cq/scripts/helpers/diff_utils.py

"""Compilation of utility functions for working with file diffs."""

from __future__ import absolute_import
import re

import six

def diff_to_comments(diff):
  r"""Converts a unified diff to a map of comments to be added to a review.

  Comments are formatted according to:
  https://gerrit-review.googlesource.com/Documentation/rest-api-changes.html#comment-input

  Example output:
  {
   'path/to/foo': [{'line': '1', 'message': '+ bar'},
                   {'line': '10', 'message': '+ baz\n- foo'}]
   'path/to/bar': [{'line': '8', 'message': '- biz'}]
  }

  Args:
    diff: output from difflib.unified_diff()

  Returns:
   A dictionary of comments to be added to a review.
   Keys are filenames and values are lists of dictionaries
   with 'line' and 'message' entries.
  """
  comments = {}
  tofile = ''
  line_range = ()
  message_lines = []

  for line in diff:
    line = line.strip()
    if line.startswith('+++'):
      if message_lines:
        # We transitioned from one file's diffs to the next,
        # so add last comment from the previous file
        _add_comment(tofile, line_range, message_lines, comments)
        message_lines = []
      tofile = _get_filename(line)
    elif line.startswith('---'):
      # fromfile name is not used
      continue
    elif line.startswith('@@') and line.endswith('@@'):
      if message_lines:
        _add_comment(tofile, line_range, message_lines, comments)
        message_lines = []
      line_range = _get_linerange(line)
    else:
      message_lines.append(line)

  if message_lines:
    _add_comment(tofile, line_range, message_lines, comments)

  return comments


def _get_filename(line):
  """Returns the filename from a unified diff line.

  The line should be of the format:
  --- filename\tdate\tdate
  +++ filename\tdate\tdate

  Args:
    line: A line from a unified diff output

  Returns:
    the filename in the line, if it exists. '' otherwise.
  """
  assert isinstance(line, six.string_types)
  assert line.startswith('---') or line.startswith('+++')

  if len(line) > 4:
    return line[4:].split('\t', 1)[0]
  else:
    return ''


def _get_linerange(line):
  """Returns a tuple of (start_line, end_line) for this change.

  The line should be of the format:
  @@ -#,# +#,# @@

  Args:
    line: A line from a unified diff output

  Returns:
    The line number of the tofile that the change starts at.
  """
  assert isinstance(line, six.string_types)
  assert line.startswith('@@')
  assert line.endswith('@@')

  matches = re.search(r'-(\d+)(,(\d+))? \+(\d+)(,(\d+))?', line)
  start_line = matches.group(4)
  diff_length = matches.group(6) or 1
  end_line = str(int(start_line) + int(diff_length) - 1)
  return (start_line, end_line)


def _add_comment(filename, line_range, message_lines, comments):
  """Add a comment to the list of comments.

  Args:
    filename: the file the comment applies to
    line_range: the (start_line, end_line) numbers that this comment applies to
    message_lines: a list of lines to put in the comment
    comments: the dictionary of comments to add this comment to
  """
  assert message_lines
  assert len(line_range) == 2

  start_line, end_line = line_range
  start_character = 0
  end_character = max(len(message_lines[-1]) - 1, 0)
  # If the message line is a diff'd line, it will have an extra character
  # at the start that the actual file line does not have.
  if message_lines[-1].startswith(('-', '+')):
    end_character -= 1
  # Prepend a space to each line so gerrit doesn't format the "-" as a list.
  message_lines = [' ' + line for line in message_lines]
  comment = {'range': {'start_line': str(start_line),
                       'start_character': str(start_character),
                       'end_line': str(end_line),
                       'end_character': str(end_character)},
             'message': '\n'.join(message_lines)}
  if filename in comments:
    comments[filename].append(comment)
  else:
    comments[filename] = [comment]