lcov_parser/changed_lines.py

"""This module parses `git diff` output to determine the added/changed
lines per file.
"""
from typing import Tuple, Text, List, Dict
import re


class GitDiffSyntaxError(Exception):
  """GitDiffSyntaxError is raised when there are formatting issues in the `git diff`
  command output.
  """
  pass


def from_diff(git_diff: Text) -> Dict[Text, List[int]]:
    """Converts `git diff` output into a map of files to their changed lines
    (in the new version).
    """

    git_diff = git_diff.strip()

    # check if `git diff` returned nothing, meaning no files had modified lines
    if git_diff == '':
        return {}

    # splits the `git diff` result into per-file data
    file_breakdown = git_diff.split('diff --git ')

    # .split() returns an empty string at index 0 if the input starts with the separator
    file_breakdown = file_breakdown[1:]

    file_to_lines = {}  # type: Dict[Text, List[int]]
    for file_diff in file_breakdown:
        file_name, changed_lines = _parse_file_diff(file_diff)

        # this might happen if only lines were deleted but none were added
        if not changed_lines:
            continue

        file_to_lines[file_name] = sorted(changed_lines)

    return file_to_lines


def _parse_file_diff(file_diff: Text) -> Tuple[Text, List[int]]:
    """Takes in a section of the `git diff` command that contains
    all the information about a given file and returns the file's name and it's associated
    changed lines.

    Raises:
        GitDiffSyntaxError if there are formatting issues parsing file name in `file_diff`.
    """

    first_line = file_diff.split('\n')[0]

    m = re.search(r'a/(\S+) b/\S+', first_line)
    if not m:
        raise GitDiffSyntaxError('Expected first line to contain `a/fileName b/fileName`,' + \
                                 'got: ' + first_line)

    file_name = m.group(1)

    hunks = file_diff.split('\n@@ ')[1:]

    changed_lines = []

    for hunk in hunks:
        changed_lines += _changed_lines_in_hunk(hunk)

    return (file_name, changed_lines)


def _changed_lines_in_hunk(hunk: Text) -> List[int]:
    """Returns the added/modified line numbers in a hunk (in the new version).

    Raises:
        GitDiffSyntaxError if there are formatting issues with the first line
            of `hunk`.
    """
    lines = hunk.split('\n')
    changed_lines = []

    m = re.search(r'-\d+,\d+ \+(\d+),\d+ @@', lines[0])
    if not m:
        raise GitDiffSyntaxError('Expected first line in hunk to be in the format:' + \
                                 '\n\t-#,# +#,# @@\ngot: ' + lines[0])

    starting_line_number = int(m.group(1))

    curr_line = starting_line_number

    for line in lines[1:]:
        if line == '' or line[0] == '-':
            continue

        if line[0] == '+':
            changed_lines.append(curr_line)

        curr_line += 1


    return sorted(changed_lines)