export only specific tabs

Author: somogyijanos

README.md

@@ -2,14 +2,14 @@
 
 This project provides a command-line interface (CLI) tool to discover and export AI chat data from [Cursor](https://cursor.sh). The tool is implemented in `chat.py` and leverages utility classes for querying the database, formatting the chat data, and saving it to files.
 
-Cursor's chat history is stored in `sqlite3` databases using `state.vscdb` files. One such file is for one workspace. All the chats for a workspace are saved here in so called *tabs*, one tab means a single chat.
+Cursor's chat history is stored in `sqlite3` databases using `state.vscdb` files. One such file is for one workspace. All the chats for a workspace are saved here in so-called *tabs*, one tab means a single chat.
 
 Also see [this](https://forum.cursor.com/t/guide-5-steps-exporting-chats-prompts-from-cursor/2825) forum post on this topic.
 
 ## Features
 
 - **Discover Chats**: Discover all chats from all workspaces and print a few lines of dialogue so one can identify which is the workspace (or chat) one is searching for. It's also possible to filter by text.
-- **Export Chats**: Export chats for the most recent (or a specific) workspace to Markdown files or print it to the command line.
+- **Export Chats**: Export chats for the most recent (or a specific) workspace to Markdown files or print them to the command line.
 
 ## Installation
 
@@ -26,7 +26,7 @@ Also see [this](https://forum.cursor.com/t/guide-5-steps-exporting-chats-prompts
 
 ## Usage
 
-First, find, where the `state.vscdb` files are located on your computer. Confirm that corresponding to your system, the right path is set in the [config.yml](./config.yml) file. Update it if not set correctly.
+First, find where the `state.vscdb` files are located on your computer. Confirm that corresponding to your system, the right path is set in the [config.yml](./config.yml) file. Update it if not set correctly.
 
 Both the `discover` and `export` commands will work with this path by default, but you can also provide a custom path any time.
 
@@ -64,6 +64,9 @@ See `./chat.py export --help` for general help. Examples:
 # Export only the latest chat of the most recent workspace
 ./chat.py export --latest-tab --output-dir "/path/to/output"
 
+# Export only chat No. 2 and 3 of the most recent workspace
+./chat.py export --tab-ids 2,3 --output-dir "/path/to/output"
+
 # Export all chats of a specifc workspace
 ./chat.py export --output-dir "/path/to/output" "/path/to/workspaces/workspace-dir/state.vscdb"
 ```

chat.py

@@ -20,7 +20,8 @@
 def export(
     db_path: str = typer.Argument(None, help="The path to the SQLite database file. If not provided, the latest workspace will be used."),
     output_dir: str = typer.Option(None, help="The directory where the output markdown files will be saved. If not provided, prints to command line."),
-    latest_tab: bool = typer.Option(False, "--latest-tab", help="Export only the latest tab. If not set, all tabs will be exported.")
+    latest_tab: bool = typer.Option(False, "--latest-tab", help="Export only the latest tab. If not set, all tabs will be exported."),
+    tab_ids: str = typer.Option(None, help="Comma-separated list of tab IDs to export. For example, '1,2,3'. If not set, all tabs will be exported.")
 ):
     """
     Export chat data from the database to markdown files or print it to the command line.
@@ -43,10 +44,14 @@ def export(
         # Convert the chat data from JSON string to dictionary
         chat_data_dict = json.loads(chat_data[0])
 
+        tab_id_list = None
         if latest_tab:
             # Get the latest tab by timestamp
             latest_tab = max(chat_data_dict['tabs'], key=lambda tab: tab.get('timestamp', 0))
             chat_data_dict['tabs'] = [latest_tab]
+        elif tab_ids:
+            # Filter tabs by provided tab IDs
+            tab_id_list = [int(ti) - 1 for ti in tab_ids.split(',')]
 
         # Check if there are any images in the chat data
         has_images = any('image' in bubble for tab in chat_data_dict['tabs'] for bubble in tab.get('bubbles', []))
@@ -60,11 +65,11 @@ def export(
             # Save the chat data
             saver = MarkdownFileSaver()
             exporter = ChatExporter(formatter, saver)
-            exporter.export(chat_data_dict, output_dir, image_dir)
+            exporter.export(chat_data_dict, output_dir, image_dir, tab_ids=tab_id_list)
             success_message = f"Chat data has been successfully exported to {output_dir}"
             logger.info(success_message)
         else:
-            formatted_chats = formatter.format(chat_data_dict, image_dir)
+            formatted_chats = formatter.format(chat_data_dict, image_dir, tab_ids=tab_id_list)
             # Print the chat data to the command line using markdown
             for formatted_data in formatted_chats:
                 console.print(Markdown(formatted_data))

src/export.py

@@ -9,15 +9,15 @@
 
 class ChatFormatter(ABC):
     @abstractmethod
-    def format(self, chat_data: dict[str, Any], image_dir: str = 'images') -> str:
+    def format(self, chat_data: dict[str, Any], image_dir: str = 'images') -> dict[int, str] | None:
         """Format the chat data into Markdown format.
 
         Args:
             chat_data (dict[str, Any]): The chat data to format.
             image_dir (str): The directory where images will be saved. Defaults to 'images'.
 
         Returns:
-            str: The formatted chat data in Markdown.
+            dict[int, str]: The formatted chat for each tab.
         """
         pass
 
@@ -61,19 +61,24 @@ def _extract_text_from_user_bubble(self, bubble: dict) -> str:
 
         return user_text_text
 
-    def format(self, chat_data: dict[str, Any], image_dir: str | None = 'images') -> str | None:
+    def format(self, chat_data: dict[str, Any], image_dir: str | None = 'images', tab_ids: list[int] | None = None) -> dict[int, str] | None:
         """Format the chat data into Markdown format.
 
         Args:
             chat_data (dict[str, Any]): The chat data to format.
             image_dir (str): The directory where images will be saved. Defaults to 'images'.
+            tab_ids (list[int]): List of tab indices to include exclusively.
 
         Returns:
-            str: The formatted chat data in Markdown.
+            dict[int, str]: The formatted chat in Markdown for each tab.
         """
         try:
-            formatted_chats = []
+            formatted_chats = {}
             for tab_index, tab in enumerate(chat_data['tabs']):
+                if tab_ids is not None:
+                    if tab_index not in tab_ids:
+                        continue
+
                 bubbles = tab['bubbles']
                 formatted_chat = [f"# Chat Transcript - Tab {tab_index + 1}\n"]
 
@@ -116,7 +121,7 @@ def format(self, chat_data: dict[str, Any], image_dir: str | None = 'images') ->
                         raw_text = re.sub(r'```python:[^\n]+', '```python', bubble['rawText'])
                         formatted_chat.append(f"## AI ({model_type}):\n\n{raw_text}\n")
 
-                formatted_chats.append("\n".join(formatted_chat))
+                formatted_chats[f"tab_{tab_index + 1}"] = "\n".join(formatted_chat)
 
             logger.success("Chats formatted.")
             return formatted_chats
@@ -163,7 +168,7 @@ def __init__(self, formatter: ChatFormatter, saver: FileSaver) -> None:
         self.formatter = formatter
         self.saver = saver
 
-    def export(self, chat_data: dict[str, Any], output_dir: str, image_dir: str) -> None:
+    def export(self, chat_data: dict[str, Any], output_dir: str, image_dir: str, tab_ids: list[int] | None = None) -> None:
         """Export the chat data by formatting and saving it.
 
         Args:
@@ -173,10 +178,10 @@ def export(self, chat_data: dict[str, Any], output_dir: str, image_dir: str) ->
         """
         try:
             os.makedirs(output_dir, exist_ok=True)
-            formatted_chats = self.formatter.format(chat_data, image_dir)
+            formatted_chats = self.formatter.format(chat_data, image_dir, tab_ids=tab_ids)
             if formatted_chats is not None:
-                for tab_index, formatted_data in enumerate(formatted_chats):
-                    tab_file_path = os.path.join(output_dir, f"tab_{tab_index + 1}.md")
+                for tab_name, formatted_data in formatted_chats.items():
+                    tab_file_path = os.path.join(output_dir, f"{tab_name}.md")
                     self.saver.save(formatted_data, tab_file_path)
         except Exception as e:
             logger.error(f"Failed to export chat data: {e}")