ownsource: try to honour .gitignore
[hippotat] / hippotatlib / ownsource.py
CommitLineData
4460af45
IJ
1# Automatic source code provision (AGPL compliance)
2
b635cd93 3import os
4460af45
IJ
4import sys
5import fnmatch
b635cd93
IJ
6import stat
7import subprocess
e68fca7c 8import tempfile
4460af45
IJ
9
10class SourceShipmentPreparer():
11 def __init__(s, destdir):
12 # caller may modify, and should read after calling generate()
13 s.output_name = 'srcbomb.tar.gz'
14 # defaults, caller can modify after creation
15 s.src_filter = s.src_filter_glob
e68fca7c 16 s.src_package_globs = ['!/usr/local/*', '/usr*']
e98894bc 17 s.src_filter_globs = ['!/etc/*']
4460af45 18 s.src_likeparent = s.src_likeparent_git
28861486 19 s.src_direxcludes = s.src_direxcludes_git
e98894bc 20 s.report_from_packages = s.report_from_packages_debian
4460af45 21 s.cwd = os.getcwd()
b635cd93 22 s.find_rune_base = "find -type f -perm -004 \! -path '*/tmp/*'"
28861486 23 s.ignores = ['*~', '*.bak', '*.tmp', '#*#', '__pycache__',
b635cd93 24 '[0-9][0-9][0-9][0-9]-src.cpio']
4460af45 25 s.rune_shell = ['/bin/bash', '-ec']
b635cd93
IJ
26 s.show_pathnames = True
27 s.rune_cpio = r'''
4460af45
IJ
28 set -o pipefail
29 (
30 %s
31 # ^ by default, is find ... -print0
32 ) | (
33 cpio -Hustar -o --quiet -0 -R 1000:1000 || \
34 cpio -Hustar -o --quiet -0
35 )
36 '''
37 s.rune_portmanteau = r'''
38 outfile=$1; shift
39 rm -f "$outfile"
16374080 40 GZIP=-1 tar zcf "$outfile" "$@"
4460af45
IJ
41 '''
42 s.manifest_name='0000-MANIFEST.txt'
43 # private
44 s._destdir = destdir
b635cd93 45 s._outcounter = 0
4460af45 46 s._manifest = []
2371516a 47 s._dirmap = { }
e98894bc 48 s._package_files = { } # map filename => infol
4460af45 49
e98894bc
IJ
50 def thing_matches_globs(s, thing, globs):
51 for pat in globs:
4460af45
IJ
52 negate = pat.startswith('!')
53 if negate: pat = pat[1:]
e98894bc 54 if fnmatch.fnmatch(thing, pat):
4460af45
IJ
55 return not negate
56 return negate
57
e98894bc 58 def src_filter_glob(s, src): # default s.src_filter
e68fca7c 59 return s.thing_matches_globs(src, s.src_filter_globs)
e98894bc 60
28861486
IJ
61 def src_direxcludes_git(s, d):
62 try:
63 excl = open(os.path.join(d, '.gitignore'))
64 except FileNotFoundError:
65 return []
66 r = []
67 for l in excl:
68 l.strip
69 if l.startswith('#'): next
70 if not len(l): next
71 r += l
72 return r
73
4460af45
IJ
74 def src_likeparent_git(s, src):
75 try:
b635cd93 76 os.stat(os.path.join(src, '.git/.'))
4460af45
IJ
77 except FileNotFoundError:
78 return False
79 else:
80 return True
81
82 def src_parentfinder(s, src, infol): # callers may monkey-patch away
83 for deref in (False,True):
84 xinfo = []
85
86 search = src
87 if deref:
88 search = os.path.realpath(search)
89
90 def ascend():
b635cd93 91 nonlocal search
4460af45
IJ
92 xinfo.append(os.path.basename(search))
93 search = os.path.dirname(search)
94
95 try:
b635cd93 96 stab = os.lstat(search)
4460af45
IJ
97 except FileNotFoundError:
98 return
99 if stat.S_ISREG(stab.st_mode):
100 ascend()
101
102 while not os.path.ismount(search):
103 if s.src_likeparent(search):
104 xinfo.reverse()
b635cd93 105 if len(xinfo): infol.append('want=' + os.path.join(*xinfo))
4460af45
IJ
106 return search
107
108 ascend()
109
110 # no .git found anywhere
b635cd93 111 return src
4460af45 112
e68fca7c 113 def path_prenormaliser(s, d, infol): # callers may monkey-patch away
4460af45
IJ
114 return os.path.join(s.cwd, os.path.abspath(d))
115
e98894bc 116 def srcdir_find_rune(s, d):
b635cd93 117 script = s.find_rune_base
28861486
IJ
118 ignores = s.ignores + [s.output_name, s.manifest_name]
119 ignores += s.src_direxcludes(d)
120 for excl in ignores:
4460af45 121 assert("'" not in excl)
28861486
IJ
122 script += r" \! -name '%s'" % excl
123 script += r" \! -path '*/%s/*'" % excl
4460af45 124 script += ' -print0'
b635cd93 125 return script
4460af45 126
2371516a 127 def manifest_append(s, name, infol):
a53aba4d 128 s._manifest.append({ 'file':name, 'info':' '.join(infol) })
2371516a 129
e68fca7c
IJ
130 def manifest_append_absentfile(s, name, infol):
131 s._manifest.append({ 'file_print':name, 'info':' '.join(infol) })
132
4460af45 133 def new_output_name(s, nametail, infol):
b635cd93
IJ
134 s._outcounter += 1
135 name = '%04d-%s' % (s._outcounter, nametail)
2371516a 136 s.manifest_append(name, infol)
4460af45
IJ
137 return name
138
139 def open_output_fh(s, name, mode):
140 return open(os.path.join(s._destdir, name), mode)
141
e98894bc 142 def src_dir(s, d, infol):
2371516a
IJ
143 try: name = s._dirmap[d]
144 except KeyError: pass
145 else:
146 s.manifest_append(name, infol)
147 return
148
b635cd93 149 if s.show_pathnames: infol.append(d)
e98894bc 150 find_rune = s.srcdir_find_rune(d)
4460af45 151 total_rune = s.rune_cpio % find_rune
2371516a
IJ
152
153 name = s.new_output_name('src.cpio', infol)
154 s._dirmap[d] = name
155 fh = s.open_output_fh(name, 'wb')
156
4460af45 157 subprocess.run(s.rune_shell + [total_rune],
b635cd93 158 cwd=d,
4460af45
IJ
159 stdin=subprocess.DEVNULL,
160 stdout=fh,
b635cd93
IJ
161 restore_signals=True,
162 check=True)
4460af45
IJ
163 fh.close()
164
e98894bc 165 def src_indir(s, d, infol):
e68fca7c 166 d = s.path_prenormaliser(d, infol)
4460af45 167 if not s.src_filter(d): return
e98894bc 168
4460af45 169 d = s.src_parentfinder(d, infol)
e68fca7c
IJ
170 if d is None: return
171 s.src_dir(d, infol)
e98894bc
IJ
172
173 def report_from_packages_debian(s, files):
e68fca7c 174 dpkg_S_in = tempfile.TemporaryFile(mode='w+')
e98894bc
IJ
175 for (file, infols) in files.items():
176 assert('\n' not in file)
177 dpkg_S_in.write(file)
178 dpkg_S_in.write('\0')
179 dpkg_S_in.seek(0)
180 cmdl = ['xargs','-0r','dpkg','-S','--']
181 dpkg_S = subprocess.Popen(cmdl,
a53aba4d
IJ
182 cwd='/',
183 stdin=dpkg_S_in,
184 stdout=subprocess.PIPE,
185 stderr=sys.stderr,
186 close_fds=False)
e68fca7c 187 dpkg_show_in = tempfile.TemporaryFile(mode='w+')
e98894bc 188 pkginfos = { }
e68fca7c
IJ
189 for l in dpkg_S.stdout:
190 l = l.strip(b'\n').decode('utf-8')
e98894bc 191 (pkgs, fname) = l.split(': ',1)
e68fca7c
IJ
192 pks = pkgs.split(', ')
193 for pk in pks:
194 pkginfos.setdefault(pk,{'files':[]})['files'].append(fname)
195 print(pk, file=dpkg_show_in)
a53aba4d
IJ
196 assert(dpkg_S.wait() == 0)
197 dpkg_show_in.seek(0)
198 cmdl = ['xargs','-r','dpkg-query',
e68fca7c 199 r'-f${binary:Package}\t${Package}\t${Architecture}\t${Version}\t${source:Package}\t${source:Version}\n',
a53aba4d
IJ
200 '--show','--']
201 dpkg_show = subprocess.Popen(cmdl,
202 cwd='/',
203 stdin=dpkg_show_in,
204 stdout=subprocess.PIPE,
205 stderr=sys.stderr,
206 close_fds=False)
207 for l in dpkg_show.stdout:
e68fca7c
IJ
208 l = l.strip(b'\n').decode('utf-8')
209 (pk,p,a,v,sp,sv) = l.split('\t')
210 pkginfos[pk]['binary'] = p
211 pkginfos[pk]['arch'] = a
212 pkginfos[pk]['version'] = v
213 pkginfos[pk]['source'] = sp
214 pkginfos[pk]['sourceversion'] = sv
a53aba4d 215 assert(dpkg_show.wait() == 0)
e68fca7c
IJ
216 for pk in sorted(pkginfos.keys()):
217 pi = pkginfos[pk]
218 debfname = '%s_%s_%s.deb' % (pi['binary'], pi['version'], pi['arch'])
a53aba4d 219 dscfname = '%s_%s.dsc' % (pi['source'], pi['sourceversion'])
e68fca7c
IJ
220 s.manifest_append_absentfile(dscfname, [debfname])
221 for fname in pi['files']:
222 infol = files[fname]
56ffddf8 223 if s.show_pathnames: infol = infol + ['loaded='+fname]
e68fca7c 224 s.manifest_append_absentfile(' \t' + debfname, infol)
e98894bc
IJ
225
226 def thing_ought_packaged(s, fname):
227 return s.thing_matches_globs(fname, s.src_package_globs)
228
e68fca7c
IJ
229 def src_file_packaged(s, fname, infol):
230 s._package_files.setdefault(fname,[]).extend(infol)
e98894bc
IJ
231
232 def src_file(s, fname, infol):
233 def fngens():
e68fca7c
IJ
234 yield (infol, fname)
235 infol_copy = infol.copy()
236 yield (infol_copy, s.path_prenormaliser(fname, infol_copy))
237 yield (infol, os.path.realpath(fname))
238
239 for (tinfol, tfname) in fngens():
240 if s.thing_ought_packaged(tfname):
241 s.src_file_packaged(tfname, tinfol)
e98894bc 242 return
4460af45 243
e98894bc
IJ
244 s.src_indir(fname, infol)
245
246 def src_argv0(s, program, infol):
e68fca7c 247 s.src_file(program, infol)
e98894bc
IJ
248
249 def src_syspath(s, fname, infol):
56ffddf8 250 if s.thing_ought_packaged(fname): return
e98894bc
IJ
251 s.src_indir(fname, infol)
252
253 def src_module(s, m, infol):
254 try: fname = m.__file__
16374080 255 except AttributeError: return
56ffddf8 256 infol.append('module='+m.__name__)
16374080 257
e98894bc
IJ
258 if s.thing_ought_packaged(fname):
259 s.src_file_packaged(fname, infol)
260 else:
e68fca7c 261 s.src_indir(fname, infol)
e98894bc
IJ
262
263 def srcs_allitems(s, dirs=sys.path):
264 s.src_argv0(sys.argv[0], ['argv[0]'])
4460af45 265 for d in sys.path:
e98894bc 266 s.src_syspath(d, ['sys.path'])
16374080 267 for m in sys.modules.values():
e98894bc 268 s.src_module(m, ['sys.modules'])
e68fca7c 269 s.report_from_packages(s._package_files)
4460af45 270
b635cd93 271 def mk_portmanteau(s):
4460af45
IJ
272 cmdl = s.rune_shell + [ s.rune_portmanteau, 'x',
273 s.output_name, s.manifest_name ]
b635cd93 274 mfh = s.open_output_fh(s.manifest_name,'w')
a53aba4d
IJ
275 for me in s._manifest:
276 try: fname = me['file']
277 except KeyError: fname = me.get('file_print','')
e68fca7c 278 else: cmdl.append(fname)
a53aba4d 279 print('%s\t%s' % (fname, me['info']), file=mfh)
4460af45 280 mfh.close()
2371516a 281 subprocess.run(cmdl,
b635cd93 282 cwd=s._destdir,
4460af45
IJ
283 stdin=subprocess.DEVNULL,
284 stdout=sys.stderr,
b635cd93
IJ
285 restore_signals=True,
286 check=True)
4460af45
IJ
287
288 def generate(s):
e98894bc 289 s.srcs_allitems()
4460af45 290 s.mk_portmanteau()