changelog.py

#!/usr/bin/python

import subprocess
import os
import re
import sys
import itertools

sprint_regex = '^sprint\-(\d+)?$'
semver_regex = '^v(\d+)\.(\d+)\.(\d+)(\-(.*))?$'
include_regex='\.(css|dust|html|js|json|less|md|conf|css|gradle|html|java|js|json|md|properties|raml|sh|sql|txt|xml|xsd|yml)$'
exclude_regex='vendor|JMeter|dist|static'

def toint(strOrNone):
  if strOrNone:
    return int(strOrNone)
  else:
    return None

def run(cmd, cwd = None, stdout = subprocess.PIPE):
  p = subprocess.Popen(cmd, cwd = cwd, env = os.environ, stdout = stdout, stderr = subprocess.PIPE)
  stdout, err = p.communicate()
  if p.wait() != 0: raise Exception(err)
  return stdout

def git_tags():
  return run(['git', 'tag']).splitlines()

def git_commits(ref_old, ref_new):
  commits_raw = run(['git', 'log', '%s..%s' % (ref_old, ref_new), '--no-merges', '--pretty=%B|<|>|']).split('|<|>|')[:-1]
  commits = filter(lambda c: c != None, map(lambda c: parse_commit(c), commits_raw))

  return commits

def git_changes_stats(ref_old, ref_new):
  global include_regex
  global exclude_regex

  lines = run(['git', 'diff', '%s..%s' % (ref_old, ref_new), '--numstat', '-w']).splitlines()
  all = filter(lambda c: c != None, map(parse_change, lines))
  filtered_1 = filter(lambda c: re.search(include_regex, c['file']) != None, all)
  filtered_2 = filter(lambda c: re.search(exclude_regex, c['file']) is None, filtered_1)

  return filtered_2

def git_changes(ref_old, ref_new):
  global include_regex
  global exclude_regex

  lines = run(['git', 'log', '%s..%s' % (ref_old, ref_new), '--pretty=tformat:', '--numstat', '-w']).splitlines()
  all = filter(lambda c: c != None, map(parse_change, lines))
  filtered_1 = filter(lambda c: re.search(include_regex, c['file']) != None, all)
  filtered_2 = filter(lambda c: re.search(exclude_regex, c['file']) is None, filtered_1)

  return filtered_2

def tags_filter(t):
  global sprint_regex
  global semver_regex

  return re.search(sprint_regex, t) or re.search(semver_regex, t)

def tags_sort(a, b):
  global sprint_regex
  global semver_regex

  sprint_a = re.match(sprint_regex, a)
  sprint_b = re.match(sprint_regex, b)
  semver_a = re.match(semver_regex, a)
  semver_b = re.match(semver_regex, b)

  if sprint_a and semver_b:
    return -1
  elif sprint_b and semver_a:
    return 1
  elif sprint_a and sprint_b:
    va = toint(sprint_a.group(1))
    vb = toint(sprint_b.group(1))
    return cmp(va, vb)
  else:
    va = [toint(semver_a.group(1)), toint(semver_a.group(2)), toint(semver_a.group(3))]
    vb = [toint(semver_b.group(1)), toint(semver_b.group(2)), toint(semver_b.group(3))]
    if cmp(va, vb) != 0:
      return cmp(va, vb)
    else:
      return -cmp(semver_a.group(5), semver_b.group(5))

def parse_change(raw_line):
  match = re.match('^(\d+)\s+(\d+)\s(.*)$', raw_line)

  if match:
    return {
      'file': match.group(3).strip(),
      'added': int(match.group(1)),
      'removed': int(match.group(2))
    }
  else:
    return None

def parse_commit(raw_body):
  subject = re.match('^([^\(]+)\(([^\)]+)\):(.*)$', raw_body.strip().splitlines()[0].strip())
  jira_ids = re.findall('UNITY\-\d+', raw_body)

  if subject:
    return {
      'raw_body': raw_body.strip(),
      'type': subject.group(1).strip(),
      'scope': subject.group(2).strip(),
      'subject': subject.group(3).strip(),
      'jira_ids': jira_ids
    }
  else:
    return None

tags = ['HEAD'] + list(reversed(sorted(filter(tags_filter, git_tags()), cmp = tags_sort)))

print '# Changelog\n'
for ref_new, ref_old in zip(tags[:-1], tags[1:]):
  print '## %s\n' % ref_new

  changes = git_changes(ref_old, ref_new)
  commits = git_commits(ref_old, ref_new)

  for type, g1 in itertools.groupby(sorted(commits, key = lambda c: c['type']), lambda c: c['type']):
    print '### %s\n' % type

    for commit in g1:
      jira_ids_md = map(lambda j: '[[%s]](https://epages.atlassian.net/browse/%s)' % (j, j), commit['jira_ids'])
      print '* (%s) %s %s' % (commit['scope'], commit['subject'], ', '.join(jira_ids_md))
    print

print '# Metrics\n'
for ref_new, ref_old in zip(tags[:-1], tags[1:]):
  print '## %s\n' % ref_new

  changes_stats = git_changes_stats(ref_old, ref_new)

  added = reduce(lambda s, c: s + c['added'], changes_stats, 0)
  removed = reduce(lambda s, c: s + c['removed'], changes_stats, 0)

  print '* Added %d' % added
  print '* Removed %d' % removed
  print