- don't read all rpm/deb tags/fields by default

osc/fetch.py

@@ -90,7 +90,7 @@ class Fetcher:
             print >>sys.stderr, '\n'.join(pac.urllist)
             sys.exit(1)
 
-        pkgq = packagequery.PackageQuery.query(pac.fullpartname)
+        pkgq = packagequery.PackageQuery.query(pac.fullpartname, extra_rpmtags=(1044, 1051, 1052))
         arch = pkgq.arch()
         # SOURCERPM = 1044
         if pkgq.filename_suffix == 'rpm' and not pkgq.getTag(1044):

osc/util/debquery.py

@@ -7,12 +7,16 @@ class DebError(packagequery.PackageError):
     pass
 
 class DebQuery(packagequery.PackageQuery):
+
+    default_tags = ('package', 'version', 'release', 'epoch', 'architecture', 'description',
+        'provides', 'depends', 'pre_depends')
+
     def __init__(self, fh):
         self.__file = fh
         self.filename_suffix = 'deb'
         self.fields = {}
 
-    def read(self):
+    def read(self, all_tags = False, *extra_tags):
         arfile = ar.Ar(fh = self.__file)
         arfile.read()
         debbin = arfile.get_file('debian-binary')
@@ -29,9 +33,9 @@ class DebQuery(packagequery.PackageQuery):
             control = tar.extractfile('./control')
         except KeyError:
             raise DebError('missing \'control\' file in control.tar.gz')
-        self.__parse_control(control)
+        self.__parse_control(control, all_tags, *extra_tags)
 
-    def __parse_control(self, control):
+    def __parse_control(self, control, all_tags = False, *extra_tags):
         data = control.readline().strip()
         while data:
             field, val = re.split(':\s*', data.strip(), 1)
@@ -39,8 +43,10 @@ class DebQuery(packagequery.PackageQuery):
             while data and re.match('\s+', data):
                 val += '\n' + data.strip()
                 data = control.readline().rstrip()
-            # a hyphen is not allowed in dict keys
-            self.fields[field.replace('-', '_').lower()] = val
+            field = field.replace('-', '_').lower()
+            if field in self.default_tags + extra_tags or all_tags:
+                # a hyphen is not allowed in dict keys
+                self.fields[field] = val
         versrel = self.fields['version'].rsplit('-', 1)
         if len(versrel) == 2:
             self.fields['version'] = versrel[0]
@@ -97,10 +103,10 @@ class DebQuery(packagequery.PackageQuery):
         return self.fields.get(num, None)
 
     @staticmethod
-    def query(filename):
+    def query(filename, all_tags = False, *extra_tags):
         f = open(filename, 'rb')
         debq = DebQuery(f)
-        debq.read()
+        debq.read(all_tags, *extra_tags)
         f.close()
         return debq
 

osc/util/packagequery.py

@@ -6,7 +6,7 @@ class PackageError(Exception):
 
 class PackageQuery:
     """abstract base class for all package types"""
-    def read(self):
+    def read(self, all_tags = False, *extra_tags):
         raise NotImplementedError
 
     def name(self):
@@ -40,20 +40,23 @@ class PackageQuery:
         raise NotImplementedError
 
     @staticmethod
-    def query(filename):
+    def query(filename, all_tags = False, extra_rpmtags = (), extra_debtags = ()):
         f = open(filename, 'rb')
         magic = f.read(7)
         f.seek(0)
+        extra_tags = ()
         pkgq = None
         if magic[:4] == '\xed\xab\xee\xdb':
             import rpmquery
             pkgq = rpmquery.RpmQuery(f)
+            extra_tags = extra_rpmtags
         elif magic == '!<arch>':
             import debquery
             pkgq = debquery.DebQuery(f)
+            extra_tags = extra_debtags
         else:
             raise PackageError('unsupported package type. magic: \'%s\' (%s)' % (magic, filename))
-        pkgq.read()
+        pkgq.read(all_tags, *extra_tags)
         f.close()
         return pkgq
 

osc/util/rpmquery.py

@@ -54,12 +54,18 @@ class RpmQuery(packagequery.PackageQuery):
     LESS = 1 << 1
     GREATER = 1 << 2
     EQUAL = 1 << 3
+
+    default_tags = (1000, 1001, 1002, 1003, 1004, 1022, 1005, 1020,
+        1047, 1112, 1113, # provides
+        1049, 1048, 1050 # requires
+    )
+
     def __init__(self, fh):
         self.__file = fh
         self.filename_suffix = 'rpm'
         self.header = None
 
-    def read(self, *tags):
+    def read(self, all_tags = False, *extra_tags):
         self.__read_lead()
         data = self.__file.read(RpmHeaderEntry.ENTRY_SIZE)
         hdrmgc, reserved, il, dl = struct.unpack('!I3i', data)
@@ -82,7 +88,7 @@ class RpmQuery(packagequery.PackageQuery):
             data = data[RpmHeaderEntry.ENTRY_SIZE:]
         data = self.__file.read(self.header.length)
         for i in self.header:
-            if i.tag in tags or len(tags) == 0:
+            if i.tag in self.default_tags + extra_tags or all_tags:
                 try: # this may fail for -debug* packages
                     self.__read_data(i, data)
                 except: pass