better title acquisition

open_notebook/domain.py

@@ -90,7 +90,6 @@ class ObjectModel(BaseModel):
 
     def _prepare_save_data(self) -> Dict[str, Any]:
         data = self.model_dump()
-        logger.debug(f"Preparing data for save: {data}")
         del data["created"]
         del data["updated"]
         return {key: value for key, value in data.items() if value is not None}
@@ -306,21 +305,22 @@ class Source(ObjectModel):
             logger.error(f"Error adding insight to source {self.id}: {str(e)}")
             raise DatabaseOperationError(e)
 
+    # todo: move this to content processing pipeline as a major graph
     def generate_toc_and_title(self) -> "Source":
         try:
             config = RunnableConfig(configurable=dict(thread_id=self.id))
             result = toc_graph.invoke({"content": self.full_text}, config=config)
-            logger.warning(result["toc"])
             self.add_insight("Table of Contents", surreal_clean(result["toc"]))
-            transformations = [
-                "Based on the Table of Contents below, please provide a Title for this content, with max 15 words"
-            ]
-            output = pattern_graph.invoke(
-                dict(content_stack=[result["toc"]], transformations=transformations)
-            )
-            logger.warning(output["output"])
-            self.title = surreal_clean(output["output"])
-            self.save()
+            if not self.title:
+                transformations = [
+                    "Based on the Table of Contents below, please provide a Title for this content, with max 15 words"
+                ]
+                output = pattern_graph.invoke(
+                    dict(content_stack=[result["toc"]], transformations=transformations)
+                )
+                logger.warning(output["output"])
+                self.title = surreal_clean(output["output"])
+                self.save()
             return self
         except Exception as e:
             logger.error(f"Error summarizing source {self.id}: {str(e)}")

open_notebook/graphs/content_process.py

@@ -4,6 +4,7 @@ import fitz  # type: ignore
 import magic
 import requests  # type: ignore
 from langgraph.graph import END, START, StateGraph
+from loguru import logger
 from typing_extensions import TypedDict
 from youtube_transcript_api import YouTubeTranscriptApi  # type: ignore
 from youtube_transcript_api.formatters import TextFormatter  # type: ignore
@@ -13,6 +14,7 @@ class SourceState(TypedDict):
     content: str
     file_path: str
     url: str
+    title: str
     source_type: str
     identified_type: str
     identified_provider: str
@@ -97,7 +99,26 @@ def extract_url(state: SourceState):
     Get the content of a URL
     """
     response = requests.get(f"https://r.jina.ai/{state.get('url')}")
-    return {"content": response.text}
+    text = response.text
+    if text.startswith("Title:") and "\n" in text:
+        title_end = text.index("\n")
+        title = text[6:title_end].strip()
+        logger.debug(f"Content has title - {title}")
+        logger.debug(text[:100])
+        content = text[title_end + 1 :].strip()
+        return {"title": title, "content": content}
+    else:
+        logger.debug("Content does not have URL")
+        return {"content": text}
+
+
+def _get_title(url):
+    """
+    Get the content of a URL
+    """
+    response = extract_url(dict(url=url))
+    if "title" in response:
+        return response["title"]
 
 
 def extract_txt(state: SourceState):
@@ -166,7 +187,8 @@ def extract_youtube_transcript(state: SourceState):
         _extract_youtube_id(state.get("url")), languages=["pt", "en"]
     )
     formatter = TextFormatter()
-    return {"content": formatter.format_transcript(transcript)}
+    title = _get_title(state.get("url"))
+    return {"content": formatter.format_transcript(transcript), "title": title}
 
 
 def should_continue(data: SourceState):

stream_app/source.py

@@ -128,6 +128,7 @@ def add_source(session_id):
             source = Source(
                 asset=Asset(url=req.get("url"), file_path=req.get("file_path")),
                 full_text=surreal_clean(result["content"]),
+                title=result.get("title"),
             )
             source.save()
             source.add_to_notebook(st.session_state[session_id]["notebook"].id)