2f710ca995b4a430594f50128b6d3f4e274d2cad
[hippotat] / hippotatlib / ownsource.py
1 # Automatic source code provision (AGPL compliance)
2
3 import os
4 import sys
5 import fnmatch
6 import stat
7 import subprocess
8
9 class SourceShipmentPreparer():
10 def __init__(s, destdir):
11 # caller may modify, and should read after calling generate()
12 s.output_name = 'srcbomb.tar.gz'
13 # defaults, caller can modify after creation
14 s.src_filter = s.src_filter_glob
15 s.src_package_globs = [!'/usr/local/*', '/usr*']
16 s.src_filter_globs = ['!/etc/*']
17 s.src_likeparent = s.src_likeparent_git
18 s.report_from_packages = s.report_from_packages_debian
19 s.cwd = os.getcwd()
20 s.find_rune_base = "find -type f -perm -004 \! -path '*/tmp/*'"
21 s.excludes = ['*~', '*.bak', '*.tmp', '#*#',
22 '[0-9][0-9][0-9][0-9]-src.cpio']
23 s.rune_shell = ['/bin/bash', '-ec']
24 s.show_pathnames = True
25 s.rune_cpio = r'''
26 set -o pipefail
27 (
28 %s
29 # ^ by default, is find ... -print0
30 ) | (
31 cpio -Hustar -o --quiet -0 -R 1000:1000 || \
32 cpio -Hustar -o --quiet -0
33 )
34 '''
35 s.rune_portmanteau = r'''
36 outfile=$1; shift
37 rm -f "$outfile"
38 GZIP=-1 tar zcf "$outfile" "$@"
39 '''
40 s.manifest_name='0000-MANIFEST.txt'
41 # private
42 s._destdir = destdir
43 s._outcounter = 0
44 s._manifest = []
45 s._dirmap = { }
46 s._package_files = { } # map filename => infol
47
48 def thing_matches_globs(s, thing, globs):
49 for pat in globs:
50 negate = pat.startswith('!')
51 if negate: pat = pat[1:]
52 if fnmatch.fnmatch(thing, pat):
53 return not negate
54 return negate
55
56 def src_filter_glob(s, src): # default s.src_filter
57 return s.thing_matches_globs(s, src, s.src_filter_globs)
58
59 def src_likeparent_git(s, src):
60 try:
61 os.stat(os.path.join(src, '.git/.'))
62 except FileNotFoundError:
63 return False
64 else:
65 return True
66
67 def src_parentfinder(s, src, infol): # callers may monkey-patch away
68 for deref in (False,True):
69 xinfo = []
70
71 search = src
72 if deref:
73 search = os.path.realpath(search)
74
75 def ascend():
76 nonlocal search
77 xinfo.append(os.path.basename(search))
78 search = os.path.dirname(search)
79
80 try:
81 stab = os.lstat(search)
82 except FileNotFoundError:
83 return
84 if stat.S_ISREG(stab.st_mode):
85 ascend()
86
87 while not os.path.ismount(search):
88 if s.src_likeparent(search):
89 xinfo.reverse()
90 if len(xinfo): infol.append('want=' + os.path.join(*xinfo))
91 return search
92
93 ascend()
94
95 # no .git found anywhere
96 return src
97
98 def src_prenormaliser(s, d, infol): # callers may monkey-patch away
99 return os.path.join(s.cwd, os.path.abspath(d))
100
101 def srcdir_find_rune(s, d):
102 script = s.find_rune_base
103 for excl in s.excludes + [s.output_name, s.manifest_name]:
104 assert("'" not in excl)
105 script += r" \! -name '%s'" % excl
106 script += ' -print0'
107 return script
108
109 def manifest_append(s, name, infol):
110 s._manifest.append((name, ' '.join(infol)))
111
112 def new_output_name(s, nametail, infol):
113 s._outcounter += 1
114 name = '%04d-%s' % (s._outcounter, nametail)
115 s.manifest_append(name, infol)
116 return name
117
118 def open_output_fh(s, name, mode):
119 return open(os.path.join(s._destdir, name), mode)
120
121 def src_dir(s, d, infol):
122 try: name = s._dirmap[d]
123 except KeyError: pass
124 else:
125 s.manifest_append(name, infol)
126 return
127
128 if s.show_pathnames: infol.append(d)
129 find_rune = s.srcdir_find_rune(d)
130 total_rune = s.rune_cpio % find_rune
131
132 name = s.new_output_name('src.cpio', infol)
133 s._dirmap[d] = name
134 fh = s.open_output_fh(name, 'wb')
135
136 subprocess.run(s.rune_shell + [total_rune],
137 cwd=d,
138 stdin=subprocess.DEVNULL,
139 stdout=fh,
140 restore_signals=True,
141 check=True)
142 fh.close()
143
144 def src_indir(s, d, infol):
145 d = s.src_prenormaliser(d, infol)
146 if not s.src_filter(d): return
147
148 d = s.src_parentfinder(d, infol)
149 s.dir(d, infol)
150
151 def report_from_packages_debian(s, files):
152 dpkg_S_in = tempfile.TemporaryFile()
153 for (file, infols) in files.items():
154 assert('\n' not in file)
155 dpkg_S_in.write(file)
156 dpkg_S_in.write('\0')
157 dpkg_S_in.seek(0)
158 cmdl = ['xargs','-0r','dpkg','-S','--']
159 dpkg_S = subprocess.Popen(cmdl,
160 cwd='/',
161 stdin=dpkg_S_in,
162 stdout=subprocess.PIPE,
163 close_fds=False)
164 dpkg_show_in = tempfile.TemporaryFile()
165 pkginfos = { }
166 for l in dpkgs.stdout:
167 (pkgs, fname) = l.split(': ',1)
168 pkgs = pkgs.split(', ')
169 for p in pkgs:
170 pkginfos[
171 print(p, file=dpkg_show_in)
172
173 dpkg-query --show PACKAGE
174
175 def thing_ought_packaged(s, fname):
176 return s.thing_matches_globs(fname, s.src_package_globs)
177
178 def src_file_packaged(s, fname);
179 try: s._package_files[fname].append(infol)
180 except KeyError: s._package_files[fname] = [infol]
181
182 def src_file(s, fname, infol):
183 def fngens():
184 yield fname
185 yield s.path_prenormaliser(fname)
186 yield os.path.realpath(fname)
187
188 for fn in fngens():
189 if s.thing_ought_packaged(fngen):
190 s.src_file_packaged(fname, infol)
191 return
192
193 s.src_indir(fname, infol)
194
195 def src_argv0(s, program, infol):
196 s.src_file(s, program, infol)
197
198 def src_syspath(s, fname, infol):
199 s.src_indir(fname, infol)
200
201 def src_module(s, m, infol):
202 try: fname = m.__file__
203 except AttributeError: return
204 infol.append(m.__name__)
205
206 if s.thing_ought_packaged(fname):
207 s.src_file_packaged(fname, infol)
208 else:
209 s.src_indir(s, fname)
210
211 def srcs_allitems(s, dirs=sys.path):
212 s.src_argv0(sys.argv[0], ['argv[0]'])
213 for d in sys.path:
214 s.src_syspath(d, ['sys.path'])
215 for m in sys.modules.values():
216 s.src_module(m, ['sys.modules'])
217 s.report_from_packages(s, s._package_files)
218
219 def mk_portmanteau(s):
220 cmdl = s.rune_shell + [ s.rune_portmanteau, 'x',
221 s.output_name, s.manifest_name ]
222 mfh = s.open_output_fh(s.manifest_name,'w')
223 for (name, info) in s._manifest:
224 if name is not None: cmdl.append(name)
225 print('%s\t%s' % (name,info), file=mfh)
226 mfh.close()
227 subprocess.run(cmdl,
228 cwd=s._destdir,
229 stdin=subprocess.DEVNULL,
230 stdout=sys.stderr,
231 restore_signals=True,
232 check=True)
233
234 def generate(s):
235 s.srcs_allitems()
236 s.mk_portmanteau()