daklib/regexes.py: updates to support new source files

The regex re_issource is modified to recognize the new compression scheme
supported by the 3.0 source package formats. Also add the
the debian.tar file used by 3.0 (quilt) and 2.0 and the
.orig-component.tar for additional upstream tarballs.

Add a new regex re_is_orig_source to know if a file correspond
to some upstream source tarballs (where the version should not have a
debian revision).

Add new regexes re_source_ext and re_orig_source_ext to match only the
extension of source files. And reuse those in the definition of
re_is_orig_source and re_issource to avoid duplication.

Signed-off-by: Raphael Hertzog <hertzog@debian.org>

diff --git a/daklib/regexes.py b/daklib/regexes.py
index d1f0d38136afcc9616bcd82e3ab50106b4bfa861..f18e2a49da6394b3facbc3561fa574c62ebd3566 100755 (executable)
--- a/daklib/regexes.py
+++ b/daklib/regexes.py
@@ -42,7 +42,11 @@ re_arch_from_filename = re.compile(r"/binary-[^/]+/")
 re_extract_src_version = re.compile (r"(\S+)\s*\((.*)\)")
 re_isadeb = re.compile (r"(.+?)_(.+?)_(.+)\.u?deb$")
 
-re_issource = re.compile (r"(.+)_(.+?)\.(orig\.tar\.gz|diff\.gz|tar\.gz|dsc)$")
+orig_source_ext_re = r"orig(?:-.+)?\.tar\.(?:gz|bz2|lzma)"
+re_orig_source_ext = re.compile(orig_source_ext_re + "$")
+re_source_ext = re.compile("(" + orig_source_ext_re + r"|debian\.tar\.(?:gz|bz2|lzma)|diff\.gz|tar\.(?:gz|bz2|lzma)|dsc)$")
+re_issource = re.compile(r"(.+)_(.+?)\." + re_source_ext.pattern)
+re_is_orig_source = re.compile (r"(.+)_(.+?)\.orig(?:-.+)?\.tar\.(?:gz|bz2|lzma)$")
 
 re_single_line_field = re.compile(r"^(\S*?)\s*:\s*(.*)")
 re_multi_line_field = re.compile(r"^\s(.*)")