Final changes to support duplicate document search

This commit is contained in:
Roberto Rosario
2011-02-21 19:32:03 -04:00
parent 5209847ab5
commit 5b4f2ecb57
4 changed files with 39 additions and 20 deletions

View File

@@ -701,28 +701,38 @@ def document_find_duplicates(request, document_id):
raise Http404(e)
document = get_object_or_404(Document, pk=document_id)
return _find_duplicate_list(request, [document])
return _find_duplicate_list(request, [document], include_source=True, confirmation=False)
def _find_duplicate_list(request, source_document_list=Document.objects.all(), include_source=False):
def _find_duplicate_list(request, source_document_list=Document.objects.all(), include_source=False, confirmation=True):
permissions = [PERMISSION_DOCUMENT_VIEW]
try:
check_permissions(request.user, 'documents', permissions)
except Unauthorized, e:
raise Http404(e)
duplicated = []
for document in source_document_list:
results = Document.objects.filter(checksum=document.checksum)
if not include_source:
results = results.exclude(id=document.id)
duplicated.extend(results)
previous = request.POST.get('previous', request.GET.get('previous', request.META.get('HTTP_REFERER', None)))
if confirmation and request.method != 'POST':
return render_to_response('generic_confirm.html', {
#'title':_(u'dupli'),
'previous':previous,
'message':_(u'On large databases this operation may take some time to execute.'),
}, context_instance=RequestContext(request))
else:
duplicated = []
for document in source_document_list:
if document not in duplicated:
results = Document.objects.filter(checksum=document.checksum).exclude(id__in=[d.id for d in duplicated]).exclude(id=document.id)
duplicated.extend(results)
return render_to_response('generic_list.html', {
'object_list':duplicated,
'title':_(u'duplicated documents'),
}, context_instance=RequestContext(request))
if include_source:
duplicated.extend(source_document_list)
return render_to_response('generic_list.html', {
'object_list':duplicated,
'title':_(u'duplicated documents'),
}, context_instance=RequestContext(request))
def document_find_all_duplicates(request):
@@ -732,4 +742,4 @@ def document_find_all_duplicates(request):
except Unauthorized, e:
raise Http404(e)
return _find_duplicate_list(request, include_source=True)
return _find_duplicate_list(request, include_source=False)

View File

@@ -12,7 +12,7 @@ check_settings = {'text':_(u'settings'), 'view':'check_settings', 'famfam':'cog'
register_menu([
{'text':_(u'home'), 'view':'home', 'famfam':'house', 'position':0},
{'text':_(u'tools'), 'view':'document_find_all_duplicates', 'links': [
{'text':_(u'tools'), 'view':'tools_menu', 'links': [
document_find_all_duplicates,
],'famfam':'wrench', 'name':'tools','position':7},

View File

@@ -3,5 +3,6 @@ from django.conf.urls.defaults import *
urlpatterns = patterns('main.views',
url(r'^$', 'home', (), 'home'),
url(r'^check_settings/$', 'check_settings', (), 'check_settings')
url(r'^check_settings/$', 'check_settings', (), 'check_settings'),
url(r'^tools_menu/$', 'blank_menu', (), 'tools_menu')
)

View File

@@ -64,8 +64,8 @@ def check_settings(request):
'hide_object':True,
'extra_columns':[
{'name':_(u'name'), 'attribute':'name'},
{'name':_(u'value'), 'attribute': lambda x: return_type(x['value'])},
{'name':_(u'exists'), 'attribute':lambda x: exists(x['value']) if 'exists' in x else ''},
{'name':_(u'value'), 'attribute': lambda x: _return_type(x['value'])},
{'name':_(u'exists'), 'attribute':lambda x: _exists(x['value']) if 'exists' in x else ''},
]
}
@@ -73,7 +73,7 @@ def check_settings(request):
context_instance=RequestContext(request))
def return_type(value):
def _return_type(value):
if isinstance(value, types.FunctionType):
return _(u'function found')
elif isinstance(value, types.ClassType):
@@ -85,8 +85,16 @@ def return_type(value):
else:
return value
def exists(path):
def _exists(path):
try:
return os.path.exists(path)
except Exception, exc:
return exc
def blank_menu(request):
return render_to_response('generic_template.html', {
'title':_(u'Tools menu'),
'content':_(u'"Find all duplicates": Search all the documents\' checksums and return a list of the exact matches.')
},
context_instance=RequestContext(request))