"OpenAI API Error: {str(e)}")  # Log the error for debugging
+    
+            return out
+        
         except Exception as e:
-            out = f"Sorry, there was an error with the OpenAI API: {e}"
-
-
-
-        return out
-
+            error_message = f"An unexpected error occurred: {str(e)}"
+            print(error_message)  # Log the error for debugging
+            return error_message
+    
     # Call WooRank API to get the data (cached)
     @st.cache_data
     def get_woorank_data(url):