Merge Hooke and BE branches of update-copyright.
authorW. Trevor King <wking@drexel.edu>
Thu, 9 Feb 2012 19:01:12 +0000 (14:01 -0500)
committerW. Trevor King <wking@drexel.edu>
Thu, 9 Feb 2012 19:01:12 +0000 (14:01 -0500)
update_copyright.py

index a600eac7e68455bb9426837137f18ec8a6b22785..4ca5e647c881c99038dacb507c16cb09626c74f3 100755 (executable)
@@ -1,26 +1,29 @@
 #!/usr/bin/python
 #
-# Copyright (C) 2009-2010 W. Trevor King <wking@drexel.edu>
+# Copyright (C) 2009-2012 W. Trevor King <wking@drexel.edu>
 #
-# This file is part of Bugs Everywhere.
+# This file is part of update-copyright.
 #
-# Bugs Everywhere is free software; you can redistribute it and/or modify it
-# under the terms of the GNU General Public License as published by the
-# Free Software Foundation, either version 2 of the License, or (at your
-# option) any later version.
+# update-copyright is free software: you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation, either version 3 of the
+# License, or (at your option) any later version.
 #
-# Bugs Everywhere is distributed in the hope that it will be useful, but
-# WITHOUT ANY WARRANTY; without even the implied warranty of
+# update-copyright is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 # General Public License for more details.
 #
 # You should have received a copy of the GNU General Public License
-# along with Bugs Everywhere.  If not, see <http://www.gnu.org/licenses/>.
+# along with update-copyright.  If not, see
+# <http://www.gnu.org/licenses/>.
 
 """Automatically update copyright boilerplate.
 
 This script is adapted from one written for `Bugs Everywhere`_. and
-later modified for `Hooke`_ before returning to `Bugs Everywhere`_.
+later modified for `Hooke`_ before returning to `Bugs Everywhere`_.  I
+finally gave up on maintaining separate versions, so here it is as a
+stand-alone module.
 
 .. _Bugs Everywhere: http://bugseverywhere.org/
 .. _Hooke: http://code.google.com/p/hooke/
@@ -30,34 +33,28 @@ import difflib
 import email.utils
 import os
 import os.path
-import re
 import sys
+import textwrap
 import time
 
 
 PROJECT_INFO = {
-    'project': 'Bugs Everywhere',
+    'project': 'update-copyright',
     'vcs': 'Git',
     }
 
 # Break "copyright" into "copy" and "right" to avoid matching the
-# REGEXP.
-COPY_RIGHT_TEXT="""
-This file is part of %(project)s.
-
-%(project)s is free software; you can redistribute it and/or modify it
-under the terms of the GNU General Public License as published by the
-Free Software Foundation, either version 2 of the License, or (at your
-option) any later version.
-
-%(project)s is distributed in the hope that it will be useful, but
-WITHOUT ANY WARRANTY; without even the implied warranty of
-MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-General Public License for more details.
-
-You should have received a copy of the GNU General Public License
-along with %(project)s.  If not, see <http://www.gnu.org/licenses/>.
-""".strip()
+# REGEXP if we decide to go back to regexps.
+COPY_RIGHT_TEXT = [
+    'This file is part of %(project)s.',
+    '%(project)s is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.',
+    '%(project)s is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more details.',
+    'You should have received a copy of the GNU General Public License along with %(project)s.  If not, see <http://www.gnu.org/licenses/>.'
+    ]
+
+SHORT_COPY_RIGHT_TEXT = [
+    '%(project)s comes with ABSOLUTELY NO WARRANTY and is licensed under the GNU General Public License.  For details, %(get-details)s.'
+    ]
 
 COPY_RIGHT_TAG='-xyz-COPY' + '-RIGHT-zyx-' # unlikely to occur in the wild :p
 
@@ -70,32 +67,14 @@ COPY_RIGHT_TAG='-xyz-COPY' + '-RIGHT-zyx-' # unlikely to occur in the wild :p
 #     }
 # Git-based projects are encouraged to use .mailmap instead of
 # ALIASES.  See git-shortlog(1) for details.
-ALIASES = {
-    'Aaron Bentley <abentley@panoramicfeedback.com>':
-        ['Aaron Bentley <aaron.bentley@utoronto.ca>'],
-    'Panometrics, Inc.':
-        ['Aaron Bentley and Panometrics, Inc.'],
-    'Ben Finney <benf@cybersource.com.au>':
-        ['Ben Finney <ben+python@benfinney.id.au>',
-         'John Doe <jdoe@example.com>'],
-    'Chris Ball <cjb@laptop.org>':
-        ['Chris Ball <cjb@thunk.printf.net>'],
-    'Gianluca Montecchi <gian@grys.it>':
-        ['gian <gian@li82-39>',
-         'gianluca <gian@galactica>'],
-    'W. Trevor King <wking@drexel.edu>':
-        ['wking <wking@mjolnir>',
-         'wking <wking@thialfi>'],
-    None:
-        ['j^ <j@oil21.org>'],
-    }
+ALIASES = {}
 
 # List of paths that should not be scanned for copyright updates.
 # IGNORED_PATHS = ['./.git/']
-IGNORED_PATHS = ['./.be/', './.git/', './build/', './doc/.build/']
+IGNORED_PATHS = ['./.git']
 # List of files that should not be scanned for copyright updates.
 # IGNORED_FILES = ['COPYING']
-IGNORED_FILES = ['COPYING', 'catmutt']
+IGNORED_FILES = ['COPYING']
 
 # Work around missing author holes in the VCS history.
 # AUTHOR_HACKS[<path tuple>] = [<missing authors]
@@ -240,13 +219,18 @@ elif PROJECT_INFO['vcs'] == 'Mercurial':
         return (tmp_stdout.getvalue().rstrip('\n'),
                 tmp_stderr.getvalue().rstrip('\n'))
 
-    def original_year(filename, year_hacks=YEAR_HACKS):
-        # shortdate filter: YEAR-MONTH-DAY
-        output,error = mercurial_cmd('log', '--follow',
-                                     '--template', '{date|shortdate}\n',
-                                     filename)
+    def original_year(filename=None, year_hacks=YEAR_HACKS):
+        args = [
+            '--template', '{date|shortdate}\n',
+            # shortdate filter: YEAR-MONTH-DAY
+            ]
+        if filename != None:
+            args.extend(['--follow', filename])
+        output,error = mercurial_cmd('log', *args)
         years = [int(line.split('-', 1)[0]) for line in output.splitlines()]
-        if splitpath(filename) in year_hacks:
+        if filename == None:
+            years.extend(year_hacks.values())
+        elif splitpath(filename) in year_hacks:
             years.append(year_hacks[splitpath(filename)])
         years.sort()
         return years[0]
@@ -301,12 +285,17 @@ elif PROJECT_INFO['vcs'] == 'Bazaar':
             authors = revision.rev.get_apparent_authors()
             self.to_file.write('\n'.join(authors)+'\n')
 
-    def original_year(filename, year_hacks=YEAR_HACKS):
+    def original_year(filename=None, year_hacks=YEAR_HACKS):
         cmd = bzrlib.builtins.cmd_log()
         cmd.outf = StringIO.StringIO()
-        cmd.run(file_list=[filename], log_format=YearLogFormatter, levels=0)
+        kwargs = {'log_format':YearLogFormatter, 'levels':0}
+        if filename != None:
+            kwargs['file_list'] = [filename]
+        cmd.run(**kwargs)
         years = [int(year) for year in set(cmd.outf.getvalue().splitlines())]
-        if splitpath(filename) in year_hacks:
+        if filename == None:
+            years.append(year_hacks.values())
+        elif splitpath(filename) in year_hacks:
             years.append(year_hacks[splitpath(filename)])
         years.sort()
         return years[0]
@@ -347,6 +336,8 @@ def _strip_email(*args):
     Examples
     --------
 
+    >>> _strip_email('J Doe')
+    ['J Doe']
     >>> _strip_email('J Doe <jdoe@a.com>')
     ['J Doe']
     >>> _strip_email('J Doe <jdoe@a.com>', 'JJJ Smith <jjjs@a.com>')
@@ -357,6 +348,8 @@ def _strip_email(*args):
         if arg == None:
             continue
         author,addr = email.utils.parseaddr(arg)
+        if author == '':
+            author = arg
         args[i] = author
     return args
 
@@ -400,23 +393,16 @@ def _replace_aliases(authors, with_email=True, aliases=None):
     ...     'JJJ Smith <jjjs@a.com>':['Jingly <jjjs@b.edu>'],
     ...     None:['Anonymous <a@a.com>'],
     ...     }
-    >>> _replace_aliases(['JJJ Smith <jjjs@a.com>', 'Johnny <jdoe@b.edu>',
-    ...                   'Jingly <jjjs@b.edu>', 'Anonymous <a@a.com>'],
-    ...                  with_email=True, aliases=aliases)
+    >>> authors = [
+    ...     'JJJ Smith <jjjs@a.com>', 'Johnny <jdoe@b.edu>',
+    ...     'Jingly <jjjs@b.edu>', 'J Doe <jdoe@a.com>', 'Anonymous <a@a.com>']
+    >>> _replace_aliases(authors, with_email=True, aliases=aliases)
     ['J Doe <jdoe@a.com>', 'JJJ Smith <jjjs@a.com>']
-    >>> _replace_aliases(['JJJ Smith', 'Johnny', 'Jingly', 'Anonymous'],
-    ...                  with_email=False, aliases=aliases)
+    >>> _replace_aliases(authors, with_email=False, aliases=aliases)
     ['J Doe', 'JJJ Smith']
-    >>> _replace_aliases(['JJJ Smith <jjjs@a.com>', 'Johnny <jdoe@b.edu>',
-    ...                   'Jingly <jjjs@b.edu>', 'J Doe <jdoe@a.com>'],
-    ...                  with_email=True, aliases=aliases)
-    ['J Doe <jdoe@a.com>', 'JJJ Smith <jjjs@a.com>']
     """
     if aliases == None:
         aliases = ALIASES
-    if with_email == False:
-        aliases = dict([(_strip_email(author)[0], _strip_email(*_aliases))
-                        for author,_aliases in aliases.items()])
     rev_aliases = _reverse_aliases(aliases)
     for i,author in enumerate(authors):
         if author in rev_aliases:
@@ -424,9 +410,39 @@ def _replace_aliases(authors, with_email=True, aliases=None):
     authors = sorted(list(set(authors)))
     if None in authors:
         authors.remove(None)
+    if with_email == False:
+        authors = _strip_email(*authors)
     return authors
 
-def _copyright_string(original_year, final_year, authors, prefix=''):
+def _long_author_formatter(copyright_year_string, authors):
+    """
+    >>> print '\\n'.join(_long_author_formatter(
+    ...     copyright_year_string='Copyright (C) 1990-2010',
+    ...     authors=['Jack', 'Jill', 'John']))
+    Copyright (C) 1990-2010 Jack
+                            Jill
+                            John
+    """
+    lines = ['%s %s' % (copyright_year_string, authors[0])]
+    for author in authors[1:]:
+        lines.append(' '*(len(copyright_year_string)+1) + author)
+    return lines
+
+def _short_author_formatter(copyright_year_string, authors):
+    """
+    >>> print '\\n'.join(_short_author_formatter(
+    ...     copyright_year_string='Copyright (C) 1990-2010',
+    ...     authors=['Jack', 'Jill', 'John']*5))
+    Copyright (C) 1990-2010 Jack, Jill, John, Jack, Jill, John, Jack, Jill, John, Jack, Jill, John, Jack, Jill, John
+    """
+    blurb = '%s %s' % (copyright_year_string, ', '.join(authors))
+    return [blurb]
+
+def _copyright_string(original_year, final_year, authors,
+                      text=COPY_RIGHT_TEXT, extra_info={},
+                      author_format_fn=_long_author_formatter,
+                      formatter_kwargs={}, prefix='', wrap=True,
+                      **wrap_kwargs):
     """
     >>> print _copyright_string(original_year=2005,
     ...                         final_year=2005,
@@ -445,20 +461,61 @@ def _copyright_string(original_year, final_year, authors, prefix=''):
                             B <b@b.edu>
     <BLANKLINE>
     This file...
+    >>> print _copyright_string(original_year=2005,
+    ...                         final_year=2005,
+    ...                         authors=['A <a@a.com>', 'B <b@b.edu>'],
+    ...                         text=SHORT_COPY_RIGHT_TEXT,
+    ...                         author_format_fn=_short_author_formatter,
+    ...                         extra_info={'get-details':'%(get-details)s'},
+    ...                         prefix='',
+    ...                         width=50,
+    ...                        )
+    Copyright (C) 2005 A <a@a.com>, B <b@b.edu>
+    <BLANKLINE>
+    update-copyright comes with ABSOLUTELY NO WARRANTY
+    and is licensed under the GNU General Public
+    License.  For details, %(get-details)s.
+
+    >>> print _copyright_string(original_year=2005,
+    ...                         final_year=2005,
+    ...                         authors=['A <a@a.com>', 'B <b@b.edu>'],
+    ...                         text=SHORT_COPY_RIGHT_TEXT,
+    ...                         extra_info={'get-details':'%(get-details)s'},
+    ...                         author_format_fn=_short_author_formatter,
+    ...                         wrap=False,
+    ...                         prefix='',
+    ...                        )
+    Copyright (C) 2005 A <a@a.com>, B <b@b.edu>
+    <BLANKLINE>
+    update-copyright comes with ABSOLUTELY NO WARRANTY and is licensed under the GNU General Public License.  For details, %(get-details)s.
     """
+    for key in ['initial_indent', 'subsequent_indent']:
+        if key not in wrap_kwargs:
+            wrap_kwargs[key] = prefix
+
     if original_year == final_year:
         date_range = '%s' % original_year
     else:
         date_range = '%s-%s' % (original_year, final_year)
-    lines = ['Copyright (C) %s %s' % (date_range, authors[0])]
-    for author in authors[1:]:
-        lines.append(' '*(len('Copyright (C) ')+len(date_range)+1) +
-                     author)
-    lines.append('')
-    lines.extend((COPY_RIGHT_TEXT % PROJECT_INFO).splitlines())
+    copyright_year_string = 'Copyright (C) %s' % date_range
+
+    lines = author_format_fn(copyright_year_string, authors,
+                             **formatter_kwargs)
     for i,line in enumerate(lines):
-        lines[i] = (prefix + line).rstrip()
-    return '\n'.join(lines)
+        lines[i] = prefix + line
+
+    info = dict(PROJECT_INFO)
+    for key,value in extra_info.items():
+        info[key] = value
+    text = [paragraph % info for paragraph in text]
+
+    if wrap == True:
+        text = [textwrap.fill(p, **wrap_kwargs) for p in text]
+    else:
+        assert wrap_kwargs['subsequent_indent'] == '', \
+            wrap_kwargs['subsequent_indent']
+    sep = '\n%s\n' % prefix.rstrip()
+    return sep.join(['\n'.join(lines)] + text)
 
 def _tag_copyright(contents):
     """
@@ -614,6 +671,41 @@ def update_files(files=None, dry_run=False, verbose=0):
             continue
         update_file(filename, dry_run=dry_run, verbose=verbose)
 
+def update_pyfile(path, original_year_fn=original_year,
+                  authors_fn=authors_list, dry_run=False, verbose=0):
+    original_year = original_year_fn()
+    current_year = time.gmtime()[0]
+    authors = authors_fn()
+    authors = _replace_aliases(authors, with_email=False, aliases=ALIASES)
+    paragraphs = _copyright_string(
+        original_year, current_year, authors,
+        text=SHORT_COPY_RIGHT_TEXT,
+        extra_info={'get-details':'%(get-details)s'},
+        author_format_fn=_short_author_formatter, wrap=False,
+        ).split('\n\n')
+    lines = [
+        _copyright_string(original_year, current_year, authors, prefix='# '),
+        '', 'import textwrap', '', '',
+        'LICENSE = """',
+        _copyright_string(original_year, current_year, authors, prefix=''),
+        '""".strip()',
+        '',
+        'def short_license(extra_info, wrap=True, **kwargs):',
+        '    paragraphs = [',
+        ]
+    for p in paragraphs:
+        lines.append("        '%s' %% extra_info," % p.replace("'", r"\'"))
+    lines.extend([
+            '        ]',
+            '    if wrap == True:',
+            '        for i,p in enumerate(paragraphs):',
+            '            paragraphs[i] = textwrap.fill(p, **kwargs)',
+            r"    return '\n\n'.join(paragraphs)",
+            ])
+    new_contents = '\n'.join(lines)+'\n'
+    _set_contents(path, new_contents, dry_run=dry_run, verbose=verbose)
+
+
 def test():
     import doctest
     doctest.testmod()
@@ -637,6 +729,9 @@ If no files are given, a list of files to update is generated
 automatically.
 """ % PROJECT_INFO
     p = optparse.OptionParser(usage)
+    p.add_option('--pyfile', dest='pyfile', default='hooke/license.py',
+                 metavar='PATH',
+                 help='Write project license info to a Python module at PATH')
     p.add_option('--test', dest='test', default=False,
                  action='store_true', help='Run internal tests and exit')
     p.add_option('--dry-run', dest='dry_run', default=False,
@@ -651,3 +746,6 @@ automatically.
 
     update_authors(dry_run=options.dry_run, verbose=options.verbose)
     update_files(files=args, dry_run=options.dry_run, verbose=options.verbose)
+    if options.pyfile != None:
+        update_pyfile(path=options.pyfile,
+                      dry_run=options.dry_run, verbose=options.verbose)