@stordata/grammars/src/visitors/CSVParserVisitor.js

A collection of ANTLR grammars used at Stordata. This project exists so that we can package the grammars (and various utilities) as a CommonJS module. The `antlr4` Javascript runtime is only available as an ES module at the time of writing.

import CSVVisitor from '../../generated/src/grammars/CSVVisitor.js';

export default class CSVParserVisitor extends CSVVisitor {

  constructor() {
    super();

    this.lines = [];
    this.headers = [];
    this.delimiter = undefined;
  }

  visitAll(ctx) {
    super.visitAll(ctx);

    return {
      lines: this.lines,
      headers: this.headers,
      delimiter: this.delimiter
    };
  }

  visitHeaders(ctx) {
    super.visitHeaders(ctx);

    this.headers = this.lines.shift();
  }

  visitLine(ctx) {
    this.line = [];

    super.visitLine(ctx);
    this.lines.push(this.line);
  }

  visitColumn(ctx) {
    let text = ctx.getText().trim();

    const isQuoted = text.startsWith('"') && text.endsWith('"');

    if (isQuoted) {
      text = text.slice(1, -1).replaceAll('""', '"');
    }

    const index = ctx.parentCtx.children.indexOf(ctx);

    if (index === 0) { // First column is always the start of a new line
      this.line.push(text);
    } else {
      const delimiter = ctx.parentCtx.children[index - 1].getText();

      if (!this.delimiter) { // This is the very first delimiter found in the file
        this.delimiter = delimiter;
      }

      if (delimiter !== this.delimiter) { // This column was split erroneously, merge it with the previous one
        this.line[this.line.length - 1] += delimiter + ctx.getText();
      } else {
        this.line.push(text);
      }
    }
  }

}