dak/examine_package.py

   1 #!/usr/bin/env python
   2
   3 # Script to automate some parts of checking NEW packages
   4 # Copyright (C) 2000, 2001, 2002, 2003, 2006  James Troup <james@nocrew.org>
   5
   6 # This program is free software; you can redistribute it and/or modify
   7 # it under the terms of the GNU General Public License as published by
   8 # the Free Software Foundation; either version 2 of the License, or
   9 # (at your option) any later version.
  10
  11 # This program is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 # GNU General Public License for more details.
  15
  16 # You should have received a copy of the GNU General Public License
  17 # along with this program; if not, write to the Free Software
  18 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19
  20 ################################################################################
  21
  22 # <Omnic> elmo wrote docs?!!?!?!?!?!?!
  23 # <aj> as if he wasn't scary enough before!!
  24 # * aj imagines a little red furry toy sitting hunched over a computer
  25 #   tapping furiously and giggling to himself
  26 # <aj> eventually he stops, and his heads slowly spins around and you
  27 #      see this really evil grin and then he sees you, and picks up a
  28 #      knife from beside the keyboard and throws it at you, and as you
  29 #      breathe your last breath, he starts giggling again
  30 # <aj> but i should be telling this to my psychiatrist, not you guys,
  31 #      right? :)
  32
  33 ################################################################################
  34
  35 import errno, os, pg, re, sys, md5
  36 import apt_pkg, apt_inst
  37 import daklib.database, daklib.utils, daklib.queue
  38
  39 ################################################################################
  40
  41 re_package = re.compile(r"^(.+?)_.*")
  42 re_doc_directory = re.compile(r".*/doc/([^/]*).*")
  43
  44 re_contrib = re.compile('^contrib/')
  45 re_nonfree = re.compile('^non\-free/')
  46
  47 re_arch = re.compile("Architecture: .*")
  48 re_builddep = re.compile("Build-Depends: .*")
  49 re_builddepind = re.compile("Build-Depends-Indep: .*")
  50
  51 re_localhost = re.compile("localhost\.localdomain")
  52 re_version = re.compile('^(.*)\((.*)\)')
  53
  54 re_newlinespace = re.compile('\n')
  55 re_spacestrip = re.compile('(\s)')
  56
  57 html_escaping = {'"':'&quot;', '&':'&amp;', '<':'&lt;', '>':'&gt;'}
  58 re_html_escaping = re.compile('|'.join(map(re.escape, html_escaping.keys())))
  59
  60 ################################################################################
  61
  62 Cnf = None
  63 projectB = None
  64
  65 Cnf = daklib.utils.get_conf()
  66 projectB = pg.connect(Cnf["DB::Name"], Cnf["DB::Host"], int(Cnf["DB::Port"]))
  67 daklib.database.init(Cnf, projectB)
  68
  69 printed_copyrights = {}
  70
  71 # default is to not output html.
  72 use_html = 0
  73
  74 ################################################################################
  75
  76 def usage (exit_code=0):
  77     print """Usage: dak examine-package [PACKAGE]...
  78 Check NEW package(s).
  79
  80   -h, --help                 show this help and exit
  81   -H, --html-output          output html page with inspection result
  82   -f, --file-name            filename for the html page
  83
  84 PACKAGE can be a .changes, .dsc, .deb or .udeb filename."""
  85
  86     sys.exit(exit_code)
  87
  88 ################################################################################
  89 # probably xml.sax.saxutils would work as well
  90
  91 def html_escape(s):
  92   return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s)
  93
  94 def escape_if_needed(s):
  95   if use_html:
  96     return re_html_escaping.sub(html_escaping.get, s)
  97   else:
  98     return s
  99
 100 def headline(s, level=2, bodyelement=None):
 101   if use_html:
 102     if bodyelement:
 103       print """<thead>
 104           <tr><th colspan="2" class="title" onclick="toggle('%(bodyelement)s', 'table-row-group', 'table-row-group')">%(title)s</th></tr>
 105         </thead>"""%{"bodyelement":bodyelement,"title":html_escape(s)}
 106     else:
 107       print "<h%d>%s</h%d>" % (level, html_escape(s), level)
 108   else:
 109     print "---- %s ----" % (s)
 110
 111 # Colour definitions, 'end' isn't really for use
 112
 113 ansi_colours = {
 114   'main': "\033[36m",
 115   'contrib': "\033[33m",
 116   'nonfree': "\033[31m",
 117   'arch': "\033[32m",
 118   'end': "\033[0m",
 119   'bold': "\033[1m",
 120   'maintainer': "\033[32m"}
 121
 122 html_colours = {
 123   'main': ('<span style="color: aqua">',"</span>"),
 124   'contrib': ('<span style="color: yellow">',"</span>"),
 125   'nonfree': ('<span style="color: red">',"</span>"),
 126   'arch': ('<span style="color: green">',"</span>"),
 127   'bold': ('<span style="font-weight: bold">',"</span>"),
 128   'maintainer': ('<span style="color: green">',"</span>")}
 129
 130 def colour_output(s, colour):
 131   if use_html:
 132     return ("%s%s%s" % (html_colours[colour][0], html_escape(s), html_colours[colour][1]))
 133   else:
 134     return ("%s%s%s" % (ansi_colours[colour], s, ansi_colours['end']))
 135
 136 def print_escaped_text(s):
 137   if use_html:
 138     print "<pre>%s</pre>" % (s)
 139   else:
 140     print s
 141
 142 def print_formatted_text(s):
 143   if use_html:
 144     print "<pre>%s</pre>" % (html_escape(s))
 145   else:
 146     print s
 147
 148 ################################################################################
 149
 150 def get_depends_parts(depend) :
 151     v_match = re_version.match(depend)
 152     if v_match:
 153         d_parts = { 'name' : v_match.group(1), 'version' : v_match.group(2) }
 154     else :
 155         d_parts = { 'name' : depend , 'version' : '' }
 156     return d_parts
 157
 158 def get_or_list(depend) :
 159     or_list = depend.split("|")
 160     return or_list
 161
 162 def get_comma_list(depend) :
 163     dep_list = depend.split(",")
 164     return dep_list
 165
 166 def split_depends (d_str) :
 167     # creates a list of lists of dictionaries of depends (package,version relation)
 168
 169     d_str = re_spacestrip.sub('',d_str)
 170     depends_tree = []
 171     # first split depends string up amongs comma delimiter
 172     dep_list = get_comma_list(d_str)
 173     d = 0
 174     while d < len(dep_list):
 175         # put depends into their own list
 176         depends_tree.append([dep_list[d]])
 177         d += 1
 178     d = 0
 179     while d < len(depends_tree):
 180         k = 0
 181         # split up Or'd depends into a multi-item list
 182         depends_tree[d] = get_or_list(depends_tree[d][0])
 183         while k < len(depends_tree[d]):
 184             # split depends into {package, version relation}
 185             depends_tree[d][k] = get_depends_parts(depends_tree[d][k])
 186             k += 1
 187         d += 1
 188     return depends_tree
 189
 190 def read_control (filename):
 191     recommends = []
 192     depends = []
 193     section = ''
 194     maintainer = ''
 195     arch = ''
 196
 197     deb_file = daklib.utils.open_file(filename)
 198     try:
 199         extracts = apt_inst.debExtractControl(deb_file)
 200         control = apt_pkg.ParseSection(extracts)
 201     except:
 202         print_formatted_text("can't parse control info")
 203         # TV-COMMENT: this will raise exceptions in two lines
 204         control = ''
 205
 206     deb_file.close()
 207
 208     control_keys = control.keys()
 209
 210     if control.has_key("Depends"):
 211         depends_str = control.Find("Depends")
 212         # create list of dependancy lists
 213         depends = split_depends(depends_str)
 214
 215     if control.has_key("Recommends"):
 216         recommends_str = control.Find("Recommends")
 217         recommends = split_depends(recommends_str)
 218
 219     if control.has_key("Section"):
 220         section_str = control.Find("Section")
 221
 222         c_match = re_contrib.search(section_str)
 223         nf_match = re_nonfree.search(section_str)
 224         if c_match :
 225             # contrib colour
 226             section = colour_output(section_str, 'contrib')
 227         elif nf_match :
 228             # non-free colour
 229             section = colour_output(section_str, 'nonfree')
 230         else :
 231             # main
 232             section = colour_output(section_str, 'main')
 233     if control.has_key("Architecture"):
 234         arch_str = control.Find("Architecture")
 235         arch = colour_output(arch_str, 'arch')
 236
 237     if control.has_key("Maintainer"):
 238         maintainer = control.Find("Maintainer")
 239         localhost = re_localhost.search(maintainer)
 240         if localhost:
 241             #highlight bad email
 242             maintainer = colour_output(maintainer, 'maintainer')
 243         else:
 244             maintainer = escape_if_needed(maintainer)
 245
 246     return (control, control_keys, section, depends, recommends, arch, maintainer)
 247
 248 def read_dsc (dsc_filename):
 249     dsc = {}
 250
 251     dsc_file = daklib.utils.open_file(dsc_filename)
 252     try:
 253         dsc = daklib.utils.parse_changes(dsc_filename)
 254     except:
 255         print_formatted_text("can't parse control info")
 256     dsc_file.close()
 257
 258     filecontents = escape_if_needed(strip_pgp_signature(dsc_filename))
 259
 260     if dsc.has_key("build-depends"):
 261         builddep = split_depends(dsc["build-depends"])
 262         builddepstr = create_depends_string(builddep)
 263         filecontents = re_builddep.sub("Build-Depends: "+builddepstr, filecontents)
 264
 265     if dsc.has_key("build-depends-indep"):
 266         builddepindstr = create_depends_string(split_depends(dsc["build-depends-indep"]))
 267         filecontents = re_builddepind.sub("Build-Depends-Indep: "+builddepindstr, filecontents)
 268
 269     if dsc.has_key("architecture") :
 270         if (dsc["architecture"] != "any"):
 271             newarch = colour_output(dsc["architecture"], 'arch')
 272             filecontents = re_arch.sub("Architecture: " + newarch, filecontents)
 273
 274     return filecontents
 275
 276 def create_depends_string (depends_tree):
 277     # just look up unstable for now. possibly pull from .changes later
 278     suite = "unstable"
 279     result = ""
 280     comma_count = 1
 281     for l in depends_tree:
 282         if (comma_count >= 2):
 283             result += ", "
 284         or_count = 1
 285         for d in l:
 286             if (or_count >= 2 ):
 287                 result += " | "
 288             # doesn't do version lookup yet.
 289
 290             q = projectB.query("SELECT DISTINCT(b.package), b.version, c.name, su.suite_name FROM  binaries b, files fi, location l, component c, bin_associations ba, suite su WHERE b.package='%s' AND b.file = fi.id AND fi.location = l.id AND l.component = c.id AND ba.bin=b.id AND ba.suite = su.id AND su.suite_name='%s' ORDER BY b.version desc" % (d['name'], suite))
 291             ql = q.getresult()
 292             if ql:
 293                 i = ql[0]
 294
 295                 adepends = d['name']
 296                 if d['version'] != '' :
 297                     adepends += " (%s)" % (d['version'])
 298
 299                 if i[2] == "contrib":
 300                     result += colour_output(adepends, "contrib")
 301                 elif i[2] == "non-free":
 302                     result += colour_output(adepends, "nonfree")
 303                 else :
 304                     result += colour_output(adepends, "main")
 305             else:
 306                 adepends = d['name']
 307                 if d['version'] != '' :
 308                     adepends += " (%s)" % (d['version'])
 309                 result += colour_output(adepends, "bold")
 310             or_count += 1
 311         comma_count += 1
 312     return result
 313
 314 def output_deb_info(filename):
 315     (control, control_keys, section, depends, recommends, arch, maintainer) = read_control(filename)
 316
 317     to_print = ""
 318     if control == '':
 319         print_formatted_text("no control info")
 320     else:
 321         for key in control_keys :
 322             output = " " + key + ": "
 323             if key == 'Depends':
 324                 output += create_depends_string(depends)
 325             elif key == 'Recommends':
 326                 output += create_depends_string(recommends)
 327             elif key == 'Section':
 328                 output += section
 329             elif key == 'Architecture':
 330                 output += arch
 331             elif key == 'Maintainer':
 332                 output += maintainer
 333             elif key == 'Description':
 334                 desc = control.Find(key)
 335                 desc = re_newlinespace.sub('\n ', desc)
 336                 output += escape_if_needed(desc)
 337             else:
 338                 output += escape_if_needed(control.Find(key))
 339             to_print += output + '\n'
 340         print_escaped_text(to_print)
 341
 342 def do_command (command, filename, escaped=0):
 343     o = os.popen("%s %s" % (command, filename))
 344     if escaped:
 345         print_escaped_text(o.read())
 346     else:
 347         print_formatted_text(o.read())
 348
 349 def do_lintian (filename):
 350     if use_html:
 351         do_command("lintian --show-overrides --color html", filename, 1)
 352     else:
 353         do_command("lintian --show-overrides --color always", filename, 1)
 354
 355 def print_copyright (deb_filename):
 356     package = re_package.sub(r'\1', deb_filename)
 357     o = os.popen("dpkg-deb -c %s | egrep 'usr(/share)?/doc/[^/]*/copyright' | awk '{print $6}' | head -n 1" % (deb_filename))
 358     copyright = o.read()[:-1]
 359
 360     if copyright == "":
 361         print_formatted_text("WARNING: No copyright found, please check package manually.")
 362         return
 363
 364     doc_directory = re_doc_directory.sub(r'\1', copyright)
 365     if package != doc_directory:
 366         print_formatted_text("WARNING: wrong doc directory (expected %s, got %s)." % (package, doc_directory))
 367         return
 368
 369     o = os.popen("dpkg-deb --fsys-tarfile %s | tar xvOf - %s 2>/dev/null" % (deb_filename, copyright))
 370     copyright = o.read()
 371     copyrightmd5 = md5.md5(copyright).hexdigest()
 372
 373     if printed_copyrights.has_key(copyrightmd5) and printed_copyrights[copyrightmd5] != "%s (%s)" % (package, deb_filename):
 374         print_formatted_text( "NOTE: Copyright is the same as %s.\n" % \
 375                 (printed_copyrights[copyrightmd5]))
 376     else:
 377         printed_copyrights[copyrightmd5] = "%s (%s)" % (package, deb_filename)
 378
 379     print_formatted_text(copyright)
 380
 381 def check_dsc (dsc_filename):
 382     headline(".dsc file for %s" % (dsc_filename))
 383     (dsc) = read_dsc(dsc_filename)
 384     print_escaped_text(dsc)
 385     headline("lintian check for %s" % (dsc_filename))
 386     do_lintian(dsc_filename)
 387
 388 def check_deb (deb_filename):
 389     filename = os.path.basename(deb_filename)
 390
 391     if filename.endswith(".udeb"):
 392         is_a_udeb = 1
 393     else:
 394         is_a_udeb = 0
 395
 396     headline("control file for %s" % (filename))
 397     #do_command ("dpkg -I", deb_filename)
 398     output_deb_info(deb_filename)
 399
 400     if is_a_udeb:
 401         headline("skipping lintian check for udeb")
 402         print
 403     else:
 404         headline("lintian check for %s" % (filename))
 405         do_lintian(deb_filename)
 406
 407     headline("contents of %s" % (filename))
 408     do_command ("dpkg -c", deb_filename)
 409
 410     if is_a_udeb:
 411         headline("skipping copyright for udeb")
 412     else:
 413         headline("copyright of %s" % (filename))
 414         print_copyright(deb_filename)
 415
 416     headline("file listing of %s" % (filename))
 417     do_command ("ls -l", deb_filename)
 418
 419 # Read a file, strip the signature and return the modified contents as
 420 # a string.
 421 def strip_pgp_signature (filename):
 422     file = daklib.utils.open_file (filename)
 423     contents = ""
 424     inside_signature = 0
 425     skip_next = 0
 426     for line in file.readlines():
 427         if line[:-1] == "":
 428             continue
 429         if inside_signature:
 430             continue
 431         if skip_next:
 432             skip_next = 0
 433             continue
 434         if line.startswith("-----BEGIN PGP SIGNED MESSAGE"):
 435             skip_next = 1
 436             continue
 437         if line.startswith("-----BEGIN PGP SIGNATURE"):
 438             inside_signature = 1
 439             continue
 440         if line.startswith("-----END PGP SIGNATURE"):
 441             inside_signature = 0
 442             continue
 443         contents += line
 444     file.close()
 445     return contents
 446
 447 # Display the .changes [without the signature]
 448 def display_changes(changes_filename):
 449     if use_html:
 450         print """<div id="changes-wrap"><a name="changes" />
 451                    <table class="infobox rfc822">"""
 452     headline(changes_filename, bodyelement="changes-body")
 453     if use_html:
 454         print """    <tbody id="changes-body" class="infobody">"""
 455         print """<tr><td>"""
 456     print_formatted_text(strip_pgp_signature(changes_filename))
 457     if use_html:
 458         print """</td></tr>"""
 459         """ <!--
 460         1. Join multiline fields
 461         2. s/^\s*\.?//gm
 462         3. s#\n#<br/>#g
 463         4. s#^\([^ :]*:\)\s*\(.*\)$#<tr><td class="key">\1</td><td class="val">\2</td></tr>#
 464         5. Special handling for homepage, distribution, bugs
 465         -->
 466           <tr><td class="key">Format:</td><td class="val">1.7</td></tr>
 467           <tr><td class="key">Date:</td><td class="val">Mon, 24 Dec 2007 15:32:08 -0200</td></tr>
 468           <tr><td class="key">Source:</td><td class="val">php-xdebug</td></tr>
 469           <tr><td class="key">Binary:</td><td class="val">php5-xdebug</td></tr>
 470           <tr><td class="key">Architecture:</td><td class="val">source amd64</td></tr>
 471           <tr><td class="key">Version:</td><td class="val">2.0.2-1</td></tr>
 472           <tr><td class="key">Distribution:</td><td class="val"><span class="dist-unstable">unstable</span></td></tr>
 473           <tr><td class="key">Urgency:</td><td class="val">low</td></tr>
 474           <tr><td class="key">Maintainer:</td><td class="val">Marcelo Jorge Vieira (metal) &lt;metal@alucinados.com&gt;</td></tr>
 475           <tr><td class="key">Changed-By:</td><td class="val">Marcelo Jorge Vieira (metal) &lt;metal@alucinados.com&gt;</td></tr>
 476           <tr><td class="key">Description:</td><td class="val">
 477               php5-xdebug - xdebug extension module for PHP5</td></tr>
 478           <tr><td class="key">Closes:</td><td class="val"><a href="http://bugs.debian.org/377348">377348</a></td></tr>
 479           <tr><td class="key">Changes:</td><td class="val">
 480               php-xdebug (2.0.2-1) unstable; urgency=low<br/>
 481               <br/>
 482               * Initial release (Closes: <a href="http://bugs.debian.org/377348">#377348</a>)</td></tr>
 483           <tr><td class="key">Files:</td><td class="val">
 484               c6ee78b58a4d70d66f8a70436b2a943c 632 web optional php-xdebug_2.0.2-1.dsc<br/>
 485               d3547f74353174884452a51ee9ca687f 279891 web optional php-xdebug_2.0.2.orig.tar.gz<br/>
 486               8e7c262113c8ac13f47781e0ac0eb4c3 4107 web optional php-xdebug_2.0.2-1.diff.gz<br/>
 487               3c6be09f23931fabf0b3048575390ed3 137930 web optional php5-xdebug_2.0.2-1_amd64.deb</td></tr>
 488               """
 489         print """</tbody></table></div>"""
 490
 491 def check_changes (changes_filename):
 492     display_changes(changes_filename)
 493
 494     changes = daklib.utils.parse_changes (changes_filename)
 495     files = daklib.utils.build_file_list(changes)
 496     for file in files.keys():
 497         if file.endswith(".deb") or file.endswith(".udeb"):
 498             check_deb(file)
 499         if file.endswith(".dsc"):
 500             check_dsc(file)
 501         # else: => byhand
 502
 503 def main ():
 504     global Cnf, projectB, db_files, waste, excluded
 505
 506 #    Cnf = daklib.utils.get_conf()
 507
 508     Arguments = [('h',"help","Examine-Package::Options::Help"),
 509                  ('H',"html-output","Examine-Package::Options::Html-Output"),
 510                 ]
 511     for i in [ "Help", "Html-Output", "partial-html" ]:
 512         if not Cnf.has_key("Examine-Package::Options::%s" % (i)):
 513             Cnf["Examine-Package::Options::%s" % (i)] = ""
 514
 515     args = apt_pkg.ParseCommandLine(Cnf,Arguments,sys.argv)
 516     Options = Cnf.SubTree("Examine-Package::Options")
 517
 518     if Options["Help"]:
 519         usage()
 520
 521     stdout_fd = sys.stdout
 522
 523     for file in args:
 524         try:
 525             if not Options["Html-Output"]:
 526                 # Pipe output for each argument through less
 527                 less_fd = os.popen("less -R -", 'w', 0)
 528                 # -R added to display raw control chars for colour
 529                 sys.stdout = less_fd
 530             try:
 531                 if file.endswith(".changes"):
 532                     check_changes(file)
 533                 elif file.endswith(".deb") or file.endswith(".udeb"):
 534                     check_deb(file)
 535                 elif file.endswith(".dsc"):
 536                     check_dsc(file)
 537                 else:
 538                     daklib.utils.fubar("Unrecognised file type: '%s'." % (file))
 539             finally:
 540                 if not Options["Html-Output"]:
 541                     # Reset stdout here so future less invocations aren't FUBAR
 542                     less_fd.close()
 543                     sys.stdout = stdout_fd
 544         except IOError, e:
 545             if errno.errorcode[e.errno] == 'EPIPE':
 546                 daklib.utils.warn("[examine-package] Caught EPIPE; skipping.")
 547                 pass
 548             else:
 549                 raise
 550         except KeyboardInterrupt:
 551             daklib.utils.warn("[examine-package] Caught C-c; skipping.")
 552             pass
 553
 554 #######################################################################################
 555
 556 if __name__ == '__main__':
 557     main()
 558