halib 0.1.61__tar.gz → 0.1.65__tar.gz

{halib-0.1.61 → halib-0.1.65}/.gitignore

@@ -51,3 +51,5 @@ Thumbs.db
 build
 dist
 data
+
+venv*/

halib-0.1.65/MANIFEST.in

@@ -0,0 +1,4 @@
+prune _archived
+prune test
+prune zout
+exclude *.toml

{halib-0.1.61 → halib-0.1.65}/PKG-INFO

@@ -1,20 +1,62 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: halib
-Version: 0.1.61
+Version: 0.1.65
 Summary: Small library for common tasks
 Author: Hoang Van Ha
 Author-email: hoangvanhauit@gmail.com
-License: UNKNOWN
-Platform: UNKNOWN
 Classifier: Programming Language :: Python :: 3
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Operating System :: OS Independent
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
+Requires-Dist: arrow
+Requires-Dist: click
+Requires-Dist: enlighten
+Requires-Dist: kaleido==0.1.*
+Requires-Dist: loguru
+Requires-Dist: more-itertools
+Requires-Dist: moviepy
+Requires-Dist: networkx
+Requires-Dist: numpy
+Requires-Dist: omegaconf
+Requires-Dist: opencv-python
+Requires-Dist: pandas
+Requires-Dist: Pillow
+Requires-Dist: Pyarrow
+Requires-Dist: pycurl
+Requires-Dist: python-telegram-bot
+Requires-Dist: requests
+Requires-Dist: rich
+Requires-Dist: scikit-learn
+Requires-Dist: matplotlib
+Requires-Dist: seaborn
+Requires-Dist: plotly
+Requires-Dist: pygwalker
+Requires-Dist: tabulate
+Requires-Dist: itables
+Requires-Dist: timebudget
+Requires-Dist: tqdm
+Requires-Dist: tube_dl
+Requires-Dist: wandb
+Requires-Dist: dataclass-wizard
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 Helper package for coding and automation
 
+**Version 0.1.65**
+
++ now use `uv` for venv management
++ `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation
+
 **Version 0.1.61**
 
 + add `util/video`: add `VideoUtils` class to handle common video-related tasks
@@ -146,5 +188,3 @@ New Features
 New Features
 
 + add support to upload local to google drive.
-
-

{halib-0.1.61 → halib-0.1.65}/README.md

@@ -1,5 +1,10 @@
 Helper package for coding and automation
 
+**Version 0.1.65**
+
++ now use `uv` for venv management
++ `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation
+
 **Version 0.1.61**
 
 + add `util/video`: add `VideoUtils` class to handle common video-related tasks

{halib-0.1.61 → halib-0.1.65}/halib/research/perfcalc.py

@@ -12,6 +12,7 @@ from abc import ABC, abstractmethod
 from ..filetype import csvfile
 from ..common import now_str
 from ..research.perftb import PerfTB
+from collections import OrderedDict
 
 # try to import torch, and torchmetrics
 try:
@@ -62,11 +63,12 @@ REQUIRED_COLS = ["experiment", "dataset"]
 CSV_FILE_POSTFIX = "__perf"
 
 class PerfCalc(ABC): # Abstract base class for performance calculation
+
     @abstractmethod
-    def get_exp_torch_metrics(self):
+    def get_experiment_name(self):
         """
-        Return a dictionary of torchmetrics to be used for performance calculation.
-        Example: {"accuracy": Accuracy(), "precision": Precision()}
+        Return the name of the experiment.
+        This function should be overridden by the subclass if needed.
         """
         pass
 
@@ -79,7 +81,22 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         pass
 
     @abstractmethod
-    def prepare_exp_data_for_metrics(self, metric_names, *args, **kwargs):
+    def get_metrics_info(self):
+        """
+        Return a list of metric names to be used for performance calculation OR a dictionaray with keys as metric names and values as metric instances of torchmetrics.Metric. For example: {"accuracy": Accuracy(), "precision": Precision()}
+
+        """
+        pass
+
+    def calc_exp_outdict_custom_fields(self, outdict, *args, **kwargs):
+        """Can be overridden by the subclass to add custom fields to the output dictionary.
+        ! must return the modified outdict, and a ordered list of custom fields to be added to the output dictionary.
+        """
+        return outdict, []
+
+    # ! can be override, but ONLY if torchmetrics are used
+    # Prepare the exp data for torch metrics.
+    def prepare_torch_metrics_exp_data(self, metric_names, *args, **kwargs):
         """
         Prepare the data for metrics.
         This function should be overridden by the subclass if needed.
@@ -88,76 +105,113 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         """
         pass
 
-    @abstractmethod
-    def get_experiment_name(self):
+    def __validate_metrics_info(self, metrics_info):
         """
-        Return the name of the experiment.
-        This function should be overridden by the subclass if needed.
+        Validate the metrics_info to ensure it is a list or a dictionary with valid metric names and instances.
         """
-        pass
+        if not isinstance(metrics_info, (list, dict)):
+            raise TypeError(f"Metrics info must be a list or a dictionary, got {type(metrics_info).__name__}")
+
+        if isinstance(metrics_info, dict):
+            for k, v in metrics_info.items():
+                if not isinstance(k, str):
+                    raise TypeError(f"Key '{k}' is not a string")
+                if not isinstance(v, Metric):
+                    raise TypeError(f"Value for key '{k}' is not a torchmetrics.Metric (got {type(v).__name__})")
+        elif isinstance(metrics_info, list):
+            for metric in metrics_info:
+                if not isinstance(metric, str):
+                    raise TypeError(f"Metric '{metric}' is not a string")
+        return metrics_info
+    def __calc_exp_perf_metrics(self, *args, **kwargs):
+        """
+        Calculate the performance metrics for the experiment.
+        """
+        metrics_info = self.__validate_metrics_info(self.get_metrics_info())
+        USED_TORCHMETRICS = isinstance(metrics_info, dict)
+        metric_names = metrics_info if isinstance(metrics_info, list) else list(metrics_info.keys())
+        out_dict = {metric: None for metric in metric_names}
+        out_dict["dataset"] = self.get_dataset_name()
+        out_dict["experiment"] = self.get_experiment_name()
+        out_dict, custom_fields = self.calc_exp_outdict_custom_fields(
+            outdict=out_dict, *args, **kwargs
+        )
+        if USED_TORCHMETRICS:
+            torch_metrics_dict = self.get_metrics_info()
+            all_metric_data = self.prepare_torch_metrics_exp_data(
+                metric_names, *args, **kwargs
+            )
+            metric_col_names = []
+            for metric in metric_names:
+                if metric not in all_metric_data:
+                    raise ValueError(f"Metric '{metric}' not found in provided data.")
+                tmetric = torch_metrics_dict[metric]  # torchmetrics instance
+                metric_data = all_metric_data[metric]  # should be a dict of args/kwargs
+                # Inspect expected parameters for the metric's update() method
+                sig = inspect.signature(tmetric.update)
+                expected_args = list(sig.parameters.values())
+                # Prepare args in correct order
+                if isinstance(metric_data, dict):
+                    # Match dict keys to parameter names
+                    args = [metric_data[param.name] for param in expected_args]
+                elif isinstance(metric_data, (list, tuple)):
+                    args = metric_data
+                else:
+                    raise TypeError(f"Unsupported data format for metric '{metric}'")
 
-    def calc_exp_outdict_custom_fields(self, outdict, *args, **kwargs):
-        """Can be overridden by the subclass to add custom fields to the output dictionary.
-        ! must return the modified outdict, and a ordered list of custom fields to be added to the output dictionary.
+                # Call update and compute
+                if len(expected_args) == 1:
+                    tmetric.update(args)  # pass as single argument
+                else:
+                    tmetric.update(*args)  # unpack multiple arguments
+                computed_value = tmetric.compute()
+                # ensure the computed value converted to a scala value or list array
+                if isinstance(computed_value, torch.Tensor):
+                    if computed_value.numel() == 1:
+                        computed_value = computed_value.item()
+                    else:
+                        computed_value = computed_value.tolist()
+                col_name = f"metric_{metric}" if "metric_" not in metric else metric
+                metric_col_names.append(col_name)
+                out_dict[col_name] = computed_value
+        else:
+            # If torchmetrics are not used, calculate metrics using the custom method
+            metric_rs_dict = self.calc_exp_perf_metrics(
+                metric_names, *args, **kwargs)
+            for metric in metric_names:
+                if metric not in metric_rs_dict:
+                    raise ValueError(f"Metric '{metric}' not found in provided data.")
+                col_name = f"metric_{metric}" if "metric_" not in metric else metric
+                out_dict[col_name] = metric_rs_dict[metric]
+            metric_col_names = [f"metric_{metric}" for metric in metric_names]
+        ordered_cols = REQUIRED_COLS + custom_fields + metric_col_names
+        # create a new ordered dictionary with the correct order
+        out_dict = OrderedDict((col, out_dict[col]) for col in ordered_cols if col in out_dict)
+        return out_dict
+
+    # ! only need to override this method if torchmetrics are not used
+    def calc_exp_perf_metrics(self, metric_names, *args, **kwargs):
         """
-        return outdict, []
+        Calculate the performance metrics for the experiment, but not using torchmetrics.
+        This function should be overridden by the subclass if needed.
+        Must return a dictionary with keys as metric names and values as the calculated metrics.
+        """
+        raise NotImplementedError("calc_exp_perf_metrics() must be overridden by the subclass if torchmetrics are not used.")
+
 
     #! custom kwargs:
     #! outfile - if provided, will save the output to a CSV file with the given path
     #! outdir - if provided, will save the output to a CSV file in the given directory with a generated filename
     #! return_df - if True, will return a DataFrame instead of a dictionary
 
-    def calculate_exp_perf_metrics(self, *args, **kwargs):
+    def calc_save_exp_perfs(self, *args, **kwargs):
         """
         Calculate the metrics.
         This function should be overridden by the subclass if needed.
         Must return a dictionary with keys as metric names and values as the calculated metrics.
         """
-        metric_names = list(self.get_exp_torch_metrics().keys())
-        out_dict = {metric: None for metric in metric_names}
-        out_dict['dataset'] = self.get_dataset_name()
-        out_dict['experiment'] = self.get_experiment_name()
-        out_dict, custom_fields = self.calc_exp_outdict_custom_fields(
-            outdict=out_dict, *args, **kwargs
-        )
-        torch_metrics_dict = self.get_exp_torch_metrics()
-        all_metric_data = self.prepare_exp_data_for_metrics(
-            metric_names, *args, **kwargs
-        )
-        metric_col_names = []
-        for metric in metric_names:
-            if metric not in all_metric_data:
-                raise ValueError(f"Metric '{metric}' not found in provided data.")
-            tmetric = torch_metrics_dict[metric] # torchmetrics instance
-            metric_data = all_metric_data[metric]  # should be a dict of args/kwargs
-            # Inspect expected parameters for the metric's update() method
-            sig = inspect.signature(tmetric.update)
-            expected_args = list(sig.parameters.values())
-            # Prepare args in correct order
-            if isinstance(metric_data, dict):
-                # Match dict keys to parameter names
-                args = [metric_data[param.name] for param in expected_args]
-            elif isinstance(metric_data, (list, tuple)):
-                args = metric_data
-            else:
-                raise TypeError(f"Unsupported data format for metric '{metric}'")
-
-            # Call update and compute
-            if len(expected_args) == 1:
-                tmetric.update(args)  # pass as single argument
-            else:
-                tmetric.update(*args)    # unpack multiple arguments
-            computed_value = tmetric.compute()
-            # ensure the computed value converted to a scala value or list array
-            if isinstance(computed_value, torch.Tensor):
-                if computed_value.numel() == 1:
-                    computed_value = computed_value.item()
-                else:
-                    computed_value = computed_value.tolist()
-            col_name = f"metric_{metric}" if 'metric_' not in metric else metric
-            metric_col_names.append(col_name)
-            out_dict[col_name] = computed_value
-
+        out_dict = self.__calc_exp_perf_metrics(*args, **kwargs)
+        # pprint(f"Output Dictionary: {out_dict}")
         # check if any kwargs named "outfile"
         csv_outfile = kwargs.get("outfile", None)
         if csv_outfile is not None:
@@ -171,7 +225,8 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
 
         # convert out_dict to a DataFrame
         df = pd.DataFrame([out_dict])
-        ordered_cols = REQUIRED_COLS + custom_fields + metric_col_names
+        # get the orders of the columns as the orders or the keys in out_dict
+        ordered_cols = list(out_dict.keys())
         df = df[ordered_cols]  # reorder columns
 
         if csv_outfile:
@@ -182,9 +237,17 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         else:
             return out_dict, csv_outfile
 
+    @staticmethod
+    def default_exp_csv_filter_fn(exp_file_name: str) -> bool:
+        """
+        Default filter function for experiments.
+        Returns True if the experiment name does not start with "test_" or "debug_".
+        """
+        return "__perf.csv" in exp_file_name
+
     @classmethod
     def gen_perf_report_for_multip_exps(
-        cls, indir: str, exp_perf_csv_pattern="__perf", csv_sep=';'
+        cls, indir: str, exp_csv_filter_fn=default_exp_csv_filter_fn, csv_sep=";"
     ) -> PerfTB:
         """
         Generate a performance report by scanning experiment subdirectories.
@@ -289,12 +352,12 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
         ]
         if len(exp_dirs) == 0:
             csv_perf_files = glob.glob(
-                os.path.join(indir, f"*{exp_perf_csv_pattern}*.csv")
+                os.path.join(indir, f"*.csv")
             )
             csv_perf_files = [
                 file_item
                 for file_item in csv_perf_files
-                if exp_perf_csv_pattern in file_item
+                if exp_csv_filter_fn(file_item)
             ]
         else:
             # multiple experiment directories found
@@ -302,33 +365,22 @@ class PerfCalc(ABC): # Abstract base class for performance calculation
             for exp_dir in exp_dirs:
                 # pprint(f"Searching in experiment directory: {exp_dir}")
                 matched = glob.glob(
-                    os.path.join(exp_dir, f"*{exp_perf_csv_pattern}*.csv")
+                    os.path.join(exp_dir, f"*.csv")
                 )
+                matched = [
+                    file_item
+                    for file_item in matched
+                    if exp_csv_filter_fn(file_item)
+                ]
                 csv_perf_files.extend(matched)
 
         assert (
             len(csv_perf_files) > 0
-        ), f"No CSV files matching pattern '{exp_perf_csv_pattern}' found in the experiment directories."
+        ), f"No CSV files matching pattern '{exp_csv_filter_fn}' found in the experiment directories."
 
-        assert len(csv_perf_files) > 0, f"No CSV files matching pattern '{exp_perf_csv_pattern}' found in the experiment directories."
+        assert len(csv_perf_files) > 0, f"No CSV files matching pattern '{exp_csv_filter_fn}' found in the experiment directories."
 
         all_exp_perf_df = get_df_for_all_exp_perf(csv_perf_files, csv_sep=csv_sep)
         csvfile.fn_display_df(all_exp_perf_df)
         perf_tb = mk_perftb_report(all_exp_perf_df)
-        return perf_tb
-
-
-def main():
-    indir = "./zreport/test"
-    report_outfile = "./zreport/all.csv"
-    exp_perf_csv_pattern = "__perf"
-    csv_sep = ";"
-    perftb = PerfCalc.gen_perf_report_for_multip_exps(
-        indir, exp_perf_csv_pattern, csv_sep
-    )
-    perftb.to_csv(report_outfile, sep=csv_sep)
-    inspect(perftb)
-    perftb.plot(save_path="./zreport/test_csv.svg", open_plot=True)
-
-if __name__ == "__main__":
-    main()
+        return perf_tb

{halib-0.1.61 → halib-0.1.65}/halib/utils/video.py

@@ -5,15 +5,17 @@ from ..system import filesys as fs
 
 
 class VideoUtils:
+
     @staticmethod
-    def get_video_meta_dict(video_path):
+    def _default_meta_extractor(video_path):
+        """Default video metadata extractor function."""
         # Open the video file
         cap = cv2.VideoCapture(video_path)
 
         # Check if the video was opened successfully
         if not cap.isOpened():
             print(f"Error: Could not open video file {video_path}")
-            return None, None
+            return None
 
         # Get the frame count
         frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
@@ -23,6 +25,7 @@ class VideoUtils:
 
         # Release the video capture object
         cap.release()
+
         meta_dict = {
             "video_path": video_path,
             "frame_count": frame_count,
@@ -31,6 +34,17 @@ class VideoUtils:
         return meta_dict
 
     @staticmethod
+    def get_video_meta_dict(video_path, meta_dict_extractor_func=None):
+        assert os.path.exists(video_path), f"Video file {video_path} does not exist"
+        if meta_dict_extractor_func and callable(meta_dict_extractor_func):
+            assert meta_dict_extractor_func.__code__.co_argcount == 1, "meta_dict_extractor_func must take exactly one argument (video_path)"
+            meta_dict = meta_dict_extractor_func(video_path)
+            assert isinstance(meta_dict, dict), "meta_dict_extractor_func must return a dictionary"
+            assert 'video_path' in meta_dict, "meta_dict must contain 'video_path'"
+        else:
+            meta_dict = VideoUtils._default_meta_extractor(video_path=video_path)
+        return  meta_dict
+    @staticmethod
     def get_video_dir_meta_df(video_dir, video_exts=['.mp4', '.avi', '.mov', '.mkv'], search_recursive=False, csv_outfile=None):
         assert os.path.exists(video_dir), f"Video directory {video_dir} does not exist"
         video_files = fs.filter_files_by_extension(video_dir, video_exts, recursive=search_recursive)

{halib-0.1.61 → halib-0.1.65}/halib.egg-info/PKG-INFO

@@ -1,20 +1,62 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: halib
-Version: 0.1.61
+Version: 0.1.65
 Summary: Small library for common tasks
 Author: Hoang Van Ha
 Author-email: hoangvanhauit@gmail.com
-License: UNKNOWN
-Platform: UNKNOWN
 Classifier: Programming Language :: Python :: 3
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Operating System :: OS Independent
 Requires-Python: >=3.8
 Description-Content-Type: text/markdown
 License-File: LICENSE.txt
+Requires-Dist: arrow
+Requires-Dist: click
+Requires-Dist: enlighten
+Requires-Dist: kaleido==0.1.*
+Requires-Dist: loguru
+Requires-Dist: more-itertools
+Requires-Dist: moviepy
+Requires-Dist: networkx
+Requires-Dist: numpy
+Requires-Dist: omegaconf
+Requires-Dist: opencv-python
+Requires-Dist: pandas
+Requires-Dist: Pillow
+Requires-Dist: Pyarrow
+Requires-Dist: pycurl
+Requires-Dist: python-telegram-bot
+Requires-Dist: requests
+Requires-Dist: rich
+Requires-Dist: scikit-learn
+Requires-Dist: matplotlib
+Requires-Dist: seaborn
+Requires-Dist: plotly
+Requires-Dist: pygwalker
+Requires-Dist: tabulate
+Requires-Dist: itables
+Requires-Dist: timebudget
+Requires-Dist: tqdm
+Requires-Dist: tube_dl
+Requires-Dist: wandb
+Requires-Dist: dataclass-wizard
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: requires-dist
+Dynamic: requires-python
+Dynamic: summary
 
 Helper package for coding and automation
 
+**Version 0.1.65**
+
++ now use `uv` for venv management
++ `research/perfcalc`: support both torchmetrics and custom metrics for performance calculation
+
 **Version 0.1.61**
 
 + add `util/video`: add `VideoUtils` class to handle common video-related tasks
@@ -146,5 +188,3 @@ New Features
 New Features
 
 + add support to upload local to google drive.
-
-

{halib-0.1.61 → halib-0.1.65}/setup.py

@@ -8,7 +8,7 @@ with open("requirements.txt") as f:
 
 setuptools.setup(
     name="halib",
-    version="0.1.61",
+    version="0.1.65",
     author="Hoang Van Ha",
     author_email="hoangvanhauit@gmail.com",
     description="Small library for common tasks",

halib-0.1.61/MANIFEST.in

@@ -1,3 +0,0 @@
-prune _archived
-prune test
-prune zout