Make options title/repository optional, add option -m/--max, major code...

Make options title/repository optional, add option -m/--max, major code cleanup, full PEP8 compliance

All changes:
- chart title and repository are now optional and must be set with two
  new options:
    -t, --title (default: hardcoded in script, depends on the chart)
    -r, --repo (default: '.', ie current directory)
- add option -m/--max: max different entries in chart: after this
  number, an entry is counted in "others" (only for charts "authors" and
  "files_type")
- if filename is "-", display SVG content on standard output
- import the new division operator and the print function
- replace %-formatting with .format()
- full PEP8 compliance
- major code cleanup: add comments, dynamically build the names for week
  days and months, use a dynamic name for functions building a chart.

README.md

@@ -18,20 +18,26 @@ installed with this command:
 Note: cairosvg is required to generate PNG files.
+## Usage
+
+See output of command:
+
+    $ python gitchart.py -h
+
 ## Examples
 Generate pie chart with authors:
-    $ python gitchart.py authors "Git authors" /path/to/gitrepo/ authors.svg
+    $ python gitchart.py -t "Git authors on project X" -r /path/to/gitrepo/ authors authors.svg
 Generate bar chart with commits by year:
-    $ python gitchart.py commits_year "Git commits by year" /path/to/gitrepo/ commits_year.svg
+    $ python gitchart.py -r /path/to/gitrepo/ commits_year commits_year.svg
 Generate bar chart with commits by version (tag):
     $ cd /path/to/gitrepo/
-    $ git tag | python /path/to/gitchart.py commits_version "Git commits by version" . /tmp/commits_version.svg
+    $ git tag | python /path/to/gitchart.py commits_version /tmp/commits_version.svg
 ## Demo

gitchart.py

@@ -21,7 +21,6 @@
 # (http://pygal.org).
 #
 # Charts supported:
-#
 #                      |       |                           | format of data
 #   name               | chart | description               | expected (stdin)
 #   -------------------+-------+---------------------------+-----------------
@@ -36,7 +35,10 @@
 #   files_type         | pie   | files by type (extension) | -
 #
+from __future__ import division, print_function
+
 import argparse
+import datetime
 import os
 import pygal
 import re
@@ -45,164 +47,178 @@ import subprocess
 import sys
 import traceback
-NAME = 'gitchart.py'
-VERSION = '0.4'
-AUTHOR = 'Sebastien Helleu <flashcode@flashtux.org>'
+VERSION = '0.5'
 class GitChart:
     """Generate a git stat chart."""
-    def __init__(self):
-        self.charts = {'authors': self._chart_authors,
-                       'commits_hour': self._chart_commits_hour,
-                       'commits_hour_week': self._chart_commits_hour_week,
-                       'commits_day': self._chart_commits_day,
-                       'commits_month': self._chart_commits_month,
-                       'commits_year': self._chart_commits_year,
-                       'commits_year_month': self._chart_commits_year_month,
-                       'commits_version': self._chart_commits_version,
-                       'files_type': self._chart_files_type,
-                       }
-        # define a Pygal style with transparent background and custom colors
-        self.style = pygal.style.Style(
-            background='transparent',
-            plot_background='transparent',
-            foreground='rgba(0, 0, 0, 0.9)',
-            foreground_light='rgba(0, 0, 0, 0.6)',
-            foreground_dark='rgba(0, 0, 0, 0.2)',
-            opacity_hover='.4',
-            colors=('#9999ff', '#8cedff', '#b6e354', '#feed6c', '#ff9966', '#ff0000', '#ff00cc', '#899ca1', '#bf4646'))
-        self.days = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
-        self.months = ['Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec']
-    def _git_command(self, repository, command, command2=None):
-        """Execute one or two piped git commands return lines as a list."""
+    charts = {
+        'authors': 'Authors',
+        'commits_hour': 'Commits by hour of day',
+        'commits_hour_week': 'Commits by hour of week',
+        'commits_day': 'Commits by day of week',
+        'commits_month': 'Commits by month of year',
+        'commits_year': 'Commits by year',
+        'commits_year_month': 'Commits by year/month',
+        'commits_version': 'Commits by version',
+        'files_type': 'Files by extension',
+    }
+    weekdays = [datetime.date(2001, 1, day).strftime('%a')
+                for day in range(1, 8)]
+    months = [datetime.date(2001, month, 1).strftime('%b')
+              for month in range(1, 13)]
+    # Pygal style with transparent background and custom colors
+    style = pygal.style.Style(background='transparent',
+                              plot_background='transparent',
+                              foreground='rgba(0, 0, 0, 0.9)',
+                              foreground_light='rgba(0, 0, 0, 0.6)',
+                              foreground_dark='rgba(0, 0, 0, 0.2)',
+                              opacity_hover='.4',
+                              colors=('#9999ff', '#8cedff', '#b6e354',
+                                      '#feed6c', '#ff9966', '#ff0000',
+                                      '#ff00cc', '#899ca1', '#bf4646'))
+
+    def __init__(self, chart_name, title=None, repository='.', output=None,
+                 max_entries=20, in_data=None):
+        self.chart_name = chart_name
+        self.title = title if title is not None else self.charts[chart_name]
+        self.title = None
+        self.repository = repository
+        self.output = output
+        self.max_entries = max_entries
+        self.in_data = in_data
+    def _git_command(self, command1, command2=None):
+        """
+        Execute one or two piped git commands.
+        Return the output lines as a list.
+        """
         if command2:
             # pipe the two commands and return output
-            p1 = subprocess.Popen(command, stdout=subprocess.PIPE, cwd=repository)
-            p2 = subprocess.Popen(command2, stdin=p1.stdout, stdout=subprocess.PIPE)
+            p1 = subprocess.Popen(command1, stdout=subprocess.PIPE,
+                                  cwd=self.repository)
+            p2 = subprocess.Popen(command2, stdin=p1.stdout,
+                                  stdout=subprocess.PIPE)
             p1.stdout.close()
             return p2.communicate()[0].decode('utf-8').strip().split('\n')
         else:
-            # execute a single git command and return output
-            p = subprocess.Popen(command, stdout=subprocess.PIPE, cwd=repository)
+            # execute a single git cmd and return output
+            p = subprocess.Popen(command1, stdout=subprocess.PIPE,
+                                 cwd=self.repository)
             return p.communicate()[0].decode('utf-8').strip().split('\n')
-    def _generate_bar_chart(self, title, data, output, sorted_keys=None, x_labels=None, x_label_rotation=0):
+    def _generate_bar_chart(self, data, sorted_keys=None, x_labels=None,
+                            x_label_rotation=0):
         """Generate a bar chart."""
-        bar_chart = pygal.Bar(style=self.style, show_legend=False, x_label_rotation=x_label_rotation, label_font_size=12)
-        bar_chart.title = title
+        bar_chart = pygal.Bar(style=self.style, show_legend=False,
+                              x_label_rotation=x_label_rotation,
+                              label_font_size=12)
+        bar_chart.title = self.title
         if not sorted_keys:
             sorted_keys = sorted(data)
         bar_chart.x_labels = x_labels if x_labels else sorted_keys
         bar_chart.add('', [data[n] for n in sorted_keys])
-        self._render(bar_chart, output)
-    def _chart_authors(self, title, repository, output, in_data=None, limit=20):
+        self._render(bar_chart)
+    def _chart_authors(self):
         """Generate pie chart with git authors."""
         # format of lines in stdout:   278  John Doe
-        stdout = self._git_command(repository, ['git', 'log', '--pretty=short'], ['git', 'shortlog', '-sn'])
-        pie_chart = pygal.Pie(style=self.style, truncate_legend=100, value_font_size=12)
-        pie_chart.title = title or 'Authors'
+        stdout = self._git_command(['git', 'log', '--pretty=short'],
+                                   ['git', 'shortlog', '-sn'])
+        pie_chart = pygal.Pie(style=self.style, truncate_legend=100,
+                              value_font_size=12)
+        pie_chart.title = self.title
         count = 0
         count_others = 0
         sum_others = 0
         for author in stdout:
             (number, name) = author.strip().split('\t', 1)
             count += 1
-            if limit <= 0 or count <= limit:
-                pie_chart.add('%s (%s)' % (name, number), int(number))
+            if self.max_entries <= 0 or count <= self.max_entries:
+                pie_chart.add(name + ' ({0})'.format(number), int(number))
             else:
                 count_others += 1
                 sum_others += int(number)
         if count_others:
-            pie_chart.add('%d others (%d)' % (count_others, sum_others), sum_others)
-        self._render(pie_chart, output)
+            pie_chart.add('{0} others ({1})'.format(count_others, sum_others),
+                          sum_others)
+        self._render(pie_chart)
         return True
-    def _chart_commits_hour(self, title, repository, output, in_data=None):
+    def _chart_commits_hour(self):
         """Generate bar chart with commits by hour of day."""
         # format of lines in stdout: 2013-03-15 18:27:55 +0100
-        stdout = self._git_command(repository, ['git', 'log', '--date=iso', '--pretty=format:%ad'])
-        commits = {}
+        stdout = self._git_command(['git', 'log', '--date=iso',
+                                    '--pretty=format:%ad'])
+        commits = {'{0:02d}'.format(hour): 0 for hour in range(0, 24)}
         for line in stdout:
-            hour = line.split()[1].split(':')[0]
-            commits[hour] = commits.get(hour, 0) + 1
-        # ensure each hour of day has a value (set 0 for hours without commits)
-        for i in range(0, 24):
-            hour = '%02d' % i
-            if not hour in commits:
-                commits[hour] = 0
-        self._generate_bar_chart(title or 'Commits by hour of day', commits, output)
+            commits[line.split()[1].split(':')[0]] += 1
+        self._generate_bar_chart(commits)
         return True
-    def _chart_commits_hour_week(self, title, repository, output, in_data=None):
+    def _chart_commits_hour_week(self):
         """Generate dot chart with commits by hour of week."""
         # format of lines in stdout: Fri, 15 Mar 2013 18:27:55 +0100
-        stdout = self._git_command(repository, ['git', 'log', '--date=rfc', '--pretty=format:%ad'])
-        commits = {}
+        stdout = self._git_command(['git', 'log', '--date=rfc',
+                                    '--pretty=format:%ad'])
+        commits = {day: {'{0:02d}'.format(hour): 0 for hour in range(0, 24)}
+                   for day in self.weekdays}
         for line in stdout:
-            items = line.split()
-            day = items[0][:-1]
-            hour = items[-2].split(':')[0]
-            if not day in commits:
-                commits[day] = {}
-            commits[day][hour] = commits[day].get(hour, 0) + 1
+            wday, _, _, _, hour, _ = line.split()
+            commits[wday[:-1]][hour.split(':')[0]] += 1
         dot_chart = pygal.Dot(style=self.style)
-        dot_chart.title = title or 'Commits by hour of week'
-        dot_chart.x_labels = ['%02d' % hour for hour in range(0, 24)]
-        for day in sorted(commits, key=self.days.index):
-            # ensure each hour of day has a value (set 0 for hours without commits)
-            for i in range(0, 24):
-                hour = '%02d' % i
-                if not hour in commits[day]:
-                    commits[day][hour] = 0
+        dot_chart.title = self.title
+        dot_chart.x_labels = ['{0:02d}'.format(hour) for hour in range(0, 24)]
+        for day in self.weekdays:
             dot_chart.add(day, commits[day])
-        self._render(dot_chart, output)
+        self._render(dot_chart)
         return True
-    def _chart_commits_day(self, title, repository, output, in_data=None):
+    def _chart_commits_day(self):
         """Generate bar chart with commits by day of week."""
         # format of lines in stdout: Fri, 15 Mar 2013 18:27:55 +0100
-        stdout = self._git_command(repository, ['git', 'log', '--date=rfc', '--pretty=format:%ad'])
-        commits = {}
+        stdout = self._git_command(['git', 'log', '--date=rfc',
+                                    '--pretty=format:%ad'])
+        commits = {day: 0 for day in self.weekdays}
         for line in stdout:
-            day = line.split(',')[0]
-            commits[day] = commits.get(day, 0) + 1
-        self._generate_bar_chart(title or 'Commits by day of week', commits, output, sorted_keys=sorted(commits, key=self.days.index))
+            commits[line.split(',')[0]] += 1
+        self._generate_bar_chart(commits,
+                                 sorted_keys=sorted(commits,
+                                                    key=self.weekdays.index))
         return True
-    def _chart_commits_month(self, title, repository, output, in_data=None):
+    def _chart_commits_month(self):
         """Generate bar chart with commits by month of year."""
         # format of lines in stdout: 2013-03-15
-        stdout = self._git_command(repository, ['git', 'log', '--date=short', '--pretty=format:%ad'])
-        commits = {}
+        stdout = self._git_command(['git', 'log', '--date=short',
+                                    '--pretty=format:%ad'])
+        commits = {month: 0 for month in self.months}
         for line in stdout:
-            month = int(line.split('-')[1])
-            commits[month] = commits.get(month, 0) + 1
-        # transform month number to name (1 => Jan, 2 => Feb,...)
-        commits2 = {}
-        for month in commits:
-            commits2[self.months[month - 1]] = commits[month]
-        self._generate_bar_chart(title or 'Commits by month of year', commits2, output, sorted_keys=sorted(commits2, key=self.months.index))
+            month = int(line.split('-')[1]) - 1
+            commits[self.months[month]] += 1
+        self._generate_bar_chart(commits,
+                                 sorted_keys=sorted(commits,
+                                                    key=self.months.index))
         return True
-    def _chart_commits_year(self, title, repository, output, in_data=None):
+    def _chart_commits_year(self):
         """Generate bar chart with commits by year."""
         # format of lines in stdout: 2013-03-15
-        stdout = self._git_command(repository, ['git', 'log', '--date=short', '--pretty=format:%ad'])
+        stdout = self._git_command(['git', 'log', '--date=short',
+                                    '--pretty=format:%ad'])
         commits = {}
         for line in stdout:
             year = line.split('-')[0]
             commits[year] = commits.get(year, 0) + 1
-        self._generate_bar_chart(title or 'Commits by year', commits, output)
+        self._generate_bar_chart(commits)
         return True
-    def _chart_commits_year_month(self, title, repository, output, in_data=None):
+    def _chart_commits_year_month(self):
         """Generate bar chart with commits by year/month."""
         # format of lines in stdout: 2013-03-15
-        stdout = self._git_command(repository, ['git', 'log', '--date=short', '--pretty=format:%ad'])
+        stdout = self._git_command(['git', 'log', '--date=short',
+                                    '--pretty=format:%ad'])
         commits = {}
         min_date = 999999
         max_date = 0
@@ -211,12 +227,12 @@ class GitChart:
             date = (int(year) * 100) + int(month)
             min_date = min(min_date, date)
             max_date = max(max_date, date)
-            year_month = '%s-%s' % (year, month)
+            year_month = '{0}-{1}'.format(year, month)
             commits[year_month] = commits.get(year_month, 0) + 1
         if min_date != 999999:
             date = min_date
             while date < max_date:
-                year_month = '%04d-%02d' % (date / 100, date % 100)
+                year_month = '{0:04d}-{1:02d}'.format(date // 100, date % 100)
                 commits[year_month] = commits.get(year_month, 0)
                 if date % 100 == 12:
                     # next year, for example: 201312 => 201401 (+89)
@@ -224,119 +240,133 @@ class GitChart:
                 else:
                     date += 1
         x_labels = sorted(commits)
-        # if there are more than 20 commits, keep one x label on 10 (starting from the end)
+        # if there are more than 20 commits, keep one x label on 10
+        # (starting from the end)
         if len(commits) > 20:
             n = 0
             for i in range(len(x_labels) - 1, -1, -1):
                 if n % 10 != 0:
                     x_labels[i] = ''
                 n += 1
-        self._generate_bar_chart(title or 'Commits by year/month', commits, output, x_labels=x_labels, x_label_rotation=45)
+        self._generate_bar_chart(commits, x_labels=x_labels,
+                                 x_label_rotation=45)
         return True
-    def _chart_commits_version(self, title, repository, output, in_data=None):
+    def _chart_commits_version(self):
         """Generate bar chart with commits by version (tag)."""
-        if not in_data:
+        if not self.in_data:
             return False
         commits = {}
         oldtag = ''
-        for tag in in_data.strip().split('\n'):
-            commits[tag] = len(self._git_command(repository, ['git', 'log', '%s..%s' % (oldtag, tag) if oldtag else tag, '--pretty=oneline']))
+        for tag in self.in_data.strip().split('\n'):
+            # transform version to keep only digits with period as separator,
+            # examples:
+            #   release-0-0-1  =>  0.0.1
+            #   v0.3.0         =>  0.3.0
+            tag2 = re.sub('([^0-9]+)', ' ', tag).strip().replace(' ', '.')
+            commits[tag2] = len(
+                self._git_command(['git', 'log',
+                                   oldtag + '..' + tag if oldtag else tag,
+                                   '--pretty=oneline']))
             oldtag = tag
-        # transform version to keep only digits with period as separator, examples:
-        #   release-0-0-1  =>  0.0.1
-        #   v0.3.0         =>  0.3.0
-        commits2 = {}
-        for tag in commits:
-            commits2[re.sub('([^0-9]+)', ' ', tag).strip().replace(' ', '.')] = commits[tag]
-        self._generate_bar_chart(title or 'Commits by version', commits2, output, x_label_rotation=90)
+        self._generate_bar_chart(commits, x_label_rotation=90)
         return True
-    def _chart_files_type(self, title, repository, output, in_data=None, limit=20):
+    def _chart_files_type(self):
         """Generate pie chart with files by extension."""
         # format of lines in stdout: path/to/file.c
-        stdout = self._git_command(repository, ['git', 'ls-tree', '-r', '--name-only', 'HEAD'])
+        stdout = self._git_command(['git', 'ls-tree', '-r', '--name-only',
+                                    'HEAD'])
         extensions = {}
         for line in stdout:
             ext = os.path.splitext(line)[1]
             if not ext:
                 ext = '(no extension)'
             extensions[ext] = extensions.get(ext, 0) + 1
-        pie_chart = pygal.Pie(style=self.style, truncate_legend=100, value_font_size=12)
-        pie_chart.title = title or 'Files by extension'
+        pie_chart = pygal.Pie(style=self.style, truncate_legend=100,
+                              value_font_size=12)
+        pie_chart.title = self.title
         count = 0
         count_others = 0
         sum_others = 0
         for ext in sorted(extensions, key=extensions.get, reverse=True):
             count += 1
-            if limit <= 0 or count <= limit:
-                pie_chart.add('%s (%d)' % (ext, extensions[ext]), extensions[ext])
+            if self.max_entries <= 0 or count <= self.max_entries:
+                pie_chart.add(ext + ' ({0})'.format(extensions[ext]),
+                              extensions[ext])
             else:
                 count_others += 1
                 sum_others += extensions[ext]
         if count_others:
-            pie_chart.add('%d others (%d)' % (count_others, sum_others), sum_others)
-        self._render(pie_chart, output)
+            pie_chart.add('{0} others ({1})'.format(count_others, sum_others),
+                          sum_others)
+        self._render(pie_chart)
         return True
-    def _render(self, chart, output):
-        if output.endswith('.png'):
-            chart.render_to_png(output)
+    def _render(self, chart):
+        if self.output == '-':
+            # display SVG on stdout
+            print(chart.render())
+        elif self.output.lower().endswith('.png'):
+            # write PNG in file
+            chart.render_to_png(self.output)
         else:
-            chart.render_to_file(output)
-    def generate(self, chart, title, repository, output, in_data=None):
+            # write SVG in file
+            chart.render_to_file(self.output)
+    def generate(self):
         """Generate a chart, and return True if OK, False if error."""
-        if chart not in self.charts:
-            return False
         try:
-            return self.charts[chart](title, repository, output, in_data)
+            # call function to build chart (name of function is dynamic)
+            return getattr(self, '_chart_' + self.chart_name)()
         except:
-            print ('Error:\n%s' % traceback.print_exc())
+            traceback.print_exc()
             return False
-# build the GitChart object
-chart = GitChart()
-
-# parse command line arguments
-parser = argparse.ArgumentParser(formatter_class=argparse.RawDescriptionHelpFormatter,
-                                 description='''
-%s %s (C) 2013 %s
-Generate statistic charts for a git repository.
-''' % (NAME, VERSION, AUTHOR),
-                                 epilog='''
-Return value:
-  0: chart successfully generated
-  1: error
-''')
-parser.add_argument('chart', metavar='chart', choices=chart.charts.keys(),
-                    help='name of chart, one of: %s' % ', '.join(chart.charts))
-parser.add_argument('title', help='title of chart')
-parser.add_argument('repository', help='directory with git repository')
-parser.add_argument('output', help='output file (/path/to/file.svg or /path/to/file.png)')
-parser.add_argument('-v', '--version', action='version', version=VERSION)
-if len(sys.argv) == 1:
-    parser.print_help()
-    sys.exit(1)
-args = parser.parse_args(sys.argv[1:])
-
-# read data on standard input
-in_data = ''
-while True:
-    inr, outr, exceptr = select.select([sys.stdin], [], [], 0)
-    if not inr:
-        break
-    data = os.read(sys.stdin.fileno(), 4096)
-    if not data:
-        break
-    in_data += data.decode('utf-8')
-# generate chart
-if chart.generate(args.chart, args.title, args.repository, args.output, in_data=in_data):
-    sys.exit(0)
-# error
-print('Failed to generate chart: %s' % args)
-sys.exit(1)
+if __name__ == "__main__":
+    # parse command line arguments
+    parser = argparse.ArgumentParser(
+        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
+        description='Generate statistic charts for a git repository.',
+        epilog='Return value: 0 = success, 1 = error.')
+    parser.add_argument('-t', '--title',
+                        help='override the default chart title')
+    parser.add_argument('-r', '--repo', default='.',
+                        help='directory with git repository')
+    parser.add_argument('-m', '--max', type=int, default=20,
+                        help='max different entries in chart: after this '
+                        'number, an entry is counted in "others" (only for '
+                        'charts "authors" and "files_type"), 0=unlimited')
+    parser.add_argument('chart', metavar='chart',
+                        choices=sorted(GitChart.charts),
+                        help='name of chart, one of: ' +
+                        ', '.join(sorted(GitChart.charts)))
+    parser.add_argument('output', help='output file (svg or png), special '
+                        'value "-" displays SVG content on standard output')
+    parser.add_argument('-v', '--version', action='version', version=VERSION)
+    if len(sys.argv) == 1:
+        parser.print_help()
+        sys.exit(1)
+    args = parser.parse_args(sys.argv[1:])
+    # read data on standard input
+    in_data = ''
+    while True:
+        inr, outr, exceptr = select.select([sys.stdin], [], [], 0)
+        if not inr:
+            break
+        data = os.read(sys.stdin.fileno(), 4096)
+        if not data:
+            break
+        in_data += data.decode('utf-8')
+    # generate chart
+    chart = GitChart(args.chart, args.title, args.repo, args.output, args.max,
+                     in_data)
+    if chart.generate():
+        sys.exit(0)
+    # error
+    print('ERROR: failed to generate chart:', vars(args))
+    sys.exit(1)