pomera-ai-commander 1.2.2 → 1.2.3

package/core/mcp/tool_registry.py

@@ -1868,13 +1868,41 @@ class ToolRegistry:
         conn.row_factory = sqlite3.Row
         return conn
     
+    def _sanitize_text(self, text: str) -> str:
+        """
+        Sanitize text by removing invalid UTF-8 surrogate characters.
+        
+        Lone surrogates (U+D800 to U+DFFF) are invalid in UTF-8 and cause
+        encoding errors. This function removes them while preserving valid content.
+        
+        Args:
+            text: Input text that may contain invalid surrogates
+            
+        Returns:
+            Sanitized text safe for UTF-8 encoding and database storage
+        """
+        if not text:
+            return text
+        
+        # Encode to UTF-8 with 'surrogatepass' to handle surrogates,
+        # then decode with 'replace' to replace invalid sequences
+        try:
+            # This two-step process handles lone surrogates:
+            # 1. surrogatepass allows encoding surrogates (normally forbidden in UTF-8)
+            # 2. errors='replace' replaces invalid sequences with replacement char
+            sanitized = text.encode('utf-8', errors='surrogatepass').decode('utf-8', errors='replace')
+            return sanitized
+        except Exception:
+            # Fallback: manually filter out surrogate characters
+            return ''.join(c for c in text if not (0xD800 <= ord(c) <= 0xDFFF))
+    
     def _handle_notes_save(self, args: Dict[str, Any]) -> str:
         """Handle saving a new note."""
         from datetime import datetime
         
-        title = args.get("title", "")
-        input_content = args.get("input_content", "")
-        output_content = args.get("output_content", "")
+        title = self._sanitize_text(args.get("title", ""))
+        input_content = self._sanitize_text(args.get("input_content", ""))
+        output_content = self._sanitize_text(args.get("output_content", ""))
         
         if not title:
             return "Error: Title is required"
@@ -2042,15 +2070,15 @@ class ToolRegistry:
             
             if "title" in args:
                 updates.append("Title = ?")
-                values.append(args["title"])
+                values.append(self._sanitize_text(args["title"]))
             
             if "input_content" in args:
                 updates.append("Input = ?")
-                values.append(args["input_content"])
+                values.append(self._sanitize_text(args["input_content"]))
             
             if "output_content" in args:
                 updates.append("Output = ?")
-                values.append(args["output_content"])
+                values.append(self._sanitize_text(args["output_content"]))
             
             if not updates:
                 conn.close()
@@ -2639,12 +2667,17 @@ class ToolRegistry:
             """Save operation to notes and return note_id."""
             try:
                 from datetime import datetime
+                # Sanitize text to prevent UTF-8 surrogate errors
+                sanitized_title = registry._sanitize_text(title)
+                sanitized_input = registry._sanitize_text(input_content)
+                sanitized_output = registry._sanitize_text(output_content)
+                
                 conn = registry._get_notes_connection()
                 now = datetime.now().strftime('%Y-%m-%d %H:%M:%S')
                 cursor = conn.execute('''
                     INSERT INTO notes (Created, Modified, Title, Input, Output)
                     VALUES (?, ?, ?, ?, ?)
-                ''', (now, now, title, input_content, output_content))
+                ''', (now, now, sanitized_title, sanitized_input, sanitized_output))
                 note_id = cursor.lastrowid
                 conn.commit()
                 conn.close()

package/mcp.json

@@ -1,6 +1,6 @@
 {
   "name": "pomera-ai-commander",
-  "version": "1.2.2",
+  "version": "1.2.3",
   "description": "Text processing toolkit with 22 MCP tools including case transformation, encoding, hashing, text analysis, and notes management for AI assistants.",
   "homepage": "https://github.com/matbanik/Pomera-AI-Commander",
   "repository": "https://github.com/matbanik/Pomera-AI-Commander",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pomera-ai-commander",
-  "version": "1.2.2",
+  "version": "1.2.3",
   "description": "Text processing toolkit with 22 MCP tools for AI assistants - case transformation, encoding, hashing, text analysis, and notes management",
   "main": "pomera_mcp_server.py",
   "bin": {

package/pomera.py

@@ -1201,7 +1201,7 @@ class PromeraAIApp(tk.Tk):
             pyproject_path = Path(__file__).parent / "pyproject.toml"
             if pyproject_path.exists():
                 content = pyproject_path.read_text()
-                match = re.search(r'version = "([^"]+)"', content)
+                match = re.search(r'version = "1.2.3"]+)"', content)
                 if match:
                     current_version = match.group(1)
         except Exception:
@@ -1212,7 +1212,7 @@ class PromeraAIApp(tk.Tk):
                 import importlib.metadata
                 current_version = importlib.metadata.version("pomera-ai-commander")
             except Exception:
-                current_version = "Unknown"
+                current_version = "1.2.3"
         
         # Detect OS for download link
         system = platform.system()
@@ -1350,11 +1350,8 @@ class PromeraAIApp(tk.Tk):
     
     def _show_about_dialog(self):
         """Show About dialog."""
-        try:
-            import importlib.metadata
-            version = importlib.metadata.version("pomera-ai-commander")
-        except Exception:
-            version = "1.2.0"
+        # Version managed by bump_version.py script
+        version = "1.2.3"
         
         messagebox.showinfo(
             "About Pomera AI Commander",

package/pomera_mcp_server.py

@@ -70,7 +70,7 @@ def main():
     parser.add_argument(
         "--version",
         action="version",
-        version="pomera-mcp-server 1.2.2"
+        version="pomera-mcp-server 1.2.3"
     )
     parser.add_argument(
         "--list-tools",
@@ -160,7 +160,7 @@ def main():
     server = StdioMCPServer(
         tool_registry=registry,
         server_name="pomera-mcp-server",
-        server_version="1.2.2"
+        server_version="1.2.3"
     )
     
     logger.info("Starting Pomera MCP Server...")

package/tools/notes_widget.py

@@ -99,6 +99,33 @@ class NotesWidget:
             if conn:
                 conn.close()
     
+    def _sanitize_text(self, text: str) -> str:
+        """
+        Sanitize text by removing invalid UTF-8 surrogate characters.
+        
+        Lone surrogates (U+D800 to U+DFFF) are invalid in UTF-8 and cause
+        encoding errors when saving to the database. This can happen when
+        pasting content from the clipboard that contains malformed data.
+        
+        Args:
+            text: Input text that may contain invalid surrogates
+            
+        Returns:
+            Sanitized text safe for UTF-8 encoding and database storage
+        """
+        if not text:
+            return text
+        
+        try:
+            # This two-step process handles lone surrogates:
+            # 1. surrogatepass allows encoding surrogates (normally forbidden in UTF-8)
+            # 2. errors='replace' replaces invalid sequences with replacement char
+            sanitized = text.encode('utf-8', errors='surrogatepass').decode('utf-8', errors='replace')
+            return sanitized
+        except Exception:
+            # Fallback: manually filter out surrogate characters
+            return ''.join(c for c in text if not (0xD800 <= ord(c) <= 0xDFFF))
+    
     def init_database(self) -> None:
         """Initialize the SQLite database and Full-Text Search (FTS5) table."""
         try:
@@ -707,10 +734,12 @@ class NotesWidget:
                     row = conn.execute('SELECT * FROM notes WHERE id = ?', (self.current_item,)).fetchone()
                     if row:
                         now = datetime.now().isoformat()
+                        # Sanitize text to prevent UTF-8 surrogate errors
                         conn.execute('''
                             INSERT INTO notes (Created, Modified, Title, Input, Output)
                             VALUES (?, ?, ?, ?, ?)
-                        ''', (now, now, row['Title'], row['Input'], row['Output']))
+                        ''', (now, now, self._sanitize_text(row['Title']), 
+                              self._sanitize_text(row['Input']), self._sanitize_text(row['Output'])))
                         conn.commit()
                         self.perform_search(select_first=True)
                         self.logger.info(f"Duplicated note {self.current_item}")
@@ -788,9 +817,10 @@ class NotesWidget:
             return
         
         try:
-            title = self.title_entry.get() if hasattr(self, 'title_entry') else ""
-            input_content = self.input_display.get(1.0, tk.END).strip()
-            output_content = self.output_display.get(1.0, tk.END).strip()
+            # Sanitize text to prevent UTF-8 surrogate errors from clipboard paste
+            title = self._sanitize_text(self.title_entry.get() if hasattr(self, 'title_entry') else "")
+            input_content = self._sanitize_text(self.input_display.get(1.0, tk.END).strip())
+            output_content = self._sanitize_text(self.output_display.get(1.0, tk.END).strip())
             
             now = datetime.now().isoformat()
             
@@ -962,12 +992,17 @@ class NotesWidget:
             The ID of the created note, or None on error
         """
         try:
+            # Sanitize text to prevent UTF-8 surrogate errors
+            sanitized_title = self._sanitize_text(title)
+            sanitized_input = self._sanitize_text(input_content)
+            sanitized_output = self._sanitize_text(output_content)
+            
             now = datetime.now().isoformat()
             with self.get_db_connection() as conn:
                 cursor = conn.execute('''
                     INSERT INTO notes (Created, Modified, Title, Input, Output)
                     VALUES (?, ?, ?, ?, ?)
-                ''', (now, now, title, input_content, output_content))
+                ''', (now, now, sanitized_title, sanitized_input, sanitized_output))
                 note_id = cursor.lastrowid
                 conn.commit()