mdw@git.distorted.org.uk Git - hippotat/blob - hippotatlib/ownsource.py

   1 # -*- python -*-
   2 #
   3 # Hippotat - Asinine IP Over HTTP program
   4 # hippotatlib/ownsource.py - Automatic source code provision (AGPL compliance)
   5 #
   6 # Copyright 2017 Ian Jackson
   7 #
   8 # This program is free software: you can redistribute it and/or modify
   9 # it under the terms of the GNU Affero General Public License as
  10 # published by the Free Software Foundation, either version 3 of the
  11 # License, or (at your option) any later version, with the "CAF Login
  12 # Exception" as published by Ian Jackson (version 2, or at your option
  13 # any later version) as an Additional Permission.
  14 #
  15 # This program is distributed in the hope that it will be useful,
  16 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 # GNU Affero General Public License for more details.
  19 #
  20 # You should have received a copy of the GNU Affero General Public
  21 # License and the CAF Login Exception along with this program, in the
  22 # file AGPLv3+CAFv2.  If not, email Ian Jackson
  23 # <ijackson@chiark.greenend.org.uk>.
  24
  25
  26 import os
  27 import sys
  28 import fnmatch
  29 import stat
  30 import subprocess
  31 import tempfile
  32
  33 class SourceShipmentPreparer():
  34   def __init__(s, destdir):
  35     # caller may modify, and should read after calling generate()
  36     s.output_name = 'srcbomb.tar.gz'
  37     # s.output_path   alternatively caller may read this
  38     # defaults, caller can modify after creation
  39     s.logger = lambda m: print('SourceShipmentPreparer',m)
  40     s.src_filter = s.src_filter_glob
  41     s.src_package_globs = ['!/usr/local/*', '/usr*']
  42     s.src_filter_globs = ['!/etc/*']
  43     s.src_likeparent = s.src_likeparent_git
  44     s.src_direxcludes = s.src_direxcludes_git
  45     s.report_from_packages = s.report_from_packages_debian
  46     s.cwd = os.getcwd()
  47     s.find_rune_base = "find -type f -perm -004 \! -path '*/tmp/*'"
  48     s.ignores = ['*~', '*.bak', '*.tmp', '#*#', '__pycache__',
  49                   '[0-9][0-9][0-9][0-9]-src.cpio']
  50     s.rune_shell = ['/bin/bash', '-ec']
  51     s.show_pathnames = True
  52     s.rune_cpio = r'''
  53             set -o pipefail
  54             (
  55              %s
  56              # ^ by default, is find ... -print0
  57             ) | (
  58              cpio -Hustar -o --quiet -0 -R 1000:1000 || \
  59              cpio -Hustar -o --quiet -0
  60             )
  61     '''
  62     s.rune_portmanteau = r'''
  63             outfile=$1; shift
  64             rm -f "$outfile"
  65             GZIP=-1 tar zcf "$outfile" "$@"
  66     '''
  67     s.manifest_name='0000-MANIFEST.txt'
  68     # private
  69     s._destdir = destdir
  70     s._outcounter = 0
  71     s._manifest = []
  72     s._dirmap = { }
  73     s._package_files = { } # map filename => infol
  74
  75   def thing_matches_globs(s, thing, globs):
  76     for pat in globs:
  77       negate = pat.startswith('!')
  78       if negate: pat = pat[1:]
  79       if fnmatch.fnmatch(thing, pat):
  80         return not negate
  81     return negate
  82
  83   def src_filter_glob(s, src): # default s.src_filter
  84     return s.thing_matches_globs(src, s.src_filter_globs)
  85
  86   def src_direxcludes_git(s, d):
  87     try:
  88       excl = open(os.path.join(d, '.gitignore'))
  89     except FileNotFoundError:
  90       return []
  91     r = []
  92     for l in excl:
  93       l.strip
  94       if l.startswith('#'): next
  95       if not len(l): next
  96       r += l
  97     return r
  98
  99   def src_likeparent_git(s, src):
 100     try:
 101       os.stat(os.path.join(src, '.git/.'))
 102     except FileNotFoundError:
 103       return False
 104     else:
 105       return True
 106
 107   def src_parentfinder(s, src, infol): # callers may monkey-patch away
 108     for deref in (False,True):
 109       xinfo = []
 110
 111       search = src
 112       if deref:
 113         search = os.path.realpath(search)
 114
 115       def ascend():
 116         nonlocal search
 117         xinfo.append(os.path.basename(search))
 118         search = os.path.dirname(search)
 119
 120       try:
 121         stab = os.lstat(search)
 122       except FileNotFoundError:
 123         return
 124       if stat.S_ISREG(stab.st_mode):
 125         ascend()
 126
 127       while not os.path.ismount(search):
 128         if s.src_likeparent(search):
 129           xinfo.reverse()
 130           if len(xinfo): infol.append('want=' + os.path.join(*xinfo))
 131           return search
 132
 133         ascend()
 134
 135     # no .git found anywhere
 136     return src
 137
 138   def path_prenormaliser(s, d, infol): # callers may monkey-patch away
 139     return os.path.join(s.cwd, os.path.abspath(d))
 140
 141   def srcdir_find_rune(s, d):
 142     script = s.find_rune_base
 143     ignores = s.ignores + [s.output_name, s.manifest_name]
 144     ignores += s.src_direxcludes(d)
 145     for excl in ignores:
 146       assert("'" not in excl)
 147       script += r" \! -name '%s'"     % excl
 148       script += r" \! -path '*/%s/*'" % excl
 149     script += ' -print0'
 150     return script
 151
 152   def manifest_append(s, name, infol):
 153     s._manifest.append({ 'file':name, 'info':' '.join(infol) })
 154
 155   def manifest_append_absentfile(s, name, infol):
 156     s._manifest.append({ 'file_print':name, 'info':' '.join(infol) })
 157
 158   def new_output_name(s, nametail, infol):
 159     s._outcounter += 1
 160     name = '%04d-%s' % (s._outcounter, nametail)
 161     s.manifest_append(name, infol)
 162     return name
 163
 164   def open_output_fh(s, name, mode):
 165     return open(os.path.join(s._destdir, name), mode)
 166
 167   def src_dir(s, d, infol):
 168     try: name = s._dirmap[d]
 169     except KeyError: pass
 170     else:
 171       s.manifest_append(name, infol)
 172       return
 173
 174     if s.show_pathnames: infol.append(d)
 175     find_rune = s.srcdir_find_rune(d)
 176     total_rune = s.rune_cpio % find_rune
 177
 178     name = s.new_output_name('src.cpio', infol)
 179     s._dirmap[d] = name
 180     fh = s.open_output_fh(name, 'wb')
 181
 182     s.logger('packing up into %s: %s (because %s)' %
 183              (name, d, ' '.join(infol)))
 184
 185     subprocess.run(s.rune_shell + [total_rune],
 186                    cwd=d,
 187                    stdin=subprocess.DEVNULL,
 188                    stdout=fh,
 189                    restore_signals=True,
 190                    check=True)
 191     fh.close()
 192
 193   def src_indir(s, d, infol):
 194     d = s.path_prenormaliser(d, infol)
 195     if not s.src_filter(d): return
 196
 197     d = s.src_parentfinder(d, infol)
 198     if d is None: return
 199     s.src_dir(d, infol)
 200
 201   def report_from_packages_debian(s, files):
 202     dpkg_S_in = tempfile.TemporaryFile(mode='w+')
 203     for (file, infols) in files.items():
 204       assert('\n' not in file)
 205       dpkg_S_in.write(file)
 206       dpkg_S_in.write('\0')
 207     dpkg_S_in.seek(0)
 208     cmdl = ['xargs','-0r','dpkg','-S','--']
 209     dpkg_S = subprocess.Popen(cmdl,
 210                               cwd='/',
 211                               stdin=dpkg_S_in,
 212                               stdout=subprocess.PIPE,
 213                               stderr=sys.stderr,
 214                               close_fds=False)
 215     dpkg_show_in = tempfile.TemporaryFile(mode='w+')
 216     pkginfos = { }
 217     for l in dpkg_S.stdout:
 218       l = l.strip(b'\n').decode('utf-8')
 219       (pkgs, fname) = l.split(': ',1)
 220       pks = pkgs.split(', ')
 221       for pk in pks:
 222         pkginfos.setdefault(pk,{'files':[]})['files'].append(fname)
 223         print(pk, file=dpkg_show_in)
 224     assert(dpkg_S.wait() == 0)
 225     dpkg_show_in.seek(0)
 226     cmdl = ['xargs','-r','dpkg-query',
 227             r'-f${binary:Package}\t${Package}\t${Architecture}\t${Version}\t${source:Package}\t${source:Version}\n',
 228             '--show','--']
 229     dpkg_show = subprocess.Popen(cmdl,
 230                                  cwd='/',
 231                                  stdin=dpkg_show_in,
 232                                  stdout=subprocess.PIPE,
 233                                  stderr=sys.stderr,
 234                                  close_fds=False)
 235     for l in dpkg_show.stdout:
 236       l = l.strip(b'\n').decode('utf-8')
 237       (pk,p,a,v,sp,sv) = l.split('\t')
 238       pkginfos[pk]['binary'] = p
 239       pkginfos[pk]['arch'] = a
 240       pkginfos[pk]['version'] = v
 241       pkginfos[pk]['source'] = sp
 242       pkginfos[pk]['sourceversion'] = sv
 243     assert(dpkg_show.wait() == 0)
 244     for pk in sorted(pkginfos.keys()):
 245       pi = pkginfos[pk]
 246       debfname = '%s_%s_%s.deb' % (pi['binary'], pi['version'], pi['arch'])
 247       dscfname = '%s_%s.dsc' % (pi['source'], pi['sourceversion'])
 248       s.manifest_append_absentfile(dscfname, [debfname])
 249       s.logger('mentioning %s and %s because %s' %
 250                (dscfname, debfname, pi['files'][0]))
 251       for fname in pi['files']:
 252         infol = files[fname]
 253         if s.show_pathnames: infol = infol + ['loaded='+fname]
 254         s.manifest_append_absentfile(' \t' + debfname, infol)
 255
 256   def thing_ought_packaged(s, fname):
 257     return s.thing_matches_globs(fname, s.src_package_globs)
 258
 259   def src_file_packaged(s, fname, infol):
 260     s._package_files.setdefault(fname,[]).extend(infol)
 261
 262   def src_file(s, fname, infol):
 263     def fngens():
 264       yield (infol, fname)
 265       infol_copy = infol.copy()
 266       yield (infol_copy, s.path_prenormaliser(fname, infol_copy))
 267       yield (infol, os.path.realpath(fname))
 268
 269     for (tinfol, tfname) in fngens():
 270       if s.thing_ought_packaged(tfname):
 271         s.src_file_packaged(tfname, tinfol)
 272         return
 273
 274     s.src_indir(fname, infol)
 275
 276   def src_argv0(s, program, infol):
 277     s.src_file(program, infol)
 278
 279   def src_syspath(s, fname, infol):
 280     if s.thing_ought_packaged(fname): return
 281     s.src_indir(fname, infol)
 282
 283   def src_module(s, m, infol):
 284     try: fname = m.__file__
 285     except AttributeError: return
 286     infol.append('module='+m.__name__)
 287
 288     if s.thing_ought_packaged(fname):
 289       s.src_file_packaged(fname, infol)
 290     else:
 291       s.src_indir(fname, infol)
 292
 293   def srcs_allitems(s, dirs=sys.path):
 294     s.logger('allitems')
 295     s.src_argv0(sys.argv[0], ['argv[0]'])
 296     for d in sys.path:
 297       s.src_syspath(d, ['sys.path'])
 298     for m in sys.modules.values():
 299       s.src_module(m, ['sys.modules'])
 300     s.report_from_packages(s._package_files)
 301     s.logger('allitems done')
 302
 303   def mk_portmanteau(s):
 304     s.logger('making portmanteau')
 305     cmdl = s.rune_shell + [ s.rune_portmanteau, 'x',
 306                             s.output_name, s.manifest_name ]
 307     mfh = s.open_output_fh(s.manifest_name,'w')
 308     for me in s._manifest:
 309       try: fname = me['file']
 310       except KeyError: fname = me.get('file_print','')
 311       else: cmdl.append(fname)
 312       print('%s\t%s' % (fname, me['info']), file=mfh)
 313     mfh.close()
 314     subprocess.run(cmdl,
 315                    cwd=s._destdir,
 316                    stdin=subprocess.DEVNULL,
 317                    stdout=sys.stderr,
 318                    restore_signals=True,
 319                    check=True)
 320     s.output_path = os.path.join(s._destdir, s.output_name)
 321     s.logger('portmanteau ready in %s' % s.output_path)
 322
 323   def generate(s):
 324     s.srcs_allitems()
 325     s.mk_portmanteau()