X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/0b7e02cf796af7d3028efabeb37473f9198d0af2..5ffbe263839acf003494225f41a3f3c95af797b7:/apps/api/handlers/library_handlers.py diff --git a/apps/api/handlers/library_handlers.py b/apps/api/handlers/library_handlers.py index 06513522..b2ff94b4 100644 --- a/apps/api/handlers/library_handlers.py +++ b/apps/api/handlers/library_handlers.py @@ -1,3 +1,4 @@ +import os.path # -*- encoding: utf-8 -*- __author__= "Łukasz Rekucki" @@ -6,19 +7,27 @@ __doc__ = "Module documentation." from piston.handler import BaseHandler, AnonymousBaseHandler - -import settings -import librarian -import api.forms as forms +import re from datetime import date from django.core.urlresolvers import reverse +from django.utils import simplejson as json -from wlrepo import MercurialLibrary, RevisionNotFound, DocumentAlreadyExists +import librarian +import librarian.html from librarian import dcparser +from wlrepo import * +from explorer.models import PullRequest, GalleryForDocument + +# internal imports +import api.forms as forms import api.response as response -from api.utils import validate_form, hglibrary +from api.utils import validate_form, hglibrary, natural_order +from api.models import PartCache + +# +import settings # # Document List Handlers @@ -44,11 +53,42 @@ class LibraryHandler(BaseHandler): def read(self, request, lib): """Return the list of documents.""" - document_list = [{ - 'url': reverse('document_view', args=[docid]), - 'name': docid } for docid in lib.documents() ] + documents = {} + + for docid in lib.documents(): + docid = docid.decode('utf-8') + documents[docid] = { + 'url': reverse('document_view', args=[docid]), + 'name': docid, + 'parts': [] + } + + parts = PartCache.objects.defer('part_id')\ + .values_list('part_id', 'document_id').distinct() + + document_tree = dict(documents) + + for part, docid in parts: + # this way, we won't display broken links + if not documents.has_key(part): + print "NOT FOUND:", part + continue + + parent = documents[docid] + child = documents[part] - return {'documents' : document_list } + # not top-level anymore + document_tree.pop(part) + parent['parts'].append(child) + + # sort the right way + + + for doc in documents.itervalues(): + doc['parts'].sort(key=natural_order(lambda d: d['name'])) + + return {'documents': sorted(document_tree.itervalues(), + key=natural_order(lambda d: d['name']) ) } @validate_form(forms.DocumentUploadForm, 'POST') @hglibrary @@ -56,7 +96,7 @@ class LibraryHandler(BaseHandler): """Create a new document.""" if form.cleaned_data['ocr_data']: - data = form.cleaned_data['ocr_data'].encode('utf-8') + data = form.cleaned_data['ocr_data'] else: data = request.FILES['ocr_file'].read().decode('utf-8') @@ -64,20 +104,35 @@ class LibraryHandler(BaseHandler): data = librarian.wrap_text(data, unicode(date.today())) docid = form.cleaned_data['bookname'] + try: - doc = lib.document_create(docid) - doc = doc.quickwrite('xml', data, '$AUTO$ XML data uploaded.', - user=request.user.username) - - url = reverse('document_view', args=[doc.id]) - - return response.EntityCreated().django_response(\ - body = { - 'url': url, - 'name': doc.id, - 'revision': doc.revision }, - url = url ) - + lock = lib.lock() + try: + print "DOCID", docid + doc = lib.document_create(docid) + # document created, but no content yet + + try: + doc = doc.quickwrite('xml', data.encode('utf-8'), + '$AUTO$ XML data uploaded.', user=request.user.username) + except Exception,e: + # rollback branch creation + lib._rollback() + raise LibraryException("Exception occured:" + repr(e)) + + url = reverse('document_view', args=[doc.id]) + + return response.EntityCreated().django_response(\ + body = { + 'url': url, + 'name': doc.id, + 'revision': doc.revision }, + url = url ) + finally: + lock.release() + except LibraryException, e: + return response.InternalError().django_response(\ + {'exception': repr(e) }) except DocumentAlreadyExists: # Document is already there return response.EntityConflict().django_response(\ @@ -98,8 +153,9 @@ class BasicDocumentHandler(AnonymousBaseHandler): result = { 'name': doc.id, - 'text_url': reverse('doctext_view', args=[doc.id,doc.revision]), - 'dc_url': reverse('docdc_view', args=[doc.id,doc.revision]), + 'html_url': reverse('dochtml_view', args=[doc.id]), + 'text_url': reverse('doctext_view', args=[doc.id]), + 'dc_url': reverse('docdc_view', args=[doc.id]), 'public_revision': doc.revision, } @@ -118,18 +174,24 @@ class DocumentHandler(BaseHandler): try: doc = lib.document(docid) udoc = doc.take(request.user.username) - except RevisionNotFound: - return request.EnityNotFound().django_response() + except RevisionNotFound, e: + return response.EntityNotFound().django_response({ + 'exception': type(e), 'message': e.message}) # is_shared = udoc.ancestorof(doc) # is_uptodate = is_shared or shared.ancestorof(document) result = { 'name': udoc.id, - 'text_url': reverse('doctext_view', args=[doc.id,doc.revision]), - 'dc_url': reverse('docdc_view', args=[doc.id,doc.revision]), + 'html_url': reverse('dochtml_view', args=[udoc.id]), + 'text_url': reverse('doctext_view', args=[udoc.id]), + 'dc_url': reverse('docdc_view', args=[udoc.id]), + 'gallery_url': reverse('docgallery_view', args=[udoc.id]), + 'merge_url': reverse('docmerge_view', args=[udoc.id]), 'user_revision': udoc.revision, - 'public_revision': doc.revision, + 'user_timestamp': udoc.revision.timestamp, + 'public_revision': doc.revision, + 'public_timestamp': doc.revision.timestamp, } return result @@ -138,34 +200,108 @@ class DocumentHandler(BaseHandler): def update(self, request, docid, lib): """Update information about the document, like display not""" return +# +# +# +class DocumentHTMLHandler(BaseHandler): + allowed_methods = ('GET') + + @hglibrary + def read(self, request, docid, lib): + """Read document as html text""" + try: + revision = request.GET.get('revision', 'latest') + + if revision == 'latest': + document = lib.document(docid) + else: + document = lib.document_for_rev(revision) + + if document.id != docid: + return response.BadRequest().django_response({'reason': 'name-mismatch', + 'message': 'Provided revision refers, to document "%s", but provided "%s"' % (document.id, docid) }) + + return librarian.html.transform(document.data('xml'), is_file=False) + except (EntryNotFound, RevisionNotFound), e: + return response.EntityNotFound().django_response({ + 'exception': type(e), 'message': e.message}) + + +# +# Image Gallery +# +from django.core.files.storage import FileSystemStorage + +class DocumentGalleryHandler(BaseHandler): + allowed_methods = ('GET') + + def read(self, request, docid): + """Read meta-data about scans for gallery of this document.""" + galleries = [] + + for assoc in GalleryForDocument.objects.filter(document=docid): + dirpath = os.path.join(settings.MEDIA_ROOT, assoc.subpath) + + if not os.path.isdir(dirpath): + print u"[WARNING]: missing gallery %s" % dirpath + continue + + gallery = {'name': assoc.name, 'pages': []} + + for file in sorted(os.listdir(dirpath), key=natural_order()): + print file + name, ext = os.path.splitext(os.path.basename(file)) + + if ext.lower() not in ['.png', '.jpeg', '.jpg']: + print "Ignoring:", name, ext + continue + + url = settings.MEDIA_URL + assoc.subpath + u'/' + file.decode('utf-8'); + gallery['pages'].append(url) + + galleries.append(gallery) + + return galleries # # Document Text View # + +XINCLUDE_REGEXP = r"""<(?:\w+:)?include\s+[^>]*?href=("|')wlrepo://(?P[^\1]+?)\1\s*[^>]*?>""" +# +# +# class DocumentTextHandler(BaseHandler): - allowed_methods = ('GET', 'PUT') + allowed_methods = ('GET', 'POST') @hglibrary - def read(self, request, docid, revision, lib): - """Read document as raw text""" + def read(self, request, docid, lib): + """Read document as raw text""" + revision = request.GET.get('revision', 'latest') try: if revision == 'latest': document = lib.document(docid) else: document = lib.document_for_rev(revision) + + if document.id != docid: + return response.BadRequest().django_response({'reason': 'name-mismatch', + 'message': 'Provided revision is not valid for this document'}) # TODO: some finer-grained access control return document.data('xml') - except RevisionNotFound: - return response.EntityNotFound().django_response() + except (EntryNotFound, RevisionNotFound), e: + return response.EntityNotFound().django_response({ + 'exception': type(e), 'message': e.message}) @hglibrary - def update(self, request, docid, revision, lib): + def create(self, request, docid, lib): try: - data = request.PUT['contents'] + data = request.POST['contents'] + revision = request.POST['revision'] - if request.PUT.has_key('message'): - msg = u"$USER$ " + request.PUT['message'] + if request.POST.has_key('message'): + msg = u"$USER$ " + request.POST['message'] else: msg = u"$AUTO$ XML content update." @@ -178,18 +314,58 @@ class DocumentTextHandler(BaseHandler): "provided_revision": orig.revision, "latest_revision": current.revision }) - ndoc = doc.quickwrite('xml', data, msg) + # try to find any Xinclude tags + includes = [m.groupdict()['link'] for m in (re.finditer(\ + XINCLUDE_REGEXP, data, flags=re.UNICODE) or []) ] + + print "INCLUDES: ", includes + + # TODO: provide useful routines to make this simpler + def xml_update_action(lib, resolve): + try: + f = lib._fileopen(resolve('parts'), 'r') + stored_includes = json.loads(f.read()) + f.close() + except: + stored_includes = [] + + if stored_includes != includes: + f = lib._fileopen(resolve('parts'), 'w+') + f.write(json.dumps(includes)) + f.close() + + lib._fileadd(resolve('parts')) + + # update the parts cache + PartCache.update_cache(docid, current.owner,\ + stored_includes, includes) + + # now that the parts are ok, write xml + f = lib._fileopen(resolve('xml'), 'w+') + f.write(data.encode('utf-8')) + f.close() + + ndoc = None + ndoc = current.invoke_and_commit(\ + xml_update_action, lambda d: (msg, current.owner) ) + + try: + # return the new revision number + return response.SuccessAllOk().django_response({ + "document": ndoc.id, + "subview": "xml", + "previous_revision": current.revision, + "revision": ndoc.revision, + 'timestamp': ndoc.revision.timestamp, + "url": reverse("doctext_view", args=[ndoc.id]) + }) + except Exception, e: + if ndoc: lib._rollback() + raise e + except RevisionNotFound, e: + return response.EntityNotFound(mimetype="text/plain").\ + django_response(e.message) - # return the new revision number - return { - "document": ndoc.id, - "subview": "xml", - "previous_revision": prev, - "updated_revision": ndoc.revision - } - - except (RevisionNotFound, KeyError): - return response.EntityNotFound().django_response() # # Dublin Core handlers @@ -197,27 +373,37 @@ class DocumentTextHandler(BaseHandler): # @requires librarian # class DocumentDublinCoreHandler(BaseHandler): - allowed_methods = ('GET', 'PUT') + allowed_methods = ('GET', 'POST') @hglibrary - def read(self, request, docid, revision, lib): + def read(self, request, docid, lib): """Read document as raw text""" try: + revision = request.GET.get('revision', 'latest') + if revision == 'latest': - document = lib.document(docid) + doc = lib.document(docid) else: - document = lib.document_for_rev(revision) + doc = lib.document_for_rev(revision) + + + if document.id != docid: + return response.BadRequest().django_response({'reason': 'name-mismatch', + 'message': 'Provided revision is not valid for this document'}) bookinfo = dcparser.BookInfo.from_string(doc.data('xml')) return bookinfo.serialize() - except RevisionNotFound: - return response.EntityNotFound().django_response() + except (EntryNotFound, RevisionNotFound), e: + return response.EntityNotFound().django_response({ + 'exception': type(e), 'message': e.message}) @hglibrary - def update(self, request, docid, revision, lib): + def create(self, request, docid, lib): try: - bi_json = request.PUT['contents'] - if request.PUT.has_key('message'): + bi_json = request.POST['contents'] + revision = request.POST['revision'] + + if request.POST.has_key('message'): msg = u"$USER$ " + request.PUT['message'] else: msg = u"$AUTO$ Dublin core update." @@ -239,26 +425,101 @@ class DocumentDublinCoreHandler(BaseHandler): document.serialize().encode('utf-8'),\ message=msg, user=request.user.username) - return { - "document": ndoc.id, - "subview": "xml", - "previous_revision": prev, - "updated_revision": ndoc.revision - } - except (RevisionNotFound, KeyError): + try: + # return the new revision number + return { + "document": ndoc.id, + "subview": "dc", + "previous_revision": current.revision, + "revision": ndoc.revision, + 'timestamp': ndoc.revision.timestamp, + "url": reverse("docdc_view", args=[ndoc.id]) + } + except Exception, e: + if ndoc: lib._rollback() + raise e + except RevisionNotFound: return response.EntityNotFound().django_response() - class MergeHandler(BaseHandler): allowed_methods = ('POST',) - @validate_form(forms.MergeRequestForm) + @validate_form(forms.MergeRequestForm, 'POST') @hglibrary def create(self, request, form, docid, lib): """Create a new document revision from the information provided by user""" - pass + target_rev = form.cleaned_data['target_revision'] - - - + doc = lib.document(docid) + udoc = doc.take(request.user.username) + + if target_rev == 'latest': + target_rev = udoc.revision + + if str(udoc.revision) != target_rev: + # user think doesn't know he has an old version + # of his own branch. + + # Updating is teorericly ok, but we need would + # have to force a refresh. Sharing may be not safe, + # 'cause it doesn't always result in update. + + # In other words, we can't lie about the resource's state + # So we should just yield and 'out-of-date' conflict + # and let the client ask again with updated info. + + # NOTE: this could result in a race condition, when there + # are 2 instances of the same user editing the same document. + # Instance "A" trying to update, and instance "B" always changing + # the document right before "A". The anwser to this problem is + # for the "A" to request a merge from 'latest' and then + # check the parent revisions in response, if he actually + # merge from where he thinks he should. If not, the client SHOULD + # update his internal state. + return response.EntityConflict().django_response({ + "reason": "out-of-date", + "provided": target_rev, + "latest": udoc.revision }) + + if not request.user.has_perm('explorer.book.can_share'): + # User is not permitted to make a merge, right away + # So we instead create a pull request in the database + prq = PullRequest( + comitter=request.user, + document=docid, + source_revision = str(udoc.revision), + status="N", + comment = form.cleaned_data['message'] or '$AUTO$ Document shared.' + ) + + prq.save() + return response.RequestAccepted().django_response(\ + ticket_status=prq.status, \ + ticket_uri=reverse("pullrequest_view", args=[prq.id]) ) + + if form.cleaned_data['type'] == 'update': + # update is always performed from the file branch + # to the user branch + success, changed = udoc.update(request.user.username) + + if form.cleaned_data['type'] == 'share': + success, changed = udoc.share(form.cleaned_data['message']) + + if not success: + return response.EntityConflict().django_response({ + 'reason': 'merge-failure', + }) + + if not changed: + return response.SuccessNoContent().django_response() + + new_udoc = udoc.latest() + + return response.SuccessAllOk().django_response({ + "name": udoc.id, + "parent_user_resivion": udoc.revision, + "parent_revision": doc.revision, + "revision": ndoc.revision, + 'timestamp': ndoc.revision.timestamp, + })