implement obsoleting languages

apply-languages.py

@@ -3,7 +3,7 @@
 #
 # apply-languages.py -- generate language subpackages from LANGUAGES comment
 #                       in RPM spec files
-# Copyright © 2012 Red Hat, Inc.
+# Copyright © 2012, 2014 Red Hat, Inc.
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
@@ -26,6 +26,7 @@ import tempfile
 import atexit
 import re
 from stat import S_IMODE
+from itertools import islice
 
 def usage():
     print >>sys.stderr, "Usage: %s <packagename>.spec" % (sys.argv[0],)
@@ -63,6 +64,10 @@ sfmode = S_IMODE(os.stat(sfpath).st_mode)
 languages_re = re.compile(
         r"^#\s*LANGUAGES:\s*(?P<languages>[^\n\r]+)\s*$", re.MULTILINE)
 langsplit_re = re.compile(r"\s+")
+begin_obsoletes_re = re.compile(
+        r"^#\s*BEGIN:\s*OBSOLETE\s+LANGUAGES\s*$", re.MULTILINE)
+end_obsoletes_re = re.compile(
+        r"^#\s*END:\s*OBSOLETE\s+LANGUAGES\s*$", re.MULTILINE)
 begin_langpkgs_re = re.compile(
         r"^#\s*BEGIN:\s*LANGUAGE\s+SUB\s+PACKAGES\s*$", re.MULTILINE)
 end_langpkgs_re = re.compile(
@@ -74,14 +79,22 @@ end_langfiles_re = re.compile(
 
 name_re = re.compile(r"^name:\s*(?P<name>[^\n\r]+)\s*$",
         re.MULTILINE | re.IGNORECASE)
+version_re = re.compile(r"^version:\s*(?P<version>[^\n\r]+)\s*$",
+        re.MULTILINE | re.IGNORECASE)
+release_re = re.compile(r"^release:\s*(?P<release>[^\n\r]+)\s*$",
+        re.MULTILINE | re.IGNORECASE)
 license_re = re.compile(r"^license:\s*(?P<license>[^\n\r]+)\s*$",
         re.MULTILINE | re.IGNORECASE)
+pkg_re = re.compile(r"^%package\s*(?P<pkg>\S+)\s*$",
+        re.MULTILINE | re.IGNORECASE)
 group_re = re.compile(r"^group:\s*(?P<group>[^\n\r]+)\s*$",
         re.MULTILINE | re.IGNORECASE)
 
 missing = False
 for what, what_re in (("name tag", name_re), ("license tag", license_re),
         ("group tag", group_re), ("LANGUAGES comment", languages_re),
+        ("BEGIN: OBSOLETE LANGUAGES comment", begin_obsoletes_re),
+        ("END: OBSOLETE LANGUAGES comment", end_obsoletes_re),
         ("BEGIN: LANGUAGE SUB PACKAGES comment", begin_langpkgs_re),
         ("END: LANGUAGE SUB PACKAGES comment", end_langpkgs_re),
         ("BEGIN: LANGUAGE FILE LISTS comment", begin_langfiles_re),
@@ -102,10 +115,20 @@ langspecs = langsplit_re.split(
 #    languages.append((langcode, langname))
 languages = [(x.split(",")[0], x.split(",")[1].replace('_', ' '))
         for x in langspecs]
+langcodes = set((x[0] for x in languages))
 name = name_re.search(sfcontent).group('name')
+version = version_re.search(sfcontent).group('version')
+release = release_re.search(sfcontent).group('release')
 license = license_re.search(sfcontent).group('license')
 group = group_re.search(sfcontent).group('group')
 
+obsoletes_re = re.compile(
+        r"^obsoletes:\s*%(name)s-(?P<lang_ver>(?P<lang>\S+)\s*.*)$" % locals(),
+        re.MULTILINE | re.IGNORECASE)
+conflicts_re = re.compile(
+        r"^conflicts:\s*%(name)s-(?P<lang_ver>(?P<lang>\S+)\s*.*)$" % locals(),
+        re.MULTILINE | re.IGNORECASE)
+
 numlang = len(languages)
 replacing = None
 
@@ -113,13 +136,81 @@ sflines = sfcontent.split("\n")
 if sflines[-1] == "":
     del sflines[-1]
 
+# handle obsoleting language subpackages
+preprocess_state_transitions = {
+    'out': (
+        begin_obsoletes_re, 'in_obsoletes', begin_langpkgs_re, 'in_langpkgs'),
+    'in_obsoletes': (end_obsoletes_re, 'out'),
+    'in_langpkgs': (end_langpkgs_re, 'out'),
+    }
+
+state = 'out'
+state_change = True
+
+found_obsoleted_langs = set()
+found_lang_pkgs = set()
+
+for line in sflines:
+    if state_change:
+        transitions = preprocess_state_transitions[state]
+        packed_transitions = zip(
+            islice(transitions, 0, None, 2), islice(transitions, 1, None, 2))
+        state_change = False
+
+    for regex, new_state in packed_transitions:
+        if regex.match(line):
+            state_change = True
+            state = new_state
+            break
+    if state_change:
+        continue
+
+    if state == 'in_obsoletes':
+        m = obsoletes_re.match(line)
+        if m:
+            found_obsoleted_langs.add(m.group('lang'))
+    elif state == 'in_langpkgs':
+        m = pkg_re.match(line)
+        if m:
+            found_lang_pkgs.add(m.group('pkg'))
+
+langcodes_to_obsolete = found_lang_pkgs - langcodes
+langcodes_to_unobsolete = found_obsoleted_langs & langcodes
+
+gobble = False
+# update spec file
 for line in sflines:
     if not replacing:
         print >>newsf, line
-        if begin_langpkgs_re.match(line):
+        if begin_obsoletes_re.match(line):
+            replacing = 'obsoletes'
+        elif begin_langpkgs_re.match(line):
             replacing = 'langpkgs'
         elif begin_langfiles_re.match(line):
             replacing = 'langfiles'
+    elif replacing == 'obsoletes':
+        om = obsoletes_re.match(line)
+        cm = conflicts_re.match(line)
+        em = end_obsoletes_re.match(line)
+        if em:
+            gobble = False
+            replacing = None
+            for lang in langcodes_to_obsolete:
+                print >>newsf, (
+                    "Obsoletes: %(name)s-%(lang)s < %(version)s-%(release)s\n"
+                    "Conflicts: %(name)s-%(lang)s < %(version)s-%(release)s"
+                    ) % locals()
+            print >>newsf, line
+        elif not (om or cm):
+            if not gobble:
+                print >>newsf, line
+        elif (
+                om and om.group('lang') not in langcodes_to_unobsolete or
+                cm and cm.group('lang') not in langcodes_to_unobsolete):
+            gobble = False
+            print >>newsf, line
+        else:
+            gobble = True
     elif replacing == 'langpkgs':
         if end_langpkgs_re.match(line):
             replacing = None