csv2yaml.py

#!/usr/bin/env python
"""
Translate a CSV document with a header to a YAML document that is a list of maps.
Uses constant memory and keeps field names in the original order.
"""

import sys
import csv

import yaml  # requires PyYAML


def read_csv(input):
    """
    :param input: stream
    :return: iterator of (field name, field value) pairs
    """
    csv_reader = csv.reader(input)
    fields = next(csv_reader)
    for row in csv_reader:
        yield zip(fields, row)


def write_yaml(pairs, output):
    """
    :param pairs: iterator of (field name, field value) pairs
    :param output: stream
    """
    yaml_writer = yaml.Dumper(output)
    yaml_writer.open()
    yaml_writer.emit(yaml.DocumentStartEvent())
    yaml_writer.emit(yaml.SequenceStartEvent(anchor=None, tag=None, implicit=True))
    for pair in pairs:
        yaml_writer.emit(yaml.MappingStartEvent(anchor=None, tag=None, implicit=True))
        for name, value in pair:
            yaml_writer.emit(yaml.ScalarEvent(anchor=None, tag=None, implicit=(True, True), value=name))
            yaml_writer.emit(yaml.ScalarEvent(anchor=None, tag=None, implicit=(True, True), value=value))
        yaml_writer.emit(yaml.MappingEndEvent())
    yaml_writer.emit(yaml.SequenceEndEvent())
    yaml_writer.emit(yaml.DocumentEndEvent())
    yaml_writer.close()


def main(input, output):
    write_yaml(read_csv(input), output)


if __name__ == '__main__':
    main(sys.stdin, sys.stdout)