opencultureconsulting
/
openrefine-client
mirror of https://github.com/opencultureconsulting/openrefine-client.git
2
0
Fork 0
Code Issues Releases Activity

Remove find-project-by-name functionality as it doesn't really work. Defer project name discovering 'til it's needed.

This commit is contained in:
Paul Makepeace 2011-04-26 02:27:22 -04:00
parent 51d2294878
commit 27586bc10c
1 changed files with 22 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
google/refine/refine.py
View File

@ -76,34 +76,26 @@ class Refine:
return self.server.urlopen_json('get-version') return self.server.urlopen_json('get-version')
def list_projects(self): def list_projects(self):
"""Return a dict of projects indexed by id & name. """Return a dict of projects indexed by id.
{u'1877818633188': { {u'1877818633188': {
'id': u'1877818633188', u'name': u'akg', 'id': u'1877818633188', u'name': u'akg',
u'modified': u'2011-04-07T12:30:07Z', u'modified': u'2011-04-07T12:30:07Z',
u'created': u'2011-04-07T12:30:07Z' u'created': u'2011-04-07T12:30:07Z'
}, },
{u'akg': { ... } } ...}""" """
projects = self.server.urlopen_json('get-all-project-metadata')['projects'] # It's tempting to add in an index by name but there can be
# Provide a way for projects to be indexed by name too # projects with the same name.
for project_id, metadata in projects.items(): return self.server.urlopen_json('get-all-project-metadata')['projects']
metadata['id'] = project_id
projects[metadata['name']] = metadata
return projects
def get_project_id_name(self, project): def get_project_name(self, project_id):
"""Returns (project_id, project_name) given either.""" """Returns project name given project_id."""
projects = self.list_projects() projects = self.list_projects()
# Is the project param an integer? If so treat as an id, else a name. return projects[project_id]['name']
if re.match(r'^\d+$', project):
return project, projects[project]['name']
else:
return projects[project]['id'], project
def open_project(self, project): def open_project(self, project_id):
"""Open a Refine project referred to by id or name.""" """Open a Refine project."""
project_id, project_name = self.get_project_id_name(project) return RefineProject(self.server, project_id)
return RefineProject(self.server, project_id, project_name)
def new_project(self, project_file=None, project_url=None, def new_project(self, project_file=None, project_url=None,
project_name=None, project_name=None,
@ -142,16 +134,17 @@ class Refine:
'filename': project_file, 'filename': project_file,
} }
if project_name is None: if project_name is None:
# strip extension and directories # make a name for itself by stripping extension and directories
project_name = (project_file or 'New project').rsplit('.', 1)[0] project_name = (project_file or 'New project').rsplit('.', 1)[0]
project_name = os.path.basename(project_name) project_name = os.path.basename(project_name)
options['project-name'] = project_name options['project-name'] = project_name
response = self.server.urlopen('create-project-from-upload', options) response = self.server.urlopen('create-project-from-upload', options)
# expecting a redirect to the new project containing the id in the url # expecting a redirect to the new project containing the id in the url
url_params = urlparse.parse_qs(urlparse.urlparse(response.geturl()).query) url_params = urlparse.parse_qs(
urlparse.urlparse(response.geturl()).query)
if 'project' in url_params: if 'project' in url_params:
project_id = url_params['project'][0] project_id = url_params['project'][0]
return RefineProject(self.server, project_id, project_name) return RefineProject(self.server, project_id)
else: else:
raise Exception('Project not created') raise Exception('Project not created')
@ -202,7 +195,7 @@ def RowsResponseFactory(column_index):
class RefineProject: class RefineProject:
"""A Google Refine project.""" """A Google Refine project."""
def __init__(self, server, project_id=None, project_name=None): def __init__(self, server, project_id):
if not isinstance(server, RefineServer): if not isinstance(server, RefineServer):
url = urlparse.urlparse(server) url = urlparse.urlparse(server)
if url.query: if url.query:
@ -212,13 +205,9 @@ class RefineProject:
url.scheme, url.netloc, '', '', '', '')) url.scheme, url.netloc, '', '', '', ''))
server = RefineServer(server) server = RefineServer(server)
self.server = server self.server = server
if not project_id and not project_name: if not project_id:
raise Exception('Missing Refine project ID and name; need at least one of those') raise Exception('Missing Refine project ID')
if not project_name or not project_id:
project_id, project_name = Refine(server).get_project_id_name(
project_name or project_id)
self.project_id = project_id self.project_id = project_id
self.project_name = project_name
self.engine = facet.Engine() self.engine = facet.Engine()
self.sorting = facet.Sorting() self.sorting = facet.Sorting()
# following filled in by get_models() # following filled in by get_models()
@ -227,6 +216,9 @@ class RefineProject:
self.rows_response_factory = None # for parsing get_rows() self.rows_response_factory = None # for parsing get_rows()
self.get_models() self.get_models()
def project_name(self):
return Refine(self.server).get_project_name(project_id)
def do_raw(self, command, data): def do_raw(self, command, data):
"""Issue a command to the server & return a response object.""" """Issue a command to the server & return a response object."""
return self.server.urlopen(command, self.project_id, data) return self.server.urlopen(command, self.project_id, data)
@ -277,7 +269,7 @@ class RefineProject:
def export(self, export_format='tsv'): def export(self, export_format='tsv'):
"""Return a fileobject of a project's data.""" """Return a fileobject of a project's data."""
url = ('export-rows/' + urllib.quote(self.project_name) + '.' + url = ('export-rows/' + urllib.quote(self.project_name()) + '.' +
export_format) export_format)
return self.do_raw(url, {'format': export_format}) return self.do_raw(url, {'format': export_format})