@thi.ng/csv/parse.js

Customizable, transducer-based CSV parser/object mapper and transformer

import { isArray } from "@thi.ng/checks/is-array";
import { isFunction } from "@thi.ng/checks/is-function";
import { isIterable } from "@thi.ng/checks/is-iterable";
import { illegalArgs } from "@thi.ng/errors/illegal-arguments";
import { ESCAPES } from "@thi.ng/strings/escape";
import { split } from "@thi.ng/strings/split";
import { compR } from "@thi.ng/transducers/compr";
import { iterator1 } from "@thi.ng/transducers/iterator";
const DEFAULT_OPTS = {
  delim: ",",
  quote: '"',
  comment: "#",
  trim: false
};
function parseCSV(opts, src) {
  return isIterable(src) ? iterator1(parseCSV(opts), src) : (rfn) => {
    const {
      all = true,
      cols,
      comment,
      delim,
      header,
      quote,
      trim
    } = {
      ...DEFAULT_OPTS,
      ...opts
    };
    const reduce = rfn[2];
    let index;
    let revIndex;
    let defaults;
    let first = true;
    let isQuoted = false;
    let record = [];
    const init = (header2) => {
      if (cols) {
        index = __initIndex(header2, cols);
        defaults = __initDefaults(cols);
      }
      if (all) revIndex = __initRevIndex(header2);
      first = false;
    };
    const collectAll = (row) => record.reduce(
      (acc, x, i) => (acc[revIndex[i]] = trim ? x.trim() : x, acc),
      row
    );
    const collectIndexed = (row) => Object.entries(index).reduce((acc, [id, { i, spec }]) => {
      let val = record[i];
      if (val !== void 0) {
        trim && (val = val.trim());
        all && spec.alias && delete acc[id];
        acc[spec.alias || id] = spec.tx ? spec.tx(val, acc) : val;
      }
      return acc;
    }, row);
    const collectDefaults = (row) => defaults.reduce((acc, { alias, default: val }) => {
      if (acc[alias] === void 0 || acc[alias] === "") {
        acc[alias] = isFunction(val) ? val(acc) : val;
      }
      return acc;
    }, row);
    header && init(header);
    return compR(rfn, (acc, line) => {
      if ((!line.length || line.startsWith(comment)) && !isQuoted)
        return acc;
      if (!first) {
        isQuoted = __parseLine(
          line,
          record,
          isQuoted,
          delim,
          quote
        );
        if (isQuoted) return acc;
        const row = {};
        all && collectAll(row);
        index && collectIndexed(row);
        defaults && collectDefaults(row);
        record = [];
        return reduce(acc, row);
      } else {
        isQuoted = __parseLine(
          line,
          record,
          isQuoted,
          delim,
          quote
        );
        if (!isQuoted) {
          init(record);
          record = [];
        }
        return acc;
      }
    });
  };
}
function parseCSVSimple(opts, src) {
  return isIterable(src) ? iterator1(parseCSVSimple(opts), src) : (rfn) => {
    const {
      header = true,
      cols,
      comment,
      delim,
      quote,
      trim
    } = {
      ...DEFAULT_OPTS,
      ...opts
    };
    const reduce = rfn[2];
    let first = header;
    let isQuoted = false;
    let record = [];
    const collect = () => cols.reduce((acc, col, i) => {
      if (col) {
        let val = record[i];
        if (val !== void 0) {
          trim && (val = val.trim());
          acc.push(isFunction(col) ? col(val, acc) : val);
        }
      }
      return acc;
    }, []);
    return compR(rfn, (acc, line) => {
      if ((!line.length || line.startsWith(comment)) && !isQuoted)
        return acc;
      if (!first) {
        isQuoted = __parseLine(
          line,
          record,
          isQuoted,
          delim,
          quote
        );
        if (isQuoted) return acc;
        const row = cols ? collect() : record;
        record = [];
        return reduce(acc, row);
      } else {
        isQuoted = __parseLine(
          line,
          record,
          isQuoted,
          delim,
          quote
        );
        first = false;
        record = [];
        return acc;
      }
    });
  };
}
const parseCSVFromString = (opts, src) => parseCSV(opts, split(src));
const parseCSVSimpleFromString = (opts, src) => parseCSVSimple(opts, split(src));
const __parseLine = (line, acc, isQuoted, delim, quote) => {
  let curr = "";
  let p = "";
  let openQuote = isQuoted;
  for (let i = 0, n = line.length; i < n; i++) {
    const c = line[i];
    if (p === "\\") {
      curr += ESCAPES[c] || c;
    } else if (c === quote) {
      if (!isQuoted) {
        p = "";
        isQuoted = true;
        continue;
      } else if (p === quote) {
        curr += quote;
        p = "";
        continue;
      } else if (line[i + 1] !== quote) isQuoted = false;
    } else if (!isQuoted && c === delim) {
      __collectCell(acc, curr, openQuote);
      openQuote = false;
      curr = "";
    } else if (c !== "\\") {
      curr += c;
    }
    p = c;
  }
  curr !== "" && __collectCell(acc, curr, openQuote);
  return isQuoted;
};
const __collectCell = (acc, curr, openQuote) => openQuote ? acc[acc.length - 1] += "\n" + curr : acc.push(curr);
const __initIndex = (line, cols) => isArray(cols) ? cols.reduce((acc, spec, i) => {
  if (spec) {
    const alias = spec.alias || line[i] || String(i);
    acc[alias] = { i, spec: { alias, ...spec } };
  }
  return acc;
}, {}) : line.reduce(
  (acc, id, i) => cols[id] ? (acc[id] = { i, spec: cols[id] }, acc) : acc,
  {}
);
const __initRevIndex = (line) => line.reduce((acc, x, i) => (acc[i] = x, acc), {});
const __initDefaults = (cols) => {
  const defaults = isArray(cols) ? cols.filter((c) => {
    if (!c || c.default == void 0) return false;
    if (!c.alias)
      illegalArgs(
        `missing column alias for default: ${c.default}`
      );
    return true;
  }) : Object.entries(cols).reduce((acc, [k, v]) => {
    if (v.default !== void 0) {
      acc.push({
        alias: v.alias ?? k,
        default: v.default
      });
    }
    return acc;
  }, []);
  return defaults.length ? defaults : void 0;
};
export {
  parseCSV,
  parseCSVFromString,
  parseCSVSimple,
  parseCSVSimpleFromString
};