[121365] branches/gsoc14-pip2port

[gaurav at macports.org]

Revision: 121365
          https://trac.macports.org/changeset/121365
Author:   gaurav at macports.org
Date:     2014-06-24 00:09:20 -0700 (Tue, 24 Jun 2014)
Log Message:
-----------
Updated the encoding to only printable strings

Modified Paths:
--------------
    branches/gsoc14-pip2port/license.py
    branches/gsoc14-pip2port/pypi2port.py

Modified: branches/gsoc14-pip2port/license.py
===================================================================
--- branches/gsoc14-pip2port/license.py	2014-06-24 03:46:27 UTC (rev 121364)
+++ branches/gsoc14-pip2port/license.py	2014-06-24 07:09:20 UTC (rev 121365)
@@ -5,6 +5,7 @@
 
 import xmlrpclib
 import sys
+import string
 #import pdb
 #from datetime import datetime
 
@@ -25,12 +26,8 @@
         if data:
 #            pdb.set_trace()
             license = data['license']
-            try:
-                license = license.decode('utf-8','ignore')
-                license = license.encode('utf-8','ignore')
-                print package_name,vers,license
-            except:
-                print package_name,vers
+            license = filter(lambda x: x in string.printable, license)
+            print package_name,vers,license
         else:
             print package_name,vers
     else:

Modified: branches/gsoc14-pip2port/pypi2port.py
===================================================================
--- branches/gsoc14-pip2port/pypi2port.py	2014-06-24 03:46:27 UTC (rev 121364)
+++ branches/gsoc14-pip2port/pypi2port.py	2014-06-24 07:09:20 UTC (rev 121365)
@@ -219,6 +219,7 @@
         license = dict['license']
         if license and not license == "UNKNOWN":
             license = license.encode('utf-8')
+            license = filter(lambda x: x in string.printable, license)
             license = license.split('\n')[0]
             license = re.sub(r'[\[\]\{\}\;\:\$\t\"\'\`\=(--)]+', ' ', license)
             license = re.sub(r'\s(\s)+', ' ', license)
@@ -261,6 +262,7 @@
         description = dict['description']
         if description:
             description = description.encode('utf-8')
+            description = filter(lambda x: x in string.printable, description)
             description = re.sub(r'[\[\]\{\}\;\:\$\t\"\'\`\=(--)]+',
                                  ' ', description)
             description = re.sub(r'\s(\s)+', ' ', description)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.macosforge.org/pipermail/macports-changes/attachments/20140624/f0dd6d30/attachment.html>