X-Git-Url: https://git.mdrn.pl/redakcja.git/blobdiff_plain/9921992e940cf9c1cc53715098233e23665d8f94..a08a59835fa7872c69ec1a662aa7945473eeaf2b:/apps/api/handlers/library_handlers.py diff --git a/apps/api/handlers/library_handlers.py b/apps/api/handlers/library_handlers.py old mode 100644 new mode 100755 index 5696cc95..e2d986ee --- a/apps/api/handlers/library_handlers.py +++ b/apps/api/handlers/library_handlers.py @@ -1,33 +1,69 @@ # -*- encoding: utf-8 -*- +import os.path -__author__= "Łukasz Rekucki" -__date__ = "$2009-09-25 15:49:50$" -__doc__ = "Module documentation." +import logging +log = logging.getLogger('platforma.api.library') from piston.handler import BaseHandler, AnonymousBaseHandler +from piston.utils import rc -import re from datetime import date from django.core.urlresolvers import reverse -from django.utils import simplejson as json +from django.db import IntegrityError import librarian import librarian.html -from librarian import dcparser +import difflib +import wlrepo -from wlrepo import RevisionNotFound, LibraryException, DocumentAlreadyExists -from explorer.models import PullRequest +from explorer.models import GalleryForDocument # internal imports import api.forms as forms import api.response as response -from api.utils import validate_form, hglibrary -from api.models import PartCache +from api.utils import validate_form, hglibrary, natural_order +from api.models import PartCache, PullRequest + +from pygments import highlight +from pygments.lexers import DiffLexer +from pygments.formatters import HtmlFormatter + +# +import settings + + +def is_prq(username): + return username.startswith('$prq-') + +def prq_for_user(username): + try: + return PullRequest.objects.get(id=int(username[5:])) + except: + return None + +def check_user(request, user): + log.info("user: %r, perm: %r" % (request.user, request.user.get_all_permissions()) ) + #pull request + if is_prq(user): + if not request.user.has_perm('api.view_prq'): + yield response.AccessDenied().django_response({ + 'reason': 'access-denied', + 'message': "You don't have enough priviliges to view pull requests." + }) + # other users + elif request.user.username != user: + if not request.user.has_perm('api.view_other_document'): + yield response.AccessDenied().django_response({ + 'reason': 'access-denied', + 'message': "You don't have enough priviliges to view other people's document." + }) + pass # # Document List Handlers # +# TODO: security check class BasicLibraryHandler(AnonymousBaseHandler): allowed_methods = ('GET',) @@ -37,10 +73,11 @@ class BasicLibraryHandler(AnonymousBaseHandler): document_list = [{ 'url': reverse('document_view', args=[docid]), 'name': docid } for docid in lib.documents() ] - return {'documents' : document_list} - +# +# This handler controlls the document collection +# class LibraryHandler(BaseHandler): allowed_methods = ('GET', 'POST') anonymous = BasicLibraryHandler @@ -51,34 +88,37 @@ class LibraryHandler(BaseHandler): documents = {} - for docid in lib.documents(): + for docid in lib.documents(): documents[docid] = { 'url': reverse('document_view', args=[docid]), 'name': docid, 'parts': [] } - related = PartCache.objects.defer('part_id')\ + parts = PartCache.objects.defer('part_id')\ .values_list('part_id', 'document_id').distinct() + + document_tree = dict(documents) - for part, docid in related: + for part, docid in parts: # this way, we won't display broken links if not documents.has_key(part): + log.info("NOT FOUND: %s", part) continue - child = documents[part] parent = documents[docid] + child = documents[part] + + # not top-level anymore + document_tree.pop(part) + parent['parts'].append(child) + + for doc in documents.itervalues(): + doc['parts'].sort(key=natural_order(lambda d: d['name'])) - 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)] - } + return {'documents': sorted(document_tree.itervalues(), + key=natural_order(lambda d: d['name']) ) } + @validate_form(forms.DocumentUploadForm, 'POST') @hglibrary @@ -90,6 +130,9 @@ class LibraryHandler(BaseHandler): else: data = request.FILES['ocr_file'].read().decode('utf-8') + if data is None: + return response.BadRequest().django_response('You must pass ocr_data or ocr_file.') + if form.cleaned_data['generate_dc']: data = librarian.wrap_text(data, unicode(date.today())) @@ -98,17 +141,17 @@ class LibraryHandler(BaseHandler): try: lock = lib.lock() try: - print "DOCID", docid + log.info("DOCID %s", 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: + import traceback # rollback branch creation lib._rollback() - raise LibraryException("Exception occured:" + repr(e)) + raise wlrepo.LibraryException(traceback.format_exc()) url = reverse('document_view', args=[doc.id]) @@ -120,247 +163,312 @@ class LibraryHandler(BaseHandler): url = url ) finally: lock.release() - except LibraryException, e: - return response.InternalError().django_response(\ - {'exception': repr(e) }) - except DocumentAlreadyExists: + except wlrepo.LibraryException, e: + import traceback + return response.InternalError().django_response({ + "reason": traceback.format_exc() + }) + except wlrepo.DocumentAlreadyExists: # Document is already there - return response.EntityConflict().django_response(\ - {"reason": "Document %s already exists." % docid}) + return response.EntityConflict().django_response({ + "reason": "already-exists", + "message": "Document already exists." % docid + }) # # Document Handlers # -class BasicDocumentHandler(AnonymousBaseHandler): - allowed_methods = ('GET',) +class DiffHandler(BaseHandler): + allowed_methods = ('GET',) + @hglibrary def read(self, request, docid, lib): - try: - doc = lib.document(docid) - except RevisionNotFound: - return rc.NOT_FOUND - - result = { - 'name': doc.id, - '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 diff between source_revision and target_revision)''' + revision = request.GET.get('revision') + if not revision: + return '' + source_document = lib.document(docid) + target_document = lib.document_for_revision(revision) + print source_document, target_document + + diff = difflib.unified_diff( + source_document.data('xml').splitlines(True), + target_document.data('xml').splitlines(True), + 'source', + 'target') + + s = ''.join(list(diff)) + return highlight(s, DiffLexer(), HtmlFormatter(cssclass="pastie")) - return result # # Document Meta Data # class DocumentHandler(BaseHandler): allowed_methods = ('GET', 'PUT') - anonymous = BasicDocumentHandler + @validate_form(forms.DocumentRetrieveForm, 'GET') @hglibrary - def read(self, request, docid, lib): + def read(self, request, form, docid, lib): """Read document's meta data""" - try: - doc = lib.document(docid) - udoc = doc.take(request.user.username) - except RevisionNotFound: - return request.EnityNotFound().django_response() - - # is_shared = udoc.ancestorof(doc) - # is_uptodate = is_shared or shared.ancestorof(document) + log.info(u"User '%s' wants to edit %s(%s) as %s" % \ + (request.user.username, docid, form.cleaned_data['revision'], form.cleaned_data['user']) ) - result = { - 'name': udoc.id, - '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, - } + user = form.cleaned_data['user'] or request.user.username + rev = form.cleaned_data['revision'] or 'latest' - return result + for error in check_user(request, user): + return error + + try: + doc = lib.document(docid, user, rev=rev) + except wlrepo.RevisionMismatch, e: + # the document exists, but the revision is bad + return response.EntityNotFound().django_response({ + 'reason': 'revision-mismatch', + 'message': e.message, + 'docid': docid, + 'user': user, + }) + except wlrepo.RevisionNotFound, e: + # the user doesn't have this document checked out + # or some other weird error occured + # try to do the checkout + try: + if user == request.user.username: + mdoc = lib.document(docid) + doc = mdoc.take(user) + elif is_prq(user): + prq = prq_for_user(user) + # commiter's document + prq_doc = lib.document_for_revision(prq.source_revision) + doc = prq_doc.take(user) + else: + return response.EntityNotFound().django_response({ + 'reason': 'document-not-found', + 'message': e.message, + 'docid': docid, + 'user': user, + }) + except wlrepo.RevisionNotFound, e: + return response.EntityNotFound().django_response({ + 'reason': 'document-not-found', + 'message': e.message, + 'docid': docid, + 'user': user + }) - @hglibrary - def update(self, request, docid, lib): - """Update information about the document, like display not""" - return + return { + 'name': doc.id, + 'user': user, + 'html_url': reverse('dochtml_view', args=[doc.id]), + 'text_url': reverse('doctext_view', args=[doc.id]), + # 'dc_url': reverse('docdc_view', args=[doc.id]), + 'gallery_url': reverse('docgallery_view', args=[doc.id]), + 'merge_url': reverse('docmerge_view', args=[doc.id]), + 'revision': doc.revision, + 'timestamp': doc.revision.timestamp, + # 'public_revision': doc.revision, + # 'public_timestamp': doc.revision.timestamp, + } + + +# @hglibrary +# def update(self, request, docid, lib): +# """Update information about the document, like display not""" +# return # # # class DocumentHTMLHandler(BaseHandler): - allowed_methods = ('GET', 'PUT') + allowed_methods = ('GET') + @validate_form(forms.DocumentRetrieveForm, 'GET') @hglibrary - def read(self, request, docid, revision, lib): + def read(self, request, form, docid, lib, stylesheet='full'): """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() + revision = form.cleaned_data['revision'] + user = form.cleaned_data['user'] or request.user.username + document = lib.document_for_revision(revision) + + if document.id != docid: + return response.BadRequest().django_response({ + 'reason': 'name-mismatch', + 'message': 'Provided revision is not valid for this document' + }) + if document.owner != user: + return response.BadRequest().django_response({ + 'reason': 'user-mismatch', + 'message': "Provided revision doesn't belong to user %s" % user + }) + for error in check_user(request, user): + return error + return librarian.html.transform(document.data('xml'), is_file=False, \ + parse_dublincore=False, stylesheet=stylesheet,\ + options={ + "with-paths": 'boolean(1)', + }) + + except (wlrepo.EntryNotFound, wlrepo.RevisionNotFound), e: + return response.EntityNotFound().django_response({ + 'reason': 'not-found', 'message': e.message}) + except librarian.ValidationError, e: + return response.InternalError().django_response({ + 'reason': 'xml-non-valid', 'message': e.message or u''}) + except librarian.ParseError, e: + return response.InternalError().django_response({ + 'reason': 'xml-parse-error', 'message': e.message or u'' }) # -# Document Text View +# Image Gallery # -XINCLUDE_REGEXP = r"""<(?:\w+:)?include\s+[^>]*?href=("|')wlrepo://(?P[^\1]+?)\1\s*[^>]*?>""" -# -# -class DocumentTextHandler(BaseHandler): - allowed_methods = ('GET', 'PUT') +class DocumentGalleryHandler(BaseHandler): + allowed_methods = ('GET', 'POST') + + + def read(self, request, docid): + """Read meta-data about scans for gallery of this document.""" + galleries = [] + from urllib import quote + + for assoc in GalleryForDocument.objects.filter(document=docid): + dirpath = os.path.join(settings.MEDIA_ROOT, assoc.subpath) + + if not os.path.isdir(dirpath): + log.warn(u"[WARNING]: missing gallery %s", dirpath) + continue - @hglibrary - def read(self, request, docid, revision, lib): - """Read document as raw text""" - try: - if revision == 'latest': - document = lib.document(docid) - else: - document = lib.document_for_rev(revision) + gallery = {'name': assoc.name, 'pages': []} - # TODO: some finer-grained access control - return document.data('xml') - except RevisionNotFound: - return response.EntityNotFound().django_response() - - @hglibrary - def update(self, request, docid, revision, lib): - try: - data = request.PUT['contents'] - - if request.PUT.has_key('message'): - msg = u"$USER$ " + request.PUT['message'] - else: - msg = u"$AUTO$ XML content update." - - current = lib.document(docid, request.user.username) - orig = lib.document_for_rev(revision) - - if current != orig: - 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 = [] + for file in os.listdir(dirpath): + if not isinstance(file, unicode): + try: + file = file.decode('utf-8') + except: + log.warn(u"File %r in gallery %r is not unicode. Ommiting."\ + % (file, dirpath) ) + file = None + + if file is not None: + name, ext = os.path.splitext(os.path.basename(file)) + + if ext.lower() not in [u'.png', u'.jpeg', u'.jpg']: + log.warn(u"Ignoring: %s %s", name, ext) + url = None + + url = settings.MEDIA_URL + assoc.subpath + u'/' + file - if stored_includes != includes: - f = lib._fileopen(resolve('parts'), 'w+') - f.write(json.dumps(includes)) - f.close() + if url is None: + url = settings.MEDIA_URL + u'/missing.png' + + gallery['pages'].append( quote(url.encode('utf-8')) ) - lib._fileadd(resolve('parts')) + gallery['pages'].sort() + galleries.append(gallery) - # update the parts cache - PartCache.update_cache(docid, current.owner,\ - stored_includes, includes) + return galleries - # now that the parts are ok, write xml - f = lib._fileopen(resolve('xml'), 'w+') - f.write(data.encode('utf-8')) - f.close() + def create(self, request, docid): + if not request.user.is_superuser: + return rc.FORBIDDEN + + new_path = request.POST.get('path') + + if new_path: + gallery, created = GalleryForDocument.objects.get_or_create( + document = docid, + defaults = { + 'subpath': new_path, + } + ) - ndoc = None - ndoc = current.invoke_and_commit(\ - xml_update_action, lambda d: (msg, current.owner) ) + if not created: + gallery.subpath = new_path + gallery.save() - try: - # return the new revision number - return response.SuccessAllOk().django_response({ - "document": ndoc.id, - "subview": "xml", - "previous_revision": current.revision, - "updated_revision": ndoc.revision, - "url": reverse("doctext_view", args=[ndoc.id, ndoc.revision]) - }) - except Exception, e: - if ndoc: lib._rollback() - raise e - except RevisionNotFound, e: - return response.EntityNotFound().django_response(e) + return rc.CREATED + + return rc.BAD_REQUEST # # Dublin Core handlers # # @requires librarian # -class DocumentDublinCoreHandler(BaseHandler): - allowed_methods = ('GET', 'PUT') - - @hglibrary - def read(self, request, docid, revision, lib): - """Read document as raw text""" - try: - 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 response.EntityNotFound().django_response() - - @hglibrary - def update(self, request, docid, revision, lib): - try: - 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(revision) - - if current != orig: - 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 - ndoc = current.quickwrite('xml', \ - document.serialize().encode('utf-8'),\ - message=msg, user=request.user.username) - - 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 DocumentDublinCoreHandler(BaseHandler): +# allowed_methods = ('GET', 'POST') +# +# @hglibrary +# def read(self, request, docid, lib): +# """Read document as raw text""" +# try: +# revision = request.GET.get('revision', 'latest') +# +# if revision == 'latest': +# doc = lib.document(docid) +# else: +# doc = lib.document_for_revision(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 (EntryNotFound, RevisionNotFound), e: +# return response.EntityNotFound().django_response({ +# 'exception': type(e), 'message': e.message}) +# +# @hglibrary +# def create(self, request, docid, lib): +# try: +# 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." +# +# current = lib.document(docid, request.user.username) +# orig = lib.document_for_revision(revision) +# +# if current != orig: +# 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 +# ndoc = current.quickwrite('xml', \ +# document.serialize().encode('utf-8'),\ +# message=msg, user=request.user.username) +# +# 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',) @@ -369,75 +477,125 @@ class MergeHandler(BaseHandler): @hglibrary def create(self, request, form, docid, lib): """Create a new document revision from the information provided by user""" + try: + revision = form.cleaned_data['revision'] - target_rev = form.cleaned_data['target_revision'] - - doc = lib.document(docid) - udoc = doc.take(request.user.username) + # fetch the main branch document + doc = lib.document(docid) - if target_rev == 'latest': - target_rev = udoc.revision + # fetch the base document + user_doc = lib.document_for_revision(revision) + base_doc = user_doc.latest() - 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({ + if base_doc != user_doc: + 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() + "provided": str(user_doc.revision), + "latest": str(base_doc.revision) + }) - return response.SuccessAllOk().django_response({ - "name": udoc.id, - "parent_user_resivion": udoc.revision, - "parent_revision": doc.revision, - "revision": udoc.revision, - }) + if form.cleaned_data['type'] == 'update': + # update is always performed from the file branch + # to the user branch + user_doc_new = base_doc.update(request.user.username) + + if user_doc_new == user_doc: + return response.SuccessAllOk().django_response({ + "result": "no-op" + }) + + # shared document is the same + doc_new = doc + + if form.cleaned_data['type'] == 'share': + if not base_doc.up_to_date(): + return response.BadRequest().django_response({ + "reason": "not-fast-forward", + "message": "You must first update your branch to the latest version." + }) + + anwser, info = base_doc.would_share() + + if not anwser: + return response.SuccessAllOk().django_response({ + "result": "no-op", "message": info + }) + + # check for unresolved conflicts + if base_doc.has_conflict_marks(): + return response.BadRequest().django_response({ + "reason": "unresolved-conflicts", + "message": "There are unresolved conflicts in your file. Fix them, and try again." + }) + + if not request.user.has_perm('api.share_document'): + # User is not permitted to make a merge, right away + # So we instead create a pull request in the database + try: + prq, created = PullRequest.objects.get_or_create( + comitter = request.user, + document = docid, + status = "N", + defaults = { + 'source_revision': str(base_doc.revision), + 'comment': form.cleaned_data['message'] or '$AUTO$ Document shared.', + } + ) + + # there can't be 2 pending request from same user + # for the same document + if not created: + prq.source_revision = str(base_doc.revision) + prq.comment = prq.comment + 'u\n\n' + (form.cleaned_data['message'] or u'') + prq.save() + + return response.RequestAccepted().django_response(\ + ticket_status=prq.status, \ + ticket_uri=reverse("pullrequest_view", args=[prq.id]) ) + except IntegrityError: + return response.EntityConflict().django_response({ + 'reason': 'request-already-exist' + }) + + changed = base_doc.share(form.cleaned_data['message']) + + # update shared version if needed + if changed: + doc_new = doc.latest() + else: + doc_new = doc + + # the user wersion is the same + user_doc_new = base_doc + + # The client can compare parent_revision to revision + # to see if he needs to update user's view + # Same goes for shared view + + return response.SuccessAllOk().django_response({ + "result": "success", + "name": user_doc_new.id, + "user": user_doc_new.owner, + + "revision": user_doc_new.revision, + 'timestamp': user_doc_new.revision.timestamp, + + "parent_revision": user_doc.revision, + "parent_timestamp": user_doc.revision.timestamp, + + "shared_revision": doc_new.revision, + "shared_timestamp": doc_new.revision.timestamp, + + "shared_parent_revision": doc.revision, + "shared_parent_timestamp": doc.revision.timestamp, + }) + except wlrepo.OutdatedException, e: + return response.BadRequest().django_response({ + "reason": "not-fast-forward", + "message": e.message + }) + except wlrepo.LibraryException, e: + return response.InternalError().django_response({ + "reason": "merge-error", + "message": e.message + })