ownsource: logging etc.
[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'
c2c0da38 14 # s.output_path alternatively caller may read this
4460af45 15 # defaults, caller can modify after creation
c2c0da38 16 s.logger = lambda m: print('SourceShipmentPreparer',m)
4460af45 17 s.src_filter = s.src_filter_glob
e68fca7c 18 s.src_package_globs = ['!/usr/local/*', '/usr*']
e98894bc 19 s.src_filter_globs = ['!/etc/*']
4460af45 20 s.src_likeparent = s.src_likeparent_git
28861486 21 s.src_direxcludes = s.src_direxcludes_git
e98894bc 22 s.report_from_packages = s.report_from_packages_debian
4460af45 23 s.cwd = os.getcwd()
b635cd93 24 s.find_rune_base = "find -type f -perm -004 \! -path '*/tmp/*'"
28861486 25 s.ignores = ['*~', '*.bak', '*.tmp', '#*#', '__pycache__',
b635cd93 26 '[0-9][0-9][0-9][0-9]-src.cpio']
4460af45 27 s.rune_shell = ['/bin/bash', '-ec']
b635cd93
IJ
28 s.show_pathnames = True
29 s.rune_cpio = r'''
4460af45
IJ
30 set -o pipefail
31 (
32 %s
33 # ^ by default, is find ... -print0
34 ) | (
35 cpio -Hustar -o --quiet -0 -R 1000:1000 || \
36 cpio -Hustar -o --quiet -0
37 )
38 '''
39 s.rune_portmanteau = r'''
40 outfile=$1; shift
41 rm -f "$outfile"
16374080 42 GZIP=-1 tar zcf "$outfile" "$@"
4460af45
IJ
43 '''
44 s.manifest_name='0000-MANIFEST.txt'
45 # private
46 s._destdir = destdir
b635cd93 47 s._outcounter = 0
4460af45 48 s._manifest = []
2371516a 49 s._dirmap = { }
e98894bc 50 s._package_files = { } # map filename => infol
4460af45 51
e98894bc
IJ
52 def thing_matches_globs(s, thing, globs):
53 for pat in globs:
4460af45
IJ
54 negate = pat.startswith('!')
55 if negate: pat = pat[1:]
e98894bc 56 if fnmatch.fnmatch(thing, pat):
4460af45
IJ
57 return not negate
58 return negate
59
e98894bc 60 def src_filter_glob(s, src): # default s.src_filter
e68fca7c 61 return s.thing_matches_globs(src, s.src_filter_globs)
e98894bc 62
28861486
IJ
63 def src_direxcludes_git(s, d):
64 try:
65 excl = open(os.path.join(d, '.gitignore'))
66 except FileNotFoundError:
67 return []
68 r = []
69 for l in excl:
70 l.strip
71 if l.startswith('#'): next
72 if not len(l): next
73 r += l
74 return r
75
4460af45
IJ
76 def src_likeparent_git(s, src):
77 try:
b635cd93 78 os.stat(os.path.join(src, '.git/.'))
4460af45
IJ
79 except FileNotFoundError:
80 return False
81 else:
82 return True
83
84 def src_parentfinder(s, src, infol): # callers may monkey-patch away
85 for deref in (False,True):
86 xinfo = []
87
88 search = src
89 if deref:
90 search = os.path.realpath(search)
91
92 def ascend():
b635cd93 93 nonlocal search
4460af45
IJ
94 xinfo.append(os.path.basename(search))
95 search = os.path.dirname(search)
96
97 try:
b635cd93 98 stab = os.lstat(search)
4460af45
IJ
99 except FileNotFoundError:
100 return
101 if stat.S_ISREG(stab.st_mode):
102 ascend()
103
104 while not os.path.ismount(search):
105 if s.src_likeparent(search):
106 xinfo.reverse()
b635cd93 107 if len(xinfo): infol.append('want=' + os.path.join(*xinfo))
4460af45
IJ
108 return search
109
110 ascend()
111
112 # no .git found anywhere
b635cd93 113 return src
4460af45 114
e68fca7c 115 def path_prenormaliser(s, d, infol): # callers may monkey-patch away
4460af45
IJ
116 return os.path.join(s.cwd, os.path.abspath(d))
117
e98894bc 118 def srcdir_find_rune(s, d):
b635cd93 119 script = s.find_rune_base
28861486
IJ
120 ignores = s.ignores + [s.output_name, s.manifest_name]
121 ignores += s.src_direxcludes(d)
122 for excl in ignores:
4460af45 123 assert("'" not in excl)
28861486
IJ
124 script += r" \! -name '%s'" % excl
125 script += r" \! -path '*/%s/*'" % excl
4460af45 126 script += ' -print0'
b635cd93 127 return script
4460af45 128
2371516a 129 def manifest_append(s, name, infol):
a53aba4d 130 s._manifest.append({ 'file':name, 'info':' '.join(infol) })
2371516a 131
e68fca7c
IJ
132 def manifest_append_absentfile(s, name, infol):
133 s._manifest.append({ 'file_print':name, 'info':' '.join(infol) })
134
4460af45 135 def new_output_name(s, nametail, infol):
b635cd93
IJ
136 s._outcounter += 1
137 name = '%04d-%s' % (s._outcounter, nametail)
2371516a 138 s.manifest_append(name, infol)
4460af45
IJ
139 return name
140
141 def open_output_fh(s, name, mode):
142 return open(os.path.join(s._destdir, name), mode)
143
e98894bc 144 def src_dir(s, d, infol):
2371516a
IJ
145 try: name = s._dirmap[d]
146 except KeyError: pass
147 else:
148 s.manifest_append(name, infol)
149 return
150
b635cd93 151 if s.show_pathnames: infol.append(d)
e98894bc 152 find_rune = s.srcdir_find_rune(d)
4460af45 153 total_rune = s.rune_cpio % find_rune
2371516a
IJ
154
155 name = s.new_output_name('src.cpio', infol)
156 s._dirmap[d] = name
157 fh = s.open_output_fh(name, 'wb')
158
c2c0da38
IJ
159 s.logger('packing up into %s: %s (because %s)' %
160 (name, d, ' '.join(infol)))
161
4460af45 162 subprocess.run(s.rune_shell + [total_rune],
b635cd93 163 cwd=d,
4460af45
IJ
164 stdin=subprocess.DEVNULL,
165 stdout=fh,
b635cd93
IJ
166 restore_signals=True,
167 check=True)
4460af45
IJ
168 fh.close()
169
e98894bc 170 def src_indir(s, d, infol):
e68fca7c 171 d = s.path_prenormaliser(d, infol)
4460af45 172 if not s.src_filter(d): return
e98894bc 173
4460af45 174 d = s.src_parentfinder(d, infol)
e68fca7c
IJ
175 if d is None: return
176 s.src_dir(d, infol)
e98894bc
IJ
177
178 def report_from_packages_debian(s, files):
e68fca7c 179 dpkg_S_in = tempfile.TemporaryFile(mode='w+')
e98894bc
IJ
180 for (file, infols) in files.items():
181 assert('\n' not in file)
182 dpkg_S_in.write(file)
183 dpkg_S_in.write('\0')
184 dpkg_S_in.seek(0)
185 cmdl = ['xargs','-0r','dpkg','-S','--']
186 dpkg_S = subprocess.Popen(cmdl,
a53aba4d
IJ
187 cwd='/',
188 stdin=dpkg_S_in,
189 stdout=subprocess.PIPE,
190 stderr=sys.stderr,
191 close_fds=False)
e68fca7c 192 dpkg_show_in = tempfile.TemporaryFile(mode='w+')
e98894bc 193 pkginfos = { }
e68fca7c
IJ
194 for l in dpkg_S.stdout:
195 l = l.strip(b'\n').decode('utf-8')
e98894bc 196 (pkgs, fname) = l.split(': ',1)
e68fca7c
IJ
197 pks = pkgs.split(', ')
198 for pk in pks:
199 pkginfos.setdefault(pk,{'files':[]})['files'].append(fname)
200 print(pk, file=dpkg_show_in)
a53aba4d
IJ
201 assert(dpkg_S.wait() == 0)
202 dpkg_show_in.seek(0)
203 cmdl = ['xargs','-r','dpkg-query',
e68fca7c 204 r'-f${binary:Package}\t${Package}\t${Architecture}\t${Version}\t${source:Package}\t${source:Version}\n',
a53aba4d
IJ
205 '--show','--']
206 dpkg_show = subprocess.Popen(cmdl,
207 cwd='/',
208 stdin=dpkg_show_in,
209 stdout=subprocess.PIPE,
210 stderr=sys.stderr,
211 close_fds=False)
212 for l in dpkg_show.stdout:
e68fca7c
IJ
213 l = l.strip(b'\n').decode('utf-8')
214 (pk,p,a,v,sp,sv) = l.split('\t')
215 pkginfos[pk]['binary'] = p
216 pkginfos[pk]['arch'] = a
217 pkginfos[pk]['version'] = v
218 pkginfos[pk]['source'] = sp
219 pkginfos[pk]['sourceversion'] = sv
a53aba4d 220 assert(dpkg_show.wait() == 0)
e68fca7c
IJ
221 for pk in sorted(pkginfos.keys()):
222 pi = pkginfos[pk]
223 debfname = '%s_%s_%s.deb' % (pi['binary'], pi['version'], pi['arch'])
a53aba4d 224 dscfname = '%s_%s.dsc' % (pi['source'], pi['sourceversion'])
e68fca7c 225 s.manifest_append_absentfile(dscfname, [debfname])
c2c0da38
IJ
226 s.logger('mentioning %s and %s because %s' %
227 (dscfname, debfname, pi['files'][0]))
e68fca7c
IJ
228 for fname in pi['files']:
229 infol = files[fname]
56ffddf8 230 if s.show_pathnames: infol = infol + ['loaded='+fname]
e68fca7c 231 s.manifest_append_absentfile(' \t' + debfname, infol)
e98894bc
IJ
232
233 def thing_ought_packaged(s, fname):
234 return s.thing_matches_globs(fname, s.src_package_globs)
235
e68fca7c
IJ
236 def src_file_packaged(s, fname, infol):
237 s._package_files.setdefault(fname,[]).extend(infol)
e98894bc
IJ
238
239 def src_file(s, fname, infol):
240 def fngens():
e68fca7c
IJ
241 yield (infol, fname)
242 infol_copy = infol.copy()
243 yield (infol_copy, s.path_prenormaliser(fname, infol_copy))
244 yield (infol, os.path.realpath(fname))
245
246 for (tinfol, tfname) in fngens():
247 if s.thing_ought_packaged(tfname):
248 s.src_file_packaged(tfname, tinfol)
e98894bc 249 return
4460af45 250
e98894bc
IJ
251 s.src_indir(fname, infol)
252
253 def src_argv0(s, program, infol):
e68fca7c 254 s.src_file(program, infol)
e98894bc
IJ
255
256 def src_syspath(s, fname, infol):
56ffddf8 257 if s.thing_ought_packaged(fname): return
e98894bc
IJ
258 s.src_indir(fname, infol)
259
260 def src_module(s, m, infol):
261 try: fname = m.__file__
16374080 262 except AttributeError: return
56ffddf8 263 infol.append('module='+m.__name__)
16374080 264
e98894bc
IJ
265 if s.thing_ought_packaged(fname):
266 s.src_file_packaged(fname, infol)
267 else:
e68fca7c 268 s.src_indir(fname, infol)
e98894bc
IJ
269
270 def srcs_allitems(s, dirs=sys.path):
c2c0da38 271 s.logger('allitems')
e98894bc 272 s.src_argv0(sys.argv[0], ['argv[0]'])
4460af45 273 for d in sys.path:
e98894bc 274 s.src_syspath(d, ['sys.path'])
16374080 275 for m in sys.modules.values():
e98894bc 276 s.src_module(m, ['sys.modules'])
e68fca7c 277 s.report_from_packages(s._package_files)
c2c0da38 278 s.logger('allitems done')
4460af45 279
b635cd93 280 def mk_portmanteau(s):
c2c0da38 281 s.logger('making portmanteau')
4460af45
IJ
282 cmdl = s.rune_shell + [ s.rune_portmanteau, 'x',
283 s.output_name, s.manifest_name ]
b635cd93 284 mfh = s.open_output_fh(s.manifest_name,'w')
a53aba4d
IJ
285 for me in s._manifest:
286 try: fname = me['file']
287 except KeyError: fname = me.get('file_print','')
e68fca7c 288 else: cmdl.append(fname)
a53aba4d 289 print('%s\t%s' % (fname, me['info']), file=mfh)
4460af45 290 mfh.close()
2371516a 291 subprocess.run(cmdl,
b635cd93 292 cwd=s._destdir,
4460af45
IJ
293 stdin=subprocess.DEVNULL,
294 stdout=sys.stderr,
b635cd93
IJ
295 restore_signals=True,
296 check=True)
c2c0da38
IJ
297 s.output_path = os.path.join(s._destdir, s.output_name)
298 s.logger('portmanteau ready in %s' % s.output_path)
4460af45
IJ
299
300 def generate(s):
e98894bc 301 s.srcs_allitems()
4460af45 302 s.mk_portmanteau()