add filter-pristine-tar

to filter upstream tarball before passing it to pristine-tar

Closes: #520722

6 files changed, 48 insertions, 1 deletions

diff --git a/docs/manpages/git-import-orig.sgml b/docs/manpages/git-import-orig.sgml
index 2c0d44da..7c19adf6 100644
--- a/docs/manpages/git-import-orig.sgml
+++ b/docs/manpages/git-import-orig.sgml
@@ -32,6 +32,7 @@
       <arg><option>--upstream-tag=</option><replaceable>tag-format</replaceable></arg>
       <arg><option>--filter=</option><replaceable>pattern</replaceable></arg>
       <arg><option>--[no-]pristine-tar</option></arg>
+      <arg><option>--[no-]filter-pristine-tar</option></arg>
       <arg choice="plain"><replaceable>upstream-source</replaceable></arg>
     </cmdsynopsis>
   </refsynopsisdiv>
@@ -136,6 +137,14 @@
           <para>generate pristine-tar delta file</para>
         </listitem>
       </varlistentry>
+      <varlistentry>
+        <term><option>--filter-pristine-tar</option>
+        </term>
+        <listitem>
+	  <para>if using a filter also filter the files out of the tarball
+	  passed to pristine tar</para>
+        </listitem>
+      </varlistentry>
     </variablelist>
   </refsect1>
   <refsect1>
diff --git a/gbp.conf b/gbp.conf
index 7a1029e6..ce6fc111 100644
--- a/gbp.conf
+++ b/gbp.conf
@@ -37,6 +37,8 @@
 #upstream-branch = newupstream
 #debian-branch = dfsgclean
 #filter = .svn
+# filter out files from tarball passed to pristine tar
+#filter-pristine-tar = True
 
 # Options only affecting git-import-dsc
 [git-import-dsc]
diff --git a/gbp/command_wrappers.py b/gbp/command_wrappers.py
index 1439cc59..8cfa7cb9 100644
--- a/gbp/command_wrappers.py
+++ b/gbp/command_wrappers.py
@@ -132,6 +132,19 @@ class UnpackTarArchive(Command):
         Command.__init__(self, 'tar', exclude + ['-C', dir, decompress, '-xf', archive ])
         self.run_error = 'Couldn\'t unpack "%s"' % self.archive
 
+class RepackTarArchive(Command):
+    """Wrap tar to Repack a gzipped tar archive"""
+    def __init__(self, archive, dir, dest):
+        self.archive = archive
+        self.dir = dir
+
+        if archive.lower().endswith(".bz2"):
+            compress = "--bzip2"
+        else:
+            compress = "--gzip"
+
+        Command.__init__(self, 'tar', ['-C', dir, compress, '-cf', archive, dest])
+        self.run_error = 'Couldn\'t repack "%s"' % self.archive
 
 class RemoveTree(Command):
     "Wrap rm to remove a whole directory tree"
diff --git a/gbp/config.py b/gbp/config.py
index 6fa223a6..0ffcc327 100644
--- a/gbp/config.py
+++ b/gbp/config.py
@@ -30,6 +30,7 @@ class GbpOptionParser(OptionParser):
                  'debian-branch'   : 'master',
                  'upstream-branch' : 'upstream',
                  'pristine-tar'    : 'False',
+                 'filter-pristine-tar' : 'False',
                  'sign-tags'       : 'False',
                  'no-create-orig'  : 'False',
                  'keyid'           : '',
@@ -66,6 +67,8 @@ class GbpOptionParser(OptionParser):
                   "GPG keyid to sign tags with, default is '%(keyid)s'",
              'pristine-tar':
                   "use pristine-tar to create .orig.tar.gz, default is '%(pristine-tar)s'",
+             'filter-pristine-tar':
+                  "Filter pristine-tar when filter option is used",
              'filter':
                   "files to filter out during import (can be given multiple times)",
              'git-author':
diff --git a/gbp/deb_utils.py b/gbp/deb_utils.py
index f20033a0..c439ba36 100644
--- a/gbp/deb_utils.py
+++ b/gbp/deb_utils.py
@@ -199,6 +199,17 @@ def unpack_orig(archive, tmpdir, filters):
         raise GbpError
     return unpackArchive.dir
 
+def repack_orig(archive, tmpdir, dest):
+    """
+    recreate a new .orig.tar.gz from tmpdir (useful when using filter option)
+    """
+    try:
+        repackArchive = gbpc.RepackTarArchive(archive, tmpdir, dest)
+        repackArchive()
+    except gbpc.CommandExecFailed:
+        print >>sys.stderr, "Failed to create %s" % archive
+        raise GbpError
+    return repackArchive.dir
 
 def tar_toplevel(dir):
     """tar archives can contain a leading directory not"""
diff --git a/git-import-orig b/git-import-orig
index acb012ce..9e47bf84 100755
--- a/git-import-orig
+++ b/git-import-orig
@@ -27,7 +27,7 @@ import subprocess
 import tarfile
 import time
 import gbp.command_wrappers as gbpc
-from gbp.deb_utils import parse_changelog, unpack_orig, NoChangelogError, has_epoch, tar_toplevel, guess_upstream_version
+from gbp.deb_utils import parse_changelog, unpack_orig, repack_orig, NoChangelogError, has_epoch, tar_toplevel, guess_upstream_version
 from gbp.git_utils import (GitRepositoryError, GitRepository, build_tag)
 from gbp.config import GbpOptionParser
 from gbp.errors import (GbpError, GbpNothingImported)
@@ -228,6 +228,7 @@ def main(argv):
     parser.add_config_file_option(option_name="upstream-tag", dest="upstream_tag")
     parser.add_config_file_option(option_name="filter", dest="filters", action="append")
     parser.add_boolean_config_file_option(option_name="pristine-tar", dest="pristine_tar")
+    parser.add_boolean_config_file_option(option_name="filter-pristine-tar", dest="filter_pristine_tar")
     (options, args) = parser.parse_args(argv[1:])
 
     if options.verbose:
@@ -292,6 +293,14 @@ on howto create it otherwise use --upstream-branch to specify it.
                 if options.verbose:
                     print "Unpacked %s to '%s'" % (archive , tmpdir)
                 orig_dir = tar_toplevel(tmpdir)
+                if options.pristine_tar and options.filter_pristine_tar and len(options.filters) > 0:
+                    if options.verbose:
+                        print "Filter pristine-tar: repacking %s from '%s'" % (archive, tmpdir)
+                    archive = os.path.join(
+                        os.path.dirname(archive),
+                        os.path.basename(archive).replace(".tar", ".gbp.tar")
+                        )
+                    repack_orig(archive, tmpdir, os.path.basename(orig_dir))
             try:
                 cp = parse_changelog('debian/changelog')
                 pristine_orig = symlink_orig(archive, cp['Source'], version)