Fix Typos and Type Definitions

README.md

@@ -27,7 +27,7 @@ iOS), web, desktop, edge devices, and IoT, effortlessly.
 
 ## Get started
 
-You can get started with MediaPipe Solutions by by checking out any of the
+You can get started with MediaPipe Solutions by checking out any of the
 developer guides for
 [vision](https://developers.google.com/mediapipe/solutions/vision/object_detector),
 [text](https://developers.google.com/mediapipe/solutions/text/text_classifier),

docs/index.md

@@ -27,7 +27,7 @@ iOS), web, desktop, edge devices, and IoT, effortlessly.
 
 ## Get started
 
-You can get started with MediaPipe Solutions by by checking out any of the
+You can get started with MediaPipe Solutions by checking out any of the
 developer guides for
 [vision](https://developers.google.com/mediapipe/solutions/vision/object_detector),
 [text](https://developers.google.com/mediapipe/solutions/text/text_classifier),

mediapipe/python/pybind/calculator_graph.cc

@@ -125,7 +125,7 @@ void CalculatorGraphSubmodule(pybind11::module* module) {
 
   Raises:
     FileNotFoundError: If the binary graph file can't be found.
-    ValueError: If the input arguments prvoided are more than needed or the
+    ValueError: If the input arguments provided are more than needed or the
       graph validation process contains error.
 )doc");
 
@@ -202,7 +202,7 @@ void CalculatorGraphSubmodule(pybind11::module* module) {
     graph.add_packet_to_input_stream(
         stream='in',
         packet=packet_creator.create_string('hello world'),
-        timstamp=1)
+        timestamp=1)
 )doc",
       py::arg("stream"), py::arg("packet"),
       py::arg("timestamp") = Timestamp::Unset());
@@ -461,7 +461,7 @@ void CalculatorGraphSubmodule(pybind11::module* module) {
       R"doc(Get output side packet by name after the graph is done.
 
   Args:
-    stream: The name of the outnput stream.
+    stream: The name of the output stream.
 
   Raises:
     RuntimeError: If the graph is still running or the output side packet is not

mediapipe/python/pybind/packet_creator.cc

@@ -377,7 +377,7 @@ void PublicPacketCreators(pybind11::module* m) {
     A MediaPipe double Packet.
 
   Raises:
-    TypeError: If the input is neither a float nore a np.double.
+    TypeError: If the input is neither a float nor a np.double.
 
   Examples:
     packet = mp.packet_creator.create_double(0.1)
@@ -702,7 +702,7 @@ void InternalPacketCreators(pybind11::module* m) {
         // TODO: Implement this.
         throw RaisePyError(PyExc_NotImplementedError,
                            "Creating a packet from a vector of proto messages "
-                           "is not supproted yet.");
+                           "is not supported yet.");
         return Packet();
       },
       py::return_value_policy::move);

mediapipe/python/pybind/packet_getter.cc

@@ -458,7 +458,7 @@ void InternalPacketGetters(pybind11::module* m) {
         // https://github.com/pybind/pybind11/issues/1236
         // However, when Pybind11 performs the C++ to Python transition, it
         // only increases the py::bytes object's ref count. See the
-        // implmentation at line 1583 in "pybind11/cast.h".
+        // implementation at line 1583 in "pybind11/cast.h".
         return py::bytes(packet.GetProtoMessageLite().SerializeAsString());
       },
       py::return_value_policy::move);

mediapipe/python/pybind/util.h

@@ -67,7 +67,7 @@ inline void RaisePyErrorIfNotOk(const absl::Status& status,
 inline void RaisePyErrorIfOverflow(int64_t value, int64_t min, int64_t max) {
   if (value > max) {
     throw RaisePyError(PyExc_OverflowError,
-                       absl::StrCat(value, " execeeds the maximum value (", max,
+                       absl::StrCat(value, " exceeds the maximum value (", max,
                                     ") the data type can have.")
                            .c_str());
   } else if (value < min) {

mediapipe/python/pybind/validated_graph_config.cc

@@ -84,7 +84,7 @@ void ValidatedGraphConfigSubmodule(pybind11::module* module) {
 
   Raises:
     FileNotFoundError: If the binary graph file can't be found.
-    ValueError: If the input arguments prvoided are more than needed or the
+    ValueError: If the input arguments provided are more than needed or the
       graph validation process contains error.
 
   Examples:

mediapipe/tasks/cc/core/model_resources.h

@@ -66,7 +66,7 @@ class ModelResources {
   // Takes the ownership of the provided ExternalFile proto and creates
   // ModelResources from the proto and an op resolver mediapipe packet. A
   // non-empty tag must be set if the ModelResources will be used through
-  // ModelResourcesCacheService. The op resolver packet, usually prvoided by a
+  // ModelResourcesCacheService. The op resolver packet, usually provided by a
   // ModelResourcesCacheService object, contains the TFLite op resolvers
   // required by the model.
   static absl::StatusOr<std::unique_ptr<ModelResources>> Create(

mediapipe/tasks/cc/vision/image_embedder/image_embedder.h

@@ -156,7 +156,7 @@ class ImageEmbedder : core::BaseVisionTaskApi {
   // sent to the object detector. The input timestamps must be monotonically
   // increasing.
   //
-  // The "result_callback" prvoides
+  // The "result_callback" provides
   //   - The embedding results as a
   //     components::containers::proto::EmbeddingResult object.
   //   - The const reference to the corresponding input image that the image

mediapipe/tasks/python/audio/__init__.py

@@ -14,20 +14,19 @@
 
 """MediaPipe Tasks Audio API."""
 
-import mediapipe.tasks.python.audio.core
-import mediapipe.tasks.python.audio.audio_classifier
-import mediapipe.tasks.python.audio.audio_embedder
+import mediapipe.tasks.python.audio.core.audio_task_running_mode as audio_task_running_mode
+import mediapipe.tasks.python.audio.audio_classifier as audio_classifier
+import mediapipe.tasks.python.audio.audio_embedder as audio_embedder
 
 AudioClassifier = audio_classifier.AudioClassifier
 AudioClassifierOptions = audio_classifier.AudioClassifierOptions
 AudioClassifierResult = audio_classifier.AudioClassifierResult
 AudioEmbedder = audio_embedder.AudioEmbedder
 AudioEmbedderOptions = audio_embedder.AudioEmbedderOptions
 AudioEmbedderResult = audio_embedder.AudioEmbedderResult
-RunningMode = core.audio_task_running_mode.AudioTaskRunningMode
+RunningMode = audio_task_running_mode.AudioTaskRunningMode
 
 # Remove unnecessary modules to avoid duplication in API docs.
 del audio_classifier
 del audio_embedder
-del core
-del mediapipe
+del audio_task_running_mode

mediapipe/tasks/python/audio/audio_classifier.py

@@ -281,7 +281,7 @@ def classify_async(self, audio_block: _AudioData, timestamp_ms: int) -> None:
     for adjacent calls of this method. This method will return immediately after
     the input audio data is accepted. The results will be available via the
     `result_callback` provided in the `AudioClassifierOptions`. The
-    `classify_async` method is designed to process auido stream data such as
+    `classify_async` method is designed to process audio stream data such as
     microphone input.
 
     The input audio data may be longer than what the model is able to process

mediapipe/tasks/python/audio/audio_embedder.py

@@ -242,7 +242,7 @@ def embed_async(self, audio_block: _AudioData, timestamp_ms: int) -> None:
     for adjacent calls of this method. This method will return immediately after
     the input audio data is accepted. The results will be available via the
     `result_callback` provided in the `AudioEmbedderOptions`. The
-    `embed_async` method is designed to process auido stream data such as
+    `embed_async` method is designed to process audio stream data such as
     microphone input.
 
     The input audio data may be longer than what the model is able to process

mediapipe/tasks/python/audio/core/base_audio_task_api.py

@@ -64,7 +64,7 @@ def __init__(
           'callback should not be provided.')
     self._runner = _TaskRunner.create(graph_config, packet_callback)
     self._running_mode = running_mode
-    self._default_sample_rate = None
+    self._default_sample_rate: Optional[float] = None
 
   def _process_audio_clip(
       self, inputs: Mapping[str, _Packet]) -> Mapping[str, _Packet]:
@@ -142,7 +142,7 @@ def create_audio_record(
 
     Note that MediaPipe Audio tasks will up/down sample automatically to fit the
     sample rate required by the model. The default sample rate of the MediaPipe
-    pretrained audio model, Yamnet is 16kHz.
+    pretrained audio model, YAMNet is 16kHz.
 
     Args:
       num_channels: The number of audio channels.

mediapipe/tasks/python/benchmark/benchmark_utils.py

@@ -43,7 +43,7 @@ def get_test_data_path(test_srcdir, file_or_dirname_path: str) -> str:
       if path.endswith(file_or_dirname_path):
         return path
   raise ValueError(
-      "No %s in test directory: %s." % (file_or_dirname_path, test_srcdir)
+      f"No {file_or_dirname_path} in test directory: {test_srcdir}."
   )
 
 
@@ -57,14 +57,14 @@ def get_model_path(custom_model, default_model_path):
   Returns:
       The path to the model to be used.
   """
-  if custom_model is not None and os.path.exists(custom_model):
-    print(f"Using provided model: {custom_model}")
-    return custom_model
-  else:
-    if custom_model is not None:
-      print(
-          f"Warning: Provided model '{custom_model}' not found. "
-          "Using default model instead."
-      )
-    print(f"Using default model: {default_model_path}")
-    return default_model_path
+  if custom_model is not None:
+    if os.path.exists(custom_model):
+      print(f"Using provided model: {custom_model}")
+      return custom_model
+    print(
+        f"Warning: Provided model '{custom_model}' not found. "
+        "Using default model instead."
+    )
+
+  print(f"Using default model: {default_model_path}")
+  return default_model_path

mediapipe/tasks/python/components/containers/__init__.py

@@ -14,15 +14,15 @@
 
 """MediaPipe Tasks Components Containers API."""
 
-import mediapipe.tasks.python.components.containers.audio_data
-import mediapipe.tasks.python.components.containers.bounding_box
-import mediapipe.tasks.python.components.containers.category
-import mediapipe.tasks.python.components.containers.classification_result
-import mediapipe.tasks.python.components.containers.detections
-import mediapipe.tasks.python.components.containers.embedding_result
-import mediapipe.tasks.python.components.containers.landmark
-import mediapipe.tasks.python.components.containers.landmark_detection_result
-import mediapipe.tasks.python.components.containers.rect
+import mediapipe.tasks.python.components.containers.audio_data as audio_data
+import mediapipe.tasks.python.components.containers.bounding_box as bounding_box
+import mediapipe.tasks.python.components.containers.category as category
+import mediapipe.tasks.python.components.containers.classification_result as classification_result
+import mediapipe.tasks.python.components.containers.detections as detections
+import mediapipe.tasks.python.components.containers.embedding_result as embedding_result
+import mediapipe.tasks.python.components.containers.landmark as landmark
+import mediapipe.tasks.python.components.containers.landmark_detection_result as landmark_detection_result
+import mediapipe.tasks.python.components.containers.rect as rect
 
 AudioDataFormat = audio_data.AudioDataFormat
 AudioData = audio_data.AudioData
@@ -50,4 +50,3 @@
 del landmark
 del landmark_detection_result
 del rect
-del mediapipe

mediapipe/tasks/python/components/processors/__init__.py

@@ -14,10 +14,9 @@
 
 """MediaPipe Tasks Components Processors API."""
 
-import mediapipe.tasks.python.components.processors.classifier_options
+import mediapipe.tasks.python.components.processors.classifier_options as classifier_options
 
 ClassifierOptions = classifier_options.ClassifierOptions
 
 # Remove unnecessary modules to avoid duplication in API docs.
 del classifier_options
-del mediapipe

mediapipe/tasks/python/components/utils/cosine_similarity.py

@@ -24,7 +24,7 @@ def _compute_cosine_similarity(u: np.ndarray, v: np.ndarray):
   """Computes cosine similarity between two embeddings."""
 
   if len(u) <= 0:
-    raise ValueError("Cannot compute cosing similarity on empty embeddings.")
+    raise ValueError("Cannot compute cosine similarity on empty embeddings.")
 
   norm_u = np.linalg.norm(u)
   norm_v = np.linalg.norm(v)

mediapipe/tasks/python/core/pybind/task_runner.cc

@@ -82,7 +82,7 @@ mode) or not (synchronous mode).)doc");
 #if !MEDIAPIPE_DISABLE_GPU
         auto gpu_resources_ = mediapipe::GpuResources::Create();
         if (!gpu_resources_.ok()) {
-          ABSL_LOG(INFO) << "GPU suport is not available: "
+          ABSL_LOG(INFO) << "GPU support is not available: "
                          << gpu_resources_.status();
           gpu_resources_ = nullptr;
         }
@@ -91,7 +91,7 @@ mode) or not (synchronous mode).)doc");
             absl::make_unique<core::MediaPipeBuiltinOpResolver>(),
             std::move(callback),
             /* default_executor= */ nullptr,
-            /* input_side_packes= */ std::nullopt, std::move(*gpu_resources_));
+            /* input_side_packets= */ std::nullopt, std::move(*gpu_resources_));
 #else
         auto task_runner = TaskRunner::Create(
             std::move(graph_config),
@@ -117,7 +117,7 @@ provide the input packets and receive the output packets.
 Raises:
   RuntimeError: Any of the following:
     a) The graph config proto is invalid.
-    b) The underlying medipaipe graph fails to initialize and start.
+    b) The underlying mediapipe graph fails to initialize and start.
 )doc",
       py::arg("graph_config"), py::arg("packets_callback") = py::none());
 
@@ -155,7 +155,7 @@ threads and to ensure that the input packet timestamps are in order.
   RuntimeError: Any of the following:
     a) TaskRunner is in the asynchronous mode (the packets callback is set).
     b) Any input stream name is not valid.
-    c) The underlying medipaipe graph occurs any error during this call.
+    c) The underlying mediapipe graph occurs any error during this call.
 )doc",
       py::arg("input_packets"));
 
@@ -191,7 +191,7 @@ the input packet timestamps are in order.
        queue size or the wrong packet type.
     d) The timestamp of any packet is invalid or is not greater than the
        previously received timestamps.
-    e) The underlying medipaipe graph occurs any error during adding input
+    e) The underlying mediapipe graph occurs any error during adding input
        packets.
 )doc",
       py::arg("input_packets"));
@@ -208,7 +208,7 @@ After the runner is closed, any calls that send input data to the runner are
 illegal and will receive errors.
 
 Raises:
-  RuntimeError: The underlying medipaipe graph fails to close any input streams
+  RuntimeError: The underlying mediapipe graph fails to close any input streams
     or calculators.
 )doc");
 
@@ -223,7 +223,7 @@ illegal and will receive errors.
 This can be useful for resetting a stateful task graph to process new data.
 
 Raises:
-  RuntimeError: The underlying medipaipe graph fails to reset and restart.
+  RuntimeError: The underlying mediapipe graph fails to reset and restart.
 )doc");
 
   task_runner.def(

mediapipe/tasks/python/genai/bundler/__init__.py

@@ -14,7 +14,7 @@
 
 """MediaPipe Tasks GenAI Bundler API."""
 
-import mediapipe.tasks.python.genai.bundler.llm_bundler
+import mediapipe.tasks.python.genai.bundler.llm_bundler as llm_bundler
 
 BundleConfig = llm_bundler.BundleConfig
 create_bundle = llm_bundler.create_bundle

mediapipe/tasks/python/genai/bundler/llm_bundler.py

@@ -18,9 +18,9 @@
 import enum
 from typing import List, Optional
 
+import sentencepiece
 from mediapipe.tasks.python.metadata.metadata_writers import model_asset_bundle_utils
 from mediapipe.tasks.cc.genai.inference.proto import llm_params_pb2
-import sentencepiece
 
 
 @dataclasses.dataclass(frozen=True)
@@ -30,7 +30,7 @@ class BundleConfig:
   Attributes:
     tflite_model: Path to the multi-signature tflite model with "prefill" and
       "decode" signatures converted using ODML Transformers APIs.
-    tokenizer_model: Path to the tokenizer model. Currently only SentencePience
+    tokenizer_model: Path to the tokenizer model. Currently only SentencePiece
       tokenizer is supported. As such, tokenizer.model proto is expected to be
       passed here.
     start_token: Token that will be used to signify the beginning of a sequence.

mediapipe/tasks/python/genai/bundler/llm_bundler_test.py

@@ -17,9 +17,9 @@
 import zipfile
 
 from absl.testing import absltest
+from sentencepiece import sentencepiece_model_pb2
 
 from mediapipe.tasks.python.genai.bundler import llm_bundler
-from sentencepiece import sentencepiece_model_pb2
 
 
 class LlmBundlerTest(absltest.TestCase):
@@ -127,7 +127,7 @@ def test_invalid_stop_tokens_raises_value_error(self):
         tflite_model=tflite_file_path,
         tokenizer_model=sp_file_path,
         start_token=self.BOS,
-        stop_tokens=self.EOS,
+        stop_tokens=[self.EOS],
         output_filename=output_file,
     )
     with self.assertRaisesRegex(

mediapipe/tasks/python/genai/converter/__init__.py

@@ -14,11 +14,10 @@
 
 """MediaPipe Tasks GenAI Converter API."""
 
-import mediapipe.tasks.python.genai.converter.llm_converter
+import mediapipe.tasks.python.genai.converter.llm_converter as llm_converter
 
 ConversionConfig = llm_converter.ConversionConfig
 convert_checkpoint = llm_converter.convert_checkpoint
 
 # Remove unnecessary modules to avoid duplication in API docs.
-del mediapipe
 del llm_converter

mediapipe/tasks/python/genai/converter/converter_base.py

@@ -14,9 +14,8 @@
 
 """Defines a couple base classes for the conversion/quantization process."""
 
-from typing import Iterator
 import os
-from typing import Dict, List, Optional, Tuple
+from typing import Dict, List, Optional, Tuple, Iterator
 import numpy as np