From 273821dc854ba53015e022600574655160ce1a40 Mon Sep 17 00:00:00 2001 From: Tibor Frank Date: Wed, 12 Apr 2023 07:55:52 +0200 Subject: C-Dash: Add coverage tables Signed-off-by: Tibor Frank Change-Id: I4f20936db02dc4b974134c676c368c479e931038 --- csit.infra.dash/app/cdash/data/data.py | 67 ++++++++++++++++++++++---------- csit.infra.dash/app/cdash/data/data.yaml | 50 +++++++++++++++++++++++- 2 files changed, 95 insertions(+), 22 deletions(-) (limited to 'csit.infra.dash/app/cdash/data') diff --git a/csit.infra.dash/app/cdash/data/data.py b/csit.infra.dash/app/cdash/data/data.py index 8537cd8db1..c8d5907200 100644 --- a/csit.infra.dash/app/cdash/data/data.py +++ b/csit.infra.dash/app/cdash/data/data.py @@ -48,7 +48,12 @@ class Data: self._data_spec = list() # Data frame to keep the data: - self._data = pd.DataFrame() + self._data = { + "statistics": pd.DataFrame(), + "trending": pd.DataFrame(), + "iterative": pd.DataFrame(), + "coverage": pd.DataFrame() + } # Read from files: try: @@ -69,11 +74,13 @@ class Data: def data(self): return self._data - def _get_list_of_files(self, - path, - last_modified_begin=None, - last_modified_end=None, - days=None) -> list: + @staticmethod + def _get_list_of_files( + path, + last_modified_begin=None, + last_modified_end=None, + days=None + ) -> list: """Get list of interested files stored in S3 compatible storage and returns it. @@ -111,11 +118,11 @@ class Data: return file_list + @staticmethod def _create_dataframe_from_parquet( - self, path, partition_filter=None, columns=None, - categories=list(), + categories=None, validate_schema=False, last_modified_begin=None, last_modified_end=None, @@ -157,7 +164,7 @@ class Data: :returns: Pandas DataFrame or None if DataFrame cannot be fetched. :rtype: DataFrame """ - df = None + df = pd.DataFrame() start = time() if days: last_modified_begin = datetime.now(tz=UTC) - timedelta(days=days) @@ -170,7 +177,7 @@ class Data: use_threads=True, dataset=True, columns=columns, - # categories=categories, + categories=categories, partition_filter=partition_filter, last_modified_begin=last_modified_begin, last_modified_end=last_modified_end @@ -180,9 +187,19 @@ class Data: f"\nCreation of dataframe {path} took: {time() - start}\n" ) except NoFilesFound as err: - logging.error(f"No parquets found.\n{err}") + logging.error( + f"No parquets found in specified time period.\n" + f"Nr of days: {days}\n" + f"last_modified_begin: {last_modified_begin}\n" + f"{err}" + ) except EmptyDataFrame as err: - logging.error(f"No data.\n{err}") + logging.error( + f"No data in parquets in specified time period.\n" + f"Nr of days: {days}\n" + f"last_modified_begin: {last_modified_begin}\n" + f"{err}" + ) return df @@ -197,11 +214,9 @@ class Data: :rtype: dict(str: pandas.DataFrame) """ - self._data = dict() - self._data["trending"] = pd.DataFrame() - self._data["iterative"] = pd.DataFrame() lst_trending = list() lst_iterative = list() + lst_coverage = list() for data_set in self._data_spec: logging.info( @@ -211,13 +226,16 @@ class Data: partition_filter = lambda part: True \ if part[data_set["partition"]] == data_set["partition_name"] \ else False - - data = self._create_dataframe_from_parquet( + if data_set["data_type"] in ("trending", "statistics"): + time_period = days + else: + time_period = None + data = Data._create_dataframe_from_parquet( path=data_set["path"], partition_filter=partition_filter, - columns=data_set.get("columns", list()), - categories=data_set.get("categories", list()), - days=None if data_set["data_type"] == "iterative" else days + columns=data_set.get("columns", None), + categories=data_set.get("categories", None), + days=time_period ) if data_set["data_type"] == "statistics": @@ -228,6 +246,10 @@ class Data: data["release"] = data_set["release"] data["release"] = data["release"].astype("category") lst_iterative.append(data) + elif data_set["data_type"] == "coverage": + data["release"] = data_set["release"] + data["release"] = data["release"].astype("category") + lst_coverage.append(data) else: raise NotImplementedError( f"The data type {data_set['data_type']} is not implemented." @@ -243,6 +265,11 @@ class Data: ignore_index=True, copy=False ) + self._data["coverage"] = pd.concat( + lst_coverage, + ignore_index=True, + copy=False + ) for key in self._data.keys(): logging.info( diff --git a/csit.infra.dash/app/cdash/data/data.yaml b/csit.infra.dash/app/cdash/data/data.yaml index 4e78323fe7..975241b84e 100644 --- a/csit.infra.dash/app/cdash/data/data.yaml +++ b/csit.infra.dash/app/cdash/data/data.yaml @@ -90,7 +90,6 @@ - passed - telemetry - test_id - - test_type - version categories: - job @@ -301,14 +300,41 @@ - passed - telemetry - test_id - - test_type - version categories: - job - build - dut_type - dut_version + - tg_type + - version +- data_type: coverage + partition: test_type + partition_name: ndrpdr + release: rls2302 + path: s3://fdio-docs-s3-cloudfront-index/csit/parquet/coverage_rls2302 + columns: + - job + - build + - dut_type + - dut_version + - tg_type + - hosts + - start_time + - passed + - test_id - version + - result_pdr_lower_rate_unit + - result_pdr_lower_rate_value + - result_ndr_lower_rate_value + - result_pdr_lower_bandwidth_value + - result_ndr_lower_bandwidth_value + - result_latency_reverse_pdr_90_hdrh + - result_latency_reverse_pdr_50_hdrh + - result_latency_reverse_pdr_10_hdrh + - result_latency_forward_pdr_90_hdrh + - result_latency_forward_pdr_50_hdrh + - result_latency_forward_pdr_10_hdrh categories: - job - build @@ -316,3 +342,23 @@ - dut_version - tg_type - version +# - data_type: coverage +# partition: test_type +# partition_name: device +# release: rls2302 +# path: s3://fdio-docs-s3-cloudfront-index/csit/parquet/coverage_rls2302 +# columns: +# - job +# - build +# - dut_type +# - dut_version +# - start_time +# - passed +# - test_id +# - version +# categories: +# - job +# - build +# - dut_type +# - dut_version +# - version -- cgit 1.2.3-korg