How to use expect_worker method in Playwright Python

Best Python code snippet using playwright-python

searcher.py

Source:searcher.py

1# Copyright 1999-2021 Alibaba Group Holding Ltd.2#3# Licensed under the Apache License, Version 2.0 (the "License");4# you may not use this file except in compliance with the License.5# You may obtain a copy of the License at6#7#      http://www.apache.org/licenses/LICENSE-2.08#9# Unless required by applicable law or agreed to in writing, software10# distributed under the License is distributed on an "AS IS" BASIS,11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.12# See the License for the specific language governing permissions and13# limitations under the License.14import itertools15import logging16import os17import pickle  # nosec  # pylint: disable=import_pickle18import random19from hashlib import md520from collections import defaultdict21import numpy as np22from .... import opcodes23from .... import tensor as mt24from ....config import options25from ....core import ENTITY_TYPE, OutputType, recursive_tile26from ....core.context import get_context27from ....core.operand import OperandStage28from ....lib.filesystem import get_fs, FileSystem29from ....serialization.serializables import (30    KeyField,31    StringField,32    Int32Field,33    Int64Field,34    DictField,35    AnyField,36    BytesField,37    BoolField,38)39from ....tensor.core import TensorOrder40from ....utils import has_unknown_shape, Timer, ceildiv41from ...operands import LearnOperand, LearnOperandMixin42from ..core import proxima, validate_tensor, get_proxima_type43logger = logging.getLogger(__name__)44class ProximaSearcher(LearnOperand, LearnOperandMixin):45    _op_type_ = opcodes.PROXIMA_SIMPLE_SEARCHER46    _tensor = KeyField("tensor")47    _distance_metric = StringField("distance_metric")48    _dimension = Int32Field("dimension")49    _row_number = Int64Field("row_number")50    _topk = Int32Field("topk")51    _threads = Int32Field("threads")52    _index = AnyField("index")53    _index_searcher = StringField("index_searcher")54    _index_searcher_params = DictField("index_searcher_params")55    _index_reformer = StringField("index_reformer")56    _index_reformer_params = DictField("index_reformer_params")57    _download_index = BoolField("download_index")58    _storage_options = BytesField(59        "storage_options", on_serialize=pickle.dumps, on_deserialize=pickle.loads60    )61    def __init__(62        self,63        tensor=None,64        distance_metric=None,65        dimension=None,66        row_number=None,67        topk=None,68        index=None,69        threads=None,70        index_searcher=None,71        index_searcher_params=None,72        index_reformer=None,73        index_reformer_params=None,74        download_index=None,75        storage_options=None,76        output_types=None,77        stage=None,78        **kw,79    ):80        super().__init__(81            _tensor=tensor,82            _distance_metric=distance_metric,83            _row_number=row_number,84            _dimension=dimension,85            _index=index,86            _threads=threads,87            _index_searcher=index_searcher,88            _index_searcher_params=index_searcher_params,89            _index_reformer=index_reformer,90            _index_reformer_params=index_reformer_params,91            _download_index=download_index,92            _output_types=output_types,93            _topk=topk,94            _storage_options=storage_options,95            **kw,96        )97        if self._output_types is None:98            self._output_types = [OutputType.tensor, OutputType.tensor]99    @property100    def tensor(self):101        return self._tensor102    @property103    def distance_metric(self):104        return self._distance_metric105    @property106    def dimension(self):107        return self._dimension108    @property109    def row_number(self):110        return self._row_number111    @property112    def topk(self):113        return self._topk114    @property115    def threads(self):116        return self._threads117    @property118    def index(self):119        return self._index120    @property121    def index_searcher(self):122        return self._index_searcher123    @property124    def index_searcher_params(self):125        return self._index_searcher_params126    @property127    def index_reformer(self):128        return self._index_reformer129    @property130    def index_reformer_params(self):131        return self._index_reformer_params132    @property133    def download_index(self):134        return self._download_index135    @property136    def storage_options(self):137        return self._storage_options138    @property139    def output_limit(self):140        return 1 if self._download_index else 2141    def _set_inputs(self, inputs):142        super()._set_inputs(inputs)143        if self.stage != OperandStage.agg and not self._download_index:144            self._tensor = self._inputs[0]145            if isinstance(self._index, ENTITY_TYPE):146                self._index = self._inputs[-1]147    def __call__(self, tensor, index):148        kws = [149            {150                "dtype": np.dtype(np.uint64),151                "shape": (tensor.shape[0], self._topk),152                "order": TensorOrder.C_ORDER,153            },154            {155                "dtype": np.dtype(np.float32),156                "shape": (tensor.shape[0], self._topk),157                "order": TensorOrder.C_ORDER,158            },159        ]160        inputs = [tensor]161        if hasattr(index, "op"):162            inputs.append(index)163        return mt.ExecutableTuple(self.new_tileables(inputs, kws=kws))164    @classmethod165    def _build_download_chunks(cls, op, indexes):166        ctx = get_context()167        workers = ctx.get_worker_addresses() or [None]168        if len(workers) < len(indexes):169            workers = [workers[i % len(workers)] for i in range(len(indexes))]170        indexes_iter = iter(itertools.cycle(indexes))171        download_chunks = defaultdict(list)172        for i, worker in enumerate(workers):173            download_op = op.copy().reset_key()174            download_op.stage = OperandStage.map175            download_op.expect_worker = worker176            download_op._download_index = True177            download_op._tensor = None178            download_op._index = next(indexes_iter)179            download_chunks[i % len(indexes)].append(180                download_op.new_chunk(181                    None, index=(i,), shape=(), dtype=op.inputs[0].dtype182                )183            )184        return download_chunks185    @classmethod186    def tile(cls, op: "ProximaSearcher"):187        tensor = op.tensor188        index = op.index189        topk = op.topk190        outs = op.outputs191        row_number = op.row_number192        ctx = get_context()193        # make sure all inputs have known chunk sizes194        if has_unknown_shape(*op.inputs):195            yield196        rechunk_size = dict()197        if tensor.chunk_shape[1] > 1:198            rechunk_size[1] = tensor.shape[1]199        if row_number is not None:200            rechunk_size[0] = tensor.shape[0] // row_number201        if len(rechunk_size) > 0:202            tensor = yield from recursive_tile(tensor.rechunk(rechunk_size))203        logger.warning(f"query chunks count: {len(tensor.chunks)} ")204        if hasattr(index, "op"):205            built_indexes = [index.chunks] * len(tensor.chunks)206        else:207            # index path208            fs: FileSystem = get_fs(index, op.storage_options)209            index_paths = [210                f for f in fs.ls(index) if f.rsplit("/", 1)[-1].startswith("proxima_")211            ]212            download_chunks = cls._build_download_chunks(op, index_paths)213            iters = [iter(itertools.cycle(i)) for i in download_chunks.values()]214            built_indexes = []215            for _ in range(len(tensor.chunks)):216                built_indexes.append([next(it) for it in iters])217        if hasattr(index, "op"):218            index_chunks_workers = [219                m["bands"][0][0]220                for m in ctx.get_chunks_meta(221                    [c.key for c in index.chunks], fields=["bands"]222                )223            ]224        else:225            index_chunks_workers = [None] * len(built_indexes[0])226        out_chunks = [], []227        for i, tensor_chunk in enumerate(tensor.chunks):228            pk_chunks, distance_chunks = [], []229            for j, chunk_index, worker in zip(230                itertools.count(), built_indexes[i], index_chunks_workers231            ):232                chunk_op = op.copy().reset_key()233                chunk_op.stage = OperandStage.map234                if hasattr(index, "op"):235                    chunk_op.expect_worker = worker236                else:237                    chunk_op.expect_worker = chunk_index.op.expect_worker238                chunk_op._index = chunk_index239                chunk_op._tensor = None240                chunk_kws = [241                    {242                        "index": (tensor_chunk.index[0], j),243                        "dtype": outs[0].dtype,244                        "shape": (tensor_chunk.shape[0], topk),245                        "order": TensorOrder.C_ORDER,246                    },247                    {248                        "index": (tensor_chunk.index[0], j),249                        "dtype": outs[1].dtype,250                        "shape": (tensor_chunk.shape[0], topk),251                        "order": TensorOrder.C_ORDER,252                    },253                ]254                chunk_inputs = [tensor_chunk, chunk_index]255                pk_chunk, distance_chunk = chunk_op.new_chunks(256                    chunk_inputs, kws=chunk_kws257                )258                pk_chunks.append(pk_chunk)259                distance_chunks.append(distance_chunk)260            if len(pk_chunks) == 1:261                out_chunks[0].append(pk_chunks[0])262                out_chunks[1].append(distance_chunks[0])263                continue264            # combine topk results265            combine_size = options.combine_size266            tensor_out_chunks = [pk_chunks, distance_chunks]267            while True:268                chunk_size = ceildiv(len(tensor_out_chunks[0]), combine_size)269                cur_out_chunks = [[], []]270                for k in range(chunk_size):271                    to_combine_pks = tensor_out_chunks[0][272                        k * combine_size : (k + 1) * combine_size273                    ]274                    to_combine_distances = tensor_out_chunks[1][275                        k * combine_size : (k + 1) * combine_size276                    ]277                    chunk_op = op.copy().reset_key()278                    chunk_op.stage = OperandStage.agg279                    chunk_op._tensor = None280                    chunk_op._index = None281                    agg_chunk_kws = [282                        {283                            "index": (i, 0),284                            "dtype": outs[0].dtype,285                            "shape": (tensor_chunk.shape[0], topk),286                            "order": outs[0].order,287                        },288                        {289                            "index": (i, 0),290                            "dtype": outs[1].dtype,291                            "shape": (tensor_chunk.shape[0], topk),292                            "order": outs[1].order,293                        },294                    ]295                    pk_result_chunk, distance_result_chunk = chunk_op.new_chunks(296                        to_combine_pks + to_combine_distances, kws=agg_chunk_kws297                    )298                    cur_out_chunks[0].append(pk_result_chunk)299                    cur_out_chunks[1].append(distance_result_chunk)300                tensor_out_chunks = cur_out_chunks301                if len(tensor_out_chunks[0]) == 1:302                    break303            out_chunks[0].append(tensor_out_chunks[0][0])304            out_chunks[1].append(tensor_out_chunks[1][0])305        kws = []306        pk_params = outs[0].params307        pk_params["chunks"] = out_chunks[0]308        pk_params["nsplits"] = (tensor.nsplits[0], (topk,))309        kws.append(pk_params)310        distance_params = outs[1].params311        distance_params["chunks"] = out_chunks[1]312        distance_params["nsplits"] = (tensor.nsplits[0], (topk,))313        kws.append(distance_params)314        new_op = op.copy()315        return new_op.new_tileables(op.inputs, kws=kws)316    @classmethod317    def _execute_download(cls, ctx, op: "ProximaSearcher"):318        index_path = op.index319        with Timer() as timer:320            fs = get_fs(index_path, op.storage_options)321            # TODO322            dirs = os.environ.get("MARS_SPILL_DIRS")323            if dirs:324                temp_dir = random.choice(dirs.split(":"))325            else:326                temp_dir = "/tmp/proxima-index/"327            local_path = os.path.join(328                temp_dir, md5(str(index_path).encode("utf-8")).hexdigest()329            )  # noqa: B303  # nosec330            exist_state = True331            if not os.path.exists(local_path):332                exist_state = False333                if not os.path.exists(local_path.rsplit("/", 1)[0]):334                    os.mkdir(local_path.rsplit("/", 1)[0])335                with open(local_path, "wb") as out_f:336                    with fs.open(index_path, "rb") as in_f:337                        # 32M338                        chunk_bytes = 32 * 1024**2339                        while True:340                            data = in_f.read(chunk_bytes)341                            if data:342                                out_f.write(data)343                            else:344                                break345        logger.warning(346            f"ReadingFromVolume({op.key}), index path: {index_path}, "347            f"local_path {local_path}"348            f"size {os.path.getsize(local_path)}, "349            f"already exist {exist_state}, "350            f"costs {timer.duration} seconds "351            f"speed {round(os.path.getsize(local_path) / (1024 ** 2) / timer.duration, 2)} MB/s"352        )353        ctx[op.outputs[0].key] = local_path354    @classmethod355    def _execute_map(cls, ctx, op: "ProximaSearcher"):356        if op.download_index:357            cls._execute_download(ctx, op)358            return359        inp = ctx[op.tensor.key]360        index_path = ctx[op.inputs[-1].key]361        with Timer() as timer:362            flow = proxima.IndexFlow(363                container_name="MMapFileContainer",364                container_params={},365                searcher_name=op.index_searcher,366                searcher_params=op.index_searcher_params,367                measure_name="",368                measure_params={},369                reformer_name=op.index_reformer,370                reformer_params=op.index_reformer_params,371            )372            flow.load(index_path)373            vecs = np.ascontiguousarray(inp)374        logger.warning(375            f"LoadIndex({op.key})  index path: {index_path}  costs {timer.duration} seconds"376        )377        logger.warning(f"threads count:{op.threads}  vecs count:{len(vecs)}")378        with Timer() as timer:379            batch = 10000380            s_idx = 0381            e_idx = min(s_idx + batch, len(vecs))382            result_pks, result_distances = None, None383            while s_idx < len(vecs):384                with Timer() as timer_s:385                    tp = get_proxima_type(vecs.dtype)386                    result_pks_b, result_distances_b = proxima.IndexUtility.ann_search(387                        searcher=flow,388                        type=tp,389                        query=vecs[s_idx:e_idx],390                        topk=op.topk,391                        threads=op.threads,392                    )393                    if result_pks is None:394                        result_pks = np.asarray(result_pks_b)395                        result_distances = np.asarray(result_distances_b)396                    else:397                        result_pks = np.concatenate(398                            (result_pks, np.asarray(result_pks_b))399                        )400                        result_distances = np.concatenate(401                            (result_distances, np.asarray(result_distances_b))402                        )403                    s_idx = e_idx404                    e_idx = min(s_idx + batch, len(vecs))405                logger.warning(406                    f"Search({op.key}) count {s_idx}/{len(vecs)}:{round(s_idx * 100 / len(vecs), 2)}%"407                    f" costs {round(timer_s.duration, 2)} seconds"408                )409        logger.warning(f"Search({op.key}) costs {timer.duration} seconds")410        ctx[op.outputs[0].key] = np.asarray(result_pks)411        ctx[op.outputs[1].key] = np.asarray(result_distances)412    @classmethod413    def _execute_agg(cls, ctx, op: "ProximaSearcher"):414        inputs_data = [ctx[inp.key] for inp in op.inputs]415        chunk_num = len(inputs_data) // 2416        pks = np.concatenate(inputs_data[:chunk_num], axis=1)417        distances = np.concatenate(inputs_data[chunk_num:], axis=1)418        n_doc = len(pks)419        topk = op.topk420        # calculate topk on rows421        if op.distance_metric == "InnerProduct":422            inds = np.argsort(distances, axis=1)[:, -1 : -topk - 1 : -1]423        else:424            inds = np.argsort(distances, axis=1)[:, :topk]425        result_pks = np.empty((n_doc, topk), dtype=pks.dtype)426        result_distances = np.empty((n_doc, topk), dtype=distances.dtype)427        rng = np.arange(n_doc)428        for i in range(topk):429            ind = inds[:, i]430            result_pks[:, i] = pks[rng, ind]431            result_distances[:, i] = distances[rng, ind]432        del rng433        ctx[op.outputs[0].key] = result_pks434        ctx[op.outputs[1].key] = result_distances435    @classmethod436    def execute(cls, ctx, op: "ProximaSearcher"):437        if op.stage != OperandStage.agg:438            return cls._execute_map(ctx, op)439        else:440            return cls._execute_agg(ctx, op)441def search_index(442    tensor,443    topk,444    index,445    threads=4,446    row_number=None,447    dimension=None,448    distance_metric=None,449    index_searcher=None,450    index_searcher_params=None,451    index_reformer=None,452    index_reformer_params=None,453    storage_options=None,454    run=True,455    session=None,456    run_kwargs=None,457):458    tensor = validate_tensor(tensor)459    if dimension is None:460        dimension = tensor.shape[1]461    if index_searcher is None:462        index_searcher = ""463    if index_searcher_params is None:464        index_searcher_params = {}465    if index_reformer is None:466        index_reformer = ""467    if index_reformer_params is None:468        index_reformer_params = {}469    if distance_metric is None:470        distance_metric = ""471    if hasattr(index, "op") and index.op.index_path is not None:472        storage_options = storage_options or index.op.storage_options473        index = index.op.index_path474    op = ProximaSearcher(475        tensor=tensor,476        distance_metric=distance_metric,477        dimension=dimension,478        row_number=row_number,479        topk=topk,480        index=index,481        threads=threads,482        index_searcher=index_searcher,483        index_searcher_params=index_searcher_params,484        index_reformer=index_reformer,485        index_reformer_params=index_reformer_params,486        storage_options=storage_options,487    )488    result = op(tensor, index)489    if run:490        return result.execute(session=session, **(run_kwargs or dict()))491    else:...

analyzer.py

Source:analyzer.py

1# Copyright 1999-2021 Alibaba Group Holding Ltd.2#3# Licensed under the Apache License, Version 2.0 (the "License");4# you may not use this file except in compliance with the License.5# You may obtain a copy of the License at6#7#      http://www.apache.org/licenses/LICENSE-2.08#9# Unless required by applicable law or agreed to in writing, software10# distributed under the License is distributed on an "AS IS" BASIS,11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.12# See the License for the specific language governing permissions and13# limitations under the License.14import logging15from collections import deque, defaultdict16from typing import Dict, List, Tuple, Type, Union17from ....config import Config18from ....core import ChunkGraph, ChunkType, enter_mode19from ....core.operand import Fetch, VirtualOperand, LogicKeyGenerator20from ....resource import Resource21from ....typing import BandType22from ....utils import build_fetch, tokenize23from ...subtask import SubtaskGraph, Subtask24from ..core import Task, new_task_id25from .assigner import AbstractGraphAssigner, GraphAssigner26from .fusion import Coloring27logger = logging.getLogger(__name__)28class GraphAnalyzer:29    def __init__(30        self,31        chunk_graph: ChunkGraph,32        band_resource: Dict[BandType, Resource],33        task: Task,34        config: Config,35        graph_assigner_cls: Type[AbstractGraphAssigner] = None,36        stage_id: str = None,37    ):38        self._chunk_graph = chunk_graph39        self._band_resource = band_resource40        self._task = task41        self._stage_id = stage_id42        self._config = config43        self._fuse_enabled = task.fuse_enabled44        self._extra_config = task.extra_config45        if graph_assigner_cls is None:46            graph_assigner_cls = GraphAssigner47        self._graph_assigner_cls = graph_assigner_cls48        self._chunk_to_copied = dict()49        self._logic_key_generator = LogicKeyGenerator()50    @classmethod51    def _iter_start_ops(cls, chunk_graph: ChunkGraph):52        visited = set()53        op_keys = set()54        start_chunks = deque(chunk_graph.iter_indep())55        stack = deque([start_chunks.popleft()])56        while stack:57            chunk = stack.popleft()58            if chunk not in visited:59                inp_chunks = chunk_graph.predecessors(chunk)60                if not inp_chunks or all(61                    inp_chunk in visited for inp_chunk in inp_chunks62                ):63                    if len(inp_chunks) == 0:64                        op_key = chunk.op.key65                        if op_key not in op_keys:66                            op_keys.add(op_key)67                            yield chunk.op68                    visited.add(chunk)69                    stack.extend(c for c in chunk_graph[chunk] if c not in visited)70                else:71                    stack.appendleft(chunk)72                    stack.extendleft(73                        reversed(74                            [75                                c76                                for c in chunk_graph.predecessors(chunk)77                                if c not in visited78                            ]79                        )80                    )81            if not stack and start_chunks:82                stack.appendleft(start_chunks.popleft())83    @classmethod84    def _gen_input_chunks(85        cls,86        inp_chunks: List[ChunkType],87        chunk_to_fetch_chunk: Dict[ChunkType, ChunkType],88    ) -> List[ChunkType]:89        # gen fetch chunks for input chunks90        inp_fetch_chunks = []91        for inp_chunk in inp_chunks:92            if inp_chunk in chunk_to_fetch_chunk:93                inp_fetch_chunks.append(chunk_to_fetch_chunk[inp_chunk])94            elif isinstance(inp_chunk.op, Fetch):95                chunk_to_fetch_chunk[inp_chunk] = inp_chunk96                inp_fetch_chunks.append(inp_chunk)97            else:98                fetch_chunk = build_fetch(inp_chunk).data99                chunk_to_fetch_chunk[inp_chunk] = fetch_chunk100                inp_fetch_chunks.append(fetch_chunk)101        return inp_fetch_chunks102    @staticmethod103    def _to_band(band_or_worker: Union[BandType, str]) -> BandType:104        if isinstance(band_or_worker, tuple) and len(band_or_worker) == 2:105            # band already106            return band_or_worker107        else:108            return band_or_worker, "numa-0"109    def _gen_subtask_info(110        self,111        chunks: List[ChunkType],112        chunk_to_subtask: Dict[ChunkType, Subtask],113        chunk_to_bands: Dict[ChunkType, BandType],114        chunk_to_fetch_chunk: Dict[ChunkType, ChunkType],115    ) -> Tuple[Subtask, List[Subtask]]:116        # gen subtask and its input subtasks117        final_result_chunks_set = set(self._chunk_graph.result_chunks)118        chunks_set = set(chunks)119        result_chunks = []120        result_chunks_set = set()121        chunk_graph = ChunkGraph(result_chunks)122        out_of_scope_chunks = []123        chunk_to_copied = self._chunk_to_copied124        # subtask properties125        band = None126        is_virtual = None127        retryable = True128        chunk_priority = None129        expect_worker = None130        bands_specified = None131        for chunk in chunks:132            if expect_worker is None:133                expect_worker = chunk.op.expect_worker134                bands_specified = expect_worker is not None135            else:  # pragma: no cover136                assert (137                    chunk.op.expect_worker is None138                    or expect_worker == chunk.op.expect_worker139                ), (140                    f"expect_worker {chunk.op.expect_worker} conflicts with chunks that have same color: "141                    f"{expect_worker}"142                )143            # process band144            chunk_band = chunk_to_bands.get(chunk)145            if chunk_band is not None:146                assert (147                    band is None or band == chunk_band148                ), "band conflicts with chunks that have same color"149                band = chunk_band150            # process is_virtual151            if isinstance(chunk.op, VirtualOperand):152                assert is_virtual is None, "only 1 virtual operand can exist"153                is_virtual = True154            else:155                is_virtual = False156            # process retryable157            if not chunk.op.retryable:158                retryable = False159            # process priority160            if chunk.op.priority is not None:161                assert (162                    chunk_priority is None or chunk_priority == chunk.op.priority163                ), "priority conflicts with chunks that have same color"164                chunk_priority = chunk.op.priority165            # process input chunks166            inp_chunks = []167            build_fetch_index_to_chunks = dict()168            for i, inp_chunk in enumerate(chunk.inputs):169                if inp_chunk in chunks_set:170                    inp_chunks.append(chunk_to_copied[inp_chunk])171                else:172                    build_fetch_index_to_chunks[i] = inp_chunk173                    inp_chunks.append(None)174                    if not isinstance(inp_chunk.op, Fetch):175                        out_of_scope_chunks.append(inp_chunk)176            fetch_chunks = self._gen_input_chunks(177                list(build_fetch_index_to_chunks.values()), chunk_to_fetch_chunk178            )179            for i, fetch_chunk in zip(build_fetch_index_to_chunks, fetch_chunks):180                inp_chunks[i] = fetch_chunk181            copied_op = chunk.op.copy()182            copied_op._key = chunk.op.key183            out_chunks = [184                c.data185                for c in copied_op.new_chunks(186                    inp_chunks, kws=[c.params.copy() for c in chunk.op.outputs]187                )188            ]189            for src_chunk, out_chunk in zip(chunk.op.outputs, out_chunks):190                out_chunk._key = src_chunk.key191                chunk_graph.add_node(out_chunk)192                chunk_to_copied[src_chunk] = out_chunk193                if chunk in final_result_chunks_set:194                    result_chunks.append(out_chunk)195                    result_chunks_set.add(out_chunk)196                if not is_virtual:197                    # skip adding fetch chunk to chunk graph when op is virtual operand198                    for c in inp_chunks:199                        if c not in chunk_graph:200                            chunk_graph.add_node(c)201                        chunk_graph.add_edge(c, out_chunk)202        # add chunks with no successors into result chunks203        result_chunks.extend(204            c205            for c in chunk_graph.iter_indep(reverse=True)206            if c not in result_chunks_set207        )208        expect_bands = (209            [self._to_band(expect_worker)]210            if bands_specified211            else ([band] if band is not None else None)212        )213        # calculate priority214        if out_of_scope_chunks:215            inp_subtasks = []216            for out_of_scope_chunk in out_of_scope_chunks:217                copied_out_of_scope_chunk = chunk_to_copied[out_of_scope_chunk]218                inp_subtask = chunk_to_subtask[out_of_scope_chunk]219                if (220                    copied_out_of_scope_chunk221                    not in inp_subtask.chunk_graph.result_chunks222                ):223                    # make sure the chunk that out of scope224                    # is in the input subtask's results,225                    # or the meta may be lost226                    inp_subtask.chunk_graph.result_chunks.append(227                        copied_out_of_scope_chunk228                    )229                inp_subtasks.append(inp_subtask)230            depth = max(st.priority[0] for st in inp_subtasks) + 1231        else:232            inp_subtasks = []233            depth = 0234        priority = (depth, chunk_priority or 0)235        subtask = Subtask(236            subtask_id=new_task_id(),237            stage_id=self._stage_id,238            logic_key=self._gen_logic_key(chunks),239            session_id=self._task.session_id,240            task_id=self._task.task_id,241            chunk_graph=chunk_graph,242            expect_bands=expect_bands,243            bands_specified=bands_specified,244            virtual=is_virtual,245            priority=priority,246            retryable=retryable,247            extra_config=self._extra_config,248        )249        return subtask, inp_subtasks250    def _gen_logic_key(self, chunks: List[ChunkType]):251        return tokenize(252            *[self._logic_key_generator.get_logic_key(chunk.op) for chunk in chunks]253        )254    @enter_mode(build=True)255    def gen_subtask_graph(self) -> SubtaskGraph:256        """257        Analyze chunk graph and generate subtask graph.258        Returns259        -------260        subtask_graph: SubtaskGraph261            Subtask graph.262        """263        reassign_worker_ops = [264            chunk.op for chunk in self._chunk_graph if chunk.op.reassign_worker265        ]266        start_ops = (267            list(self._iter_start_ops(self._chunk_graph))268            if len(self._chunk_graph) > 0269            else []270        )271        # assign start chunks272        to_assign_ops = start_ops + reassign_worker_ops273        assigner = self._graph_assigner_cls(274            self._chunk_graph, to_assign_ops, self._band_resource275        )276        # assign expect workers277        cur_assigns = {278            op.key: self._to_band(op.expect_worker)279            for op in start_ops280            if op.expect_worker is not None281        }282        logger.debug(283            "Start to assign %s start chunks for task %s",284            len(start_ops),285            self._task.task_id,286        )287        chunk_to_bands = assigner.assign(cur_assigns=cur_assigns)288        logger.debug(289            "Assigned %s start chunks for task %s", len(start_ops), self._task.task_id290        )291        # assign expect workers for those specified with `expect_worker`292        # skip `start_ops`, which have been assigned before293        for chunk in self._chunk_graph:294            if chunk not in start_ops and chunk.op.expect_worker is not None:295                chunk_to_bands[chunk] = self._to_band(chunk.op.expect_worker)296        # color nodes297        if self._fuse_enabled:298            logger.debug("Start to fuse chunks for task %s", self._task.task_id)299            # sort start chunks in coloring as start_ops300            op_key_to_chunks = defaultdict(list)301            for chunk in self._chunk_graph:302                op_key_to_chunks[chunk.op.key].append(chunk)303            init_chunk_to_bands = dict()304            for start_op in start_ops:305                for start_chunk in op_key_to_chunks[start_op.key]:306                    init_chunk_to_bands[start_chunk] = chunk_to_bands[start_chunk]307            coloring = Coloring(308                self._chunk_graph,309                list(self._band_resource),310                init_chunk_to_bands,311                initial_same_color_num=getattr(312                    self._config, "initial_same_color_num", None313                ),314                as_broadcaster_successor_num=getattr(315                    self._config, "as_broadcaster_successor_num", None316                ),317            )318            chunk_to_colors = coloring.color()319        else:320            # if not fuse enabled, color all chunks with different colors321            chunk_to_colors = {322                c: i for i, c in enumerate(self._chunk_graph.topological_iter())323            }324        color_to_chunks = defaultdict(list)325        for chunk, color in chunk_to_colors.items():326            color_to_chunks[color].append(chunk)327        # gen subtask graph328        subtask_graph = SubtaskGraph()329        chunk_to_fetch_chunk = dict()330        chunk_to_subtask = dict()331        # states332        visited = set()333        logic_key_to_subtasks = defaultdict(list)334        for chunk in self._chunk_graph.topological_iter():335            if chunk in visited:336                continue337            color = chunk_to_colors[chunk]338            same_color_chunks = color_to_chunks[color]339            if all(isinstance(c.op, Fetch) for c in same_color_chunks):340                # all fetch ops, no need to gen subtask341                continue342            subtask, inp_subtasks = self._gen_subtask_info(343                same_color_chunks,344                chunk_to_subtask,345                chunk_to_bands,346                chunk_to_fetch_chunk,347            )348            subtask_graph.add_node(subtask)349            logic_key_to_subtasks[subtask.logic_key].append(subtask)350            for inp_subtask in inp_subtasks:351                subtask_graph.add_edge(inp_subtask, subtask)352            for c in same_color_chunks:353                chunk_to_subtask[c] = subtask354            visited.update(same_color_chunks)355        for subtasks in logic_key_to_subtasks.values():356            for logic_index, subtask in enumerate(subtasks):357                subtask.logic_index = logic_index358                subtask.logic_parallelism = len(subtasks)...

builder.py

Source:builder.py

1# Copyright 1999-2021 Alibaba Group Holding Ltd.2#3# Licensed under the Apache License, Version 2.0 (the "License");4# you may not use this file except in compliance with the License.5# You may obtain a copy of the License at6#7#      http://www.apache.org/licenses/LICENSE-2.08#9# Unless required by applicable law or agreed to in writing, software10# distributed under the License is distributed on an "AS IS" BASIS,11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.12# See the License for the specific language governing permissions and13# limitations under the License.14import itertools15import logging16import os17import pickle  # nosec  # pylint: disable=import_pickle18import tempfile19import uuid20import numpy as np21from .... import opcodes22from .... import tensor as mt23from ....lib.filesystem import get_fs24from ....core import OutputType25from ....core.context import get_context26from ....core.operand import OperandStage27from ....serialization.serializables import (28    StringField,29    Int32Field,30    Int64Field,31    DictField,32    BytesField,33    TupleField,34    DataTypeField,35)36from ....utils import has_unknown_shape, Timer37from ...operands import LearnOperand, LearnOperandMixin38from ..core import (39    proxima,40    get_proxima_type,41    validate_tensor,42    available_numpy_dtypes,43    rechunk_tensor,44    build_mmap_chunks,45)46logger = logging.getLogger(__name__)47DEFAULT_INDEX_SIZE = 5 * 10**648class ProximaBuilder(LearnOperand, LearnOperandMixin):49    _op_type_ = opcodes.PROXIMA_SIMPLE_BUILDER50    _distance_metric = StringField("distance_metric")51    _dimension = Int32Field("dimension")52    _column_number = Int64Field("column_number")53    _index_path = StringField("index_path")54    _index_builder = StringField("index_builder")55    _index_builder_params = DictField("index_builder_params")56    _index_converter = StringField("index_converter")57    _index_converter_params = DictField("index_converter_params")58    _topk = Int32Field("topk")59    _storage_options = BytesField(60        "storage_options", on_serialize=pickle.dumps, on_deserialize=pickle.loads61    )62    # only for chunk63    _array_shape = TupleField("array_shape")64    _array_dtype = DataTypeField("array_dtype")65    _offset = Int64Field("offset")66    def __init__(67        self,68        distance_metric=None,69        index_path=None,70        dimension=None,71        column_number=None,72        index_builder=None,73        index_builder_params=None,74        index_converter=None,75        index_converter_params=None,76        array_shape=None,77        array_dtype=None,78        offset=None,79        topk=None,80        storage_options=None,81        output_types=None,82        **kw,83    ):84        super().__init__(85            _distance_metric=distance_metric,86            _index_path=index_path,87            _dimension=dimension,88            _column_number=column_number,89            _index_builder=index_builder,90            _index_builder_params=index_builder_params,91            _array_shape=array_shape,92            _array_dtype=array_dtype,93            _offset=offset,94            _index_converter=index_converter,95            _index_converter_params=index_converter_params,96            _topk=topk,97            _storage_options=storage_options,98            _output_types=output_types,99            **kw,100        )101        if self._output_types is None:102            self._output_types = [OutputType.object]103    @property104    def distance_metric(self):105        return self._distance_metric106    @property107    def column_number(self):108        return self._column_number109    @property110    def index_path(self):111        return self._index_path112    @property113    def dimension(self):114        return self._dimension115    @property116    def index_builder(self):117        return self._index_builder118    @property119    def index_builder_params(self):120        return self._index_builder_params121    @property122    def index_converter(self):123        return self._index_converter124    @property125    def index_converter_params(self):126        return self._index_converter_params127    @property128    def topk(self):129        return self._topk130    @property131    def storage_options(self):132        return self._storage_options133    @property134    def array_shape(self):135        return self._array_shape136    @property137    def array_dtype(self):138        return self._array_dtype139    @property140    def offset(self):141        return self._offset142    def __call__(self, tensor):143        return self.new_tileable([tensor])144    @classmethod145    def _get_atleast_topk_nsplit(cls, nsplit, topk):146        new_nsplit = []147        i = 0148        while i < len(nsplit):149            cur = nsplit[i]150            i += 1151            if cur >= topk:152                new_nsplit.append(cur)153            else:154                while i < len(nsplit):155                    cur += nsplit[i]156                    i += 1157                    if cur >= topk:158                        break159                if cur < topk and len(new_nsplit) > 0:160                    new_nsplit[-1] += cur161                elif cur >= topk:162                    new_nsplit.append(cur)163        new_nsplit = tuple(new_nsplit)164        assert sum(new_nsplit) == sum(nsplit), (165            f"sum of nsplit not equal, " f"old: {nsplit}, new: {new_nsplit}"166        )167        return new_nsplit168    @classmethod169    def tile(cls, op):170        tensor = op.inputs[0]171        out = op.outputs[0]172        index_path = op.index_path173        ctx = get_context()174        fs = None175        if index_path is not None:176            fs = get_fs(index_path, op.storage_options)177        if index_path is not None:178            # check if the index path is empty179            try:180                files = [f for f in fs.ls(index_path) if "proxima_" in f]181                if files:182                    raise ValueError(183                        f"Directory {index_path} contains built proxima index, "184                        f"clean them to perform new index building"185                    )186            except FileNotFoundError:187                # if not exist, create directory188                fs.mkdir(index_path)189        # make sure all inputs have known chunk sizes190        if has_unknown_shape(*op.inputs):191            yield192        if op.column_number:193            index_chunk_size = op.inputs[0].shape[0] // op.column_number194        else:195            worker_num = len(ctx.get_worker_addresses() or [])196            if worker_num > 0:197                index_chunk_size = max(198                    op.inputs[0].shape[0] // worker_num, DEFAULT_INDEX_SIZE199                )200            else:201                index_chunk_size = DEFAULT_INDEX_SIZE202        if op.topk is not None:203            index_chunk_size = cls._get_atleast_topk_nsplit(index_chunk_size, op.topk)204        # build chunks for writing tensors to mmap files.205        worker_iter = iter(itertools.cycle(ctx.get_worker_addresses() or [None]))206        chunk_groups = rechunk_tensor(tensor, index_chunk_size)207        out_chunks = []208        offsets = []209        offset = 0210        for chunk_group in chunk_groups:211            offsets.append(offset)212            file_prefix = f"proxima-build-{str(uuid.uuid4())}"213            out_chunks.append(214                build_mmap_chunks(215                    chunk_group, next(worker_iter), file_prefix=file_prefix216                )217            )218            offset += sum(c.shape[0] for c in chunk_group)219        final_out_chunks = []220        for j, chunks in enumerate(out_chunks):221            chunk_op = op.copy().reset_key()222            chunk_op.stage = OperandStage.map223            chunk_op.expect_worker = chunks[0].op.expect_worker224            chunk_op._array_shape = chunks[0].op.total_shape225            chunk_op._array_dtype = chunks[0].dtype226            chunk_op._offset = offsets[j]227            out_chunk = chunk_op.new_chunk(chunks, index=(j,))228            final_out_chunks.append(out_chunk)229        logger.warning(f"index chunks count: {len(final_out_chunks)} ")230        params = out.params231        params["chunks"] = final_out_chunks232        params["nsplits"] = ((1,) * len(final_out_chunks),)233        new_op = op.copy()234        return new_op.new_tileables(op.inputs, kws=[params])235    @classmethod236    def _execute_map(cls, ctx, op: "ProximaBuilder"):237        mmap_path = ctx[op.inputs[0].key]238        out = op.outputs[0]239        data = np.memmap(240            mmap_path, dtype=op.array_dtype, mode="r", shape=op.array_shape241        )242        proxima_type = get_proxima_type(op.array_dtype)243        offset = op.offset244        # holder245        with Timer() as timer:246            holder = proxima.IndexHolder(247                type=proxima_type, dimension=op.dimension, shallow=True248            )249            holder.mount(data, key_base=offset)250        logger.warning(f"Holder({op.key}) costs {timer.duration} seconds")251        # converter252        meta = proxima.IndexMeta(253            proxima_type, dimension=op.dimension, measure_name=op.distance_metric254        )255        if op.index_converter is not None:256            with Timer() as timer:257                converter = proxima.IndexConverter(258                    name=op.index_converter, meta=meta, params=op.index_converter_params259                )260                converter.train_and_transform(holder)261                holder = converter.result()262                meta = converter.meta()263            logger.warning(f"Converter({op.key}) costs {timer.duration} seconds")264        # builder265        with Timer() as timer:266            builder = proxima.IndexBuilder(267                name=op.index_builder, meta=meta, params=op.index_builder_params268            )269            builder = builder.train_and_build(holder)270        logger.warning(f"Builder({op.key}) costs {timer.duration} seconds")271        # remove mmap file272        os.remove(mmap_path)273        # dumper274        with Timer() as timer:275            path = tempfile.mkstemp(prefix="proxima-", suffix=".index")[1]276            dumper = proxima.IndexDumper(name="FileDumper", path=path)277            builder.dump(dumper)278            dumper.close()279        logger.warning(f"Dumper({op.key}) costs {timer.duration} seconds")280        if op.index_path is None:281            ctx[out.key] = path282        else:283            # write to external file284            with Timer() as timer:285                fs = get_fs(op.index_path, op.storage_options)286                filename = f"proxima_{out.index[0]}_index"287                out_path = f'{op.index_path.rstrip("/")}/{filename}'288                def write_index():289                    with fs.open(out_path, "wb") as out_f:290                        with open(path, "rb") as in_f:291                            # 128M292                            chunk_bytes = 128 * 1024**2293                            while True:294                                data = in_f.read(chunk_bytes)295                                if data:296                                    out_f.write(data)297                                else:298                                    break299                # retry 3 times300                for _ in range(3):301                    try:302                        write_index()303                        break304                    except:  # noqa: E722  # nosec  # pylint: disable=bare-except305                        fs.delete(out_path)306                        continue307            logger.warning(308                f"WritingToVolume({op.key}), out path: {out_path}, "309                f"size {os.path.getsize(path)}, "310                f"costs {timer.duration} seconds "311                f"speed {round(os.path.getsize(path) / (1024 ** 2) / timer.duration, 2)} MB/s"312            )313            ctx[out.key] = filename314    @classmethod315    def _execute_agg(cls, ctx, op: "ProximaBuilder"):316        paths = [ctx[inp.key] for inp in op.inputs]317        ctx[op.outputs[0].key] = paths318    @classmethod319    def execute(cls, ctx, op: "ProximaBuilder"):320        if op.stage != OperandStage.agg:321            return cls._execute_map(ctx, op)322        else:323            return cls._execute_agg(ctx, op)324    @classmethod325    def concat_tileable_chunks(cls, tileable):326        assert not tileable.is_coarse()327        op = cls(stage=OperandStage.agg)328        chunk = cls(stage=OperandStage.agg).new_chunk(tileable.chunks)329        return op.new_tileable([tileable], chunks=[chunk], nsplits=((1,),))330def build_index(331    tensor,332    dimension=None,333    index_path=None,334    column_number=None,335    need_shuffle=False,336    distance_metric="SquaredEuclidean",337    index_builder="SsgBuilder",338    index_builder_params=None,339    index_converter=None,340    index_converter_params=None,341    topk=None,342    storage_options=None,343    run=True,344    session=None,345    run_kwargs=None,346):347    tensor = validate_tensor(tensor)348    if tensor.dtype not in available_numpy_dtypes:349        raise ValueError(350            f"Dtype to build index should be one of {available_numpy_dtypes}, "351            f"got {tensor.dtype}"352        )353    if dimension is None:354        dimension = tensor.shape[1]355    if index_builder_params is None:356        index_builder_params = {}357    if index_converter_params is None:358        index_converter_params = {}359    if need_shuffle:360        tensor = mt.random.permutation(tensor)361    op = ProximaBuilder(362        distance_metric=distance_metric,363        index_path=index_path,364        dimension=dimension,365        column_number=column_number,366        index_builder=index_builder,367        index_builder_params=index_builder_params,368        index_converter=index_converter,369        index_converter_params=index_converter_params,370        topk=topk,371        storage_options=storage_options,372    )373    result = op(tensor)374    if run:375        return result.execute(session=session, **(run_kwargs or dict()))376    else:...

test_worker.py

Source:test_worker.py

...15from asyncio.futures import Future16import pytest17from playwright.async_api import Error, Page, Worker18async def test_workers_page_workers(page, server):19    async with page.expect_worker() as worker_info:20        await page.goto(server.PREFIX + "/worker/worker.html")21    worker = await worker_info.value22    assert "worker.js" in worker.url23    assert (24        await worker.evaluate('() => self["workerFunction"]()')25        == "worker function result"26    )27    await page.goto(server.EMPTY_PAGE)28    assert len(page.workers) == 029async def test_workers_should_emit_created_and_destroyed_events(page: Page):30    worker_obj = None31    async with page.expect_event("worker") as event_info:32        worker_obj = await page.evaluate_handle(33            "() => new Worker(URL.createObjectURL(new Blob(['1'], {type: 'application/javascript'})))"34        )35    worker = await event_info.value36    worker_this_obj = await worker.evaluate_handle("() => this")37    worker_destroyed_promise: Future[Worker] = asyncio.Future()38    worker.once("close", lambda w: worker_destroyed_promise.set_result(w))39    await page.evaluate("workerObj => workerObj.terminate()", worker_obj)40    assert await worker_destroyed_promise == worker41    with pytest.raises(Error) as exc:42        await worker_this_obj.get_property("self")43    assert "Most likely the worker has been closed." in exc.value.message44async def test_workers_should_report_console_logs(page):45    async with page.expect_console_message() as message_info:46        await page.evaluate(47            '() => new Worker(URL.createObjectURL(new Blob(["console.log(1)"], {type: "application/javascript"})))'48        )49    message = await message_info.value50    assert message.text == "1"51@pytest.mark.skip_browser("firefox")  # TODO: investigate further @pavelfeldman52async def test_workers_should_have_JSHandles_for_console_logs(page):53    log_promise = asyncio.Future()54    page.on("console", lambda m: log_promise.set_result(m))55    await page.evaluate(56        "() => new Worker(URL.createObjectURL(new Blob(['console.log(1,2,3,this)'], {type: 'application/javascript'})))"57    )58    log = await log_promise59    assert log.text == "1 2 3 JSHandle@object"60    assert len(log.args) == 461    assert await (await log.args[3].get_property("origin")).json_value() == "null"62async def test_workers_should_evaluate(page):63    async with page.expect_event("worker") as event_info:64        await page.evaluate(65            "() => new Worker(URL.createObjectURL(new Blob(['console.log(1)'], {type: 'application/javascript'})))"66        )67    worker = await event_info.value68    assert await worker.evaluate("1+1") == 269async def test_workers_should_report_errors(page):70    error_promise = asyncio.Future()71    page.on("pageerror", lambda e: error_promise.set_result(e))72    await page.evaluate(73        """() => new Worker(URL.createObjectURL(new Blob([`74      setTimeout(() => {75        // Do a console.log just to check that we do not confuse it with an error.76        console.log('hey');77        throw new Error('this is my error');78      })79    `], {type: 'application/javascript'})))"""80    )81    error_log = await error_promise82    assert "this is my error" in error_log.message83@pytest.mark.skip_browser("firefox")  # TODO: fails upstream84async def test_workers_should_clear_upon_navigation(server, page):85    await page.goto(server.EMPTY_PAGE)86    async with page.expect_event("worker") as event_info:87        await page.evaluate(88            '() => new Worker(URL.createObjectURL(new Blob(["console.log(1)"], {type: "application/javascript"})))'89        )90    worker = await event_info.value91    assert len(page.workers) == 192    destroyed = []93    worker.once("close", lambda _: destroyed.append(True))94    await page.goto(server.PREFIX + "/one-style.html")95    assert destroyed == [True]96    assert len(page.workers) == 097@pytest.mark.skip_browser("firefox")  # TODO: fails upstream98async def test_workers_should_clear_upon_cross_process_navigation(server, page):99    await page.goto(server.EMPTY_PAGE)100    async with page.expect_event("worker") as event_info:101        await page.evaluate(102            "() => new Worker(URL.createObjectURL(new Blob(['console.log(1)'], {type: 'application/javascript'})))"103        )104    worker = await event_info.value105    assert len(page.workers) == 1106    destroyed = []107    worker.once("close", lambda _: destroyed.append(True))108    await page.goto(server.CROSS_PROCESS_PREFIX + "/empty.html")109    assert destroyed == [True]110    assert len(page.workers) == 0111async def test_workers_should_report_network_activity(page, server):112    async with page.expect_worker() as worker_info:113        await page.goto(server.PREFIX + "/worker/worker.html"),114    worker = await worker_info.value115    url = server.PREFIX + "/one-style.css"116    async with page.expect_request(url) as request_info, page.expect_response(117        url118    ) as response_info:119        await worker.evaluate(120            "url => fetch(url).then(response => response.text()).then(console.log)", url121        )122    request = await request_info.value123    response = await response_info.value124    assert request.url == url125    assert response.request == request126    assert response.ok127async def test_workers_should_report_network_activity_on_worker_creation(page, server):128    # Chromium needs waitForDebugger enabled for this one.129    await page.goto(server.EMPTY_PAGE)130    url = server.PREFIX + "/one-style.css"131    async with page.expect_request(url) as request_info, page.expect_response(132        url133    ) as response_info:134        await page.evaluate(135            """url => new Worker(URL.createObjectURL(new Blob([`136        fetch("${url}").then(response => response.text()).then(console.log);137        `], {type: 'application/javascript'})))""",138            url,139        )140    request = await request_info.value141    response = await response_info.value142    assert request.url == url143    assert response.request == request144    assert response.ok145async def test_workers_should_format_number_using_context_locale(browser, server):146    context = await browser.new_context(locale="ru-RU")147    page = await context.new_page()148    await page.goto(server.EMPTY_PAGE)149    async with page.expect_worker() as worker_info:150        await page.evaluate(151            "() => new Worker(URL.createObjectURL(new Blob(['console.log(1)'], {type: 'application/javascript'})))"152        )153    worker = await worker_info.value154    assert await worker.evaluate("() => (10000.20).toLocaleString()") == "10\u00A0000,2"...

Playwright tutorial

LambdaTest’s Playwright tutorial will give you a broader idea about the Playwright automation framework, its unique features, and use cases with examples to exceed your understanding of Playwright testing. This tutorial will give A to Z guidance, from installing the Playwright framework to some best practices and advanced concepts.

Chapters:

What is Playwright : Playwright is comparatively new but has gained good popularity. Get to know some history of the Playwright with some interesting facts connected with it.
How To Install Playwright : Learn in detail about what basic configuration and dependencies are required for installing Playwright and run a test. Get a step-by-step direction for installing the Playwright automation framework.
Playwright Futuristic Features: Launched in 2020, Playwright gained huge popularity quickly because of some obliging features such as Playwright Test Generator and Inspector, Playwright Reporter, Playwright auto-waiting mechanism and etc. Read up on those features to master Playwright testing.
What is Component Testing: Component testing in Playwright is a unique feature that allows a tester to test a single component of a web application without integrating them with other elements. Learn how to perform Component testing on the Playwright automation framework.
Inputs And Buttons In Playwright: Every website has Input boxes and buttons; learn about testing inputs and buttons with different scenarios and examples.
Functions and Selectors in Playwright: Learn how to launch the Chromium browser with Playwright. Also, gain a better understanding of some important functions like “BrowserContext,” which allows you to run multiple browser sessions, and “newPage” which interacts with a page.
Handling Alerts and Dropdowns in Playwright : Playwright interact with different types of alerts and pop-ups, such as simple, confirmation, and prompt, and different types of dropdowns, such as single selector and multi-selector get your hands-on with handling alerts and dropdown in Playright testing.
Playwright vs Puppeteer: Get to know about the difference between two testing frameworks and how they are different than one another, which browsers they support, and what features they provide.
Run Playwright Tests on LambdaTest: Playwright testing with LambdaTest leverages test performance to the utmost. You can run multiple Playwright tests in Parallel with the LammbdaTest test cloud. Get a step-by-step guide to run your Playwright test on the LambdaTest platform.
Playwright Python Tutorial: Playwright automation framework support all major languages such as Python, JavaScript, TypeScript, .NET and etc. However, there are various advantages to Python end-to-end testing with Playwright because of its versatile utility. Get the hang of Playwright python testing with this chapter.
Playwright End To End Testing Tutorial: Get your hands on with Playwright end-to-end testing and learn to use some exciting features such as TraceViewer, Debugging, Networking, Component testing, Visual testing, and many more.
Playwright Video Tutorial: Watch the video tutorials on Playwright testing from experts and get a consecutive in-depth explanation of Playwright automation testing.