publish.midas3.mdwsgi: allow for unicode in filenames

python/nistoar/pdr/preserv/bagger/midas3.py

@@ -28,7 +28,7 @@
 from ....nerdm import utils as nerdutils
 from ... import def_merge_etcdir, utils, ARK_NAAN, PDR_PUBLIC_SERVER
 from .. import (SIPDirectoryError, SIPDirectoryNotFound, AIPValidationError,
-                ConfigurationException, StateException, PODError,
+                ConfigurationException, StateException, PODError, NERDError,
                 PreservationStateError)
 from .... import pdr
 from .prepupd import UpdatePrepService
@@ -324,7 +324,7 @@ def _filepaths_in_pod(self):
 
         pod = self._pod_rec()
 
-        return [self._distsvcurl.sub('', urllib.unquote(d['downloadURL']))
+        return [self._distsvcurl.sub('', urllib.unquote(str(d['downloadURL'])).decode('utf8'))
                 for d in pod.get('distribution',[]) if 'downloadURL' in d]
 
 

python/nistoar/pdr/publish/midas3/mdwsgi.py

@@ -26,7 +26,7 @@
              .getChild(pdrsys.subsystem_abbrev) \
              .getChild("m3mdserv")
 
-DEF_BASE_PATH = "/midas/"
+DEF_BASE_PATH = u"/midas/"
 
 class MIDAS3DataAccessApp(object):
     """
@@ -114,11 +114,15 @@ def __init__(self, app, wsgienv, start_resp):
         self._midascl = app._midascl
 
     def send_error(self, code, message):
+        if isinstance(message, unicode):
+            message = message.encode('utf8')
         status = "{0} {1}".format(str(code), message)
         self._start(status, [], sys.exc_info())
         return []
 
     def send_ok(self, message="OK"):
+        if isinstance(message, unicode):
+            message = message.encode('utf8')
         status = "{0} {1}".format(str(code), message)
         self._start(status, [], None)
         return []
@@ -143,9 +147,11 @@ def handle(self):
         meth_handler = 'do_'+self._meth
 
         path = self._env.get('PATH_INFO', '/')
+        if isinstance(path, str):
+            path = path.decode('utf8')
         if '/' not in path:
             path += '/'
-        if not path.startswith(self.app.base_path):
+        if not path.startswith(self.app.base_path):    # path and base_path must both be unicode
             return self.send_error(404, "Resource not found")
         path = path[len(self.app.base_path):].rstrip('/')
 
@@ -278,9 +284,12 @@ def get_metadata(self, dsid):
                 if not dir:
                     continue
                 mdfile = os.path.join(dir, dsid+".json")
+                if isinstance(mdfile, unicode):
+                    mdfile = mdfile.encode('utf8')
                 if os.path.isfile(mdfile):
                     mdata = read_json(mdfile)
-                    log.info("Retrieving metadata record for id=%s from %s", dsid, mdfile)
+                    log.info("Retrieving metadata record for id=%s from %s",
+                             dsid, mdfile.decode('utf8'))
 
         except ValueError as ex:
             log.exception("Internal error while parsing JSON file, %s: %s", mdfile, str(ex))
@@ -296,7 +305,7 @@ def send_auto_readme(self, dsid, withprompts=True, bebrief=False):
             if mdata is None:
                 log.info("Metadata record not found for ID="+dsid)
                 return self.send_error(404,
-                                       "Dataset with ID={0} not being edited".format(dsid))
+                                       "Dataset with ID=%s not being edited" % dsid)
         except ValueError as ex:
             return self.send_error(500, "Internal parsing error")
         except Exception as ex:
@@ -326,7 +335,7 @@ def send_metadata(self, dsid):
             if mdata is None:
                 log.info("Metadata record not found for ID="+dsid)
                 return self.send_error(404,
-                                       "Dataset with ID={0} not being edited".format(dsid))
+                                       "Dataset with ID=%s not being edited" % dsid)
         except ValueError as ex:
             return self.send_error(500, "Internal parsing error")
         except Exception as ex:
@@ -374,13 +383,13 @@ def send_datafile(self, id, filepath):
             mdata = self.get_metadata(id)
             if mdata is None:
                 log.info("send_datafile: Metadata record not found for ID="+id)
-                return self.send_error(404,"Dataset with ID={0} not available".format(id))
+                return self.send_error(404,"Dataset with ID=%s not available" % id)
             sip = MIDASSIP.fromNERD(mdata, self.app.revdir, self.app.upldir)
 
         except SIPDirectoryNotFound as ex:
             # shouldn't happen
             log.warn("No SIP directories for ID="+dsid)
-            self.send_error(404,"Dataset with ID={0} not available".format(id))
+            self.send_error(404,"Dataset with ID=%s not available" % id)
             return []
         except Exception as ex:
             log.exception("Internal error: "+str(ex))

python/tests/nistoar/pdr/publish/midas3/test_mdwsgi.py

@@ -68,6 +68,17 @@ def test_bad_id(self):
         self.assertIn("404", self.resp[0])
         self.assertIn('asdifuiad', self.resp[0])
 
+    def test_bad_id_unicode(self):
+        req = {
+            'PATH_INFO': '/asdifui\xce\xb1d',
+            'REQUEST_METHOD': 'GET'
+        }
+        body = self.svc(req, self.start)
+
+        self.assertGreater(len(self.resp), 0)
+        self.assertIn("404", self.resp[0])
+        self.assertIn('asdifui\xce\xb1d', self.resp[0])
+
     def test_ark_id(self):
         req = {
             'PATH_INFO': '/ark:/88434/mds4-29sd17',