v0.16.5 - Markdown and Code Rendering

RELEASE.md

@@ -1,5 +1,31 @@
 # Releases
 
+## 0.16.5 - Chatbot Updates
+
+Image Edit
+
+* SwarmUI: Added support for sending an image along with the prompt (image+prompt / img2img). If an image is uploaded in the chat UI, `/image {prompt}` will pass it as an init image to SwarmUI, keeping external link behavior. Backwards compatible when no image is attached. Internally, we now populate `rawInput` with `initimage`/`maskimage` data-URLs for the Swarm API.
+* New command: `/image edit {prompt}` explicitly selects edit mode. Requires an uploaded image and applies a default edit noise/strength (configurable via `IMAGE_EDIT_NOISE`, default 0.35). Helpful for workflows like object additions, style changes, or masked edits when combined with a `maskimage` in future UI updates.
+
+Markdown Rendering in Chat UI
+
+* Full Markdown Support: Integrated `marked.js` library (v5+) with GitHub-flavored markdown for comprehensive rendering of LLM responses.
+* Supported Elements: Headers (h1-h6 with borders), bold/italic text, tables (auto-width with tight padding), bullet/numbered lists (compact spacing), blockquotes, horizontal rules, inline code, code blocks, and hyperlinks.
+* Live Streaming: Real-time markdown rendering during token streaming with 300ms debounced parsing for smooth display.
+* Code Block Features:
+  - Language detection with "Copy [language]" buttons positioned at top-right of code blocks.
+  - Event delegation pattern ensures copy functionality persists across all interactions.
+* Raw/Rendered Toggle: Blue "View Raw" button appears on hover at bottom-right of responses, allowing users to switch between rendered markdown and raw text. Toggle persists across multiple conversation turns.
+* Optimized Styling: Table styling with `width: auto` and `table-layout: auto` for content-fitted display, compact list spacing (`line-height: 1.3`, `margin: 0` on list items), and consistent formatting across all markdown elements.
+
+Static Assets & UI Improvements
+
+* Local JavaScript Libraries: Moved to local copies of `marked.js` and `socket.io.js` for air-gapped deployment support. All static assets now served from `/static/` directory.
+* Consolidated Static File Serving: Single `/static/{filename}` route handles all static assets (JS, CSS, SVG) with automatic MIME type detection.
+* Custom Favicon: Added SVG favicon (blue circle with white "T") for professional browser tab appearance.
+* Proper MIME Types: Added support for `.svg` files with `image/svg+xml` content type to ensure proper rendering in browsers.
+
+
 ## 0.16.4 - Document Generation & Security
 
 * Chatbot - Document Generation: End-to-end document creation (PDF, DOCX, XLSX, PPTX) with download links.
@@ -36,7 +62,6 @@ Example Usage
 
 > Write an algorithm in python to sort a list of number. Create a PDF to describe how it works.
 
-
 ## 0.16.2 - Repetition Filter Settings
 
 * Added environment variables for repetition filter: `REPEAT_WINDOW` and `REPEAT_COUNT`.

chatbot/README.md

@@ -417,3 +417,13 @@ The Chatbot can use this information if you send the prompt command:
 /rag records How much did we donate to charity in 2022?
 /rag blog 5 List some facts about solar energy.
 ```
+
+## Credits
+
+This project uses the following open-source libraries:
+
+- **[Prism.js](https://prismjs.com/)** - Syntax highlighting (MIT License)
+- **[marked.js](https://marked.js.org/)** - Markdown parsing (MIT License)
+- **[Socket.IO](https://socket.io/)** - Real-time communication (MIT License)
+
+See [THIRD-PARTY-LICENSES.md](THIRD-PARTY-LICENSES.md) for full license details.

chatbot/RELEASE.md

@@ -4,6 +4,15 @@
 
 The TinyLLM Chatbot docker container is available at: [jasonacox/chatbot](https://hub.docker.com/r/jasonacox/chatbot).
 
+## 0.16.5 - Enhanced Markdown and Code Rendering
+
+* Added syntax highlighting for code blocks using Prism.js with support for Python, JavaScript, and HTML.
+* Improved streaming output rendering with proper HTML escaping to prevent code interpretation during display.
+* Enhanced code block styling with dark theme background and optimized font size and line spacing.
+* Fixed content jump issue when streaming text transitions to rendered markdown by adding consistent margin spacing.
+* Added comprehensive third-party license attribution for all frontend and backend libraries.
+* All static assets (Prism.js, marked.js, Socket.IO) now served locally to support air-gapped deployments.
+
 ## 0.16.4 - Document Generation Support
 
 * Added comprehensive document generation functionality with support for PDF, Word, Excel, and PowerPoint formats.

chatbot/THIRD-PARTY-LICENSES.md

@@ -0,0 +1,170 @@
+# Third-Party Software Licenses
+
+This project uses the following third-party libraries:
+
+## Frontend Libraries (JavaScript/CSS)
+
+### Prism.js
+- **License**: MIT License
+- **Copyright**: Copyright (c) 2012 Lea Verou
+- **Website**: https://prismjs.com/
+- **Purpose**: Syntax highlighting for code blocks in the chat interface
+- **Files**: 
+  - `app/static/prism.min.js`
+  - `app/static/prism.min.css`
+  - `app/static/prism-python.min.js`
+  - `app/static/prism-javascript.min.js`
+  - `app/static/prism-markup.min.js`
+
+### marked.js
+- **License**: MIT License
+- **Copyright**: Copyright (c) 2018+, MarkedJS (https://github.com/markedjs/)
+- **Website**: https://marked.js.org/
+- **Purpose**: Markdown parsing and rendering in the chat interface
+- **Files**: `app/static/marked.min.js`
+
+### Socket.IO
+- **License**: MIT License
+- **Copyright**: Copyright (c) 2014-present Automattic <dev@cloudup.com>
+- **Website**: https://socket.io/
+- **Purpose**: Real-time bidirectional communication between client and server
+- **Files**: `app/static/socket.io.js`
+
+## Backend Libraries (Python)
+
+### FastAPI
+- **License**: MIT License
+- **Copyright**: Copyright (c) 2018 Sebastián Ramírez
+- **Website**: https://fastapi.tiangolo.com/
+- **Purpose**: Modern web framework for building APIs
+
+### Uvicorn
+- **License**: BSD License
+- **Website**: https://www.uvicorn.org/
+- **Purpose**: ASGI web server implementation
+
+### Python-SocketIO
+- **License**: MIT License
+- **Website**: https://python-socketio.readthedocs.io/
+- **Purpose**: Server-side Socket.IO implementation
+
+### OpenAI Python Client
+- **License**: MIT License
+- **Website**: https://github.com/openai/openai-python
+- **Purpose**: OpenAI API client for LLM interaction
+
+### Beautiful Soup 4 (bs4)
+- **License**: MIT License
+- **Website**: https://www.crummy.com/software/BeautifulSoup/
+- **Purpose**: HTML/XML parsing for web scraping
+
+### PyPDF
+- **License**: BSD License
+- **Website**: https://github.com/py-pdf/pypdf
+- **Purpose**: PDF file reading and text extraction
+
+### Pillow & pillow-heif
+- **License**: HPND License (Historical Permission Notice and Disclaimer)
+- **Website**: https://python-pillow.org/
+- **Purpose**: Image processing and HEIF format support
+
+### Weaviate Client
+- **License**: BSD-3-Clause License
+- **Website**: https://weaviate.io/
+- **Purpose**: Vector database client for RAG functionality
+
+### ReportLab
+- **License**: BSD License
+- **Website**: https://www.reportlab.com/opensource/
+- **Purpose**: PDF document generation
+
+### python-docx
+- **License**: MIT License
+- **Website**: https://python-docx.readthedocs.io/
+- **Purpose**: Microsoft Word document generation
+
+### python-pptx
+- **License**: MIT License
+- **Website**: https://python-pptx.readthedocs.io/
+- **Purpose**: Microsoft PowerPoint document generation
+
+### openpyxl
+- **License**: MIT License
+- **Website**: https://openpyxl.readthedocs.io/
+- **Purpose**: Excel spreadsheet generation
+
+### pypandoc
+- **License**: MIT License
+- **Website**: https://github.com/bebraw/pypandoc
+- **Purpose**: Document format conversion
+
+### aiohttp
+- **License**: Apache License 2.0
+- **Website**: https://docs.aiohttp.org/
+- **Purpose**: Async HTTP client/server framework
+
+### Requests
+- **License**: Apache License 2.0
+- **Website**: https://requests.readthedocs.io/
+- **Purpose**: HTTP library for API calls
+
+### lxml
+- **License**: BSD License
+- **Website**: https://lxml.de/
+- **Purpose**: XML and HTML processing
+
+### Jinja2
+- **License**: BSD License
+- **Website**: https://jinja.palletsprojects.com/
+- **Purpose**: Template engine for HTML rendering
+
+### Pydantic
+- **License**: MIT License
+- **Website**: https://pydantic-docs.helpmanual.io/
+- **Purpose**: Data validation using Python type annotations
+
+### python-dotenv
+- **License**: BSD License
+- **Website**: https://github.com/theskumar/python-dotenv
+- **Purpose**: Environment variable management
+
+### pandas
+- **License**: BSD License
+- **Website**: https://pandas.pydata.org/
+- **Purpose**: Data manipulation and analysis
+
+---
+
+## MIT License Text
+
+```
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+```
+
+## Apache License 2.0
+
+Libraries using Apache License 2.0 (aiohttp, requests) are licensed under terms that allow free use, modification, and distribution. Full license text available at: https://www.apache.org/licenses/LICENSE-2.0
+
+## BSD License
+
+Libraries using BSD License (Uvicorn, PyPDF, ReportLab, Jinja2, python-dotenv, pandas, lxml, Weaviate Client) are licensed under permissive terms similar to MIT. Each has slight variations but all allow free use, modification, and distribution with attribution.
+
+---
+
+All third-party libraries are used in accordance with their respective licenses.

chatbot/app/api/routes.py

@@ -27,7 +27,7 @@
 
 # Third-party imports
 from fastapi import FastAPI, Request, File, UploadFile, Form, HTTPException
-from fastapi.responses import HTMLResponse, FileResponse
+from fastapi.responses import HTMLResponse, FileResponse, Response
 from fastapi.templating import Jinja2Templates
 from PIL import Image
 import pillow_heif
@@ -153,10 +153,28 @@ async def lifespan(app: FastAPI): # pylint: disable=unused-argument,redefined-ou
 async def index(request: Request):
     return templates.TemplateResponse(request, "index.html")
 
-# Serve static socket.io.js
-@app.get("/socket.io.js")
-def serve_socket_io_js():
-    return FileResponse("app/templates/socket.io.js", media_type="application/javascript")
+# Serve favicon.ico (return 204 No Content to avoid console errors)
+@app.get("/favicon.ico")
+async def favicon():
+    # Serve the SVG favicon
+    return FileResponse("app/static/favicon.svg", media_type="image/svg+xml")
+
+# Serve static files from app/static directory
+@app.get("/static/{filename}")
+def serve_static_files(filename: str):
+    file_path = f"app/static/{filename}"
+    if os.path.exists(file_path):
+        # Determine media type based on file extension
+        if filename.endswith('.js'):
+            media_type = "application/javascript"
+        elif filename.endswith('.css'):
+            media_type = "text/css"
+        elif filename.endswith('.svg'):
+            media_type = "image/svg+xml"
+        else:
+            media_type = "application/octet-stream"
+        return FileResponse(file_path, media_type=media_type)
+    raise HTTPException(status_code=404, detail="File not found")
 
 # Display settings and stats
 @app.get("/stats")
@@ -204,6 +222,12 @@ async def home(format: str = None):
         "Image Provider (IMAGE_PROVIDER)": IMAGE_PROVIDER,
         "SwarmUI Host (SWARMUI)": SWARMUI if IMAGE_PROVIDER == "swarmui" else "Not Used",
     }
+    # Image edit config
+    try:
+        from app.core.config import IMAGE_EDIT_NOISE
+        data["Image Edit Default Noise (IMAGE_EDIT_NOISE)"] = IMAGE_EDIT_NOISE
+    except Exception:
+        pass
     # Add supported document formats (aliases) to status page
     try:
         doc_gen = get_document_generator()
@@ -787,7 +811,7 @@ async def handle_url_prompt(session_id, p):
 async def handle_command(session_id, p):
     command = p[1:].split(" ")[0].lower()
     if command == "":
-        await sio.emit('update', {'update': '[Commands: /image /intent /model /news /rag /reset /search /sessions /stock /think /version /weather /repeat]', 'voice': 'user'}, room=session_id)
+        await sio.emit('update', {'update': '[Commands: /image (or /image edit) /intent /model /news /rag /reset /search /sessions /stock /think /version /weather /repeat]', 'voice': 'user'}, room=session_id)
         client[session_id]["prompt"] = ''
     elif command == "repeat":
         await handle_repeat_command(session_id, p)
@@ -1051,17 +1075,73 @@ async def handle_image_command(session_id, p):
     if not current_image_generator:
         await sio.emit('update', {'update': '[Image Generation is not enabled]', 'voice': 'user'}, room=session_id)
         return
-    prompt = p[6:].strip()
-    if not prompt:
-        await sio.emit('update', {'update': '[Usage: /image {prompt}] - Generate image for prompt.', 'voice': 'user'}, room=session_id)
+    raw_args = p[6:].strip()
+    if not raw_args:
+        await sio.emit('update', {'update': '[Usage: /image {prompt} | /image edit {prompt}]', 'voice': 'user'}, room=session_id)
         return
+    # Support explicit edit intent: "/image edit {prompt}"
+    is_edit = False
+    prompt = raw_args
+    parts = raw_args.split()
+    if parts and parts[0].lower() in ["edit", "eidt"]:
+        is_edit = True
+        prompt = ' '.join(parts[1:]).strip()
+        if not prompt:
+            await sio.emit('update', {'update': '[Usage: /image edit {prompt}] - Provide a prompt describing the edit.', 'voice': 'user'}, room=session_id)
+            return
     await sio.emit('update', {'update': '%s [Generating Image...]' % p, 'voice': 'user'}, room=session_id)
     debug(f"Image Prompt: {prompt}")
     client[session_id]["visible"] = False
     client[session_id]["remember"] = True
     # Generate image from current_image_generator
+    # If the user has uploaded an image already, pass it along as an init image (image+prompt)
+    gen_kwargs = {}
+    try:
+        client_image_data = client[session_id].get("image_data")
+    except Exception:  # pragma: no cover - defensive
+        client_image_data = None
+    if client_image_data:
+        # client_image_data is base64 without header from our upload route
+        # Some frontends may include data URLs already; handle both
+        if client_image_data.startswith("data:image/"):
+            gen_kwargs["init_image_data_url"] = client_image_data
+            # Debug: show a short preview of the image data URL to keep logs readable
+            try:
+                preview = client_image_data[:80]
+                debug(f"Init image (data URL) preview: {preview}... (len={len(client_image_data)})")
+            except Exception:
+                pass
+        else:
+            gen_kwargs["init_image_b64"] = client_image_data
+            # Debug: show a short preview of the raw base64 image payload
+            try:
+                preview = client_image_data[:80]
+                debug(f"Init image (base64) preview: {preview}... (len={len(client_image_data)})")
+            except Exception:
+                pass
+        debug("Passing init image to generator for image+prompt mode")
+    elif is_edit:
+        # Edit mode requires an init image
+        await sio.emit('update', {'update': '[Image Edit requires an uploaded image. Drag-and-drop an image, then run: /image edit {prompt}]', 'voice': 'user'}, room=session_id)
+        return
+
+    # Apply default edit noise for explicit edit flow
+    if is_edit:
+        try:
+            from app.core.config import IMAGE_EDIT_NOISE
+            gen_kwargs["init_image_noise"] = IMAGE_EDIT_NOISE
+        except Exception:
+            pass
+        # If using Flux base models, stronger edit often needs max creativity (1.0) or dedicated edit models
+        try:
+            model_name = getattr(current_image_generator, "model", "") or IMAGE_MODEL
+            if isinstance(model_name, str) and "flux" in model_name.lower():
+                gen_kwargs["init_image_noise"] = 1.0
+                await sio.emit('update', {'update': '[Note: Flux base models may ignore edits without a mask. Consider Flux Kontext/Fill or SDXL Inpainting for stronger edits.]', 'voice': 'user'}, room=session_id)
+        except Exception:
+            pass
 
-    image_encoded = await current_image_generator.generate(prompt)
+    image_encoded = await current_image_generator.generate(prompt, **gen_kwargs)
     if image_encoded:
         # Handle different response formats - OpenAI returns raw base64, SwarmUI returns data URI
         if "," in image_encoded: