Work in progress

script.py

@@ -1,14 +1,105 @@
-import os
+import re
 import sys
 from subprocess import check_output
+from datetime import datetime
 
 from jinja2 import Environment, FileSystemLoader
 
 env = Environment(loader=FileSystemLoader('templates'))
 
 
-class GitLog:
-    MARKER = '--PYCHOLOG MARKER--'
+class Issue:
+    def __init__(self, number='', url=''):
+        self.number = number
+        self.url = url
+
+
+class Commit:
+    subject_regex = re.compile(
+        r'^(?P<type>((add(ed|s)?)|(change[ds]?)|(fix(es|ed)?)|(remove[sd]?)|(merged?)))',
+        re.IGNORECASE)
+    body_regex = re.compile(r'(?P<issues>#\d+)')
+    break_regex = re.compile(r'^break(s|ing changes)?[ :].+$')
+    types = {
+        'add': 'Added',
+        'fix': 'Fixed',
+        'change': 'Changed',
+        'remove': 'Removed',
+        'merge': 'Merged'
+    }
+
+    def __init__(
+            self, hash, author_name='', author_email='', author_date='',
+            committer_name='', committer_email='', committer_date='',
+            tag='', subject='', body=None, url=''):
+        self.hash = hash
+        self.author_name = author_name
+        self.author_email = author_email
+        self.author_date = datetime.utcfromtimestamp(float(author_date))
+        self.committer_name = committer_name
+        self.committer_email = committer_email
+        self.committer_date = datetime.utcfromtimestamp(float(committer_date))
+        self.tag = self.version = tag.replace('tag: ', '')
+        self.subject = subject
+        self.body = body or []
+        self.type = ''
+        self.url = url
+        self.issues = []
+
+        self.has_breaking_change = None
+        self.adds_feature = None
+
+        self.parse_message()
+        self.fix_type()
+
+    def parse_message(self):
+        subject_match = self.subject_regex.match(self.subject)
+        if subject_match is not None:
+            for group, value in subject_match.groupdict().items():
+                setattr(self, group, value)
+        body_match = self.body_regex.match('\n'.join(self.body))
+        if body_match is not None:
+            for group, value in body_match.groupdict().items():
+                setattr(self, group, value)
+
+    def fix_type(self):
+        for k, v in self.types.items():
+            if self.type.lower().startswith(k):
+                self.type = v
+
+    def build_issues(self, url):
+        self.issues = [Issue(number=issue, url=url + issue)
+                       if isinstance(issue, str) else issue
+                       for issue in self.issues]
+
+    @property
+    def is_patch(self):
+        return not any((self.is_minor, self.is_major))
+
+    @property
+    def is_minor(self):
+        return self.type.lower() in ('add', 'adds', 'added') and not self.is_major
+
+    @property
+    def is_major(self):
+        return bool(self.break_regex.match('\n'.join(self.body)))
+
+
+class AngularCommit(Commit):
+    subject_regex = re.compile(
+        r'^(?P<type>(feat|fix|docs|style|refactor|test|chore))'
+        r'(?P<scope>\(.+\))?: (?P<subject>.+)$')
+
+    @property
+    def is_minor(self):
+        return self.type.lower() == 'feat' and not self.is_major
+
+    def fix_type(self):
+        pass
+
+
+class Gitolog:
+    MARKER = '--GITOLOG MARKER--'
     FORMAT = (
         '%H%n'  # commit hash
         '%an%n'  # author name
@@ -22,14 +113,59 @@ class GitLog:
         '%b%n' + MARKER  # body
     )
 
-    COMMAND = ['git', 'log', '--tags', '--date=raw', '--format=' + FORMAT]
+    COMMAND = ['git', 'log', '--date=unix', '--format=' + FORMAT]
+    DEFAULT_STYLE = 'basic'
+    STYLE = {
+        'basic': Commit,
+        'angular': AngularCommit
+    }
 
-    def __init__(self, repository):
+    def __init__(
+            self, repository,
+            project_url='', commit_url='', issue_url='', compare_url='',
+            style=DEFAULT_STYLE):
         self.repository = repository
-        self.raw_log = self.get()
+        self.project_url = project_url if project_url else self.get_url()
+
+        if not commit_url:
+            commit_url = self.project_url + '/commit/'
+        if not issue_url:
+            issue_url = self.project_url + '/issues/'
+        if not compare_url:
+            compare_url = self.project_url + '/compare/'
+
+        self.commit_url = commit_url
+        self.issue_url = issue_url
+        self.compare_url = compare_url
+
+        self.raw_log = self.get_log()
         self.commits = self.parse_commits()
 
-    def get(self):
+        dates = self.apply_versions_to_commits()
+        versions = self.group_commits_by_version(dates)
+        self.versions_list = versions['as_list']
+        self.versions_dict = versions['as_dict']
+
+        if isinstance(style, str):
+            try:
+                style = self.STYLE[style]
+            except KeyError:
+                print('no such style available: %s' % style, file=sys.stderr)
+                print('using default style: %s' % self.DEFAULT_STYLE, file=sys.stderr)
+                style = self.STYLE[self.DEFAULT_STYLE]
+
+        self.style = style
+
+    def get_url(self):
+        git_url = str(check_output(
+            ['git', 'config', '--get', 'remote.origin.url'],
+            cwd=self.repository))[2:-1].rstrip('\\n')
+        if git_url.startswith('git@'):
+            split = git_url.replace('git@', '', 1).split(':', 1)
+            git_url = 'https://' + split[0] + '/' + split[1]
+        return git_url
+
+    def get_log(self):
         # remove enclosing b-quotes (b'' or b"")
         return str(check_output(self.COMMAND, cwd=self.repository))[2:-1].replace("\\'", "'")
 
@@ -40,7 +176,7 @@ def parse_commits(self):
         pos = 0
         while pos < size:
             commit = Commit(
-                commit_hash=lines[pos],
+                hash=lines[pos],
                 author_name=lines[pos+1],
                 author_email=lines[pos+2],
                 author_date=lines[pos+3],
@@ -49,40 +185,85 @@ def parse_commits(self):
                 committer_date=lines[pos+6],
                 tag=lines[pos+7],
                 subject=lines[pos+8],
-                body=[lines[pos+9]]
+                body=[lines[pos+9]],
+                url=self.commit_url + lines[pos]
             )
             commits.append(commit)
+            commit.build_issues(url=self.issue_url)
             nbl_index = 10
             while lines[pos+nbl_index] != self.MARKER:
                 commit.body.append(lines[pos+nbl_index])
                 nbl_index += 1
             pos += nbl_index + 1
         return commits
 
+    def apply_versions_to_commits(self):
+        versions_dates = {}
+        version = None
+        for commit in self.commits:
+            if commit.version:
+                version = commit.version
+                versions_dates[version] = commit.committer_date.date()
+            elif version:
+                commit.version = version
+        return versions_dates
 
-class Commit:
-    def __init__(self, commit_hash,
-                 author_name=None, author_email=None, author_date=None,
-                 committer_name=None, committer_email=None, committer_date=None,
-                 tag=None, subject=None, body=None):
-        self.commit_hash = commit_hash
-        self.author_name = author_name
-        self.author_email = author_email
-        self.author_date = author_date
-        self.committer_name = committer_name
-        self.committer_email = committer_email
-        self.committer_date = committer_date
+    def group_commits_by_version(self, dates):
+        versions_list = []
+        versions_dict = {}
+        versions_types_dict = {}
+        next_version = None
+        for commit in self.commits:
+            if commit.version not in versions_dict:
+                version = versions_dict[commit.version] = Version(
+                    tag=commit.version, date=dates[commit.version])
+                if next_version:
+                    version.next_version = next_version
+                    next_version.previous_version = version
+                next_version = version
+                versions_list.append(version)
+                versions_types_dict[commit.version] = {}
+            versions_dict[commit.version].commits.append(commit)
+            if commit.type not in versions_types_dict[commit.version]:
+                section = versions_types_dict[commit.version][commit.type] = Section(
+                    type=commit.type)
+                versions_dict[commit.version].sections_list.append(section)
+                versions_dict[commit.version].sections_dict = versions_types_dict[commit.version]
+            versions_types_dict[commit.version][commit.type].commits.append(commit)
+        return {'as_list': versions_list, 'as_dict': versions_dict}
+
+
+class Version:
+    def __init__(self, tag='', date='', sections=None, commits=None, url=''):
         self.tag = tag
-        self.subject = subject
-        self.body = body or []
+        self.date = date
+
+        self.sections_list = sections or []
+        self.sections_dict = {s.type: s for s in self.sections_list}
+        self.commits = commits or []
+        self.url = url
+        self.previous_version = None
+        self.next_version = None
+
+    @property
+    def typed_sections(self):
+        return [s for s in self.sections_list if s.type]
+
+    @property
+    def untyped_section(self):
+        return self.sections_dict.get('', None)
+
+
+class Section:
+    def __init__(self, type='', commits=None):
+        self.type = type
+        self.commits = commits or []
 
 
 if __name__ == '__main__':
-    commits = GitLog('example2').commits
+    gitolog = Gitolog('example2')
     template = env.get_template('changelog.md')
-    rendered = template.render(commits=commits)
+    rendered = template.render(gitolog=gitolog)
     with open('output.md', 'w') as stream:
         stream.write(rendered)
-    for commit in commits:
-        print(commit.subject)
     sys.exit(0)

templates/changelog.md

@@ -1,5 +1,13 @@
-This is my changelog
+# Changelog
+All notable changes to this project will be documented in this file.
 
-{% for commit in commits -%}
-  {% include 'commit.md' with context %}
-{% endfor %}
+The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/)
+and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.html).
+
+{% for version in gitolog.versions_list -%}
+{% include 'version.md' with context %}
+{% endfor -%}
+
+{% for version in gitolog.versions_list -%}
+{% include 'version_link.md' with context %}
+{% endfor -%}

templates/commit.md

@@ -1,12 +1 @@
-commit {{ commit.commit_hash }}
-Author name: {{ commit.author_name }}
-Author email: {{ commit.author_email }}
-Author date: {{ commit.author_date }}
-Committer name: {{ commit.committer_name }}
-Committer email: {{ commit.committer_email }}
-Committer date: {{ commit.committer_date }}
-{% if commit.tag %}Tag: {{ commit.tag }}{% endif %}
-Subject: {{ commit.subject }}
-Body: {% for line in commit.body %}
-{% if line %}{{ line }}{% endif %}
-{% endfor %}
+- {{ commit.subject }} ([{{ commit.hash }}]({{ commit.url }}))

templates/section.md

@@ -0,0 +1,4 @@
+### {{ section.type or "Misc" }}
+{% for commit in section.commits -%}
+{% include 'commit.md' with context %}
+{% endfor %}

templates/version.md

@@ -0,0 +1,12 @@
+## [{{ version.tag }}]{% if version.date %} - {{ version.date }}{% endif %}
+
+{% for type, section in version.sections_dict|dictsort -%}
+{%- if type and type != 'Merged' -%}
+{% include 'section.md' with context %}
+{% endif -%}
+{%- endfor -%}
+{%- if version.untyped_section -%}
+{%- with section = version.untyped_section -%}
+{% include 'section.md' with context %}
+{% endwith -%}
+{%- endif -%}

templates/version_link.md

@@ -0,0 +1,7 @@
+{%- if version.previous_version -%}
+{%- if version.date -%}
+[{{ version.tag }}]: {{ gitolog.compare_url }}{{ version.previous_version.tag }}...{{ version.tag}}
+{%- else -%}
+[{{ version.tag }}]: {{ gitolog.compare_url }}{{ version.previous_version.tag }}...HEAD
+{% endif -%}
+{%- endif -%}