def _verify_tag_path_exist(adls_file_system_client: core.AzureDLFileSystem,
path: str):
"""
Verify that the tag path exists, if not the `adls_file_system_client.info` will raise a FileNotFound error.
Parameters
----------
adls_file_system_client: core.AzureDLFileSystem
the AzureDLFileSystem client to use
path : str
Path of tag to be checked if exists.
"""
adls_file_system_client.info(f"{path}")
def read_tag_files(adls_file_system_client: core.AzureDLFileSystem,
tag: SensorTag, years: range) -> pd.Series:
"""
Download tag files for the given years into dataframes,
and return as one dataframe.
Parameters
----------
adls_file_system_client: core.AzureDLFileSystem
the AzureDLFileSystem client to use
tag: SensorTag
the tag to download data for
years: range
range object providing years to include
Returns
-------
pd.Series:
Series with all years for one tag.
"""
tag_base_path = NcsReader.base_path_from_asset(tag.asset)
if not tag_base_path:
raise ValueError(f"Unable to find base path from tag {tag} ")
all_years = []
for year in years:
file_path = tag_base_path + f"/{tag.name}/{tag.name}_{year}.csv"
logger.info(f"Parsing file {file_path}")
info = adls_file_system_client.info(file_path)
file_size = info.get("length") / (1024**2)
logger.info(f"File size: {file_size:.2f}MB")
with adls_file_system_client.open(file_path, "rb") as f:
df = pd.read_csv(
f,
sep=";",
header=None,
names=["Sensor", tag.name, "Timestamp", "Status"],
usecols=[tag.name, "Timestamp"],
dtype={tag.name: np.float32},
parse_dates=["Timestamp"],
date_parser=lambda col: pd.to_datetime(col, utc=True),
index_col="Timestamp",
)
all_years.append(df)
logger.info(f"Done parsing file {file_path}")
combined = pd.concat(all_years)
# There often comes duplicated timestamps, keep the last
if combined.index.duplicated().any():
combined = combined[~combined.index.duplicated(keep="last")]
return combined[tag.name]
def read_tag_files(
self,
adls_file_system_client: core.AzureDLFileSystem,
tag: SensorTag,
years: range,
dry_run: Optional[bool] = False,
remove_status_codes: Optional[list] = [0],
dl_base_path: Optional[str] = None,
) -> pd.Series:
"""
Download tag files for the given years into dataframes,
and return as one dataframe.
Parameters
----------
adls_file_system_client: core.AzureDLFileSystem
the AzureDLFileSystem client to use
tag: SensorTag
the tag to download data for
years: range
range object providing years to include
dry_run: Optional[bool]
if True, don't download data, just check info, log, and return
remove_status_codes: Optional[list]
Removes data with Status code(s) in the list. By default it removes data
with Status code 0.
dl_base_path: Optional[str]
Base bath used to override the asset to path dictionary. Useful for demos
and other non-production settings.
Returns
-------
pd.Series:
Series with all years for one tag.
"""
tag_base_path = (dl_base_path if dl_base_path else
NcsReader.base_path_from_asset(tag.asset))
if not tag_base_path:
raise ValueError(f"Unable to find base path from tag {tag} ")
all_years = []
logger.info(f"Downloading tag: {tag} for years: {years}")
tag_name_encoded = quote(tag.name, safe=" ")
NcsReader._verify_tag_path_exist(
adls_file_system_client, f"{tag_base_path}/{tag_name_encoded}/")
dir_path = f"{tag_base_path}/{tag_name_encoded}"
for year in years:
file_path = None
file_lookup = None
for v in self.file_lookups:
file_path = v.lookup(adls_file_system_client, dir_path,
tag_name_encoded, year)
if file_path is not None:
file_lookup = v
break
if file_lookup is None:
continue
file_type = file_lookup.file_type
logger.info(f"Parsing file {file_path}")
try:
info = adls_file_system_client.info(file_path)
file_size = info.get("length") / (1024**2)
logger.debug(
f"File size for file {file_path}: {file_size:.2f}MB")
if dry_run:
logger.info("Dry run only, returning empty frame early")
return pd.Series()
before_downloading = timeit.default_timer()
with adls_file_system_client.open(file_path, "rb") as f:
df = file_type.read_df(f)
df = df.rename(columns={"Value": tag.name})
df = df[~df["Status"].isin(remove_status_codes)]
df.sort_index(inplace=True)
all_years.append(df)
logger.info(
f"Done in {(timeit.default_timer()-before_downloading):.2f} sec {file_path}"
)
except FileNotFoundError as e:
logger.debug(f"{file_path} not found, skipping it: {e}")
try:
combined = pd.concat(all_years)
except Exception as e:
logger.debug(f"Not able to concatinate all years: {e}.")
return pd.Series(name=tag.name, data=None)
# There often comes duplicated timestamps, keep the last
if combined.index.duplicated().any():
combined = combined[~combined.index.duplicated(keep="last")]
return combined[tag.name]
def is_file(client: core.AzureDLFileSystem, path: str):
try:
info = client.info(path)
except FileNotFoundError:
return False
return info["type"] == "FILE"
def read_tag_files(
adls_file_system_client: core.AzureDLFileSystem,
tag: SensorTag,
years: range,
dry_run: Optional[bool] = False,
remove_status_codes: Optional[list] = [0],
dl_base_path: Optional[str] = None,
) -> pd.Series:
"""
Download tag files for the given years into dataframes,
and return as one dataframe.
Parameters
----------
adls_file_system_client: core.AzureDLFileSystem
the AzureDLFileSystem client to use
tag: SensorTag
the tag to download data for
years: range
range object providing years to include
dry_run: Optional[bool]
if True, don't download data, just check info, log, and return
remove_status_codes: Optional[list]
Removes data with Status code(s) in the list. By default it removes data
with Status code 0.
dl_base_path: Optional[str]
Base bath used to override the asset to path dictionary. Useful for demos
and other non-production settings.
Returns
-------
pd.Series:
Series with all years for one tag.
"""
tag_base_path = (
dl_base_path if dl_base_path else NcsReader.base_path_from_asset(tag.asset)
)
if not tag_base_path:
raise ValueError(f"Unable to find base path from tag {tag} ")
all_years = []
logger.info(f"Downloading tag: {tag} for years: {years}")
tag_name_encoded = quote(tag.name, safe=" ")
NcsReader._verify_tag_path_exist(
adls_file_system_client, f"{tag_base_path}/{tag_name_encoded}/"
)
for year in years:
file_path = (
f"{tag_base_path}/{tag_name_encoded}/{tag_name_encoded}_{year}.csv"
)
logger.info(f"Parsing file {file_path}")
try:
info = adls_file_system_client.info(file_path)
file_size = info.get("length") / (1024 ** 2)
logger.debug(f"File size for file {file_path}: {file_size:.2f}MB")
if dry_run:
logger.info("Dry run only, returning empty frame early")
return pd.Series()
with adls_file_system_client.open(file_path, "rb") as f:
df = pd.read_csv(
f,
sep=";",
header=None,
names=["Sensor", tag.name, "Timestamp", "Status"],
usecols=[tag.name, "Timestamp", "Status"],
dtype={tag.name: np.float32},
parse_dates=["Timestamp"],
date_parser=lambda col: pd.to_datetime(col, utc=True),
index_col="Timestamp",
)
df = df[~df["Status"].isin(remove_status_codes)]
all_years.append(df)
logger.info(f"Done parsing file {file_path}")
except FileNotFoundError as e:
logger.debug(f"{file_path} not found, skipping it: {e}")
try:
combined = pd.concat(all_years)
except Exception as e:
logger.debug(f"Not able to concatinate all years: {e}.")
return pd.Series(name=tag.name, data=None)
# There often comes duplicated timestamps, keep the last
if combined.index.duplicated().any():
combined = combined[~combined.index.duplicated(keep="last")]
return combined[tag.name]