Show More
@@ -0,0 +1,89 b'' | |||
|
1 | # -*- coding: utf-8 -*- | |
|
2 | ||
|
3 | # Some committers are so wrong that it doesn't point at any contributor: | |
|
4 | total_ignore = set() | |
|
5 | total_ignore.add('*** failed to import extension hggit: No module named hggit') | |
|
6 | total_ignore.add('<>') | |
|
7 | ||
|
8 | # Normalize some committer names where people have contributed under different | |
|
9 | # names or email addresses: | |
|
10 | name_fixes = {} | |
|
11 | name_fixes['Andrew Shadura'] = "Andrej Shadura <andrew@shadura.me>" | |
|
12 | name_fixes['aparkar'] = "Aparkar <aparkar@icloud.com>" | |
|
13 | name_fixes['Aras Pranckevicius'] = "Aras PranckeviΔius <aras@unity3d.com>" | |
|
14 | name_fixes['Augosto Hermann'] = "Augusto Herrmann <augusto.herrmann@planejamento.gov.br>" | |
|
15 | name_fixes['"Bradley M. Kuhn" <bkuhn@ebb.org>'] = "Bradley M. Kuhn <bkuhn@sfconservancy.org>" | |
|
16 | name_fixes['dmitri.kuznetsov'] = "Dmitri Kuznetsov" | |
|
17 | name_fixes['Dmitri Kuznetsov'] = "Dmitri Kuznetsov" | |
|
18 | name_fixes['domruf'] = "Dominik Ruf <dominikruf@gmail.com>" | |
|
19 | name_fixes['Ingo von borstel'] = "Ingo von Borstel <kallithea@planetmaker.de>" | |
|
20 | name_fixes['Jan Heylen'] = "Jan Heylen <heyleke@gmail.com>" | |
|
21 | name_fixes['Jason F. Harris'] = "Jason Harris <jason@jasonfharris.com>" | |
|
22 | name_fixes['Jelmer Vernooij'] = "Jelmer VernooΔ³ <jelmer@samba.org>" | |
|
23 | name_fixes['jfh <jason@jasonfharris.com>'] = "Jason Harris <jason@jasonfharris.com>" | |
|
24 | name_fixes['Leonardo Carneiro<leonardo@unity3d.com>'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
25 | name_fixes['leonardo'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
26 | name_fixes['Leonardo <leo@unity3d.com>'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
27 | name_fixes['Les Peabody'] = "Les Peabody <lpeabody@gmail.com>" | |
|
28 | name_fixes['"Lorenzo M. Catucci" <lorenzo@sancho.ccd.uniroma2.it>'] = "Lorenzo M. Catucci <lorenzo@sancho.ccd.uniroma2.it>" | |
|
29 | name_fixes['Lukasz Balcerzak'] = "Εukasz Balcerzak <lukaszbalcerzak@gmail.com>" | |
|
30 | name_fixes['mao <mao@lins.fju.edu.tw>'] = "Ching-Chen Mao <mao@lins.fju.edu.tw>" | |
|
31 | name_fixes['marcink'] = "Marcin KuΕΊmiΕski <marcin@python-works.com>" | |
|
32 | name_fixes['Marcin Kuzminski'] = "Marcin KuΕΊmiΕski <marcin@python-works.com>" | |
|
33 | name_fixes['nansenat16@null.tw'] = "nansenat16 <nansenat16@null.tw>" | |
|
34 | name_fixes['Peter Vitt'] = "Peter Vitt <petervitt@web.de>" | |
|
35 | name_fixes['philip.j@hostdime.com'] = "Philip Jameson <philip.j@hostdime.com>" | |
|
36 | name_fixes['SΓΈren LΓΈvborg'] = "SΓΈren LΓΈvborg <sorenl@unity3d.com>" | |
|
37 | name_fixes['Thomas De Schampheleire'] = "Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>" | |
|
38 | name_fixes['Weblate'] = "<>" | |
|
39 | name_fixes['xpol'] = "xpol <xpolife@gmail.com>" | |
|
40 | name_fixes['Lars <devel@sumpfralle.de>'] = "Lars Kruse <devel@sumpfralle.de>" | |
|
41 | ||
|
42 | # Some committer email address domains that indicate that another entity might | |
|
43 | # hold some copyright too: | |
|
44 | domain_extra = {} | |
|
45 | domain_extra['unity3d.com'] = "Unity Technologies" | |
|
46 | domain_extra['rhodecode.com'] = "RhodeCode GmbH" | |
|
47 | ||
|
48 | # Repository history show some old contributions that traditionally hasn't been | |
|
49 | # listed in about.html - preserve that: | |
|
50 | no_about = set(total_ignore) | |
|
51 | # The following contributors were traditionally not listed in about.html and it | |
|
52 | # seems unclear if the copyright is personal or belongs to a company. | |
|
53 | no_about.add(('Thayne Harbaugh <thayne@fusionio.com>', '2011')) | |
|
54 | no_about.add(('Dies Koper <diesk@fast.au.fujitsu.com>', '2012')) | |
|
55 | no_about.add(('Erwin Kroon <e.kroon@smartmetersolutions.nl>', '2012')) | |
|
56 | no_about.add(('Vincent Caron <vcaron@bearstech.com>', '2012')) | |
|
57 | # These contributors' contributions might be too small to be copyrightable: | |
|
58 | no_about.add(('philip.j@hostdime.com', '2012')) | |
|
59 | no_about.add(('Stefan Engel <mail@engel-stefan.de>', '2012')) | |
|
60 | no_about.add(('Ton Plomp <tcplomp@gmail.com>', '2013')) | |
|
61 | # Was reworked and contributed later and shadowed by other contributions: | |
|
62 | no_about.add(('Sean Farley <sean.michael.farley@gmail.com>', '2013')) | |
|
63 | ||
|
64 | # Preserve contributors listed in about.html but not appearing in repository | |
|
65 | # history: | |
|
66 | other_about = [ | |
|
67 | ("2011", "Aparkar <aparkar@icloud.com>"), | |
|
68 | ("2010", "RhodeCode GmbH"), | |
|
69 | ("2011", "RhodeCode GmbH"), | |
|
70 | ("2012", "RhodeCode GmbH"), | |
|
71 | ("2013", "RhodeCode GmbH"), | |
|
72 | ] | |
|
73 | ||
|
74 | # Preserve contributors listed in CONTRIBUTORS but not appearing in repository | |
|
75 | # history: | |
|
76 | other_contributors = [ | |
|
77 | ("", "Andrew Kesterson <andrew@aklabs.net>"), | |
|
78 | ("", "cejones"), | |
|
79 | ("", "David A. SjΓΈen <david.sjoen@westcon.no>"), | |
|
80 | ("", "James Rhodes <jrhodes@redpointsoftware.com.au>"), | |
|
81 | ("", "Jonas Oberschweiber <jonas.oberschweiber@d-velop.de>"), | |
|
82 | ("", "larikale"), | |
|
83 | ("", "RhodeCode GmbH"), | |
|
84 | ("", "Sebastian Kreutzberger <sebastian@rhodecode.com>"), | |
|
85 | ("", "Steve Romanow <slestak989@gmail.com>"), | |
|
86 | ("", "SteveCohen"), | |
|
87 | ("", "Thomas <thomas@rhodecode.com>"), | |
|
88 | ("", "Thomas Waldmann <tw-public@gmx.de>"), | |
|
89 | ] |
@@ -1,254 +1,164 b'' | |||
|
1 | 1 | #!/usr/bin/env python2 |
|
2 | 2 | # -*- coding: utf-8 -*- |
|
3 | 3 | |
|
4 | 4 | """ |
|
5 | 5 | Kallithea script for maintaining contributor lists from version control |
|
6 | 6 | history. |
|
7 | 7 | |
|
8 | 8 | This script and the data in it is a best effort attempt at reverse engineering |
|
9 | 9 | previous attributions and correlate that with version control history while |
|
10 | 10 | preserving all existing copyright statements and attribution. This script is |
|
11 | 11 | processing and summarizing information found elsewhere - it is not by itself |
|
12 | 12 | making any claims. Comments in the script are an attempt at reverse engineering |
|
13 | 13 | possible explanations - they are not showing any intent or confirming it is |
|
14 | 14 | correct. |
|
15 | 15 | |
|
16 | 16 | Three files are generated / modified by this script: |
|
17 | 17 | |
|
18 | 18 | kallithea/templates/about.html claims to show copyright holders, and the GPL |
|
19 | 19 | license requires such existing "legal notices" to be preserved. We also try to |
|
20 | 20 | keep it updated with copyright holders, but do not claim it is a correct list. |
|
21 | 21 | |
|
22 | 22 | CONTRIBUTORS has the purpose of giving credit where credit is due and list all |
|
23 | 23 | the contributor names in the source. |
|
24 | 24 | |
|
25 | 25 | kallithea/templates/base/base.html contains the copyright years in the page |
|
26 | 26 | footer. |
|
27 | 27 | |
|
28 | 28 | Both make a best effort of listing all copyright holders, but revision control |
|
29 | 29 | history might be a better and more definitive source. |
|
30 | 30 | |
|
31 | 31 | Contributors are sorted "fairly" by copyright year and amount of |
|
32 | 32 | contribution. |
|
33 | 33 | |
|
34 | 34 | New contributors are listed, without considering if the contribution contains |
|
35 | 35 | copyrightable work. |
|
36 | 36 | |
|
37 | 37 | When the copyright might belong to a different legal entity than the |
|
38 | 38 | contributor, the legal entity is given credit too. |
|
39 | 39 | """ |
|
40 | 40 | |
|
41 | ||
|
42 | # Some committers are so wrong that it doesn't point at any contributor: | |
|
43 | total_ignore = set() | |
|
44 | total_ignore.add('*** failed to import extension hggit: No module named hggit') | |
|
45 | total_ignore.add('<>') | |
|
46 | ||
|
47 | # Normalize some committer names where people have contributed under different | |
|
48 | # names or email addresses: | |
|
49 | name_fixes = {} | |
|
50 | name_fixes['Andrew Shadura'] = "Andrej Shadura <andrew@shadura.me>" | |
|
51 | name_fixes['aparkar'] = "Aparkar <aparkar@icloud.com>" | |
|
52 | name_fixes['Aras Pranckevicius'] = "Aras PranckeviΔius <aras@unity3d.com>" | |
|
53 | name_fixes['Augosto Hermann'] = "Augusto Herrmann <augusto.herrmann@planejamento.gov.br>" | |
|
54 | name_fixes['"Bradley M. Kuhn" <bkuhn@ebb.org>'] = "Bradley M. Kuhn <bkuhn@sfconservancy.org>" | |
|
55 | name_fixes['dmitri.kuznetsov'] = "Dmitri Kuznetsov" | |
|
56 | name_fixes['Dmitri Kuznetsov'] = "Dmitri Kuznetsov" | |
|
57 | name_fixes['domruf'] = "Dominik Ruf <dominikruf@gmail.com>" | |
|
58 | name_fixes['Ingo von borstel'] = "Ingo von Borstel <kallithea@planetmaker.de>" | |
|
59 | name_fixes['Jan Heylen'] = "Jan Heylen <heyleke@gmail.com>" | |
|
60 | name_fixes['Jason F. Harris'] = "Jason Harris <jason@jasonfharris.com>" | |
|
61 | name_fixes['Jelmer Vernooij'] = "Jelmer VernooΔ³ <jelmer@samba.org>" | |
|
62 | name_fixes['jfh <jason@jasonfharris.com>'] = "Jason Harris <jason@jasonfharris.com>" | |
|
63 | name_fixes['Leonardo Carneiro<leonardo@unity3d.com>'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
64 | name_fixes['leonardo'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
65 | name_fixes['Leonardo <leo@unity3d.com>'] = "Leonardo Carneiro <leonardo@unity3d.com>" | |
|
66 | name_fixes['Les Peabody'] = "Les Peabody <lpeabody@gmail.com>" | |
|
67 | name_fixes['"Lorenzo M. Catucci" <lorenzo@sancho.ccd.uniroma2.it>'] = "Lorenzo M. Catucci <lorenzo@sancho.ccd.uniroma2.it>" | |
|
68 | name_fixes['Lukasz Balcerzak'] = "Εukasz Balcerzak <lukaszbalcerzak@gmail.com>" | |
|
69 | name_fixes['mao <mao@lins.fju.edu.tw>'] = "Ching-Chen Mao <mao@lins.fju.edu.tw>" | |
|
70 | name_fixes['marcink'] = "Marcin KuΕΊmiΕski <marcin@python-works.com>" | |
|
71 | name_fixes['Marcin Kuzminski'] = "Marcin KuΕΊmiΕski <marcin@python-works.com>" | |
|
72 | name_fixes['nansenat16@null.tw'] = "nansenat16 <nansenat16@null.tw>" | |
|
73 | name_fixes['Peter Vitt'] = "Peter Vitt <petervitt@web.de>" | |
|
74 | name_fixes['philip.j@hostdime.com'] = "Philip Jameson <philip.j@hostdime.com>" | |
|
75 | name_fixes['SΓΈren LΓΈvborg'] = "SΓΈren LΓΈvborg <sorenl@unity3d.com>" | |
|
76 | name_fixes['Thomas De Schampheleire'] = "Thomas De Schampheleire <thomas.de_schampheleire@nokia.com>" | |
|
77 | name_fixes['Weblate'] = "<>" | |
|
78 | name_fixes['xpol'] = "xpol <xpolife@gmail.com>" | |
|
79 | name_fixes['Lars <devel@sumpfralle.de>'] = "Lars Kruse <devel@sumpfralle.de>" | |
|
80 | ||
|
81 | ||
|
82 | # Some committer email address domains that indicate that another entity might | |
|
83 | # hold some copyright too: | |
|
84 | domain_extra = {} | |
|
85 | domain_extra['unity3d.com'] = "Unity Technologies" | |
|
86 | domain_extra['rhodecode.com'] = "RhodeCode GmbH" | |
|
87 | ||
|
88 | # Repository history show some old contributions that traditionally hasn't been | |
|
89 | # listed in about.html - preserve that: | |
|
90 | no_about = set(total_ignore) | |
|
91 | # The following contributors were traditionally not listed in about.html and it | |
|
92 | # seems unclear if the copyright is personal or belongs to a company. | |
|
93 | no_about.add(('Thayne Harbaugh <thayne@fusionio.com>', '2011')) | |
|
94 | no_about.add(('Dies Koper <diesk@fast.au.fujitsu.com>', '2012')) | |
|
95 | no_about.add(('Erwin Kroon <e.kroon@smartmetersolutions.nl>', '2012')) | |
|
96 | no_about.add(('Vincent Caron <vcaron@bearstech.com>', '2012')) | |
|
97 | # These contributors' contributions might be too small to be copyrightable: | |
|
98 | no_about.add(('philip.j@hostdime.com', '2012')) | |
|
99 | no_about.add(('Stefan Engel <mail@engel-stefan.de>', '2012')) | |
|
100 | no_about.add(('Ton Plomp <tcplomp@gmail.com>', '2013')) | |
|
101 | # Was reworked and contributed later and shadowed by other contributions: | |
|
102 | no_about.add(('Sean Farley <sean.michael.farley@gmail.com>', '2013')) | |
|
103 | ||
|
104 | # Preserve contributors listed in about.html but not appearing in repository | |
|
105 | # history: | |
|
106 | other_about = [ | |
|
107 | ("2011", "Aparkar <aparkar@icloud.com>"), | |
|
108 | ("2010", "RhodeCode GmbH"), | |
|
109 | ("2011", "RhodeCode GmbH"), | |
|
110 | ("2012", "RhodeCode GmbH"), | |
|
111 | ("2013", "RhodeCode GmbH"), | |
|
112 | ] | |
|
113 | ||
|
114 | # Preserve contributors listed in CONTRIBUTORS but not appearing in repository | |
|
115 | # history: | |
|
116 | other_contributors = [ | |
|
117 | ("", "Andrew Kesterson <andrew@aklabs.net>"), | |
|
118 | ("", "cejones"), | |
|
119 | ("", "David A. SjΓΈen <david.sjoen@westcon.no>"), | |
|
120 | ("", "James Rhodes <jrhodes@redpointsoftware.com.au>"), | |
|
121 | ("", "Jonas Oberschweiber <jonas.oberschweiber@d-velop.de>"), | |
|
122 | ("", "larikale"), | |
|
123 | ("", "RhodeCode GmbH"), | |
|
124 | ("", "Sebastian Kreutzberger <sebastian@rhodecode.com>"), | |
|
125 | ("", "Steve Romanow <slestak989@gmail.com>"), | |
|
126 | ("", "SteveCohen"), | |
|
127 | ("", "Thomas <thomas@rhodecode.com>"), | |
|
128 | ("", "Thomas Waldmann <tw-public@gmx.de>"), | |
|
129 | ] | |
|
130 | ||
|
131 | ||
|
132 | 41 | import os |
|
133 | 42 | import re |
|
134 | 43 | from collections import defaultdict |
|
44 | import contributor_data | |
|
135 | 45 | |
|
136 | 46 | |
|
137 | 47 | def sortkey(x): |
|
138 | 48 | """Return key for sorting contributors "fairly": |
|
139 | 49 | * latest contribution |
|
140 | 50 | * first contribution |
|
141 | 51 | * number of contribution years |
|
142 | 52 | * name (with some unicode normalization) |
|
143 | 53 | The entries must be 2-tuples of a list of string years and the unicode name""" |
|
144 | 54 | return (x[0] and -int(x[0][-1]), |
|
145 | 55 | x[0] and int(x[0][0]), |
|
146 | 56 | -len(x[0]), |
|
147 | 57 | x[1].decode('utf-8').lower().replace(u'\xe9', u'e').replace(u'\u0142', u'l') |
|
148 | 58 | ) |
|
149 | 59 | |
|
150 | 60 | |
|
151 | 61 | def nice_years(l, dash='-', join=' '): |
|
152 | 62 | """Convert a list of years into brief range like '1900-1901, 1921'.""" |
|
153 | 63 | if not l: |
|
154 | 64 | return '' |
|
155 | 65 | start = end = int(l[0]) |
|
156 | 66 | ranges = [] |
|
157 | 67 | for year in l[1:] + [0]: |
|
158 | 68 | year = int(year) |
|
159 | 69 | if year == end + 1: |
|
160 | 70 | end = year |
|
161 | 71 | continue |
|
162 | 72 | if start == end: |
|
163 | 73 | ranges.append('%s' % start) |
|
164 | 74 | else: |
|
165 | 75 | ranges.append('%s%s%s' % (start, dash, end)) |
|
166 | 76 | start = end = year |
|
167 | 77 | assert start == 0 and end == 0, (start, end) |
|
168 | 78 | return join.join(ranges) |
|
169 | 79 | |
|
170 | 80 | |
|
171 | 81 | def insert_entries( |
|
172 | 82 | filename, |
|
173 | 83 | all_entries, |
|
174 | 84 | no_entries, |
|
175 | 85 | domain_extra, |
|
176 | 86 | split_re, |
|
177 | 87 | normalize_name, |
|
178 | 88 | format_f): |
|
179 | 89 | """Update file with contributor information. |
|
180 | 90 | all_entries: list of tuples with year and name |
|
181 | 91 | no_entries: set of names or name and year tuples to ignore |
|
182 | 92 | domain_extra: map domain name to extra credit name |
|
183 | 93 | split_re: regexp matching the part of file to rewrite |
|
184 | 94 | normalize_name: function to normalize names for grouping and display |
|
185 | 95 | format_f: function formatting year list and name to a string |
|
186 | 96 | """ |
|
187 | 97 | name_years = defaultdict(set) |
|
188 | 98 | |
|
189 | 99 | for year, name in all_entries: |
|
190 | 100 | if name in no_entries or (name, year) in no_entries: |
|
191 | 101 | continue |
|
192 | 102 | domain = name.split('@', 1)[-1].rstrip('>') |
|
193 | 103 | if domain in domain_extra: |
|
194 | 104 | name_years[domain_extra[domain]].add(year) |
|
195 | 105 | name_years[normalize_name(name)].add(year) |
|
196 | 106 | |
|
197 | 107 | l = [(list(sorted(year for year in years if year)), name) |
|
198 | 108 | for name, years in name_years.items()] |
|
199 | 109 | l.sort(key=sortkey) |
|
200 | 110 | |
|
201 | 111 | with open(filename) as f: |
|
202 | 112 | pre, post = re.split(split_re, f.read()) |
|
203 | 113 | |
|
204 | 114 | with open(filename, 'w') as f: |
|
205 | 115 | f.write(pre + |
|
206 | 116 | ''.join(format_f(years, name) for years, name in l) + |
|
207 | 117 | post) |
|
208 | 118 | |
|
209 | 119 | |
|
210 | 120 | def main(): |
|
211 | 121 | repo_entries = [ |
|
212 | (year, name_fixes.get(name) or name_fixes.get(name.rsplit('<', 1)[0].strip()) or name) | |
|
122 | (year, contributor_data.name_fixes.get(name) or contributor_data.name_fixes.get(name.rsplit('<', 1)[0].strip()) or name) | |
|
213 | 123 | for year, name in |
|
214 | 124 | (line.strip().split(' ', 1) |
|
215 | 125 | for line in os.popen("""hg log -r '::.' -T '{date(date,"%Y")} {author}\n'""").readlines()) |
|
216 | 126 | ] |
|
217 | 127 | |
|
218 | 128 | insert_entries( |
|
219 | 129 | filename='kallithea/templates/about.html', |
|
220 | all_entries=repo_entries + other_about, | |
|
221 | no_entries=no_about, | |
|
222 | domain_extra=domain_extra, | |
|
130 | all_entries=repo_entries + contributor_data.other_about, | |
|
131 | no_entries=contributor_data.no_about, | |
|
132 | domain_extra=contributor_data.domain_extra, | |
|
223 | 133 | split_re=r'(?: <li>Copyright © [^\n]*</li>\n)*', |
|
224 | 134 | normalize_name=lambda name: name.split('<', 1)[0].strip(), |
|
225 | 135 | format_f=lambda years, name: ' <li>Copyright © %s, %s</li>\n' % (nice_years(years, '–', ', '), name), |
|
226 | 136 | ) |
|
227 | 137 | |
|
228 | 138 | insert_entries( |
|
229 | 139 | filename='CONTRIBUTORS', |
|
230 | all_entries=repo_entries + other_contributors, | |
|
231 | no_entries=total_ignore, | |
|
232 | domain_extra=domain_extra, | |
|
140 | all_entries=repo_entries + contributor_data.other_contributors, | |
|
141 | no_entries=contributor_data.total_ignore, | |
|
142 | domain_extra=contributor_data.domain_extra, | |
|
233 | 143 | split_re=r'(?: [^\n]*\n)*', |
|
234 | 144 | normalize_name=lambda name: name, |
|
235 | 145 | format_f=lambda years, name: (' %s%s%s\n' % (name, ' ' if years else '', nice_years(years))), |
|
236 | 146 | ) |
|
237 | 147 | |
|
238 | 148 | insert_entries( |
|
239 | 149 | filename='kallithea/templates/base/base.html', |
|
240 | 150 | all_entries=repo_entries, |
|
241 | no_entries=total_ignore, | |
|
151 | no_entries=contributor_data.total_ignore, | |
|
242 | 152 | domain_extra={}, |
|
243 | 153 | split_re=r'(?<=©) .* (?=by various authors)', |
|
244 | 154 | normalize_name=lambda name: '', |
|
245 | 155 | format_f=lambda years, name: ' ' + nice_years(years, '–', ', ') + ' ', |
|
246 | 156 | ) |
|
247 | 157 | |
|
248 | 158 | |
|
249 | 159 | if __name__ == '__main__': |
|
250 | 160 | main() |
|
251 | 161 | |
|
252 | 162 | |
|
253 | 163 | # To list new contributors since last tagging: |
|
254 | 164 | # { hg log -r '::tagged()' -T ' {author}\n {author}\n'; hg log -r '::.' -T ' {author}\n' | sort | uniq; } | sort | uniq -u |
General Comments 0
You need to be logged in to leave comments.
Login now