Exemple #1
0
    def __call__(self):
        # super().__call__()
        ep = EndPoint()
        session = ep.connect()

        if self.container == '.':
            ep.set_workdir(session['owner'])
            return

        if self.container == '..':
            resp = ep.get(url=NODE.format(session['cwd']))
            context = resp.json()
            if context.get('parentId'):
                ep.set_workdir(context.get('parentId'))
            else:
                print('You are at /osdr/home/{}'.format(session['username']))
            return

        list_url = CONTENTS.format(session['cwd'])

        record = ep.get_container_by_id(self.container)
        if not record:
            records = ep.get_containers(list_url)
            record = ep.get_uniq_container(records, self.container)

        ep.set_workdir(record['id'])
Exemple #2
0
    def __call__(self):
        # super().__call__()
        ep = EndPoint()
        session = ep.connect()

        list_url = CONTENTS.format(session['cwd'])

        record = ep.get_container_by_id(self.container)
        if not record:
            records = ep.get_containers(list_url)
            record = ep.get_uniq_container(records, self.container)

        folder = '.'
        file_name = record['name']
        if self.output:
            if os.path.isdir(self.output):
                folder = self.output
            elif not os.path.dirname(self.output):
                file_name = self.output
            else:
                file_name = os.path.basename(self.output)
                folder = os.path.dirname(self.output)
                assert os.path.isdir(folder), 'No such folder: {}'.format(folder)

        path = os.path.join(folder, file_name)
        if not self.overwrite:
            assert not os.path.isfile(path), "File {path} exists. To overwrite use -f key.".format(path=path)
        print(record)
        ep.download(record, path)
Exemple #3
0
    def __call__(self):
        # super().__call__()
        ep = EndPoint()
        session = ep.connect()
        mime_type = 'multipart/mixed'

        list_url = CONTENTS.format(session['cwd'])
        if self.container == '.':
            self.container = session['cwd']
        else:
            record = ep.get_container_by_id(self.container)
            if not record:
                records = ep.get_containers(list_url)
                record = ep.get_uniq_container(records, self.container)

            assert record['type'] in ('User', 'Folder'), \
                "Container '{name}' is not a folder".format(**record)
            self.container = session['cwd'] = record['id']

        cwd = session['cwd']
        owner = session['owner']
        files = {}
        meta = {}

        # prepare model data for uploading
        if self.meta:
            meta = MACHINE_LEARNING_MODEL
            meta.update({
                'UserId': cwd,
                'ParentId': cwd,
            })
            model_meta = self.get_meta(filepath=self.meta)
            meta['ModelInfo'] = json.dumps(model_meta)

        # prepare file data for upload
        if self.meta:
            self.file = self.file[:1]  # only one model allowed
            mime_type = 'application/x-spss-sav'
        total_names = self.name and len(self.name) or 0
        for idx, file in enumerate(self.file):
            if not os.path.isfile(file):
                raise IOError('File %s not found' % file)
            fh = open(file, 'rb')
            if idx < total_names and self.name[idx]:
                filename = self.name[idx]
            else:
                filename = os.path.basename(file)
            if self.meta:
                filename = model_meta.get('ModelName', filename)
            files.update({'file%d' % idx: (filename, fh, mime_type)})

        url = UPLOAD.format(id=owner)
        meta.update({
            'parentId': cwd,
        })

        resp = ep.post(url, meta, files=files)
Exemple #4
0
    def __call__(self):
        # super().__call__()
        ep = EndPoint()
        session = ep.connect()
        self.container = self.container[0]

        list_url = CONTENTS.format(session['cwd'])
        record = ep.get_container_by_id(self.container)
        if not record:
            records = ep.get_containers(list_url)
            record = ep.get_uniq_container(records, self.container)

        ep.remove(record)
Exemple #5
0
    def __call__(self):
        ep = EndPoint()
        session = ep.connect()

        # get c sdf file
        list_url = CONTENTS.format(session['cwd'])
        record = ep.get_container_by_id(self.container)
        if not record:
            records = ep.get_containers(list_url)
            record = ep.get_uniq_container(records, self.container)

        assert record.get('status') == 'Processed', \
            "File '{}' has not been processed yet".format(record['name'])

        assert record['name'].lower().endswith('sdf'), \
            "Unsupported file format"

        model_name = self.__set_model_meta(record['name'])

        # check folder name
        records = ep.get_containers(list_url)
        try:
            ep.get_uniq_container(records, model_name)
            raise RuntimeError(
                "Model container '{}' already exists".format(model_name))
        except AssertionError:
            pass

        placement = {
            "parentId": session['cwd'],
            "userId": session['owner'],
            "sourceBlobId": record['blob']['id'],
            "sourceBucket": record['blob']['bucket'],
        }

        self.model_meta.update(placement)
        pprint.pprint(self.model_meta)

        url = TRAIN
        resp = ep.post(url, self.model_meta)
        result = resp.json()
Exemple #6
0
    def __call__(self):
        ep = EndPoint()
        session = ep.connect()
        list_url = CONTENTS.format(session['cwd'])

        # Container 
        records = ep.get_containers(list_url)
        try:
            ep.get_uniq_container(records, self.folder)
            raise RuntimeError(
                "Model container '{}' already exists".format(self.folder))
        except AssertionError:
            pass

        # Recordset
        resp = ep.get(NODE.format(self.recordset))
        recordset = resp.json()
        assert recordset['subType'] == 'Records'

        # Model
        resp = ep.get(NODE.format(self.model))
        model = resp.json()
        assert model['subType'] == 'Model'

        query = {
            'UserId': session['owner'],
            'ParentId': session['cwd'],
            'FolderName': self.folder,
            'DatasetBlobId': recordset['blob']['id'],
            'DatasetBucket': recordset['blob']['bucket'],
            'ModelBlobId': model['blob']['id'],
            'ModelBucket': model['blob']['bucket'],
        }
        # print(query)

        resp = ep.post(PREDICT, data=query)
        result = resp.json()
        assert result['modelFolderId'], \
            "Error training model '{}'".format(model_name)
Exemple #7
0
    def __call__(self):

        assert os.path.isdir(self.folder), \
            "'{folder}' is not a folder".format(folder=self.folder)

        ep = EndPoint()
        session = ep.connect()

        # Local files
        lfiles = ListHelper(path=self.folder,
                            update=self.update_remote)

        for file in os.listdir(self.folder):
            path = os.path.join(self.folder, file)
            rec = LocalFiles(name=os.path.basename(path),
                             mtime=os.path.getmtime(path))
            if self._is_local_file(path):
                lfiles.list.append(rec)

        # remote folder id
        list_url = CONTENTS.format(session['cwd'])
        if self.container == '.':
            self.container = session['cwd']
        else:
            record = ep.get_container_by_id(self.container)
            if not record:
                records = ep.get_containers(list_url)
                record = ep.get_uniq_container(records, self.container)

            assert record['type'] in ('User', 'Folder'), \
                "Container '{name}' is not a folder".format(**record)
            self.container = session['cwd'] = record['id']

        list_url = CONTENTS.format(self.container)

        # remote files
        records = ep.get_containers(list_url)
        records = list(records)

        rfiles = ListHelper(self.folder, update=self.update_local)
        rfiles.list = [RemoteFiles(name=rec['name'], id=rec['id'],
                                   version=rec['version'],
                                   length=rec['blob']['length'],
                                   bucket=rec['blob']['bucket'],
                                   bid=rec['blob']['id'])
                       for rec in filter(self._is_remote_file, records)]

        # # files to download
        print('\n\nDownloading...')
        for file in rfiles - lfiles:
            rec = {'type': 'File', 'name': file.name, 'length': file.length,
                   'blob': {'id': file.bid, 'bucket': file.bucket,
                            'length': file.length}}
            path = os.path.join(self.folder, file.name)
            try:
                ep.download(rec, path=path)
                lfiles.log(path=path, file=file)
            except Exception as e:
                print(e)

        # file to upload
        print('Uploading...')
        for file in lfiles - rfiles:
            path = os.path.join(self.folder, file.name)
            try:
                print('Uploading %s' % path)
                ep.upload(session, path)
                lfiles.log(path=path, file=file)
            except Exception as e:
                print(e)
        lfiles.store_log()