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