[Improvement] TextSplitter: return chunk start and end position

[chuang8511]

Background

According to LangChain return sources function, we can know letting users know the source is important.

The UX Part to be Improved

However, by the current way, it only returns the raw chunk text rather than pointing out the position from the raw document.
When the users want to check the raw chunk of the raw document, it can not be achieved by the way LangChain provides.

Solution

1. Return start & end position for each chunk when the RAG system providers splits the text.
2. The RAG system providers can use this start & end position to trace back.

Further information

Currently, I am working on another project to achieve this system.
Because I refer to LangChainGo a lot, I think it will be also helpful for the users in this domain.

If you think it is good, I will take time to integrate the logic into langchaingo.