Need help using FigmaFileLoader for large Figma designs

[dangn511]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from langchain_openai import OpenAIEmbeddings
from langchain.indexes import VectorstoreIndexCreator
from langchain_community.document_loaders.figma import FigmaFileLoader

figma_loader = FigmaFileLoader(
    os.environ.get("ACCESS_TOKEN"),
    os.environ.get("NODE_IDS"),
    os.environ.get("FILE_KEY"),
)

embeddings = OpenAIEmbeddings()
index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([figma_loader])
figma_doc_retriever = index.vectorstore.as_retriever()

Description

I'm following the tutorial here https://python.langchain.com/docs/integrations/document_loaders/figma/ for making an agent to write code from Figma design. I'm having trouble following the tutorial, as my Figma design is rather large, and the serialized data exceeded the token limit for the embedding.

Here's the console output after running the above sample code on my Figma design:

C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain\indexes\vectorstore.py:171: UserWarning: Using InMemoryVectorStore as the default vectorstore.This memory store won't persist data. You should explicitlyspecify a vectorstore when using VectorstoreIndexCreator
  warnings.warn(
Traceback (most recent call last):
  File "C:\Users\User\PycharmProjects\figma_json_coder\figma_test_scratch.py", line 96, in <module>
    index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([figma_loader])
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain\indexes\vectorstore.py", line 206, in from_loaders
    return self.from_documents(docs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain\indexes\vectorstore.py", line 233, in from_documents
    vectorstore = self.vectorstore_cls.from_documents(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_core\vectorstores\base.py", line 844, in from_documents
    return cls.from_texts(texts, embedding, metadatas=metadatas, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_core\vectorstores\in_memory.py", line 521, in from_texts
    store.add_texts(texts=texts, metadatas=metadatas, **kwargs)
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_core\vectorstores\base.py", line 112, in add_texts
    return self.add_documents(docs, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_core\vectorstores\in_memory.py", line 187, in add_documents
    vectors = self.embedding.embed_documents(texts)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_openai\embeddings\base.py", line 588, in embed_documents
    return self._get_len_safe_embeddings(texts, engine=engine)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\langchain_openai\embeddings\base.py", line 483, in _get_len_safe_embeddings
    response = self.client.create(
               ^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\openai\resources\embeddings.py", line 128, in create
    return self._post(
           ^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\openai\_base_client.py", line 1242, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\openai\_base_client.py", line 919, in request
    return self._request(
           ^^^^^^^^^^^^^^
  File "C:\Users\User\PycharmProjects\figma_json_coder\venv\Lib\site-packages\openai\_base_client.py", line 1023, in _request
    raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': 'Requested 374225 tokens, max 300000 tokens per request', 'type': 'max_tokens_per_request', 'param': None, 'code': 'max_tokens_per_request'}}

Process finished with exit code 1

I've been trying to work around it by building my own VectorStore indexer builder, however it seems like FigmaFileLoader is not returning a JSON formatted output.

document = figma_loader.load()
print(document[0].page_content)

gives me a string of text that is not in JSON format, like so

id: 302:1290
name: 65%
type: TEXT
scrollBehavior: SCROLLS
blendMode: PASS_THROUGH
fills: 
blendMode: NORMAL
type: SOLID
color: 
r: 1.0
g: 1.0
b: 1.0
a: 1.0


strokes: 
strokeWeight: 1.0
strokeAlign: OUTSIDE
absoluteBoundingBox: 
x: 4294.0
y: 1378.0
width: 62.0
height: 41.0

absoluteRenderBounds: 
x: 4295.6689453125
y: 1386.62890625
width: 59.18359375
height: 24.7861328125
[......]

I'd like to know

1. Is there another approach to using built in VectorstoreIndexCreator to handle very large Figma outputs?
2. If not, how can I get my Figmal design outputted into a proper JSON format in order to write my own splitter and embedder?

System Info

langchain = 0.3.20
langchain_community = 0.3.19
Platform: Windows 11
Python 3.11.0