{
"cells": [
{
"attachments": {},
"cell_type": "markdown",
"id": "33205b12",
"metadata": {},
"source": [
"# Figma\n",
"\n",
"This notebook covers how to load data from the Figma REST API into a format that can be ingested into LangChain, along with example usage for code generation."
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "90b69c94",
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"\n",
"\n",
"from langchain.document_loaders.figma import FigmaFileLoader\n",
"\n",
"from langchain.text_splitter import CharacterTextSplitter\n",
"from langchain.chat_models import ChatOpenAI\n",
"from langchain.indexes import VectorstoreIndexCreator\n",
"from langchain.chains import ConversationChain, LLMChain\n",
"from langchain.memory import ConversationBufferWindowMemory\n",
"from langchain.prompts.chat import (\n",
" ChatPromptTemplate,\n",
" SystemMessagePromptTemplate,\n",
" AIMessagePromptTemplate,\n",
" HumanMessagePromptTemplate,\n",
")"
]
},
{
"attachments": {},
"cell_type": "markdown",
"id": "d809744a",
"metadata": {},
"source": [
"The Figma API Requires an access token, node_ids, and a file key.\n",
"\n",
"The file key can be pulled from the URL. https://www.figma.com/file/{filekey}/sampleFilename\n",
"\n",
"Node IDs are also available in the URL. Click on anything and look for the '?node-id={node_id}' param.\n",
"\n",
"Access token instructions are in the Figma help center article: https://help.figma.com/hc/en-us/articles/8085703771159-Manage-personal-access-tokens"
]
},
{
"cell_type": "code",
"execution_count": 2,
"id": "13deb0f5",
"metadata": {},
"outputs": [],
"source": [
"figma_loader = FigmaFileLoader(\n",
" os.environ.get('ACCESS_TOKEN'),\n",
" os.environ.get('NODE_IDS'),\n",
" os.environ.get('FILE_KEY')\n",
")"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "9ccc1e2f",
"metadata": {},
"outputs": [],
"source": [
"# see https://python.langchain.com/en/latest/modules/indexes/getting_started.html for more details\n",
"index = VectorstoreIndexCreator().from_loaders([figma_loader])\n",
"figma_doc_retriever = index.vectorstore.as_retriever()"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "3e64cac2",
"metadata": {},
"outputs": [],
"source": [
"def generate_code(human_input):\n",
" # I have no idea if the Jon Carmack thing makes for better code. YMMV.\n",
" # See https://python.langchain.com/en/latest/modules/models/chat/getting_started.html for chat info\n",
" system_prompt_template = \"\"\"You are expert coder Jon Carmack. Use the provided design context to create idomatic HTML/CSS code as possible based on the user request.\n",
" Everything must be inline in one file and your response must be directly renderable by the browser.\n",
" Figma file nodes and metadata: {context}\"\"\"\n",
"\n",
" human_prompt_template = \"Code the {text}. Ensure it's mobile responsive\"\n",
" system_message_prompt = SystemMessagePromptTemplate.from_template(system_prompt_template)\n",
" human_message_prompt = HumanMessagePromptTemplate.from_template(human_prompt_template)\n",
" # delete the gpt-4 model_name to use the default gpt-3.5 turbo for faster results\n",
" gpt_4 = ChatOpenAI(temperature=.02, model_name='gpt-4')\n",
" # Use the retriever's 'get_relevant_documents' method if needed to filter down longer docs\n",
" relevant_nodes = figma_doc_retriever.get_relevant_documents(human_input)\n",
" conversation = [system_message_prompt, human_message_prompt]\n",
" chat_prompt = ChatPromptTemplate.from_messages(conversation)\n",
" response = gpt_4(chat_prompt.format_prompt( \n",
" context=relevant_nodes, \n",
" text=human_input).to_messages())\n",
" return response"
]
},
{
"cell_type": "code",
"execution_count": null,
"id": "36a96114",
"metadata": {},
"outputs": [],
"source": [
"response = generate_code(\"page top header\")"
]
},
{
"attachments": {},
"cell_type": "markdown",
"id": "baf9b2c9",
"metadata": {},
"source": [
"Returns the following in `response.content`:\n",
"```\n",
"\\n\\n\\n \\n \\n \\n\\n\\n \\n\\n\n",
"```"
]
},
{
"cell_type": "markdown",
"id": "38827110",
"metadata": {},
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.10"
}
},
"nbformat": 4,
"nbformat_minor": 5
}