Merge pull request #16 from jhakulin/jhakulin/v2-migration-file-search

Jhakulin/v2 migration file search

README.md

@@ -26,7 +26,7 @@
 💬 Inbuilt thread and memory management <br>
 📊 Advanced Data Analysis, create data visualizations and solving complex code and math problems with **Code Interpreter**<br>
 🚀 Build your own tools or call external tools and APIs with **Function Calling**<br>
-📚 Retrieval Augmented Generation with **Retrieval** tool (coming soon to Azure OpenAI Assistants)<br>
+📚 Retrieval Augmented Generation with **File Search** tool (coming soon to Azure OpenAI Assistants)<br>
 🎤📢 Speech transcription and synthesis using Azure CognitiveServices Speech SDK<br>
 📤 Exporting the assistant configuration into simple CLI application
 
@@ -92,7 +92,7 @@ Build the wheel for `azure.ai.assistant` library using the following instruction
 - Go to the`sdk/azure-ai-assistant` folder
 - Build the wheel using following command: `python setup.py sdist bdist_wheel`
 - Go to generated `dist` folder
-- Install the generated wheel using following command: `pip install --force-reinstall azure_ai_assistant-0.2.12a1-py3-none-any.whl`
+- Install the generated wheel using following command: `pip install --force-reinstall azure_ai_assistant-0.3.0a1-py3-none-any.whl`
   - This installation will pick the necessary dependencies for the library (openai, python-Levenshtein, fuzzywuzzy, Pillow, requests)
 
 ### Step 4: Install Python UI libraries

gui/conversation.py

@@ -318,23 +318,42 @@ def replace_with_link(match):
         return url_regex.sub(replace_with_link, text)
 
     def format_file_links(self, text):
-        markdown_link_pattern = r'\[([^\]]+)\]\(sandbox:/mnt/data/([^)]+)\)'
-
+        # Pattern to find citations in the form [Download text]( [index])
+        citation_link_pattern = r'\[([^\]]+)\]\(\s*\[(\d+)\]\s*\)'
+        # Dictionary to store file paths indexed by the citation index
+        citation_to_filename = {}
+
+        # First, extract all file citations like "[0] finance_sector_revenue_chart.png"
+        file_citations = re.findall(r'\[(\d+)\]\s*(.+)', text)
+        for index, filename in file_citations:
+            citation_to_filename[index] = filename
+
+        # Function to replace citation links with clickable HTML links
         def replace_with_clickable_text(match):
             link_text = match.group(1)
-            file_name = match.group(2)
-            local_file_path = os.path.normpath(os.path.join(self.file_path, file_name))
+            citation_index = match.group(2)
+            file_name = citation_to_filename.get(citation_index)
+
+            if file_name:
+                local_file_path = os.path.normpath(os.path.join(self.file_path, file_name))
+
+                if link_text in self.text_to_url_map:
+                    link_text = f"{link_text} {len(self.text_to_url_map) + 1}"
+
+                # Store the file path
+                self.text_to_url_map[link_text] = {"path": local_file_path}
+
+                # Return the HTML link and the local file path in separate inline-block divs
+                return (f'<div style="display: inline-block;"><a href="{local_file_path}" style="color:green; text-decoration: underline;" download="{file_name}">{link_text}</a></div>'
+                        f'<div style="display: inline-block; color:gray;">{local_file_path}</div>')
 
-            if link_text in self.text_to_url_map:
-                link_text = f"{link_text} {len(self.text_to_url_map) + 1}"
-            # Store the file path
-            self.text_to_url_map[link_text] = {"path": local_file_path}
+        # Replace links in the original text
+        updated_text = re.sub(citation_link_pattern, replace_with_clickable_text, text)
 
-            # Return the HTML link and the local file path in separate inline-block divs
-            return (f'<div style="display: inline-block;"><a href="#" style="color:green; text-decoration: underline;" onclick="return false;">{link_text}</a></div>'
-                    f'<div style="display: inline-block; color:gray;">{local_file_path}</div>')
+        # Remove the original citation lines
+        updated_text = re.sub(r'\[\d+\]\s*[^ ]+\.png', '', updated_text)
 
-        return re.sub(markdown_link_pattern, replace_with_clickable_text, text)
+        return updated_text
 
     def parse_message(self, message):
         """Parse the message into a list of (is_code, text) tuples."""

gui/conversation_sidebar.py

@@ -45,17 +45,15 @@ def __init__(self, parent=None):
         super().__init__(parent)
         self.setAcceptDrops(True)
         self.itemToFileMap = {}  # Maps list items to attached file paths
-        self.itemToInstructionsMap = {}  # Maps list items to additional instructions
 
-    def clear_files_and_instructions(self):
+    def clear_files(self):
         self.itemToFileMap.clear()
-        self.itemToInstructionsMap.clear()
 
     def contextMenuEvent(self, event):
         context_menu = QMenu(self)
-        attach_file_action = context_menu.addAction("Attach File")
-        add_instructions_action = context_menu.addAction("Additional Instructions")
-
+        attach_file_search_action = context_menu.addAction("Attach File for File Search")
+        attach_file_code_action = context_menu.addAction("Attach File for Code Interpreter")
+        
         current_item = self.currentItem()
         remove_file_menu = None
         if current_item:
@@ -64,39 +62,24 @@ def contextMenuEvent(self, event):
                 remove_file_menu = context_menu.addMenu("Remove File")
                 for file_info in self.itemToFileMap[row]:
                     actual_file_path = file_info['file_path']
-                    action = remove_file_menu.addAction(os.path.basename(actual_file_path))
+                    tool_type = file_info['tools'][0]['type']
+
+                    file_label = f"{os.path.basename(actual_file_path)} ({tool_type})"
+                    action = remove_file_menu.addAction(file_label)
                     action.setData(file_info)
 
         selected_action = context_menu.exec_(self.mapToGlobal(event.pos()))
 
-        if selected_action == attach_file_action:
-            self.attach_file_to_selected_item()
-        elif selected_action == add_instructions_action:
-            self.add_edit_instructions()
+        if selected_action == attach_file_search_action:
+            self.attach_file_to_selected_item("file_search")
+        elif selected_action == attach_file_code_action:
+            self.attach_file_to_selected_item("code_interpreter")
         elif remove_file_menu and isinstance(selected_action, QAction) and selected_action.parent() == remove_file_menu:
             file_info = selected_action.data()
             self.remove_specific_file_from_selected_item(file_info, row)
 
-    def add_edit_instructions(self):
-        current_item = self.currentItem()
-        if current_item:
-            row = self.row(current_item)
-            current_instructions = self.itemToInstructionsMap.get(row, "")
-            text, ok = QInputDialog.getText(self, "Additional Instructions",
-                                            "Enter instructions:", text=current_instructions)
-            if ok:
-                self.itemToInstructionsMap[row] = text
-                attached_files = self.itemToFileMap.get(row, [])
-                self.update_item_icon(current_item, attached_files, text)
-
-    def get_instructions_for_selected_item(self):
-        current_item = self.currentItem()
-        if current_item:
-            row = self.row(current_item)
-            return self.itemToInstructionsMap.get(row, "")
-        return ""
-
-    def attach_file_to_selected_item(self):
+    def attach_file_to_selected_item(self, mode):
+        """Attaches a file to the selected item with a specified mode indicating its intended use."""
         file_dialog = QFileDialog(self)
         file_path, _ = file_dialog.getOpenFileName(self, "Select File")
         if file_path:
@@ -106,65 +89,90 @@ def attach_file_to_selected_item(self):
                 if row not in self.itemToFileMap:
                     self.itemToFileMap[row] = []
 
-                temp_file_id = "temp_" + os.path.basename(file_path)
-                self.itemToFileMap[row].append({"file_id": temp_file_id, "file_path": file_path})
+                self.itemToFileMap[row].append({
+                    "file_id": None,  # This will be updated later
+                    "file_path": file_path,
+                    "tools": [{"type": mode}]  # Store the tool type for later use
+                })
 
-                current_instructions = self.itemToInstructionsMap.get(row, "")
-                self.update_item_icon(current_item, self.itemToFileMap[row], current_instructions)
+                self.update_item_icon(current_item, self.itemToFileMap[row])
 
     def remove_specific_file_from_selected_item(self, file_info, row):
+        """Removes a specific file from the selected item based on the file info provided."""
         if row in self.itemToFileMap:
             file_path_to_remove = file_info['file_path']
             self.itemToFileMap[row] = [fi for fi in self.itemToFileMap[row] if fi['file_path'] != file_path_to_remove]
-            
+
             current_item = self.item(row)
-            current_instructions = self.itemToInstructionsMap.get(row, "")
-            if not self.itemToFileMap[row] and not current_instructions:
+            if not self.itemToFileMap[row]:
                 current_item.setIcon(QIcon())
             else:
-                self.update_item_icon(current_item, self.itemToFileMap[row], current_instructions)
+                self.update_item_icon(current_item, self.itemToFileMap[row])
 
-    def update_item_icon(self, item, files, instructions):
-        if files or instructions:
+    def update_item_icon(self, item, files):
+        """Updates the list item's icon based on whether there are attached files."""
+        if files:
             item.setIcon(QIcon("gui/images/paperclip_icon.png"))
         else:
             item.setIcon(QIcon())
 
-    def get_attached_files_for_selected_item(self):
-        """Return the file paths of files attached to the currently selected item."""
+    def get_attachments_for_selected_item(self):
+        """Return the details of files attached to the currently selected item including file path and specific tool usage."""
         current_item = self.currentItem()
         if current_item:
             row = self.row(current_item)
             attached_files_info = self.itemToFileMap.get(row, [])
-            # Extract and return only the file paths
-            return [file_info['file_path'] for file_info in attached_files_info]
+            attachments = []
+            for file_info in attached_files_info:
+                file_path = file_info['file_path']
+                file_name = os.path.basename(file_path)
+                file_id = file_info.get('file_id', None)
+                tools = file_info.get('tools', [])
+
+                # Create a structured entry for the attachments list including file_path
+                attachments.append({
+                    "file_name": file_name,
+                    "file_id": file_id,
+                    "file_path": file_path,  # Include the full file path for upload or further processing
+                    "tools": tools
+                })
+            return attachments
         return []
+
+    def set_attachments_for_selected_item(self, attachments):
+        """Set the attachments for the currently selected item."""
+        current_item = self.currentItem()
+        if current_item is not None:
+            row = self.row(current_item)
+            self.itemToFileMap[row] = attachments[:]
+            self.update_item_icon(current_item, attachments)
+        else:
+            logger.warning("No item is currently selected.")
 
-    def load_threads_with_files_and_instructions(self, threads):
-        """Load threads into the list widget, adding icons for attached files and instructions."""
+    def load_threads_with_attachments(self, threads):
+        """Load threads into the list widget, adding icons for attached files only, based on attachments info."""
         for thread in threads:
             item = QListWidgetItem(thread['thread_name'])
             self.addItem(item)
-            thread_tooltip_text = f"You can add/remove files and instructions by right-clicking this item."
+            thread_tooltip_text = "You can add/remove files by right-clicking this item. NOTE: ChatAssistant will not be able to access the files."
             item.setToolTip(thread_tooltip_text)
 
-            # Get file paths and instructions from the thread data
-            file_paths = thread.get('file_ids', [])
-            additional_instructions = thread.get('additional_instructions', "")
+            # Get attachments from the thread data
+            attachments = thread.get('attachments', [])
 
-            # Update the item with the paperclip icon if there are attached files or instructions
-            self.update_item_with_files_and_instructions(item, file_paths, additional_instructions)
+            # Update the item to reflect any attachments
+            self.update_item_with_attachments(item, attachments)
 
-    def update_item_with_files_and_instructions(self, item, file_paths, additional_instructions=""):
-        """Update the given item with a paperclip icon if there are attached files or instructions."""
+    def update_item_with_attachments(self, item, attachments):
+        """Update the given item with a paperclip icon if there are attachments."""
         row = self.row(item)
-        if file_paths or additional_instructions:
+        if attachments:
             item.setIcon(QIcon("gui/images/paperclip_icon.png"))
         else:
             item.setIcon(QIcon())
 
-        self.itemToFileMap[row] = file_paths
-        self.itemToInstructionsMap[row] = additional_instructions
+        # Store complete attachment information in the mapping
+        self.itemToFileMap[row] = attachments[:]
 
     def keyPressEvent(self, event):
         if event.key() in (Qt.Key_Delete, Qt.Key_Backspace):
@@ -250,6 +258,12 @@ def __init__(self, main_window):
 
         # Create a list widget for displaying the threads
         self.threadList = CustomListWidget(self)
+        self.threadList.setStyleSheet("QListWidget {"
+            "  border-style: solid;"
+            "  border-width: 1px;"
+            "  border-color: #a0a0a0 #ffffff #ffffff #a0a0a0;"  # Light on top and left, dark on bottom and right
+            "  padding: 1px;"
+            "}")
         self.threadList.setFont(QFont("Arial", 11))
 
         # Create connections for the thread and button
@@ -263,6 +277,12 @@ def __init__(self, main_window):
         self.assistantList = QListWidget(self)
         self.assistantList.setFont(QFont("Arial", 11))
         self.assistantList.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Preferred)
+        self.assistantList.setStyleSheet("QListWidget {"
+            "  border-style: solid;"
+            "  border-width: 1px;"
+            "  border-color: #a0a0a0 #ffffff #ffffff #a0a0a0;"  # Light on top and left, dark on bottom and right
+            "  padding: 1px;"
+            "}")
         self.assistantList.itemDoubleClicked.connect(self.on_assistant_double_clicked)
         self.assistantList.setToolTip("Select assistants to use in the conversation or double-click to edit the selected assistant.")
         self.threadList.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Preferred)
@@ -317,6 +337,7 @@ def on_assistant_config_submitted(self, assistant_config_json, ai_client_type, a
             self.assistant_client_manager.register_client(assistant_client.name, assistant_client)
             client_type = AIClientType[ai_client_type]
             self.main_window.conversation_sidebar.load_assistant_list(client_type)
+            self.dialog.update_assistant_combobox()
         except Exception as e:
             QMessageBox.warning(self.main_window, "Error", f"An error occurred while creating/updating the assistant: {e}")
 
@@ -410,18 +431,21 @@ def on_ai_client_type_changed(self, index):
 
             # Clear the existing items in the thread list
             self.threadList.clear()
-            self.threadList.clear_files_and_instructions()
+            self.threadList.clear_files()
 
             # Get the threads for the selected AI client type
             threads_client = ConversationThreadClient.get_instance(self._ai_client_type)
-            # TODO use ai_client_type to retrieve threads from cloud API
             threads = threads_client.get_conversation_threads()
-            self.threadList.load_threads_with_files_and_instructions(threads)
+            self.threadList.load_threads_with_attachments(threads)
         except Exception as e:
             logger.error(f"Error while changing AI client type: {e}")
         finally:
             self.main_window.set_active_ai_client_type(self._ai_client_type)
 
+    def set_attachments_for_selected_thread(self, attachments):
+        """Set the attachments for the currently selected item."""
+        self.threadList.set_attachments_for_selected_item(attachments)
+
     def on_add_thread_button_clicked(self):
         """Handle clicks on the add thread button."""
         # Get the selected assistant
@@ -449,7 +473,7 @@ def create_conversation_thread(self, threads_client : ConversationThreadClient,
             logger.debug(f"Total time taken to create a new conversation thread: {end_time - start_time} seconds")
             new_item = QListWidgetItem(unique_thread_name)
             self.threadList.addItem(new_item)
-            thread_tooltip_text = f"You can add/remove files and instructions by right-clicking this item."
+            thread_tooltip_text = f"You can add/remove files by right-clicking this item. NOTE: ChatAssistant will not be able to access the files."
             new_item.setToolTip(thread_tooltip_text)
 
             if not is_scheduled_task: