Correctly match version part of source tarballs

[dak.git] / daklib / regexes.py

diff --git a/daklib/regexes.py b/daklib/regexes.py
index 47e6ab2c9f255402d23878eccfb0be903abf1761..3da3b298793f89e74ec621108034dadede600a40 100755 (executable)
--- a/daklib/regexes.py
+++ b/daklib/regexes.py
@@ -49,11 +49,11 @@ re_arch_from_filename = re.compile(r"/binary-[^/]+/")
 re_extract_src_version = re.compile (r"(\S+)\s*\((.*)\)")
 re_isadeb = re.compile (r"(.+?)_(.+?)_(.+)\.u?deb$")
 
-orig_source_ext_re = r"orig(?:-.+)?\.tar\.(?:gz|bz2)"
+orig_source_ext_re = r"orig(?:-.+)?\.tar\.(?:gz|bz2|xz)"
 re_orig_source_ext = re.compile(orig_source_ext_re + "$")
-re_source_ext = re.compile("(" + orig_source_ext_re + r"|debian\.tar\.(?:gz|bz2)|diff\.gz|tar\.(?:gz|bz2)|dsc)$")
+re_source_ext = re.compile("(" + orig_source_ext_re + r"|debian\.tar\.(?:gz|bz2|xz)|diff\.gz|tar\.(?:gz|bz2|xz)|dsc)$")
 re_issource = re.compile(r"(.+)_(.+?)\." + re_source_ext.pattern)
-re_is_orig_source = re.compile (r"(.+)_(.+?)\.orig(?:-.+)?\.tar\.(?:gz|bz2)$")
+re_is_orig_source = re.compile (r"(.+)_(.+?)\.orig(?:-.+)?\.tar\.(?:gz|bz2|xz)$")
 #re_is_orig_source = re.compile (r"(.+)_(.+?)\.(?:orig\.)?tar\.(?:gz|bz2)$")
 
 re_single_line_field = re.compile(r"^(\S*?)\s*:\s*(.*)")
@@ -127,4 +127,53 @@ re_match_expired = re.compile(r"^The key used to sign .+ has expired on .+$")
 
 # in generate-releases
 re_gensubrelease = re.compile (r".*/(binary-[0-9a-z-]+|source)$")
-re_includeinrelease = re.compile (r"(Contents-[0-9a-z-]+.gz|Index|Packages(.gz|.bz2)?|Sources(.gz|.bz2)?|MD5SUMS|Release)$")
+re_includeinrelease = re.compile (r"(Translation-[a-zA-Z_]+\.(?:bz2|xz)|Contents-[0-9a-z-]+.gz|Index|Packages(.gz|.bz2)?|Sources(.gz|.bz2)?|MD5SUMS|SHA256SUMS|Release)$")
+
+# in generate_index_diffs
+re_includeinpdiff = re.compile(r"(Translation-[a-zA-Z_]+\.(?:bz2|xz))")
+
+
+######################################################################
+# Patterns matching filenames                                        #
+######################################################################
+
+# Match safe filenames
+re_file_safe = re.compile(r'^[a-zA-Z0-9][a-zA-Z0-9_.:~+-]*$')
+
+# Prefix of binary and source filenames
+_re_file_prefix = '^(?P<package>[a-z0-9][a-z0-9.+-]+)_(?P<version>[A-Za-z0-9.:~+-]+?)'
+
+# Match binary packages
+# Groups: package, version, architecture, type
+re_file_binary = re.compile(_re_file_prefix + '_(?P<architecture>[a-z0-9]+)\.(?P<type>u?deb)$')
+
+# Match dsc files
+# Groups: package, version
+re_file_dsc = re.compile(_re_file_prefix + r'\.dsc$')
+
+# Match other source files
+# Groups: package, version
+re_file_source = re.compile(_re_file_prefix + r'(?:(?:\.orig(?:-[a-zA-Z0-9-]+)?|\.debian)?\.tar\.(?:bz2|gz|xz)|\.diff\.gz)$')
+
+# Match upstream tarball
+# Groups: package, version
+re_file_orig = re.compile(_re_file_prefix + r'\.orig(?:-[a-zA-Z0-9-]+)?\.tar\.(?:bz2|gz|xz)')
+
+######################################################################
+# Patterns matching fields                                           #
+######################################################################
+
+# Match package name
+re_field_package = re.compile(r'^[a-z0-9][a-z0-9.+-]+$')
+
+# Match version
+# Groups: without-epoch
+re_field_version = re.compile(r'^(?:[0-9]+:)?(?P<without_epoch>[A-Za-z0-9.:~+-]+)$')
+
+# Extract upstream version
+# Groups: upstream
+re_field_version_upstream = re.compile(r'^(?:[0-9]+:)?(?P<upstream>.*)-[^-]*$')
+
+# Match source field
+# Groups: package, version
+re_field_source = re.compile(r'^(?P<package>[a-z0-9][a-z0-9.+-]+)(:?\s*\((?P<version>[A-Za-z0-9.:~+-])\))?')