From 00e2f9d6c35fd655fbaca1480f3350de6e474666 Mon Sep 17 00:00:00 2001
From: star-nox <dabholkar.asmita@gmail.com>
Date: Wed, 29 Nov 2023 12:48:14 -0600
Subject: [PATCH] added similar format_for_json for MQR

---
 ai_ta_backend/parallel_context_processing.py |  8 +++-
 ai_ta_backend/vector_database.py             | 44 +++++++++++++++-----
 2 files changed, 40 insertions(+), 12 deletions(-)

diff --git a/ai_ta_backend/parallel_context_processing.py b/ai_ta_backend/parallel_context_processing.py
index d0297933..61793eb7 100644
--- a/ai_ta_backend/parallel_context_processing.py
+++ b/ai_ta_backend/parallel_context_processing.py
@@ -54,13 +54,15 @@ def qdrant_context_processing(doc, course_name, result_contexts):
     }
     if 'url' in doc.metadata.keys():
       context_dict['url'] = doc.metadata['url']
+    else:
+      context_dict['url'] = ''
         
     result_contexts.append(context_dict)
     return result_contexts    
 
 def supabase_context_padding(doc, course_name, result_docs):
     """
-    Does context padding for given doc. Used with context_padding()
+    Does context padding for given doc.
     """
         
     # query by url or s3_path
@@ -98,6 +100,7 @@ def supabase_context_padding(doc, course_name, result_docs):
         pagenumber = doc.metadata['pagenumber']
             
         for context in contexts:
+          # pad contexts belonging to same page number
           if int(context['pagenumber']) == pagenumber:
             context['readable_filename'] = filename
             context['course_name'] = course_name
@@ -119,6 +122,9 @@ def supabase_context_padding(doc, course_name, result_docs):
         }
         if 'url' in doc.metadata.keys():
           context_dict['url'] = doc.metadata['url']
+        else:
+          context_dict['url'] = ''
+
         result_docs.append(context_dict)
     
     
diff --git a/ai_ta_backend/vector_database.py b/ai_ta_backend/vector_database.py
index 282b0979..78eee5f9 100644
--- a/ai_ta_backend/vector_database.py
+++ b/ai_ta_backend/vector_database.py
@@ -1409,7 +1409,7 @@ def getTopContextsWithMQR(self, search_query: str, course_name: str, token_limit
       print(f"⏰ ^^ Runtime of getTopContextsWithMQR: {(time.monotonic() - start_time_overall):.2f} seconds")
       if len(valid_docs) == 0:
         return []
-      return self.format_for_json(valid_docs)
+      return self.format_for_json_mqr(valid_docs)
     except Exception as e:
       # return full traceback to front end
       err: str = f"ERROR: In /getTopContextsWithMQR. Course: {course_name} ||| search_query: {search_query}\nTraceback: {traceback.format_exc()}❌❌ Error in {inspect.currentframe().f_code.co_name}:\n{e}"  # type: ignore
@@ -1577,6 +1577,29 @@ def get_stuffed_prompt(self, search_query: str, course_name: str, token_limit: i
       print(err)
       return err
 
+  
+  def format_for_json_mqr(self, found_docs) -> List[Dict]:
+    """
+    Same as format_for_json, but for the new MQR pipeline.
+    """
+    for found_doc in found_docs:
+      if "pagenumber" not in found_doc.keys():
+        print("found no pagenumber")
+        found_doc['pagenumber'] = found_doc['pagenumber_or_timestamp']
+
+    contexts = [{
+        'text': doc['text'],
+        'readable_filename': doc['readable_filename'],
+        'course_name ': doc['course_name'],
+        's3_path': doc['s3_path'],
+        'pagenumber': doc['pagenumber'], 
+        'url': doc['url'], # wouldn't this error out?
+        'base_url': doc['base_url'],
+    } for doc in found_docs]
+
+    return contexts
+  
+
   def format_for_json(self, found_docs: List[Document]) -> List[Dict]:
     """Formatting only.
       {'course_name': course_name, 'contexts': [{'source_name': 'Lumetta_notes', 'source_location': 'pg. 19', 'text': 'In FSM, we do this...'}, {'source_name': 'Lumetta_notes', 'source_location': 'pg. 20', 'text': 'In Assembly language, the code does that...'},]}
@@ -1590,21 +1613,20 @@ def format_for_json(self, found_docs: List[Document]) -> List[Dict]:
     Returns:
         List[Dict]: _description_
     """
-    
     for found_doc in found_docs:
-      if "pagenumber" not in found_doc.keys():
+      if "pagenumber" not in found_doc.metadata.keys():
         print("found no pagenumber")
-        found_doc['pagenumber'] = found_doc['pagenumber_or_timestamp']
+        found_doc.metadata['pagenumber'] = found_doc.metadata['pagenumber_or_timestamp']
 
     contexts = [{
-        'text': doc['text'],
-        'readable_filename': doc['readable_filename'],
-        'course_name ': doc['course_name'],
-        's3_path': doc['s3_path'],
-        'pagenumber': doc['pagenumber'], # this because vector db schema is older...
+        'text': doc.page_content,
+        'readable_filename': doc.metadata['readable_filename'],
+        'course_name ': doc.metadata['course_name'],
+        's3_path': doc.metadata['s3_path'],
+        'pagenumber': doc.metadata['pagenumber'], # this because vector db schema is older...
         # OPTIONAL PARAMS...
-        'url': doc['url'], # wouldn't this error out?
-        'base_url': doc['base_url'],
+        'url': doc.metadata.get('url'), # wouldn't this error out?
+        'base_url': doc.metadata.get('base_url'),
     } for doc in found_docs]
 
     return contexts