X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/352b8591bd1c7163835a6fa1db34d3e2861c1071..ad481d9335cb91f56adfffda3999284f37d37942:/apps/api/handlers/library_handlers.py diff --git a/apps/api/handlers/library_handlers.py b/apps/api/handlers/library_handlers.py index b47d41cc..ae61389a 100644 --- a/apps/api/handlers/library_handlers.py +++ b/apps/api/handlers/library_handlers.py @@ -5,71 +5,128 @@ __date__ = "$2009-09-25 15:49:50$" __doc__ = "Module documentation." from piston.handler import BaseHandler, AnonymousBaseHandler -from piston.utils import rc -import settings -import librarian -import api.forms as forms +import re from datetime import date from django.core.urlresolvers import reverse -from wlrepo import MercurialLibrary, RevisionNotFound +from django.utils import simplejson as json +import librarian +import librarian.html from librarian import dcparser +from wlrepo import RevisionNotFound, LibraryException, DocumentAlreadyExists +from explorer.models import PullRequest + +# internal imports +import api.forms as forms +import api.response as response +from api.utils import validate_form, hglibrary +from api.models import PartCache + # # Document List Handlers # class BasicLibraryHandler(AnonymousBaseHandler): allowed_methods = ('GET',) - def read(self, request): - """Return the list of documents.""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) - + @hglibrary + 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() ] return {'documents' : document_list} + class LibraryHandler(BaseHandler): allowed_methods = ('GET', 'POST') anonymous = BasicLibraryHandler - def read(self, request): + @hglibrary + def read(self, request, lib): """Return the list of documents.""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) - document_list = [{ - 'url': reverse('document_view', args=[docid]), - 'name': docid } for docid in lib.documents() ] - - return {'documents' : document_list } - - def create(self, request): - """Create a new document.""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) + documents = {} + + for docid in lib.documents(): + documents[docid] = { + 'url': reverse('document_view', args=[docid]), + 'name': docid, + 'parts': [] + } + + related = PartCache.objects.defer('part_id')\ + .values_list('part_id', 'document_id').distinct() + + for part, docid in related: + # this way, we won't display broken links + if not documents.has_key(part): + continue + + child = documents[part] + parent = documents[docid] + + if isinstance(parent, dict): # the parent is top-level + documents.pop(part) + parent['parts'].append(child) + documents[part] = child['parts'] + else: # not top-level + parent.append(child) + + return { + 'documents': [d for d in documents.itervalues() if isinstance(d, dict)] + } - form = forms.DocumentUploadForm(request.POST, request.FILES) - if not form.is_valid(): - return rc.BAD_REQUEST + @validate_form(forms.DocumentUploadForm, 'POST') + @hglibrary + def create(self, request, form, lib): + """Create a new document.""" - f = request.FILES['ocr'] - data = f.read().decode('utf-8') + if form.cleaned_data['ocr_data']: + data = form.cleaned_data['ocr_data'] + else: + data = request.FILES['ocr_file'].read().decode('utf-8') if form.cleaned_data['generate_dc']: data = librarian.wrap_text(data, unicode(date.today())) - # TODO: what if the file exists ? - doc = lib.document_create(form.cleaned_data['bookname']) - doc.quickwrite('xml', data, '$AUTO$ XML data uploaded.', - user=request.user.username) + docid = form.cleaned_data['bookname'] - return { - 'url': reverse('document_view', args=[doc.id]), - 'name': doc.id, - 'revision': doc.revision } + try: + 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(\ + {"reason": "Document %s already exists." % docid}) # # Document Handlers @@ -77,20 +134,19 @@ class LibraryHandler(BaseHandler): class BasicDocumentHandler(AnonymousBaseHandler): allowed_methods = ('GET',) - def read(self, request, docid): - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) - - opts = forms.DocumentGetForm(request.GET) - if not opts.is_valid(): - return rc.BAD_REQUEST - - doc = lib.document(docid) + @hglibrary + def read(self, request, docid, lib): + try: + doc = lib.document(docid) + except RevisionNotFound: + return rc.NOT_FOUND result = { 'name': doc.id, - 'text_url': reverse('doctext_view', args=[doc.id]), - 'dc_url': reverse('docdc_view', docid=doc.id), - 'latest_rev': doc.revision, + 'html_url': reverse('dochtml_view', args=[doc.id,doc.revision]), + 'text_url': reverse('doctext_view', args=[doc.id,doc.revision]), + 'dc_url': reverse('docdc_view', args=[doc.id,doc.revision]), + 'public_revision': doc.revision, } return result @@ -102,58 +158,83 @@ class DocumentHandler(BaseHandler): allowed_methods = ('GET', 'PUT') anonymous = BasicDocumentHandler - def read(self, request, docid): - """Read document's meta data""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) - - opts = forms.DocumentGetForm(request.GET) - if not opts.is_valid(): - return rc.BAD_REQUEST - + @hglibrary + def read(self, request, docid, lib): + """Read document's meta data""" try: doc = lib.document(docid) udoc = doc.take(request.user.username) except RevisionNotFound: - return rc.NOT_HERE + return request.EnityNotFound().django_response() # is_shared = udoc.ancestorof(doc) # is_uptodate = is_shared or shared.ancestorof(document) result = { 'name': udoc.id, - 'text_url': reverse('doctext_view', args=[udoc.id]), - 'dc_url': reverse('docdc_view', args=[udoc.id]), - 'parts_url': reverse('docparts_view', args=[udoc.id]), - 'latest_rev': udoc.revision, - 'latest_shared_rev': doc.revision, - # 'shared': is_shared, - # 'up_to_date': is_uptodate, - } - - #if request.GET.get('with_part', 'no') == 'yes': - # result['parts'] = document.parts() + 'html_url': reverse('dochtml_view', args=[udoc.id,udoc.revision]), + 'text_url': reverse('doctext_view', args=[udoc.id,udoc.revision]), + 'dc_url': reverse('docdc_view', args=[udoc.id,udoc.revision]), + 'user_revision': udoc.revision, + 'public_revision': doc.revision, + } return result + @hglibrary + def update(self, request, docid, lib): + """Update information about the document, like display not""" + return +# +# +# +class DocumentHTMLHandler(BaseHandler): + allowed_methods = ('GET', 'PUT') + + @hglibrary + def read(self, request, docid, revision, lib): + """Read document as html text""" + try: + if revision == 'latest': + document = lib.document(docid) + else: + document = lib.document_for_rev(revision) + + return librarian.html.transform(document.data('xml'), is_file=False) + except RevisionNotFound: + return response.EntityNotFound().django_response() + + + + # # Document Text View # + +XINCLUDE_REGEXP = r"""<(?:\w+:)?include\s+[^>]*?href=("|')wlrepo://(?P[^\1]+?)\1\s*[^>]*?>""" +# +# class DocumentTextHandler(BaseHandler): allowed_methods = ('GET', 'PUT') - def read(self, request, docid): - """Read document as raw text""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) + @hglibrary + def read(self, request, docid, revision, lib): + """Read document as raw text""" try: - return lib.document(docid, request.user.username).data('xml') + if revision == 'latest': + document = lib.document(docid) + else: + document = lib.document_for_rev(revision) + + # TODO: some finer-grained access control + return document.data('xml') except RevisionNotFound: - return rc.NOT_HERE + return response.EntityNotFound().django_response() - def update(self, request, docid): - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) + @hglibrary + def update(self, request, docid, revision, lib): try: - data = request.PUT['contents'] - prev = request.PUT['revision'] + data = request.PUT['contents'] if request.PUT.has_key('message'): msg = u"$USER$ " + request.PUT['message'] @@ -161,16 +242,57 @@ class DocumentTextHandler(BaseHandler): msg = u"$AUTO$ XML content update." current = lib.document(docid, request.user.username) - orig = lib.document_for_rev(prev) + orig = lib.document_for_rev(revision) if current != orig: - return rc.DUPLICATE_ENTRY - - doc.quickwrite('xml', data, msg) - - return rc.ALL_OK - except (RevisionNotFound, KeyError): - return rc.NOT_HERE + return response.EntityConflict().django_response({ + "reason": "out-of-date", + "provided_revision": orig.revision, + "latest_revision": current.revision }) + + # try to find any Xinclude tags + includes = [m.groupdict()['link'] for m in (re.finditer(\ + XINCLUDE_REGEXP, data, flags=re.UNICODE) or []) ] + + # 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() + + # 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) + f.close() + + ndoc = current.invoke_and_commit(\ + xml_update_action, lambda d: (msg, current.owner) ) + + try: + # return the new revision number + return { + "document": ndoc.id, + "subview": "xml", + "previous_revision": current.revision, + "updated_revision": ndoc.revision + } + except Exception, e: + lib._rollback() + raise e + except RevisionNotFound, e: + return response.EntityNotFound().django_response(e) # # Dublin Core handlers @@ -180,40 +302,138 @@ class DocumentTextHandler(BaseHandler): class DocumentDublinCoreHandler(BaseHandler): allowed_methods = ('GET', 'PUT') - def read(self, request, docid): - """Read document as raw text""" - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) + @hglibrary + def read(self, request, docid, revision, lib): + """Read document as raw text""" try: - doc = lib.document(docid, request.user.username).data('xml') - bookinfo = dcparser.BookInfo.from_string(doc.read()) - + if revision == 'latest': + doc = lib.document(docid) + else: + doc = lib.document_for_rev(revision) + + bookinfo = dcparser.BookInfo.from_string(doc.data('xml')) return bookinfo.serialize() except RevisionNotFound: - return rc.NOT_HERE + return response.EntityNotFound().django_response() - def update(self, request, docid): - lib = MercurialLibrary(path=settings.REPOSITORY_PATH) + @hglibrary + def update(self, request, docid, revision, lib): try: - bi_json = request.PUT['contents'] - prev = request.PUT['revision'] + bi_json = request.PUT['contents'] if request.PUT.has_key('message'): msg = u"$USER$ " + request.PUT['message'] else: msg = u"$AUTO$ Dublin core update." current = lib.document(docid, request.user.username) - orig = lib.document_for_rev(prev) + orig = lib.document_for_rev(revision) if current != orig: - return rc.DUPLICATE_ENTRY + return response.EntityConflict().django_response({ + "reason": "out-of-date", + "provided": orig.revision, + "latest": current.revision }) xmldoc = parser.WLDocument.from_string(current.data('xml')) document.book_info = dcparser.BookInfo.from_json(bi_json) # zapisz - current.quickwrite('xml', document.serialize().encode('utf-8'),\ + ndoc = current.quickwrite('xml', \ + document.serialize().encode('utf-8'),\ message=msg, user=request.user.username) - return rc.ALL_OK - except (RevisionNotFound, KeyError): - return rc.NOT_HERE + try: + # return the new revision number + return { + "document": ndoc.id, + "subview": "dc", + "previous_revision": current.revision, + "updated_revision": ndoc.revision + } + except Exception, e: + lib._rollback() + raise e + except RevisionNotFound: + return response.EntityNotFound().django_response() + + + +class MergeHandler(BaseHandler): + allowed_methods = ('POST',) + + @validate_form(forms.MergeRequestForm, 'POST') + @hglibrary + def create(self, request, form, docid, lib): + """Create a new document revision from the information provided by user""" + + 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['comment'] 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['comment']) + + if not success: + return response.EntityConflict().django_response() + + 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": udoc.revision, + }) \ No newline at end of file