JS Oracle Support for parsing and loading

[AshwinM1523]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

We want to integrate Oracle databases that defines tools to load, parse, and process Oracle documents within LangChain JS version as found in Python. It would features classes and methods that handle document reading, metadata extraction, object ID generation, and text splitting.

Motivation

The motivation for creating a custom Oracle document loader and parser in JavaScript revolves around the need for compatibility with JavaScript-based ecosystems and LangChain.js.

While the Python-based LangChain library includes support for custom loaders that access and process such documents, the JS version currently lacks equivalent functionality. This inconsistency limits the ability of JS developers.

Proposal (If applicable)

Here is a basic pesudocode implementation of the main loader.

class OracleDocLoader {
  constructor(connection, params) {
    this.conn = connection; 
    this.params = params;
    this.reader = new OracleDocReader(connection);
  }

  async loadDocuments() {
    const documents = [];

    // Load from a single file if specified
    if (this.params.file) {
      const doc = await this.reader.readFile(this.params.file);
      if (doc) documents.push(doc);
    }

    // Load all files from a directory if specified
    if (this.params.dir) {
      const files = await fs.readdir(this.params.dir);
      for (const file of files) {
        const filePath = `${this.params.dir}/${file}`;
        const doc = await this.reader.readFile(filePath);
        if (doc) documents.push(doc);
      }
    }

    return documents;
}

This is a basic pseudocode implementation of the loader, we would want need to create OracleDocReader class. Furthermore, we would want to create a parser as well in similar fashion to the python implementation. This can be found in:
LangChain-OracleAi

Example usage

try {
    const connection = await oracledb.getConnection({
      user: 'your_username',
      password: 'your_password',
      connectionString: 'localhost/XE',
    });

    const loader = new OracleDocLoader(connection, { dir: './oracle_docs' });

    const documents = await loader.loadDocuments();
    console.log('Loaded Documents:', documents);

    await connection.close();
  } catch (error) {
    console.error('Error loading Oracle documents:', error);
  }

[AshwinM1523]

Hey @jacoblee93, was wondering if you think this is something we can create an issue for? We are willing to work on it and have started coding it

[jacoblee93]

Sure!

[AshwinM1523]

Thank you so much!