package.py

import logging
from urllib import urlencode
import datetime

from pylons import config
from genshi.template import MarkupTemplate
from genshi.template.text import NewTextTemplate
from paste.deploy.converters import asbool

import ckan.logic as logic
import ckan.lib.base as base
import ckan.lib.maintain as maintain
import ckan.lib.package_saver as package_saver
import ckan.lib.i18n as i18n
import ckan.lib.navl.dictization_functions as dict_fns
import ckan.lib.accept as accept
import ckan.lib.helpers as h
import ckan.model as model
import ckan.lib.datapreview as datapreview
import ckan.lib.plugins
import ckan.plugins as p

from ckan.common import OrderedDict, _, json, request, c, g, response
from home import CACHE_PARAMETERS

log = logging.getLogger(__name__)

render = base.render
abort = base.abort
redirect = base.redirect

NotFound = logic.NotFound
NotAuthorized = logic.NotAuthorized
ValidationError = logic.ValidationError
check_access = logic.check_access
get_action = logic.get_action
tuplize_dict = logic.tuplize_dict
clean_dict = logic.clean_dict
parse_params = logic.parse_params
flatten_to_string_key = logic.flatten_to_string_key

lookup_package_plugin = ckan.lib.plugins.lookup_package_plugin


def _encode_params(params):
    return [(k, v.encode('utf-8') if isinstance(v, basestring) else str(v))
            for k, v in params]


def url_with_params(url, params):
    params = _encode_params(params)
    return url + u'?' + urlencode(params)


def search_url(params, package_type=None):
    if not package_type or package_type == 'dataset':
        url = h.url_for(controller='package', action='search')
    else:
        url = h.url_for('{0}_search'.format(package_type))
    return url_with_params(url, params)


class PackageController(base.BaseController):

    def _package_form(self, package_type=None):
        return lookup_package_plugin(package_type).package_form()

    def _setup_template_variables(self, context, data_dict, package_type=None):
        return lookup_package_plugin(package_type).\
            setup_template_variables(context, data_dict)

    def _new_template(self, package_type):
        return lookup_package_plugin(package_type).new_template()

    def _edit_template(self, package_type):
        return lookup_package_plugin(package_type).edit_template()

    def _search_template(self, package_type):
        return lookup_package_plugin(package_type).search_template()

    def _read_template(self, package_type):
        return lookup_package_plugin(package_type).read_template()

    def _history_template(self, package_type):
        return lookup_package_plugin(package_type).history_template()

    def _guess_package_type(self, expecting_name=False):
        """
            Guess the type of package from the URL handling the case
            where there is a prefix on the URL (such as /data/package)
        """

        # Special case: if the rot URL '/' has been redirected to the package
        # controller (e.g. by an IRoutes extension) then there's nothing to do
        # here.
        if request.path == '/':
            return 'dataset'

        parts = [x for x in request.path.split('/') if x]

        idx = -1
        if expecting_name:
            idx = -2

        pt = parts[idx]
        if pt == 'package':
            pt = 'dataset'

        return pt

    def search(self):
        from ckan.lib.search import SearchError

        package_type = self._guess_package_type()

        try:
            context = {'model': model, 'user': c.user or c.author}
            check_access('site_read', context)
        except NotAuthorized:
            abort(401, _('Not authorized to see this page'))

        # unicode format (decoded from utf8)
        q = c.q = request.params.get('q', u'')
        c.query_error = False
        try:
            page = int(request.params.get('page', 1))
        except ValueError, e:
            abort(400, ('"page" parameter must be an integer'))
        limit = g.datasets_per_page

        # most search operations should reset the page counter:
        params_nopage = [(k, v) for k, v in request.params.items()
                         if k != 'page']

        def drill_down_url(alternative_url=None, **by):
            return h.add_url_param(alternative_url=alternative_url,
                                   controller='package', action='search',
                                   new_params=by)

        c.drill_down_url = drill_down_url

        def remove_field(key, value=None, replace=None):
            return h.remove_url_param(key, value=value, replace=replace,
                                  controller='package', action='search')

        c.remove_field = remove_field

        sort_by = request.params.get('sort', None)
        params_nosort = [(k, v) for k, v in params_nopage if k != 'sort']

        def _sort_by(fields):
            """
            Sort by the given list of fields.

            Each entry in the list is a 2-tuple: (fieldname, sort_order)

            eg - [('metadata_modified', 'desc'), ('name', 'asc')]

            If fields is empty, then the default ordering is used.
            """
            params = params_nosort[:]

            if fields:
                sort_string = ', '.join('%s %s' % f for f in fields)
                params.append(('sort', sort_string))
            return search_url(params, package_type)

        c.sort_by = _sort_by
        if sort_by is None:
            c.sort_by_fields = []
        else:
            c.sort_by_fields = [field.split()[0]
                                for field in sort_by.split(',')]

        def pager_url(q=None, page=None):
            params = list(params_nopage)
            params.append(('page', page))
            return search_url(params, package_type)

        c.search_url_params = urlencode(_encode_params(params_nopage))

        try:
            c.fields = []
            # c.fields_grouped will contain a dict of params containing
            # a list of values eg {'tags':['tag1', 'tag2']}
            c.fields_grouped = {}
            search_extras = {}
            fq = ''
            for (param, value) in request.params.items():
                if param not in ['q', 'page', 'sort'] \
                        and len(value) and not param.startswith('_'):
                    if not param.startswith('ext_'):
                        c.fields.append((param, value))
                        fq += ' %s:"%s"' % (param, value)
                        if param not in c.fields_grouped:
                            c.fields_grouped[param] = [value]
                        else:
                            c.fields_grouped[param].append(value)
                    else:
                        search_extras[param] = value

            context = {'model': model, 'session': model.Session,
                       'user': c.user or c.author, 'for_view': True}

            if package_type and package_type != 'dataset':
                # Only show datasets of this particular type
                fq += ' +dataset_type:{type}'.format(type=package_type)
            else:
                # Unless changed via config options, don't show non standard
                # dataset types on the default search page
                if not asbool(config.get('ckan.search.show_all_types', 'False')):
                    fq += ' +dataset_type:dataset'

            facets = OrderedDict()

            default_facet_titles = {
                    'organization': _('Organizations'),
                    'groups': _('Groups'),
                    'tags': _('Tags'),
                    'res_format': _('Formats'),
                    'license_id': _('License'),
                    }

            for facet in g.facets:
                if facet in default_facet_titles:
                    facets[facet] = default_facet_titles[facet]
                else:
                    facets[facet] = facet

            # Facet titles
            for plugin in p.PluginImplementations(p.IFacets):
                facets = plugin.dataset_facets(facets, package_type)

            c.facet_titles = facets

            data_dict = {
                'q': q,
                'fq': fq.strip(),
                'facet.field': facets.keys(),
                'rows': limit,
                'start': (page - 1) * limit,
                'sort': sort_by,
                'extras': search_extras
            }

            query = get_action('package_search')(context, data_dict)
            c.sort_by_selected = query['sort']

            c.page = h.Page(
                collection=query['results'],
                page=page,
                url=pager_url,
                item_count=query['count'],
                items_per_page=limit
            )
            c.facets = query['facets']
            c.search_facets = query['search_facets']
            c.page.items = query['results']
        except SearchError, se:
            log.error('Dataset search error: %r', se.args)
            c.query_error = True
            c.facets = {}
            c.search_facets = {}
            c.page = h.Page(collection=[])
        c.search_facets_limits = {}
        for facet in c.search_facets.keys():
            try:
                limit = int(request.params.get('_%s_limit' % facet,
                                               g.facets_default_number))
            except ValueError:
                abort(400, _('Parameter "{parameter_name}" is not '
                             'an integer').format(
                                 parameter_name='_%s_limit' % facet
                             ))
            c.search_facets_limits[facet] = limit

        maintain.deprecate_context_item(
          'facets',
          'Use `c.search_facets` instead.')

        self._setup_template_variables(context, {},
                                       package_type=package_type)

        return render(self._search_template(package_type))

    def _content_type_from_extension(self, ext):
        ct, mu, ext = accept.parse_extension(ext)
        if not ct:
            return None, None, None,
        return ct, ext, (NewTextTemplate, MarkupTemplate)[mu]

    def _content_type_from_accept(self):
        """
        Given a requested format this method determines the content-type
        to set and the genshi template loader to use in order to render
        it accurately.  TextTemplate must be used for non-xml templates
        whilst all that are some sort of XML should use MarkupTemplate.
        """
        ct, mu, ext = accept.parse_header(request.headers.get('Accept', ''))
        return ct, ext, (NewTextTemplate, MarkupTemplate)[mu]

    def read(self, id, format='html'):
        if not format == 'html':
            ctype, extension, loader = \
                self._content_type_from_extension(format)
            if not ctype:
                # An unknown format, we'll carry on in case it is a
                # revision specifier and re-constitute the original id
                id = "%s.%s" % (id, format)
                ctype, format, loader = "text/html; charset=utf-8", "html", \
                    MarkupTemplate
        else:
            ctype, format, loader = self._content_type_from_accept()

        response.headers['Content-Type'] = ctype

        package_type = self._get_package_type(id.split('@')[0])
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author, 'for_view': True}
        data_dict = {'id': id}

        # interpret @<revision_id> or @<date> suffix
        split = id.split('@')
        if len(split) == 2:
            data_dict['id'], revision_ref = split
            if model.is_id(revision_ref):
                context['revision_id'] = revision_ref
            else:
                try:
                    date = h.date_str_to_datetime(revision_ref)
                    context['revision_date'] = date
                except TypeError, e:
                    abort(400, _('Invalid revision format: %r') % e.args)
                except ValueError, e:
                    abort(400, _('Invalid revision format: %r') % e.args)
        elif len(split) > 2:
            abort(400, _('Invalid revision format: %r') %
                  'Too many "@" symbols')

        # check if package exists
        try:
            c.pkg_dict = get_action('package_show')(context, data_dict)
            c.pkg = context['package']
        except NotFound:
            abort(404, _('Dataset not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % id)

        # used by disqus plugin
        c.current_package_id = c.pkg.id
        c.related_count = c.pkg.related_count

        # can the resources be previewed?
        for resource in c.pkg_dict['resources']:
            resource['can_be_previewed'] = self._resource_preview(
                {'resource': resource, 'package': c.pkg_dict})

        self._setup_template_variables(context, {'id': id},
                                       package_type=package_type)

        package_saver.PackageSaver().render_package(c.pkg_dict, context)

        template = self._read_template(package_type)
        template = template[:template.index('.') + 1] + format

        return render(template, loader_class=loader)

    def history(self, id):
        package_type = self._get_package_type(id.split('@')[0])

        if 'diff' in request.params or 'selected1' in request.params:
            try:
                params = {'id': request.params.getone('pkg_name'),
                          'diff': request.params.getone('selected1'),
                          'oldid': request.params.getone('selected2'),
                          }
            except KeyError, e:
                if 'pkg_name' in dict(request.params):
                    id = request.params.getone('pkg_name')
                c.error = \
                    _('Select two revisions before doing the comparison.')
            else:
                params['diff_entity'] = 'package'
                h.redirect_to(controller='revision', action='diff', **params)

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}
        data_dict = {'id': id}
        try:
            c.pkg_dict = get_action('package_show')(context, data_dict)
            c.pkg_revisions = get_action('package_revision_list')(context,
                                                                  data_dict)
            # TODO: remove
            # Still necessary for the authz check in group/layout.html
            c.pkg = context['package']

        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % '')
        except NotFound:
            abort(404, _('Dataset not found'))

        format = request.params.get('format', '')
        if format == 'atom':
            # Generate and return Atom 1.0 document.
            from webhelpers.feedgenerator import Atom1Feed
            feed = Atom1Feed(
                title=_(u'CKAN Dataset Revision History'),
                link=h.url_for(controller='revision', action='read',
                               id=c.pkg_dict['name']),
                description=_(u'Recent changes to CKAN Dataset: ') +
                (c.pkg_dict['title'] or ''),
                language=unicode(i18n.get_lang()),
            )
            for revision_dict in c.pkg_revisions:
                revision_date = h.date_str_to_datetime(
                    revision_dict['timestamp'])
                try:
                    dayHorizon = int(request.params.get('days'))
                except:
                    dayHorizon = 30
                dayAge = (datetime.datetime.now() - revision_date).days
                if dayAge >= dayHorizon:
                    break
                if revision_dict['message']:
                    item_title = u'%s' % revision_dict['message'].\
                        split('\n')[0]
                else:
                    item_title = u'%s' % revision_dict['id']
                item_link = h.url_for(controller='revision', action='read',
                                      id=revision_dict['id'])
                item_description = _('Log message: ')
                item_description += '%s' % (revision_dict['message'] or '')
                item_author_name = revision_dict['author']
                item_pubdate = revision_date
                feed.add_item(
                    title=item_title,
                    link=item_link,
                    description=item_description,
                    author_name=item_author_name,
                    pubdate=item_pubdate,
                )
            feed.content_type = 'application/atom+xml'
            return feed.writeString('utf-8')

        c.related_count = c.pkg.related_count
        return render(self._history_template(c.pkg_dict.get('type',
                                                            package_type)))

    def new(self, data=None, errors=None, error_summary=None):
        package_type = self._guess_package_type(True)

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author,
                   'save': 'save' in request.params}

        # Package needs to have a organization group in the call to
        # check_access and also to save it
        try:
            check_access('package_create', context)
        except NotAuthorized:
            abort(401, _('Unauthorized to create a package'))

        if context['save'] and not data:
            return self._save_new(context, package_type=package_type)

        data = data or clean_dict(dict_fns.unflatten(tuplize_dict(parse_params(
            request.params, ignore_keys=CACHE_PARAMETERS))))
        c.resources_json = h.json.dumps(data.get('resources', []))
        # convert tags if not supplied in data
        if data and not data.get('tag_string'):
            data['tag_string'] = ', '.join(
                h.dict_list_reduce(data.get('tags', {}), 'name'))

        errors = errors or {}
        error_summary = error_summary or {}
        # in the phased add dataset we need to know that
        # we have already completed stage 1
        stage = ['active']
        if data.get('state') == 'draft':
            stage = ['active', 'complete']
        elif data.get('state') == 'draft-complete':
            stage = ['active', 'complete', 'complete']

        # if we are creating from a group then this allows the group to be
        # set automatically
        data['group_id'] = request.params.get('group') or \
            request.params.get('groups__0__id')

        vars = {'data': data, 'errors': errors,
                'error_summary': error_summary,
                'action': 'new', 'stage': stage}
        c.errors_json = h.json.dumps(errors)

        self._setup_template_variables(context, {},
                                       package_type=package_type)

        # TODO: This check is to maintain backwards compatibility with the
        # old way of creating custom forms. This behaviour is now deprecated.
        if hasattr(self, 'package_form'):
            c.form = render(self.package_form, extra_vars=vars)
        else:
            c.form = render(self._package_form(package_type=package_type),
                            extra_vars=vars)
        return render(self._new_template(package_type),
                      extra_vars={'stage': stage})

    def resource_edit(self, id, resource_id, data=None, errors=None,
                      error_summary=None):
        if request.method == 'POST' and not data:
            data = data or clean_dict(dict_fns.unflatten(tuplize_dict(parse_params(
                request.POST))))
            # we don't want to include save as it is part of the form
            del data['save']

            context = {'model': model, 'session': model.Session,
                       'api_version': 3,
                       'user': c.user or c.author}

            data['package_id'] = id
            try:
                if resource_id:
                    data['id'] = resource_id
                    get_action('resource_update')(context, data)
                else:
                    get_action('resource_create')(context, data)
            except ValidationError, e:
                errors = e.error_dict
                error_summary = e.error_summary
                return self.resource_edit(id, resource_id, data,
                                          errors, error_summary)
            except NotAuthorized:
                abort(401, _('Unauthorized to edit this resource'))
            redirect(h.url_for(controller='package', action='resource_read',
                               id=id, resource_id=resource_id))

        context = {'model': model, 'session': model.Session,
                   'api_version': 3,
                   'user': c.user or c.author,}
        pkg_dict = get_action('package_show')(context, {'id': id})
        if pkg_dict['state'].startswith('draft'):
            # dataset has not yet been fully created
            resource_dict = get_action('resource_show')(context, {'id': resource_id})
            fields = ['url', 'url_type', 'resource_type', 'format', 'name', 'description', 'id']
            data = {}
            for field in fields:
                data[field] = resource_dict[field]
            return self.new_resource(id, data=data)
        # resource is fully created
        try:
            resource_dict = get_action('resource_show')(context, {'id': resource_id})
        except NotFound:
            abort(404, _('Resource not found'))
        c.pkg_dict = pkg_dict
        c.resource = resource_dict
        # set the form action
        c.form_action = h.url_for(controller='package',
                                  action='resource_edit',
                                  resource_id=resource_id,
                                  id=id)
        if not data:
            data = resource_dict

        errors = errors or {}
        error_summary = error_summary or {}
        vars = {'data': data, 'errors': errors,
                'error_summary': error_summary, 'action': 'new'}
        return render('package/resource_edit.html', extra_vars=vars)

    def new_resource(self, id, data=None, errors=None, error_summary=None):
        '''New resources are created before resource upload in the pending
        state which will be made active by the edit action'''
        data = {'package_id': id, 'url': 'pending', 'state': 'pending'}
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}
        resource = get_action('resource_create')(context, data)
        redirect(h.url_for(controller='package', action='resource_edit',
            resource_id=resource.get('id'), id=id))

    def new_metadata(self, id, data=None, errors=None, error_summary=None):
        ''' FIXME: This is a temporary action to allow styling of the
        forms. '''
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        if request.method == 'POST' and not data:
            save_action = request.params.get('save')
            data = data or clean_dict(dict_fns.unflatten(tuplize_dict(parse_params(
                request.POST))))
            # we don't want to include save as it is part of the form
            del data['save']

            data_dict = get_action('package_show')(context, {'id': id})

            data_dict['id'] = id
            # update the state
            if save_action == 'finish':
                # we want this to go live when saved
                data_dict['state'] = 'active'
            elif save_action in ['go-resources', 'go-dataset']:
                data_dict['state'] = 'draft-complete'
            # allow the state to be changed
            context['allow_state_change'] = True
            data_dict.update(data)
            try:
                get_action('package_update')(context, data_dict)
            except ValidationError, e:
                errors = e.error_dict
                error_summary = e.error_summary
                return self.new_metadata(id, data, errors, error_summary)
            except NotAuthorized:
                abort(401, _('Unauthorized to update dataset'))
            if save_action == 'go-resources':
                # we want to go back to the add resources form stage
                redirect(h.url_for(controller='package',
                                   action='new_resource', id=id))
            elif save_action == 'go-dataset':
                # we want to go back to the add dataset stage
                redirect(h.url_for(controller='package',
                                   action='edit', id=id))

            redirect(h.url_for(controller='package', action='read', id=id))

        if not data:
            data = get_action('package_show')(context, {'id': id})
        errors = errors or {}
        error_summary = error_summary or {}
        vars = {'data': data, 'errors': errors, 'error_summary': error_summary}
        vars['pkg_name'] = id

        package_type = self._get_package_type(id)
        self._setup_template_variables(context, {},
                                       package_type=package_type)

        return render('package/new_package_metadata.html', extra_vars=vars)

    def edit(self, id, data=None, errors=None, error_summary=None):
        package_type = self._get_package_type(id)
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author,
                   'save': 'save' in request.params,
                   'moderated': config.get('moderated'),
                   'pending': True}

        if context['save'] and not data:
            return self._save_edit(id, context, package_type=package_type)
        try:
            c.pkg_dict = get_action('package_show')(context, {'id': id})
            context['for_edit'] = True
            old_data = get_action('package_show')(context, {'id': id})
            # old data is from the database and data is passed from the
            # user if there is a validation error. Use users data if there.
            if data:
                old_data.update(data)
            data = old_data
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % '')
        except NotFound:
            abort(404, _('Dataset not found'))
        # are we doing a multiphase add?
        if data.get('state', '').startswith('draft'):
            c.form_action = h.url_for(controller='package', action='new')
            c.form_style = 'new'
            return self.new(data=data, errors=errors,
                            error_summary=error_summary)

        c.pkg = context.get("package")
        c.resources_json = h.json.dumps(data.get('resources', []))

        try:
            check_access('package_update', context)
        except NotAuthorized, e:
            abort(401, _('User %r not authorized to edit %s') % (c.user, id))
        # convert tags if not supplied in data
        if data and not data.get('tag_string'):
            data['tag_string'] = ', '.join(h.dict_list_reduce(
                c.pkg_dict.get('tags', {}), 'name'))
        errors = errors or {}
        vars = {'data': data, 'errors': errors,
                'error_summary': error_summary, 'action': 'edit'}
        c.errors_json = h.json.dumps(errors)

        self._setup_template_variables(context, {'id': id},
                                       package_type=package_type)
        c.related_count = c.pkg.related_count

        # we have already completed stage 1
        vars['stage'] = ['active']
        if data.get('state') == 'draft':
            vars['stage'] = ['active', 'complete']
        elif data.get('state') == 'draft-complete':
            vars['stage'] = ['active', 'complete', 'complete']

        # TODO: This check is to maintain backwards compatibility with the
        # old way of creating custom forms. This behaviour is now deprecated.
        if hasattr(self, 'package_form'):
            c.form = render(self.package_form, extra_vars=vars)
        else:
            c.form = render(self._package_form(package_type=package_type),
                            extra_vars=vars)

        return render(self._edit_template(package_type),
                      extra_vars={'stage': vars['stage']})

    def read_ajax(self, id, revision=None):
        package_type = self._get_package_type(id)
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author,
                   'revision_id': revision}
        try:
            data = get_action('package_show')(context, {'id': id})
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % '')
        except NotFound:
            abort(404, _('Dataset not found'))

        data.pop('tags')
        data = flatten_to_string_key(data)
        response.headers['Content-Type'] = 'application/json;charset=utf-8'
        return h.json.dumps(data)

    def history_ajax(self, id):

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}
        data_dict = {'id': id}
        try:
            pkg_revisions = get_action('package_revision_list')(
                context, data_dict)
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % '')
        except NotFound:
            abort(404, _('Dataset not found'))

        data = []
        approved = False
        for num, revision in enumerate(pkg_revisions):
            if not approved and revision['approved_timestamp']:
                current_approved, approved = True, True
            else:
                current_approved = False

            data.append({'revision_id': revision['id'],
                         'message': revision['message'],
                         'timestamp': revision['timestamp'],
                         'author': revision['author'],
                         'approved': bool(revision['approved_timestamp']),
                         'current_approved': current_approved})

        response.headers['Content-Type'] = 'application/json;charset=utf-8'
        return h.json.dumps(data)

    def _get_package_type(self, id):
        """
        Given the id of a package it determines the plugin to load
        based on the package's type name (type). The plugin found
        will be returned, or None if there is no plugin associated with
        the type.
        """
        pkg = model.Package.get(id)
        if pkg:
            return pkg.type or 'dataset'
        return None

    def _tag_string_to_list(self, tag_string):
        ''' This is used to change tags from a sting to a list of dicts '''
        out = []
        for tag in tag_string.split(','):
            tag = tag.strip()
            if tag:
                out.append({'name': tag,
                            'state': 'active'})
        return out

    def _save_new(self, context, package_type=None):
        # The staged add dataset used the new functionality when the dataset is
        # partially created so we need to know if we actually are updating or
        # this is a real new.
        is_an_update = False
        ckan_phase = request.params.get('_ckan_phase')
        from ckan.lib.search import SearchIndexError
        try:
            data_dict = clean_dict(dict_fns.unflatten(
                tuplize_dict(parse_params(request.POST))))
            if ckan_phase:
                # prevent clearing of groups etc
                context['allow_partial_update'] = True
                # sort the tags
                data_dict['tags'] = self._tag_string_to_list(
                    data_dict['tag_string'])
                if data_dict.get('pkg_name'):
                    is_an_update = True
                    # This is actually an update not a save
                    data_dict['id'] = data_dict['pkg_name']
                    del data_dict['pkg_name']
                    # this is actually an edit not a save
                    pkg_dict = get_action('package_update')(context, data_dict)

                    if request.params['save'] == 'go-metadata':
                        # redirect to add metadata
                        url = h.url_for(controller='package',
                                        action='new_metadata',
                                        id=pkg_dict['name'])
                    else:
                        # redirect to add dataset resources
                        url = h.url_for(controller='package',
                                        action='new_resource',
                                        id=pkg_dict['name'])
                    redirect(url)
                # Make sure we don't index this dataset
                if request.params['save'] not in ['go-resource', 'go-metadata']:
                    data_dict['state'] = 'draft'
                # allow the state to be changed
                context['allow_state_change'] = True

            data_dict['type'] = package_type
            context['message'] = data_dict.get('log_message', '')
            pkg_dict = get_action('package_create')(context, data_dict)

            if ckan_phase:
                # redirect to add dataset resources
                url = h.url_for(controller='package',
                                action='new_resource',
                                id=pkg_dict['name'])
                redirect(url)

            self._form_save_redirect(pkg_dict['name'], 'new', package_type=package_type)
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % '')
        except NotFound, e:
            abort(404, _('Dataset not found'))
        except dict_fns.DataError:
            abort(400, _(u'Integrity Error'))
        except SearchIndexError, e:
            try:
                exc_str = unicode(repr(e.args))
            except Exception:  # We don't like bare excepts
                exc_str = unicode(str(e))
            abort(500, _(u'Unable to add package to search index.') + exc_str)
        except ValidationError, e:
            errors = e.error_dict
            error_summary = e.error_summary
            if is_an_update:
                # we need to get the state of the dataset to show the stage we
                # are on.
                pkg_dict = get_action('package_show')(context, data_dict)
                data_dict['state'] = pkg_dict['state']
                return self.edit(data_dict['id'], data_dict,
                                 errors, error_summary)
            data_dict['state'] = 'none'
            return self.new(data_dict, errors, error_summary)

    def _save_edit(self, name_or_id, context, package_type=None):
        from ckan.lib.search import SearchIndexError
        log.debug('Package save request name: %s POST: %r',
                  name_or_id, request.POST)
        try:
            data_dict = clean_dict(dict_fns.unflatten(
                tuplize_dict(parse_params(request.POST))))
            if '_ckan_phase' in data_dict:
                # we allow partial updates to not destroy existing resources
                context['allow_partial_update'] = True
                data_dict['tags'] = self._tag_string_to_list(
                    data_dict['tag_string'])
                del data_dict['_ckan_phase']
                del data_dict['save']
            context['message'] = data_dict.get('log_message', '')
            if not context['moderated']:
                context['pending'] = False
            data_dict['id'] = name_or_id
            pkg = get_action('package_update')(context, data_dict)
            if request.params.get('save', '') == 'Approve':
                get_action('make_latest_pending_package_active')(
                    context, data_dict)
            c.pkg = context['package']
            c.pkg_dict = pkg

            self._form_save_redirect(pkg['name'], 'edit', package_type=package_type)
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % id)
        except NotFound, e:
            abort(404, _('Dataset not found'))
        except dict_fns.DataError:
            abort(400, _(u'Integrity Error'))
        except SearchIndexError, e:
            try:
                exc_str = unicode(repr(e.args))
            except Exception:  # We don't like bare excepts
                exc_str = unicode(str(e))
            abort(500, _(u'Unable to update search index.') + exc_str)
        except ValidationError, e:
            errors = e.error_dict
            error_summary = e.error_summary
            return self.edit(name_or_id, data_dict, errors, error_summary)

    def _form_save_redirect(self, pkgname, action, package_type=None):
        '''This redirects the user to the CKAN package/read page,
        unless there is request parameter giving an alternate location,
        perhaps an external website.
        @param pkgname - Name of the package just edited
        @param action - What the action of the edit was
        '''
        assert action in ('new', 'edit')
        url = request.params.get('return_to') or \
            config.get('package_%s_return_url' % action)
        if url:
            url = url.replace('<NAME>', pkgname)
        else:
            if package_type is None or package_type == 'dataset':
                url = h.url_for(controller='package', action='read', id=pkgname)
            else:
                url = h.url_for('{0}_read'.format(package_type), id=pkgname)
        redirect(url)

    def _adjust_license_id_options(self, pkg, fs):
        options = fs.license_id.render_opts['options']
        is_included = False
        for option in options:
            license_id = option[1]
            if license_id == pkg.license_id:
                is_included = True
        if not is_included:
            options.insert(1, (pkg.license_id, pkg.license_id))

    def delete(self, id):

        if 'cancel' in request.params:
            h.redirect_to(controller='package', action='edit', id=id)

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        try:
            check_access('package_delete', context, {'id': id})
        except NotAuthorized:
            abort(401, _('Unauthorized to delete package %s') % '')

        try:
            if request.method == 'POST':
                get_action('package_delete')(context, {'id': id})
                h.flash_notice(_('Dataset has been deleted.'))
                h.redirect_to(controller='package', action='search')
            c.pkg_dict = get_action('package_show')(context, {'id': id})
        except NotAuthorized:
            abort(401, _('Unauthorized to delete package %s') % '')
        except NotFound:
            abort(404, _('Dataset not found'))
        return render('package/confirm_delete.html')

    def resource_delete(self, id, resource_id):

        if 'cancel' in request.params:
            h.redirect_to(controller='package', action='resource_edit', resource_id=resource_id, id=id)

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        try:
            check_access('package_delete', context, {'id': id})
        except NotAuthorized:
            abort(401, _('Unauthorized to delete package %s') % '')

        try:
            if request.method == 'POST':
                get_action('resource_delete')(context, {'id': resource_id})
                h.flash_notice(_('Resource has been deleted.'))
                h.redirect_to(controller='package', action='read', id=id)
            c.resource_dict = get_action('resource_show')(context, {'id': resource_id})
            c.pkg_id = id
        except NotAuthorized:
            abort(401, _('Unauthorized to delete resource %s') % '')
        except NotFound:
            abort(404, _('Resource not found'))
        return render('package/confirm_delete_resource.html')

    def autocomplete(self):
        # DEPRECATED in favour of /api/2/util/dataset/autocomplete
        q = unicode(request.params.get('q', ''))
        if not len(q):
            return ''

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        data_dict = {'q': q}
        packages = get_action('package_autocomplete')(context, data_dict)

        pkg_list = []
        for pkg in packages:
            pkg_list.append('%s|%s' % (pkg['match_displayed'].
                                       replace('|', ' '), pkg['name']))
        return '\n'.join(pkg_list)

    def _render_edit_form(self, fs, params={}, clear_session=False):
        # errors arrive in c.error and fs.errors
        c.log_message = params.get('log_message', '')
        # rgrp: expunge everything from session before dealing with
        # validation errors) so we don't have any problematic saves
        # when the fs.render causes a flush.
        # seb: If the session is *expunged*, then the form can't be
        # rendered; I've settled with a rollback for now, which isn't
        # necessarily what's wanted here.
        # dread: I think this only happened with tags because until
        # this changeset, Tag objects were created in the Renderer
        # every time you hit preview. So I don't believe we need to
        # clear the session any more. Just in case I'm leaving it in
        # with the log comments to find out.
        if clear_session:
            # log to see if clearing the session is ever required
            if model.Session.new or model.Session.dirty or \
                    model.Session.deleted:
                log.warn('Expunging session changes which were not expected: '
                         '%r %r %r', (model.Session.new, model.Session.dirty,
                                      model.Session.deleted))
            try:
                model.Session.rollback()
            except AttributeError:
                # older SQLAlchemy versions
                model.Session.clear()
        edit_form_html = fs.render()
        c.form = h.literal(edit_form_html)
        return h.literal(render('package/edit_form.html'))

    def _update_authz(self, fs):
        validation = fs.validate()
        if not validation:
            c.form = self._render_edit_form(fs, request.params)
            raise package_saver.ValidationException(fs)
        try:
            fs.sync()
        except Exception, inst:
            model.Session.rollback()
            raise
        else:
            model.Session.commit()

    def resource_read(self, id, resource_id):
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        try:
            c.resource = get_action('resource_show')(context,
                                                     {'id': resource_id})
            c.package = get_action('package_show')(context, {'id': id})
            # required for nav menu
            c.pkg = context['package']
            c.pkg_dict = c.package
        except NotFound:
            abort(404, _('Resource not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read resource %s') % id)
        # get package license info
        license_id = c.package.get('license_id')
        try:
            c.package['isopen'] = model.Package.\
                get_license_register()[license_id].isopen()
        except KeyError:
            c.package['isopen'] = False

        # TODO: find a nicer way of doing this
        c.datastore_api = '%s/api/action' % config.get('ckan.site_url', '').rstrip('/')

        c.related_count = c.pkg.related_count

        c.resource['can_be_previewed'] = self._resource_preview(
            {'resource': c.resource, 'package': c.package})
        return render('package/resource_read.html')

    def _resource_preview(self, data_dict):
        return bool(datapreview.res_format(data_dict['resource'])
                    in datapreview.direct() + datapreview.loadable()
                    or datapreview.get_preview_plugin(
                        data_dict, return_first=True))

    def resource_download(self, id, resource_id):
        """
        Provides a direct download by redirecting the user to the url stored
        against this resource.
        """
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        try:
            rsc = get_action('resource_show')(context, {'id': resource_id})
            pkg = get_action('package_show')(context, {'id': id})
        except NotFound:
            abort(404, _('Resource not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read resource %s') % id)

        if not 'url' in rsc:
            abort(404, _('No download is available'))
        redirect(rsc['url'])

    def follow(self, id):
        '''Start following this dataset.'''
        context = {'model': model,
                   'session': model.Session,
                   'user': c.user or c.author}
        data_dict = {'id': id}
        try:
            get_action('follow_dataset')(context, data_dict)
            package_dict = get_action('package_show')(context, data_dict)
            h.flash_success(_("You are now following {0}").format(
                package_dict['title']))
        except ValidationError as e:
            error_message = (e.extra_msg or e.message or e.error_summary
                    or e.error_dict)
            h.flash_error(error_message)
        except NotAuthorized as e:
            h.flash_error(e.extra_msg)
        h.redirect_to(controller='package', action='read', id=id)

    def unfollow(self, id):
        '''Stop following this dataset.'''
        context = {'model': model,
                   'session': model.Session,
                   'user': c.user or c.author}
        data_dict = {'id': id}
        try:
            get_action('unfollow_dataset')(context, data_dict)
            package_dict = get_action('package_show')(context, data_dict)
            h.flash_success(_("You are no longer following {0}").format(
                package_dict['title']))
        except ValidationError as e:
            error_message = (e.extra_msg or e.message or e.error_summary
                    or e.error_dict)
            h.flash_error(error_message)
        except (NotFound, NotAuthorized) as e:
            error_message = e.extra_msg or e.message
            h.flash_error(error_message)
        h.redirect_to(controller='package', action='read', id=id)

    def followers(self, id=None):
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author, 'for_view': True}
        data_dict = {'id': id}
        try:
            c.pkg_dict = get_action('package_show')(context, data_dict)
            c.pkg = context['package']
            c.followers = get_action('dataset_follower_list')(context,
                    {'id': c.pkg_dict['id']})

            c.related_count = c.pkg.related_count
        except NotFound:
            abort(404, _('Dataset not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read package %s') % id)

        return render('package/followers.html')

    def activity(self, id):
        '''Render this package's public activity stream page.'''

        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author, 'for_view': True}
        data_dict = {'id': id}
        try:
            c.pkg_dict = get_action('package_show')(context, data_dict)
            c.pkg = context['package']
            c.package_activity_stream = get_action(
                    'package_activity_list_html')(context,
                            {'id': c.pkg_dict['id']})
            c.related_count = c.pkg.related_count
        except NotFound:
            abort(404, _('Dataset not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read dataset %s') % id)

        return render('package/activity.html')

    def resource_embedded_dataviewer(self, id, resource_id,
                                     width=500, height=500):
        """
        Embeded page for a read-only resource dataview. Allows
        for width and height to be specified as part of the
        querystring (as well as accepting them via routes).
        """
        context = {'model': model, 'session': model.Session,
                   'user': c.user or c.author}

        try:
            c.resource = get_action('resource_show')(context,
                                                     {'id': resource_id})
            c.package = get_action('package_show')(context, {'id': id})
            c.resource_json = h.json.dumps(c.resource)

            # double check that the resource belongs to the specified package
            if not c.resource['id'] in [r['id']
                                        for r in c.package['resources']]:
                raise NotFound

        except NotFound:
            abort(404, _('Resource not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read resource %s') % id)

        # Construct the recline state
        state_version = int(request.params.get('state_version', '1'))
        recline_state = self._parse_recline_state(request.params)
        if recline_state is None:
            abort(400, ('"state" parameter must be a valid recline '
                        'state (version %d)' % state_version))

        c.recline_state = h.json.dumps(recline_state)

        c.width = max(int(request.params.get('width', width)), 100)
        c.height = max(int(request.params.get('height', height)), 100)
        c.embedded = True

        return render('package/resource_embedded_dataviewer.html')

    def _parse_recline_state(self, params):
        state_version = int(request.params.get('state_version', '1'))
        if state_version != 1:
            return None

        recline_state = {}
        for k, v in request.params.items():
            try:
                v = h.json.loads(v)
            except ValueError:
                pass
            recline_state[k] = v

        recline_state.pop('width', None)
        recline_state.pop('height', None)
        recline_state['readOnly'] = True

        # previous versions of recline setup used elasticsearch_url attribute
        # for data api url - see http://trac.ckan.org/ticket/2639
        # fix by relocating this to url attribute which is the default location
        if 'dataset' in recline_state and 'elasticsearch_url' in recline_state['dataset']:
            recline_state['dataset']['url'] = recline_state['dataset']['elasticsearch_url']

        # Ensure only the currentView is available
        # default to grid view if none specified
        if not recline_state.get('currentView', None):
            recline_state['currentView'] = 'grid'
        for k in recline_state.keys():
            if k.startswith('view-') and \
                    not k.endswith(recline_state['currentView']):
                recline_state.pop(k)
        return recline_state

    def resource_datapreview(self, id, resource_id):
        '''
        Embeded page for a resource data-preview.

        Depending on the type, different previews are loaded.  This could be an
        img tag where the image is loaded directly or an iframe that embeds a
        webpage, recline or a pdf preview.
        '''
        context = {
            'model': model,
            'session': model.Session,
            'user': c.user or c.author
        }

        try:
            c.resource = get_action('resource_show')(context,
                                                     {'id': resource_id})
            c.package = get_action('package_show')(context, {'id': id})

            data_dict = {'resource': c.resource, 'package': c.package}

            preview_plugin = datapreview.get_preview_plugin(data_dict)

            if preview_plugin is None:
                abort(409, _('No preview has been defined.'))

            preview_plugin.setup_template_variables(context, data_dict)
            c.resource_json = json.dumps(c.resource)
        except NotFound:
            abort(404, _('Resource not found'))
        except NotAuthorized:
            abort(401, _('Unauthorized to read resource %s') % id)
        else:
            return render(preview_plugin.preview_template(context, data_dict))