gistfile1.py

import re


diff_re = re.compile(
    "@@ \-(?P<removed_start>\d+),(?P<removed_length>\d+) "
    "\+(?P<added_start>\d+),(?P<added_length>\d+) @@"
)

class DiffContext:

    def __init__(self, lines, path):
        self.lines = lines
        self.path = path


class DiffContextParser:

    def __init__(self, diff_text):
        self.diff_text = diff_text

    def parse(self):
        before_contexts, after_contexts = [], []
        lines_before, lines_after = [], []
        before_line_number, after_line_number = 0, 0
        before_path = after_path = ''
        position = 0
        for line in self.diff_text.splitlines()[2:]:
            header = diff_re.search(line)
            if header is not None:
                before_line_number = int(header.group('removed_start'))
                after_line_number = int(header.group('added_start'))
                if lines_before:
                    before_contexts.append(DiffContext(lines_before, before_path))
                    before_lines = []
                if lines_after:
                    after_contexts.append(DiffContext(lines_after, after_path))
                    after_lines = []
                continue
            elif line.startswith('---'):
                before_path = line[5:]
                continue
            elif line.startswith('+++'):
                after_path = line[5:]
                continue
            position += 1
            if line.startswith('-'):
                lines_before.append((before_line_number, position, line[1:]))
                before_line_number += 1
            elif line.startswith('+'):
                lines_after.append((before_line_number, position, line[1:]))
                after_line_number += 1
            else:
                lines_before.append((before_line_number, position, line))
                lines_after.append((after_line_number, position, line))
                before_line_number += 1
                after_line_number += 1
        if before_lines:
            before_contexts.append(DiffContext(lines_before, before_path))
        if after_lines:
            after_contexts.append(DiffContext(lines_after, after_path))
        return before_contexts, after_contexts