extend processing of office files: error handling, handle description property as comments

integrator/README.txt

@@ -231,7 +231,7 @@ Extracting Document Properties from MS Office Files
   >>> path = os.path.join(dataDir, 'office')
   >>> fn = os.path.join(path, 'example.docx')
   >>> os.path.getsize(fn)
-  20337...
+  23561...
 
   >>> officeFile = addAndConfigureObject(resources, Resource, 'test.docx',
   ...                    title=u'Example Word File', resourceType=tOfficeFile,
@@ -241,7 +241,7 @@ Extracting Document Properties from MS Office Files
 
   >>> content = aOfficeFile.data
   >>> len(content)
-  17409
+  20327
 
   Clean up:
   >>> shutil.copy(fn + '.sav', fn)

integrator/browser.py

@@ -40,8 +40,8 @@ class ExternalCollectionView(ConceptView):
     def update(self):
         if 'update' in self.request.form:
             cta = adapted(self.context)
-            if cta is not None:
+            cta.request = self.request
-                cta.update()
+            cta.update()
             if cta.updateMessage is not None:
                 self.request.form['message'] = cta.updateMessage
         return True

integrator/collection.py

@@ -101,6 +101,7 @@ class ExternalCollectionAdapter(AdapterBase):
                     adobj = adapted(obj)
                     directory = provider.getDirectory(self)
                     adobj.storageParams=dict(subdirectory=directory)
+                    adobj.request = self.request
                     adobj.externalAddress = addr
                     # collect error information
                     if adobj.processingErrors:
@@ -209,6 +210,7 @@ class DirectoryCollectionProvider(object):
                             contentType=contentType,
             )
             adobj = adapted(obj)
+            adobj.request = client.request
             adobj.externalAddress = addr     # must be set last
             # collect error information
             if adobj.processingErrors:

integrator/office/base.py

@@ -52,16 +52,12 @@ class OfficeFile(ExternalFileAdapter):
 
     implements(IOfficeFile)
 
-    _adapterAttributes = ExternalFileAdapter._adapterAttributes + (
-                                'processingErrors',)
-
     propertyMap = {u'Revision:': 'version'}
     propFileName = 'docProps/custom.xml'
+    corePropFileName = 'docProps/core.xml'
     fileExtensions = ('.docm', '.docx', 'dotm', 'dotx', 'pptx', 'potx', 'ppsx',
                       '.xlsm', '.xlsx', '.xltm', '.xltx')
 
-    processingErrors = []
-
     @Lazy
     def logger(self):
         return getLogger('loops.integrator.office.base.OfficeFile')
@@ -91,20 +87,32 @@ class OfficeFile(ExternalFileAdapter):
             from logging import getLogger
             self.logger.warn(e)
             return []
+        if self.corePropFileName not in zf.namelist():
+            self.logger.warn('Core properties not found in file %s.' %
+                             self.externalAddress)
         if self.propFileName not in zf.namelist():
             self.logger.warn('Custom properties not found in file %s.' %
                              self.externalAddress)
         propsXml = zf.read(self.propFileName)
+        corePropsXml = zf.read(self.corePropFileName)
+        # TODO: read core.xml, return both trees in dictionary
         zf.close()
-        return etree.fromstring(propsXml)
+        return {'custom': etree.fromstring(propsXml),
+                'core': etree.fromstring(corePropsXml)}
 
     def getDocProperty(self, pname):
-        for p in self.docPropertyDom:
+        for p in self.docPropertyDom['custom']:
             name = p.attrib.get('name')
             if name == pname:
                 return p[0].text
         return None
 
+    def getCoreProperty(self, pname):
+        for p in self.docPropertyDom['core']:
+            if p.tag.endswith(pname):
+                return p.text
+        return None
+
     def processDocument(self):
         changed = False
         docVersion = None
@@ -112,11 +120,14 @@ class OfficeFile(ExternalFileAdapter):
         strType = ('{http://schemas.openxmlformats.org/'
                    'officeDocument/2006/docPropsVTypes}lpwstr')
         attributes = {}
-        dom = self.docPropertyDom
+        # get dc:description from core.xml
+        desc = self.getCoreProperty('description')
+        if desc is not None:
+            attributes['comments'] = desc
+        dom = self.docPropertyDom['custom']
         for p in dom:
             name = p.attrib.get('name')
             value = p[0].text
-            #print '***', self.externalAddress, name, value, p[0].tag
             attr = self.propertyMap.get(name)
             if attr == 'version':
                 docVersion = value

resource.py

@@ -1,5 +1,5 @@
 #
-#  Copyright (c) 2011 Helmut Merz helmutm@cy55.de
+#  Copyright (c) 2012 Helmut Merz helmutm@cy55.de
 #
 #  This program is free software; you can redistribute it and/or modify
 #  it under the terms of the GNU General Public License as published by
@@ -18,8 +18,6 @@
 
 """
 Definition of the Concept class.
-
-$Id$
 """
 
 from cStringIO import StringIO
@@ -352,7 +350,10 @@ class ExternalFileAdapter(FileAdapter):
     implements(IExternalFile)
 
     _adapterAttributes = (FileAdapter._adapterAttributes
-                       + ('storageParams', 'externalAddress', 'uniqueAddress'))
+                        + ('storageParams', 'externalAddress', 'uniqueAddress',
+                           'processingErrors'))
+
+    processingErrors = []
 
     def getStorageParams(self):
         params = getattr(self.context, '_storageParams', None)