[dak/bpo] utf-8

[Joerg Jaspert <joerg@debian.org>]

enforce properly encoded changes/dsc files. if they do not pass
unicode() they wont get in.

Signed-off-by: Joerg Jaspert <joerg@debian.org>
---
 dak/check_archive.py          |    3 +++
 dak/check_proposed_updates.py |    3 +++
 dak/examine_package.py        |    5 ++++-
 dak/process_unchecked.py      |    6 ++++++
 daklib/dak_exceptions.py      |    3 ++-
 daklib/utils.py               |    4 ++++
 6 files changed, 22 insertions(+), 2 deletions(-)

diff --git a/dak/check_archive.py b/dak/check_archive.py
index 8078290..bed974d 100755
--- a/dak/check_archive.py
+++ b/dak/check_archive.py
@@ -161,6 +161,9 @@ def check_dscs():
             except InvalidDscError, line:
                 utils.warn("syntax error in .dsc file '%s', line %s." % (f, line))
                 count += 1
+            except ChangesUnicodeError:
+                utils.warn("found invalid changes file, not properly utf-8 encoded")
+                count += 1
 
     if count:
         utils.warn("Found %s invalid .dsc files." % (count))
diff --git a/dak/check_proposed_updates.py b/dak/check_proposed_updates.py
index f12df87..16a9876 100755
--- a/dak/check_proposed_updates.py
+++ b/dak/check_proposed_updates.py
@@ -176,6 +176,9 @@ def check_changes (filename):
     try:
         changes = utils.parse_changes(filename)
         files = utils.build_file_list(changes)
+    except ChangesUnicodeError:
+        utils.warn("Improperly encoded changes file, not utf-8")
+        return
     except:
         utils.warn("Error parsing changes file '%s'" % (filename))
         return
diff --git a/dak/examine_package.py b/dak/examine_package.py
index d0e1e53..eb60279 100755
--- a/dak/examine_package.py
+++ b/dak/examine_package.py
@@ -470,7 +470,10 @@ def display_changes(suite, changes_filename):
     foldable_output(changes_filename, "changes", changes, norow=True)
 
 def check_changes (changes_filename):
-    changes = utils.parse_changes (changes_filename)
+    try:
+        changes = utils.parse_changes (changes_filename)
+    except ChangesUnicodeError:
+        utils.warn("Encoding problem with changes file %s" % (changes_filename))
     display_changes(changes['distribution'], changes_filename)
 
     files = utils.build_file_list(changes)
diff --git a/dak/process_unchecked.py b/dak/process_unchecked.py
index 2e60435..288563a 100755
--- a/dak/process_unchecked.py
+++ b/dak/process_unchecked.py
@@ -186,6 +186,9 @@ def check_changes():
     except ParseChangesError, line:
         reject("%s: parse error, can't grok: %s." % (filename, line))
         return 0
+    except ChangesUnicodeError:
+        reject("%s: changes file not proper utf-8" % (filename))
+        return 0
 
     # Parse the Files field from the .changes into another dictionary
     try:
@@ -695,6 +698,9 @@ def check_dsc():
         reject("%s: parse error, can't grok: %s." % (dsc_filename, line))
     except InvalidDscError, line:
         reject("%s: syntax error on line %s." % (dsc_filename, line))
+    except ChangesUnicodeError:
+        reject("%s: dsc file not proper utf-8." % (dsc_filename))
+
     # Build up the file list of files mentioned by the .dsc
     try:
         dsc_files.update(utils.build_file_list(dsc, is_a_dsc=1))
diff --git a/daklib/dak_exceptions.py b/daklib/dak_exceptions.py
index d18bee1..33fa5ad 100755
--- a/daklib/dak_exceptions.py
+++ b/daklib/dak_exceptions.py
@@ -58,7 +58,8 @@ dakerrors = {
     "NoFreeFilenameError": """Exception raised when no alternate filename was found.""",
     "TransitionsError":    """Exception raised when transitions file can't be parsed.""",
     "NoSourceFieldError":  """Exception raised - we cant find the source - wtf?""",
-    "DBUpdateError":       """Exception raised - could not update the database"""
+    "DBUpdateError":       """Exception raised - could not update the database""",
+    "ChangesUnicodeError": """Exception raised - changes file not properly utf-8 encoded"""
 } #: All dak exceptions
 
 def construct_dak_exception(name, description):
diff --git a/daklib/utils.py b/daklib/utils.py
index b20a063..e0bdfe8 100755
--- a/daklib/utils.py
+++ b/daklib/utils.py
@@ -234,6 +234,10 @@ def parse_changes(filename, signing_rules=0):
     changes_in = open_file(filename)
     content = changes_in.read()
     changes_in.close()
+    try:
+        unicode(content, 'utf-8')
+    except UnicodeError:
+        raise ChangesUnicodeError, "Changes file not proper utf-8"
     return parse_deb822(content, signing_rules)
 
 ################################################################################
-- 
1.5.6.5