Merge branch 'dev' into 224-prevent-use-of-run-update-methods-when-the-run-instance-has-not-been-initialised

Author: kzscisoft

.github/workflows/test_client_macos.yml

@@ -28,4 +28,4 @@ jobs:
         export SIMVUE_URL=${{ secrets.SIMVUE_URL }}
         export SIMVUE_TOKEN=${{ secrets.SIMVUE_TOKEN }}
         poetry install --all-extras
-        poetry run pytest tests/unit/ tests/refactor/
+        poetry run pytest tests/unit/ tests/refactor/ -m 'not scenario'

.github/workflows/test_client_ubuntu.yml

@@ -30,7 +30,7 @@ jobs:
         export SIMVUE_URL=${{ secrets.SIMVUE_URL }}
         export SIMVUE_TOKEN=${{ secrets.SIMVUE_TOKEN }}
         poetry install --all-extras
-        poetry run pytest --cov --cov-report=xml tests/unit/ tests/refactor/
+        poetry run pytest --cov --cov-report=xml tests/unit/ tests/refactor/ -m 'not scenario'
     - name: Upload coverage reports to Codecov
       run: |
         curl -Os https://uploader.codecov.io/latest/linux/codecov

.github/workflows/test_client_windows.yml

@@ -29,4 +29,4 @@ jobs:
         export SIMVUE_URL=${{ secrets.SIMVUE_URL }}
         export SIMVUE_TOKEN=${{ secrets.SIMVUE_TOKEN }}
         poetry install --all-extras
-        poetry run pytest tests/unit/ tests/refactor/
+        poetry run pytest tests/unit/ tests/refactor/ -m 'not scenario'

.github/workflows/test_multiple_python.yml

@@ -32,4 +32,4 @@ jobs:
         export SIMVUE_URL=${{ secrets.SIMVUE_URL }}
         export SIMVUE_TOKEN=${{ secrets.SIMVUE_TOKEN }}
         poetry install --all-extras
-        poetry run pytest --cov --cov-report=xml tests/unit/ tests/refactor/
+        poetry run pytest tests/unit/ tests/refactor/ -m 'not scenario'

poetry.lock

@@ -39,19 +39,25 @@ tenacity = "^8.2.3"
 PyJWT = "^2.8.0"
 psutil = "^5.9.8"
 pydantic = "^2.5.3"
-pandas = {version = "^2.2.0", optional = true}
+pandas = "^2.2.0"
 plotly = {version = "^5.18.0", optional = true}
-numpy = {version = "^1.26.3", optional = true}
+numpy = "^1.26.3"
 matplotlib = {version = "^3.8.2", optional = true}
 typing_extensions = { version = "^4.11.0", python = "<3.10" }
 toml = "^0.10.2"
 click = "^8.1.7"
+gitpython = "^3.1.43"
 humanfriendly = "^10.0"
 tabulate = "^0.9.0"
+torch = [
+    {platform = "linux", version = "^2.0.0+cpu", source = "pytorch-cpu", optional=true},
+    {platform = "darwin", version="^2.0.0", source="PyPI", optional=true},
+    {platform = "win32", version="^2.0.0", source="PyPI", optional=true},
+]
 
 [tool.poetry.extras]
-dataset = ["pandas", "numpy"]
 plot = ["matplotlib", "plotly"]
+torch = ["torch"]
 
 [tool.poetry.scripts]
 simvue_sender = "simvue.bin.sender:run"
@@ -63,11 +69,7 @@ ruff = ">=0.3.2,<0.5.0"
 pytest-cov = ">=4.1,<6.0"
 pytest-mock = "^3.14.0"
 pytest-sugar = "^1.0.0"
-torch = [
-    {platform = "linux", version = "^2.0.0+cpu", source = "pytorch-cpu"},
-    {platform = "darwin", version="^2.0.0", source="PyPI"},
-    {platform = "win32", version="^2.0.0", source="PyPI"},
-]
+pytest-xdist = "^3.6.1"
 
 [build-system]
 requires = ["poetry-core"]
@@ -89,7 +91,7 @@ lint.extend-select = ["C901"]
 lint.mccabe.max-complexity = 11
 
 [tool.pytest.ini_options]
-addopts = "-p no:warnings --cov=simvue"
+addopts = "-p no:warnings --cov=simvue -n auto"
 testpaths = [
     "tests"
 ]
@@ -100,7 +102,8 @@ markers = [
     "run: test the simvue Run class",
     "utilities: test simvue utilities module",
     "scenario: test scenarios",
-    "executor: tests of executors"
+    "executor: tests of executors",
+    "api: tests of RestAPI functionality"
 ]
 
 [tool.interrogate]

pyproject.toml

@@ -12,7 +12,13 @@
 import typing
 
 import requests
-from tenacity import retry, stop_after_attempt, wait_exponential
+from tenacity import (
+    retry,
+    stop_after_attempt,
+    wait_exponential,
+    retry_if_exception_type,
+)
+from .utilities import parse_validation_response
 
 DEFAULT_API_TIMEOUT = 10
 RETRY_MULTIPLIER = 1
@@ -34,6 +40,7 @@ def set_json_header(headers: dict[str, str]) -> dict[str, str]:
 @retry(
     wait=wait_exponential(multiplier=RETRY_MULTIPLIER, min=RETRY_MIN, max=RETRY_MAX),
     stop=stop_after_attempt(RETRY_STOP),
+    retry=retry_if_exception_type(RuntimeError),
     reraise=True,
 )
 def post(
@@ -73,6 +80,12 @@ def post(
             f"Authorization error [{response.status_code}]: {response.text}"
         )
 
+    if response.status_code == 422:
+        _parsed_response = parse_validation_response(response.json())
+        raise ValueError(
+            f"Validation error [{response.status_code}]:\n{_parsed_response}"
+        )
+
     if response.status_code not in (200, 201, 409):
         raise RuntimeError(f"HTTP error [{response.status_code}]: {response.text}")
 
@@ -81,6 +94,7 @@ def post(
 
 @retry(
     wait=wait_exponential(multiplier=RETRY_MULTIPLIER, min=RETRY_MIN, max=RETRY_MAX),
+    # retry=retry_if_not_exception_message(match="Validation error"), # if validation failed no point in retrying
     stop=stop_after_attempt(RETRY_STOP),
 )
 def put(

simvue/api.py

@@ -1015,7 +1015,6 @@ def get_metric_values(
             )
 
     @check_extra("plot")
-    @check_extra("dataset")
     def plot_metrics(
         self,
         run_ids: list[str],

simvue/client.py

@@ -7,12 +7,11 @@
 """
 
 import typing
+import pandas
 
 if typing.TYPE_CHECKING:
     from pandas import DataFrame
 
-from .utilities import check_extra
-
 
 def aggregated_metrics_to_dataframe(
     request_response_data: dict[str, list[dict[str, float]]],
@@ -78,9 +77,6 @@ def aggregated_metrics_to_dataframe(
                     )
 
     if parse_to == "dataframe":
-        check_extra("dataset")
-        import pandas
-
         _data_frame = pandas.DataFrame(result_dict)
         _data_frame.index.name = xaxis
         return _data_frame
@@ -174,9 +170,6 @@ def parse_run_set_metrics(
                     result_dict[metric_name][step, run_label] = next_item.get("value")
 
     if parse_to == "dataframe":
-        check_extra("dataset")
-        import pandas
-
         _data_frame = pandas.DataFrame(
             result_dict,
             index=pandas.MultiIndex.from_product(
@@ -194,7 +187,6 @@ def to_dataframe(data):
     """
     Convert runs to dataframe
     """
-    import pandas as pd
 
     metadata = []
     for run in data:
@@ -236,7 +228,7 @@ def to_dataframe(data):
                 else:
                     columns["metadata.%s" % item].append(None)
 
-    return pd.DataFrame(data=columns)
+    return pandas.DataFrame(data=columns)
 
 
 def metric_time_series_to_dataframe(
@@ -260,11 +252,10 @@ def metric_time_series_to_dataframe(
     DataFrame
         a Pandas DataFrame containing values for the metric and run at each
     """
-    import pandas as pd
 
     _df_dict: dict[str, list[float]] = {
         xaxis: [v[xaxis] for v in data],
         name or "value": [v["value"] for v in data],
     }
 
-    return pd.DataFrame(_df_dict)
+    return pandas.DataFrame(_df_dict)

simvue/converters.py

@@ -0,0 +1,74 @@
+"""
+Metadata
+========
+
+Contains functions for extracting additional metadata about the current project
+
+"""
+
+import typing
+import git
+import json
+
+
+def git_info(repository: str) -> dict[str, typing.Any]:
+    """Retrieves metadata for the target git repository
+
+    This is a passive function which returns an empty dictionary if any
+    metadata is missing. Exceptions are raised only if the repository
+    does not have the expected form assumed by this retrieval method.
+
+    Parameters
+    ----------
+    repository : str
+        root directory of the git repository
+
+    Returns
+    -------
+    dict[str, typing.Any]
+        metadata for the target repository
+    """
+    try:
+        git_repo = git.Repo(repository, search_parent_directories=True)
+    except git.InvalidGitRepositoryError:
+        return {}
+    current_commit: git.Commit = git_repo.head.commit
+    author_list: set[str] = set(
+        email
+        for commit in git_repo.iter_commits("--all")
+        if "noreply" not in (email := (commit.author.email or ""))
+        and "[bot]" not in (commit.author.name or "")
+    )
+
+    ref: str = current_commit.hexsha
+
+    # In the case where the repository is dirty blame should point to the
+    # current developer, not the person responsible for the latest commit
+    if dirty := git_repo.is_dirty():
+        blame = git_repo.config_reader().get_value("user", "email", "unknown")
+    else:
+        blame = current_commit.committer.email
+
+    for tag in git_repo.tags:
+        if tag.commit == current_commit:
+            ref = tag.name
+            break
+
+    return {
+        "git.authors": json.dumps(list(author_list)),
+        "git.ref": ref,
+        "git.msg": current_commit.message.strip(),
+        "git.time_stamp": current_commit.committed_datetime.strftime(
+            "%Y-%m-%d %H:%M:%S %z UTC"
+        ),
+        "git.blame": blame,
+        "git.url": git_repo.remote().url,
+        "git.dirty": dirty,
+    }
+
+
+if __name__ in "__main__":
+    import os.path
+    import json
+
+    print(json.dumps(git_info(os.path.dirname(__file__)), indent=2))