Bug 1869647 - Mark hasStorageAccess.sub.https.window.html as intermittent after wpt...
[gecko.git] / intl / icu_sources_data.py
blobde394eaa3f784b9e89d5b883c8e1a8d81418a63c
1 #!/usr/bin/env python
3 # Any copyright is dedicated to the Public Domain.
4 # http://creativecommons.org/publicdomain/zero/1.0/
6 # Generate SOURCES in sources.mozbuild files from ICU's Makefile.in
7 # files, and also build a standalone copy of ICU using its build
8 # system to generate a new copy of the in-tree ICU data file.
10 # This script expects to be run from `update-icu.sh` after the in-tree
11 # copy of ICU has been updated.
13 import glob
14 import multiprocessing
15 import os
16 import shutil
17 import subprocess
18 import sys
19 import tempfile
21 from mozpack import path as mozpath
23 # The following files have been determined to be dead/unused by a
24 # semi-automated analysis. You can just remove any of the files below
25 # if you need them. However, files marked with a "Cluster" comment
26 # can only be removed together, as they have (directional) dependencies.
27 # If you want to rerun this analysis, contact :decoder.
28 UNUSED_SOURCES = set(
30 "intl/icu/source/common/bytestrieiterator.cpp",
31 "intl/icu/source/common/cstr.cpp",
32 "intl/icu/source/common/cwchar.cpp",
33 "intl/icu/source/common/icudataver.cpp",
34 "intl/icu/source/common/icuplug.cpp",
35 "intl/icu/source/common/pluralmap.cpp",
36 "intl/icu/source/common/ucat.cpp",
37 "intl/icu/source/common/ucnv2022.cpp",
38 "intl/icu/source/common/ucnv_ct.cpp",
39 "intl/icu/source/common/ucnvdisp.cpp",
40 "intl/icu/source/common/ucnv_ext.cpp",
41 "intl/icu/source/common/ucnvhz.cpp",
42 "intl/icu/source/common/ucnvisci.cpp",
43 "intl/icu/source/common/ucnv_lmb.cpp",
44 "intl/icu/source/common/ucnvmbcs.cpp",
45 "intl/icu/source/common/uidna.cpp",
46 "intl/icu/source/common/unorm.cpp",
47 "intl/icu/source/common/usc_impl.cpp",
48 "intl/icu/source/common/ustr_wcs.cpp",
49 "intl/icu/source/common/util_props.cpp",
50 "intl/icu/source/i18n/anytrans.cpp",
51 "intl/icu/source/i18n/brktrans.cpp",
52 "intl/icu/source/i18n/casetrn.cpp",
53 "intl/icu/source/i18n/cpdtrans.cpp",
54 "intl/icu/source/i18n/esctrn.cpp",
55 "intl/icu/source/i18n/fmtable_cnv.cpp",
56 "intl/icu/source/i18n/funcrepl.cpp",
57 "intl/icu/source/i18n/gender.cpp",
58 "intl/icu/source/i18n/name2uni.cpp",
59 "intl/icu/source/i18n/nortrans.cpp",
60 "intl/icu/source/i18n/nultrans.cpp",
61 "intl/icu/source/i18n/quant.cpp",
62 "intl/icu/source/i18n/rbt.cpp",
63 "intl/icu/source/i18n/rbt_data.cpp",
64 "intl/icu/source/i18n/rbt_pars.cpp",
65 "intl/icu/source/i18n/rbt_rule.cpp",
66 "intl/icu/source/i18n/rbt_set.cpp",
67 "intl/icu/source/i18n/regexcmp.cpp",
68 "intl/icu/source/i18n/regeximp.cpp",
69 "intl/icu/source/i18n/regexst.cpp",
70 "intl/icu/source/i18n/regextxt.cpp",
71 "intl/icu/source/i18n/rematch.cpp",
72 "intl/icu/source/i18n/remtrans.cpp",
73 "intl/icu/source/i18n/repattrn.cpp",
74 "intl/icu/source/i18n/scientificnumberformatter.cpp",
75 "intl/icu/source/i18n/strmatch.cpp",
76 "intl/icu/source/i18n/strrepl.cpp",
77 "intl/icu/source/i18n/titletrn.cpp",
78 "intl/icu/source/i18n/tolowtrn.cpp",
79 "intl/icu/source/i18n/toupptrn.cpp",
80 "intl/icu/source/i18n/translit.cpp",
81 "intl/icu/source/i18n/transreg.cpp",
82 "intl/icu/source/i18n/tridpars.cpp",
83 "intl/icu/source/i18n/unesctrn.cpp",
84 "intl/icu/source/i18n/uni2name.cpp",
85 "intl/icu/source/i18n/uregexc.cpp",
86 "intl/icu/source/i18n/uregex.cpp",
87 "intl/icu/source/i18n/uregion.cpp",
88 "intl/icu/source/i18n/uspoof_build.cpp",
89 "intl/icu/source/i18n/uspoof_conf.cpp",
90 "intl/icu/source/i18n/utrans.cpp",
91 "intl/icu/source/i18n/vzone.cpp",
92 "intl/icu/source/i18n/zrule.cpp",
93 "intl/icu/source/i18n/ztrans.cpp",
94 # Cluster
95 "intl/icu/source/common/resbund_cnv.cpp",
96 "intl/icu/source/common/ures_cnv.cpp",
97 # Cluster
98 "intl/icu/source/common/propsvec.cpp",
99 "intl/icu/source/common/ucnvsel.cpp",
100 "intl/icu/source/common/ucnv_set.cpp",
101 # Cluster
102 "intl/icu/source/common/ubiditransform.cpp",
103 "intl/icu/source/common/ushape.cpp",
104 # Cluster
105 "intl/icu/source/i18n/csdetect.cpp",
106 "intl/icu/source/i18n/csmatch.cpp",
107 "intl/icu/source/i18n/csr2022.cpp",
108 "intl/icu/source/i18n/csrecog.cpp",
109 "intl/icu/source/i18n/csrmbcs.cpp",
110 "intl/icu/source/i18n/csrsbcs.cpp",
111 "intl/icu/source/i18n/csrucode.cpp",
112 "intl/icu/source/i18n/csrutf8.cpp",
113 "intl/icu/source/i18n/inputext.cpp",
114 "intl/icu/source/i18n/ucsdet.cpp",
115 # Cluster
116 "intl/icu/source/i18n/alphaindex.cpp",
117 "intl/icu/source/i18n/ulocdata.cpp",
120 # We use the version of double-conversion vendored in mfbt instead
122 "intl/icu/source/i18n/double-conversion-bignum-dtoa.cpp",
123 "intl/icu/source/i18n/double-conversion-bignum-dtoa.h",
124 "intl/icu/source/i18n/double-conversion-bignum.cpp",
125 "intl/icu/source/i18n/double-conversion-bignum.h",
126 "intl/icu/source/i18n/double-conversion-cached-powers.cpp",
127 "intl/icu/source/i18n/double-conversion-cached-powers.h",
128 "intl/icu/source/i18n/double-conversion-diy-fp.h",
129 "intl/icu/source/i18n/double-conversion-double-to-string.cpp",
130 "intl/icu/source/i18n/double-conversion-double-to-string.h",
131 "intl/icu/source/i18n/double-conversion-fast-dtoa.cpp",
132 "intl/icu/source/i18n/double-conversion-fast-dtoa.h",
133 "intl/icu/source/i18n/double-conversion-ieee.h",
134 "intl/icu/source/i18n/double-conversion-string-to-double.cpp",
135 "intl/icu/source/i18n/double-conversion-string-to-double.h",
136 "intl/icu/source/i18n/double-conversion-strtod.cpp",
137 "intl/icu/source/i18n/double-conversion-strtod.h",
138 "intl/icu/source/i18n/double-conversion-utils.h",
139 "intl/icu/source/i18n/double-conversion.h",
144 def ensure_source_file_exists(dir, filename):
145 f = mozpath.join(dir, filename)
146 if os.path.isfile(f):
147 return f
148 raise Exception("Couldn't find source file for: %s" % filename)
151 def get_sources(sources_file):
152 srcdir = os.path.dirname(sources_file)
153 with open(sources_file) as f:
154 return sorted(
155 (ensure_source_file_exists(srcdir, name.strip()) for name in f),
156 key=lambda x: x.lower(),
160 def list_headers(path):
161 result = []
162 for name in os.listdir(path):
163 f = mozpath.join(path, name)
164 if os.path.isfile(f):
165 result.append(f)
166 return sorted(result, key=lambda x: x.lower())
169 def write_sources(mozbuild, sources, headers):
170 with open(mozbuild, "w", newline="\n", encoding="utf-8") as f:
171 f.write(
172 "# THIS FILE IS GENERATED BY /intl/icu_sources_data.py " + "DO NOT EDIT\n"
175 def write_list(name, content):
176 if content:
177 f.write("%s %s [\n" % (name, "=" if name.islower() else "+="))
178 f.write("".join(' "/%s",\n' % s for s in content))
179 f.write("]\n")
181 write_list("sources", [s for s in sources if s not in UNUSED_SOURCES])
182 write_list("other_sources", [s for s in sources if s in UNUSED_SOURCES])
183 write_list("EXPORTS.unicode", headers)
186 def update_sources(topsrcdir):
187 print("Updating ICU sources lists...")
188 for d in ["common", "i18n", "tools/toolutil", "tools/icupkg"]:
189 base_path = mozpath.join(topsrcdir, "intl/icu/source/%s" % d)
190 sources_file = mozpath.join(base_path, "sources.txt")
191 mozbuild = mozpath.join(
192 topsrcdir, "config/external/icu/%s/sources.mozbuild" % mozpath.basename(d)
194 sources = [mozpath.relpath(s, topsrcdir) for s in get_sources(sources_file)]
195 unicode_dir = mozpath.join(base_path, "unicode")
196 if os.path.exists(unicode_dir):
197 headers = [
198 mozpath.normsep(os.path.relpath(s, topsrcdir))
199 for s in list_headers(unicode_dir)
201 else:
202 headers = None
203 write_sources(mozbuild, sources, headers)
206 def try_run(name, command, cwd=None, **kwargs):
207 try:
208 with tempfile.NamedTemporaryFile(prefix=name, delete=False) as f:
209 subprocess.check_call(
210 command, cwd=cwd, stdout=f, stderr=subprocess.STDOUT, **kwargs
212 except subprocess.CalledProcessError:
213 print(
214 """Error running "{}" in directory {}
215 See output in {}""".format(
216 " ".join(command), cwd, f.name
218 file=sys.stderr,
220 return False
221 else:
222 os.unlink(f.name)
223 return True
226 def get_data_file(data_dir):
227 files = glob.glob(mozpath.join(data_dir, "icudt*.dat"))
228 return files[0] if files else None
231 def update_data_file(topsrcdir):
232 objdir = tempfile.mkdtemp(prefix="icu-obj-")
233 configure = mozpath.join(topsrcdir, "intl/icu/source/configure")
234 env = dict(os.environ)
235 # bug 1262101 - these should be shared with the moz.build files
236 env.update(
238 "CPPFLAGS": (
239 "-DU_NO_DEFAULT_INCLUDE_UTF_HEADERS=1 "
240 + "-DU_HIDE_OBSOLETE_UTF_OLD_H=1 "
241 + "-DUCONFIG_NO_LEGACY_CONVERSION "
242 + "-DUCONFIG_NO_TRANSLITERATION "
243 + "-DUCONFIG_NO_REGULAR_EXPRESSIONS "
244 + "-DUCONFIG_NO_BREAK_ITERATION "
245 + "-DU_CHARSET_IS_UTF8 "
250 # Exclude data that we currently don't need.
252 # The file format for ICU's data build tool is described at
253 # <https://github.com/unicode-org/icu/blob/master/docs/userguide/icu_data/buildtool.md>.
254 env["ICU_DATA_FILTER_FILE"] = mozpath.join(topsrcdir, "intl/icu/data_filter.json")
256 print("Running ICU configure...")
257 if not try_run(
258 "icu-configure",
260 "sh",
261 configure,
262 "--with-data-packaging=archive",
263 "--enable-static",
264 "--disable-shared",
265 "--disable-extras",
266 "--disable-icuio",
267 "--disable-layout",
268 "--disable-layoutex",
269 "--disable-tests",
270 "--disable-samples",
271 "--disable-strict",
273 cwd=objdir,
274 env=env,
276 return False
277 print("Running ICU make...")
278 if not try_run(
279 "icu-make",
280 ["make", "--jobs=%d" % multiprocessing.cpu_count(), "--output-sync"],
281 cwd=objdir,
283 return False
284 print("Copying ICU data file...")
285 tree_data_path = mozpath.join(topsrcdir, "config/external/icu/data/")
286 old_data_file = get_data_file(tree_data_path)
287 if not old_data_file:
288 print("Error: no ICU data file in %s" % tree_data_path, file=sys.stderr)
289 return False
290 new_data_file = get_data_file(mozpath.join(objdir, "data/out"))
291 if not new_data_file:
292 print("Error: no ICU data in ICU objdir", file=sys.stderr)
293 return False
294 if os.path.basename(old_data_file) != os.path.basename(new_data_file):
295 # Data file name has the major version number embedded.
296 os.unlink(old_data_file)
297 shutil.copy(new_data_file, tree_data_path)
298 try:
299 shutil.rmtree(objdir)
300 except Exception:
301 print("Warning: failed to remove %s" % objdir, file=sys.stderr)
302 return True
305 def main():
306 if len(sys.argv) != 2:
307 print("Usage: icu_sources_data.py <mozilla topsrcdir>", file=sys.stderr)
308 sys.exit(1)
310 topsrcdir = mozpath.abspath(sys.argv[1])
311 update_sources(topsrcdir)
312 if not update_data_file(topsrcdir):
313 print("Error updating ICU data file", file=sys.stderr)
314 sys.exit(1)
317 if __name__ == "__main__":
318 main()