Merge branch 'master' of stigma.nowoczesnapolska.org.pl:platforma
[redakcja.git] / apps / api / handlers / library_handlers.py
index 45dff83..488c2d4 100644 (file)
@@ -1,4 +1,6 @@
 # -*- encoding: utf-8 -*-
+import os.path
+import logging
 
 __author__= "Ɓukasz Rekucki"
 __date__ = "$2009-09-25 15:49:50$"
@@ -6,83 +8,150 @@ __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 django.db import IntegrityError
 
-from wlrepo import MercurialLibrary, RevisionNotFound, DocumentAlreadyExists
-from librarian import dcparser
+import librarian
+import librarian.html
+from librarian import dcparser, parser
+
+from wlrepo import *
+from explorer.models import PullRequest, GalleryForDocument
 
+# internal imports
+import api.forms as forms
 import api.response as response
-from api.response import validate_form
+from api.utils import validate_form, hglibrary, natural_order
+from api.models import PartCache
+
+#
+import settings
+
+
+log = logging.getLogger('platforma.api')
+
 
 #
 # Document List Handlers
 #
+# TODO: security check
 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}
+        
 
+#
+# This handler controlls the document collection
+#
 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() ]
+        documents = {}
+        
+        for docid in lib.documents():            
+            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)
 
-        return {'documents' : document_list }
+        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
+
+            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']))
+            
+        return {'documents': sorted(document_tree.itervalues(),
+            key=natural_order(lambda d: d['name']) ) }
 
     @validate_form(forms.DocumentUploadForm, 'POST')
-    def create(self, request, form):
-        """Create a new document."""
-        lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
+    @hglibrary
+    def create(self, request, form, lib):
+        """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'].read().decode('utf-8')
+            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()))
 
         docid = form.cleaned_data['bookname']
-        try:
-            doc = lib.document_create(docid)
-            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 )
-                
+        try:
+            lock = lib.lock()            
+            try:
+                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(traceback.format_exc())
+
+                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:
+            import traceback
+            return response.InternalError().django_response({
+                "reason": traceback.format_exc()
+            })
         except 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
@@ -90,17 +159,18 @@ class LibraryHandler(BaseHandler):
 class BasicDocumentHandler(AnonymousBaseHandler):
     allowed_methods = ('GET',)
 
-    def read(self, request, docid):
-        try:
-            lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
+    @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]),
             'text_url': reverse('doctext_view', args=[doc.id]),
-            'dc_url': reverse('docdc_view', docid=doc.id),
+            'dc_url': reverse('docdc_view', args=[doc.id]),
             'public_revision': doc.revision,
         }
 
@@ -113,65 +183,177 @@ 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)
-        
+    @hglibrary
+    def read(self, request, docid, lib):
+        """Read document's meta data"""       
+        log.info(u"Read %s (%s)" % (docid, type(docid)) )
         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,
+                'docid': docid })
 
         # is_shared = udoc.ancestorof(doc)
         # is_uptodate = is_shared or shared.ancestorof(document)
 
         result = {
             'name': udoc.id,
+            'html_url': reverse('dochtml_view', args=[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]),
+            '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
 
+    @hglibrary
+    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, stylesheet='partial'):
+        """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, \
+                parse_dublincore=False, stylesheet=stylesheet,\
+                options={
+                    "with-paths": 'boolean(1)',                    
+                })
+                
+        except (EntryNotFound, RevisionNotFound), e:
+            return response.EntityNotFound().django_response({
+                'reason': 'not-found', 'message': e.message})
+        except librarian.ParseError, e:
+            return response.InternalError().django_response({
+                'reason': 'xml-parse-error', 'message': e.message })
+
+#
+# Image Gallery
+#
+
+class DocumentGalleryHandler(BaseHandler):
+    allowed_methods = ('GET')
+    
+    
+    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
+
+            gallery = {'name': assoc.name, 'pages': []}
+            
+            for file in os.listdir(dirpath):
+                if not isinstance(file, unicode):
+                    log.warn(u"File %r is gallery %r is not unicode. Ommiting."\
+                        % (file, dirpath) )
+                    continue
+                               
+                name, ext = os.path.splitext(os.path.basename(file))
+
+                if ext.lower() not in [u'.png', u'.jpeg', u'.jpg']:
+                    log.info(u"Ignoring: %s %s", name, ext)
+                    continue
+
+                url = settings.MEDIA_URL + assoc.subpath + u'/' + file;
+                gallery['pages'].append( quote(url.encode('utf-8')) )
+
+            gallery['pages'].sort()
+            galleries.append(gallery)
+
+        return galleries                      
+
 #
 # Document Text View
 #
+
+XINCLUDE_REGEXP = r"""<(?:\w+:)?include\s+[^>]*?href=("|')wlrepo://(?P<link>[^\1]+?)\1\s*[^>]*?>"""
+#
+#
+#
+
 class DocumentTextHandler(BaseHandler):
-    allowed_methods = ('GET', 'PUT')
+    allowed_methods = ('GET', 'POST')
 
-    @validate_form(forms.DocumentEntryRequest)
-    def read(self, request, form, docid):
-        """Read document as raw text"""        
-        lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
+    @hglibrary
+    def read(self, request, docid, lib):
+        """Read document as raw text"""
+        revision = request.GET.get('revision', 'latest')
+        part = request.GET.get('part', False)
         
         try:
-            if request.GET['revision'] == 'latest':
+            if revision == 'latest':
                 document = lib.document(docid)
             else:
-                document = lib.document_for_rev(request.GET['revision'])
+                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()
-
-    def update(self, request, docid):
-        lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
-        try:
-            data = request.PUT['contents']
-            prev = request.PUT['revision']
-
-            if request.PUT.has_key('message'):
-                msg = u"$USER$ " + request.PUT['message']
+            if part is False:
+                # we're done :)
+                return document.data('xml')
             else:
-                msg = u"$AUTO$ XML content update."
+                xdoc = parser.WLDocument.from_string(document.data('xml'),\
+                    parse_dublincore=False)
+                ptext = xdoc.part_as_text(part)
+
+                if ptext is None:
+                    return response.EntityNotFound().django_response({
+                      'reason': 'no-part-in-document'                     
+                    })
+
+                return ptext
+        except librarian.ParseError:
+            return response.EntityNotFound().django_response({
+                'reason': 'invalid-document-state',
+                'exception': type(e), 'message': e.message
+            })
+        except (EntryNotFound, RevisionNotFound), e:
+            return response.EntityNotFound().django_response({
+                'reason': 'not-found',
+                'exception': type(e), 'message': e.message
+            })   
+
+    @hglibrary
+    def create(self, request, docid, lib):
+        try:
+            revision = request.POST['revision']
 
             current = lib.document(docid, request.user.username)
-            orig = lib.document_for_rev(prev)
+            orig = lib.document_for_rev(revision)
 
             if current != orig:
                 return response.EntityConflict().django_response({
@@ -179,18 +361,85 @@ class DocumentTextHandler(BaseHandler):
                         "provided_revision": orig.revision,
                         "latest_revision": current.revision })
 
-            ndoc = doc.quickwrite('xml', data, msg)
+            if request.POST.has_key('message'):
+                msg = u"$USER$ " + request.POST['message']
+            else:
+                msg = u"$AUTO$ XML content update."
+
+            if request.POST.has_key('contents'):
+                data = request.POST['contents']
+            else:
+                if not request.POST.has_key('chunks'):
+                    # bad request
+                    return response.BadRequest().django_response({'reason': 'invalid-arguments',
+                        'message': 'No contents nor chunks specified.'})
+
+                    # TODO: validate
+                parts = json.loads(request.POST['chunks'])                    
+                xdoc = parser.WLDocument.from_string(current.data('xml'))
+                   
+                errors = xdoc.merge_chunks(parts)
+
+                if len(errors):
+                    return response.EntityConflict().django_response({
+                            "reason": "invalid-chunks",
+                            "message": "Unable to merge following parts into the document: %s " % ",".join(errors)
+                    })
+
+                data = xdoc.serialize()
+
+            # try to find any Xinclude tags
+            includes = [m.groupdict()['link'] for m in (re.finditer(\
+                XINCLUDE_REGEXP, data, flags=re.UNICODE) or []) ]
+
+            log.info("INCLUDES: %s", 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
@@ -198,35 +447,43 @@ class DocumentTextHandler(BaseHandler):
 # @requires librarian
 #
 class DocumentDublinCoreHandler(BaseHandler):
-    allowed_methods = ('GET', 'PUT')
+    allowed_methods = ('GET', 'POST')
 
-    @validate_form(forms.DocumentEntryRequest)
-    def read(self, request, docid):
-        """Read document as raw text"""
-        lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
+    @hglibrary
+    def read(self, request, docid, lib):
+        """Read document as raw text"""        
         try:
-            if request.GET['revision'] == 'latest':
-                document = lib.document(docid)
+            revision = request.GET.get('revision', 'latest')
+
+            if revision == 'latest':
+                doc = lib.document(docid)
             else:
-                document = lib.document_for_rev(request.GET['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})
 
-    def update(self, request, docid):
-        lib = MercurialLibrary(path=settings.REPOSITORY_PATH)
+    @hglibrary
+    def create(self, request, docid, lib):
         try:
-            bi_json = request.PUT['contents']
-            prev = request.PUT['revision']
-            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."
 
             current = lib.document(docid, request.user.username)
-            orig = lib.document_for_rev(prev)
+            orig = lib.document_for_rev(revision)
 
             if current != orig:
                 return response.EntityConflict().django_response({
@@ -242,11 +499,107 @@ 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, '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 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':        
+            if not request.user.has_perm('explorer.document.can_share'):
+                # 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(
+                        source_revision = str(udoc.revision),
+                        defaults = {
+                            'comitter': request.user,
+                            'document': docid,
+                            'status': "N",
+                            'comment': form.cleaned_data['message'] or '$AUTO$ Document shared.',
+                        }
+                    )
+
+                    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'
+                    })
+            else:
+                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()
+
+        nudoc = udoc.latest()
+
+        return response.SuccessAllOk().django_response({
+            "name": nudoc.id,
+            "parent_user_resivion": udoc.revision,
+            "parent_revision": doc.revision,
+            "revision": nudoc.revision,
+            'timestamp': nudoc.revision.timestamp,
+        })