Merge commit 'origin/master' into content_generation

[dak.git] / daklib / utils.py

diff --git a/daklib/utils.py b/daklib/utils.py
index 8e06cf35cc6dbbc795b008865c64991b2d9dbfc3..47c80f3c137975468569f7b8cb354688448edfce 100755 (executable)
--- a/daklib/utils.py
+++ b/daklib/utils.py
@@ -50,6 +50,9 @@ re_verwithext = re.compile(r"^(\d+)(?:\.(\d+))(?:\s+\((\S+)\))?$")
 
 re_srchasver = re.compile(r"^(\S+)\s+\((\S+)\)$")
 
+html_escaping = {'"':'&quot;', '&':'&amp;', '<':'&lt;', '>':'&gt;'}
+re_html_escaping = re.compile('|'.join(map(re.escape, html_escaping.keys())))
+
 default_config = "/etc/dak/dak.conf"
 default_apt_config = "/etc/dak/apt.conf"
 
@@ -62,6 +65,11 @@ known_hashes = [("sha1", apt_pkg.sha1sum, (1, 8)),
 
 ################################################################################
 
+def html_escape(s):
+    return re_html_escaping.sub(lambda x: html_escaping.get(x.group(0)), s)
+
+################################################################################
+
 def open_file(filename, mode='r'):
     try:
         f = open(filename, mode)
@@ -252,22 +260,23 @@ def check_hash(where, files, hashname, hashfunc):
     for f in files.keys():
         file_handle = None
         try:
-            file_handle = open_file(f)
-
-            # Check for the hash entry, to not trigger a KeyError.
-            if not files[f].has_key(hash_key(hashname)):
-                rejmsg.append("%s: misses %s checksum in %s" % (f, hashname,
-                    where))
+            try:
+                file_handle = open_file(f)
+    
+                # Check for the hash entry, to not trigger a KeyError.
+                if not files[f].has_key(hash_key(hashname)):
+                    rejmsg.append("%s: misses %s checksum in %s" % (f, hashname,
+                        where))
+                    continue
+    
+                # Actually check the hash for correctness.
+                if hashfunc(file_handle) != files[f][hash_key(hashname)]:
+                    rejmsg.append("%s: %s check failed in %s" % (f, hashname,
+                        where))
+            except CantOpenError:
+                # TODO: This happens when the file is in the pool.
+                # warn("Cannot open file %s" % f)
                 continue
-
-            # Actually check the hash for correctness.
-            if hashfunc(file_handle) != files[f][hash_key(hashname)]:
-                rejmsg.append("%s: %s check failed in %s" % (f, hashname,
-                    where))
-        except CantOpenError:
-            # TODO: This happens when the file is in the pool.
-            warn("Cannot open file %s" % f)
-            continue
         finally:
             if file_handle:
                 file_handle.close()
@@ -394,11 +403,15 @@ def parse_checksums(where, files, manifest, hashname):
             break
         hash, size, file = line.strip().split(' ')
         if not files.has_key(file):
-            rejmsg.append("%s: not present in files but in checksums-%s in %s" %
-                (file, hashname, where))
+        # TODO: check for the file's entry in the original files dict, not
+        # the one modified by (auto)byhand and other weird stuff
+        #    rejmsg.append("%s: not present in files but in checksums-%s in %s" %
+        #        (file, hashname, where))
+            continue
         if not files[file]["size"] == size:
             rejmsg.append("%s: size differs for files and checksums-%s entry "\
                 "in %s" % (file, hashname, where))
+            continue
         files[file][hash_key(hashname)] = hash
     for f in files.keys():
         if not files[f].has_key(hash_key(hashname)):
@@ -431,7 +444,10 @@ def build_file_list(changes, is_a_dsc=0, field="files", hashname="md5sum"):
         format = format[:2]
 
     if is_a_dsc:
-        if format != (1,0):
+        # format = (1,0) are the only formats we currently accept,
+        # format = (0,0) are missing format headers of which we still
+        # have some in the archive.
+        if format != (1,0) and format != (0,0):
             raise UnknownFormatError, "%s" % (changes.get("format","0.0"))
     else:
         if (format < (1,5) or format > (1,8)):