def get_data_co2_temp():
data_url = 'https://datahub.io/core/global-temp/datapackage.json'
# to load Data Package into storage
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
data_temperature = []
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
data_temperature.append(data)
data_temperature_annual = data_temperature[0]
data_temperature_annual = data_temperature_annual[
data_temperature_annual['Year'] >= 1980]
data_temperature_annual = data_temperature_annual.sort_values(by=['Year'])
data_temperature_annual = data_temperature_annual[
data_temperature_annual['Source'] == 'GISTEMP']
#get data co2
data_url = 'https://datahub.io/core/co2-ppm/datapackage.json'
# to load Data Package into storage
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
data_co2 = []
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
data_co2.append(data)
data_co2_annual = data_co2[-2]
data_co2_annual = data_co2_annual[data_co2_annual['Year'] < 2017]
data_annual_gistemp = data_co2_annual.rename(columns={'Mean': 'CO2'})
data_annual_gistemp = pd.merge(
data_annual_gistemp, data_temperature_annual,
on='Year').rename(columns={'Mean': 'Temperature'})
data_annual_gistemp['Group'] = 0
i = 1
for year in range(1980, 2016, 6):
if year != 2010:
data_annual_gistemp.loc[(data_annual_gistemp['Year'] >= year) &
(data_annual_gistemp['Year'] <=
(year + 5)),
'Group'] = str(year) + '-' + str(year + 5)
else:
data_annual_gistemp.loc[(data_annual_gistemp['Year'] >= year) &
(data_annual_gistemp['Year'] <=
(year + 6)),
'Group'] = str(year) + '-' + str(year + 6)
i += 1
return data_annual_gistemp
def test_uk_steel_delivered_to_uk_is_consistent():
"""Two tables provide this: (A) the difference between UK production and
exports in Tables 16 and 17; and (B) Table 18 "supply to UK market".
This test checks they match.
"""
# Load data
p = datapackage.Package(DATAPACKAGE)
dfs = {r.name: pd.DataFrame(r.read(keyed=True)) for r in p.resources}
# Version A: production - exports
prod1 = dfs['production_ecsc'].set_index(['year', 'product'])['mass']
prod2 = dfs['production_derived'].set_index(['year', 'product'])['mass']
exports = dfs['exports'].set_index(['year', 'product'])['mass']
prod = pd.concat([prod1, prod2])
A = (prod - exports).loc[2016]
# Version B: supply
supply = dfs['supply'].groupby(['product']).sum()
B = supply[['uk_production_to_stockholders', 'uk_production_to_industry']] \
.sum(axis='columns')
# Compare
df = pd.concat({'A': A, 'B': B}, axis=1)
df['diff'] = df['A'].astype(float) - df['B']
problems = (abs(df['diff']) > 0.5) | (pd.isnull(df['diff']))
assert not any(problems), \
'Differences found:\n\n%s\n' % df[problems]
def __init__(self, descriptor_file):
self._datapackage = datapackage.Package(descriptor_file)
self.__descriptor_file = descriptor_file
self.__base_path = os.path.dirname(
os.path.abspath(self.__descriptor_file))
# Index resources by name
self.__resources = {r.descriptor['name']: r
for r in self._datapackage.resources}
self.__tabular_resources = {k: self._sanitize_resource(r)
for (k, r) in self.__resources.items()
if r.tabular and
r.descriptor['path'].startswith('data')}
self.__invalid_schemas = [] # Resource names with invalid schemas
# All formats
self.raw_data = LazyLoadedDict.from_keys(
self.__resources.keys(),
self._load_raw_data,
'bytes')
# Tabular formats
self.tables = LazyLoadedDict.from_keys(
self.__tabular_resources.keys(),
self._load_table,
type_hint='list of rows')
self.dataframes = LazyLoadedDict.from_keys(
self.__tabular_resources.keys(),
self._load_dataframe,
type_hint='pandas.DataFrame')
def load_dataframe(filename, resource):
"""Load one table from a datapackage."""
package = datapackage.Package(filename)
r = package.get_resource(resource)
if r is None:
raise KeyError('No resource found: %s' % resource)
return pd.DataFrame(r.read(), columns=r.headers)
def test_it(client, doi):
jsr = client.remote_datapackage_json(doi)
dp = datapackage.Package(jsr)
if not dp.valid:
raise AssertionError(
f"Invalid datapackage.json found for {doi} "
f"({jsr['name']}).")
def getdata_old(data_url='https://datahub.io/core/covid-19/datapackage.json',
resourcename='countries-aggregated_csv'):
"""
Get data from the web.
datahub is outdated. Rewrote code to read directly from CSSE.
Parameters:
===========
data_url : string with url of the data from datahub
resourcename : resource to use
Output:
=======
pd.DataFrame with the epidemic statistics data
"""
# to load Data Package into storage
package = datapackage.Package(data_url)
resources = package.resources
for resource in resources:
if resource.name == resourcename:
url = resource.descriptor['path']
print('Importing', url)
s = requests.get(url).text
data = pd.read_csv(StringIO(s))
return data
def extract_airports_data(source, target):
"""
Description:
This function is to extract airports codes data file in csv format and load into s3 data lakes in parquet format using pandas.
Arguments:
source: location for source json file
target: location for output parquet file
Returns:
None
"""
print("INFO: Extracting and loading airports data")
# to load Data Package into storage
package = datapackage.Package(source)
# to load only tabular data
resources = package.resources
for resource in resources:
if resource.descriptor['datahub']['type'] == 'derived/csv':
parquet_file_name = resource.name.split('_')[0] + ".parquet"
parquet_file_path = os.path.join(target, parquet_file_name)
fs = s3fs.S3FileSystem(anon=False)
if fs.exists(parquet_file_path):
print("INFO: {} already processed".format(parquet_file_name))
else:
df = pd.read_csv(resource.descriptor['path'])
df.to_parquet(parquet_file_path)
def validate_save_pkg(pkg_descriptor, pkg_dir):
"""
Validate a data package descriptor and save it to a json file.
Args:
pkg_descriptor (dict):
pkg_dir (path-like):
Returns:
report
"""
# Use that descriptor to instantiate a Package object
data_pkg = datapackage.Package(pkg_descriptor)
# Validate the data package descriptor before we go to
if not data_pkg.valid:
logger.error(f"""
Invalid tabular data package: {data_pkg.descriptor["name"]}
Errors: {data_pkg.errors}""")
# pkg_json is the datapackage.json that we ultimately output:
pkg_json = os.path.join(pkg_dir, "datapackage.json")
data_pkg.save(pkg_json)
logger.info('Validating the data package...')
# Validate the data within the package using goodtables:
report = goodtables.validate(pkg_json, row_limit=1000)
if not report['valid']:
logger.error("Data package validation failed.")
else:
logger.info('Congrats! You made a valid data package!')
return report
def prepare(self, stream, schema, extra):
# Prepare package
if 'datapackage' not in extra or 'resource-name' not in extra:
return False
descriptor = extra['datapackage']
if descriptor.strip().startswith('{'):
descriptor = json.loads(descriptor)
self.__package = datapackage.Package(descriptor)
# Prepare schema
if not schema:
return False
if not schema.foreign_keys:
return False
self.__schema = schema
# Prepare foreign keys values
try:
self.__relations = _get_relations(
self.__package,
self.__schema,
current_resource_name=extra['resource-name'])
self.__foreign_keys_values = _get_foreign_keys_values(
self.__schema, self.__relations)
self.__relations_exception = None
except _ReferenceTableError as exception:
self.__relations_exception = exception
return True
def generate_metadata(pkg_settings, tables, pkg_dir, uuid_pkgs=uuid.uuid4()):
# pkg_json is the datapackage.json that we ultimately output:
pkg_json = os.path.join(pkg_dir, "datapackage.json")
# Create a tabular data resource for each of the tables.
resources = []
for t in tables:
resources.append(get_tabular_data_resource_2(t, pkg_dir=pkg_dir))
data_sources = pudl.helpers.data_sources_from_tables_pkg(tables)
sources = []
for src in data_sources:
if src in pudl.constants.data_sources:
sources.append({"title": src, "path": pc.base_data_urls[src]})
contributors = set()
for src in data_sources:
for c in pudl.constants.contributors_by_source[src]:
contributors.add(c)
pkg_descriptor = {
"name":
pkg_settings["name"],
"profile":
"tabular-data-package",
"title":
pkg_settings["title"],
"id":
uuid_pkgs,
"description":
pkg_settings["description"],
# "keywords": pkg_settings["keywords"],
"homepage":
"https://catalyst.coop/pudl/",
"created":
(datetime.datetime.utcnow().replace(microsecond=0).isoformat() + 'Z'),
"contributors": [pudl.constants.contributors[c] for c in contributors],
"sources":
sources,
"licenses": [pudl.constants.licenses["cc-by-4.0"]],
"resources":
resources,
}
# Use that descriptor to instantiate a Package object
data_pkg = datapackage.Package(pkg_descriptor)
# Validate the data package descriptor before we go to
if not data_pkg.valid:
logger.warning(f"""
Invalid tabular data package: {data_pkg.descriptor["name"]}
Errors: {data_pkg.errors}""")
data_pkg.save(pkg_json)
# Validate the data within the package using goodtables:
report = goodtables.validate(pkg_json, row_limit=1000)
if not report['valid']:
logger.warning("Data package data validation failed.")
return data_pkg, report
def _validate_datapackage(self, datapackage_json: dict):
"""Checks the correctness of datapackage.json metadata. Throws ValueError if invalid."""
dp = datapackage.Package(datapackage_json)
if not dp.valid:
msg = f"Found {len(dp.errors)} datapackage validation errors:\n"
for e in dp.errors:
msg = msg + f" * {e}\n"
raise ValueError(msg)
def __new__(cls):
data_url = 'https://datahub.io/core/s-and-p-500-companies-financials/datapackage.json'
package = datapackage.Package(data_url)
resources = package.resources
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
return data
def load_datapackage_tables(filename):
"""Load all the tables from a datapackage."""
package = datapackage.Package(filename)
tables = {
r.name: pd.DataFrame(r.read(), columns=r.headers)
for r in package.resources
}
return {k: df.set_index(['year', 'product']) for k, df in tables.items()}
def get_dataset():
data_url = 'https://datahub.io/machine-learning/iris/datapackage.json'
package = datapackage.Package(data_url)
resources = package.resources
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
return data
def downloadData(url: str, descriptor: str):
"""
Returns first corrence of provided descriptor from provided url as a file handler
"""
assets = datapackage.Package(url).resources
for data in filter(
lambda x: x.tabular and x.descriptor['name'] == descriptor,
assets):
response = requests.get(data.descriptor['path'])
return io.StringIO(response.content.decode('utf-8'))
def test_cli_init():
resource_path = 'data/valid.csv'
result = CliRunner().invoke(init, [resource_path])
assert result.exit_code == 0
dp = datapackage.Package(json.loads(result.output), strict=True)
resource = dp.resources[0]
assert resource.descriptor['path'] == resource_path
assert 'schema' in resource.descriptor
def get_s_and_p_symbols(self):
package = dp.Package(self.url)
# print list of all resources:
print(package.resource_names)
storage_dict = self.convert_to_dfs(package)
df = storage_dict['constituents']
return df['Symbol'].drop_duplicates()
def get(market):
data_url = 'https://datahub.io/core/' + market + '/datapackage.json'
# to load Data Package into storage
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
return data
def __load_data(self, csv_source):
start = time.time()
if 'http' in csv_source:
import datapackage
package = datapackage.Package(csv_source)
resources = package.resources
for resource in resources:
if resource.tabular:
return pd.read_csv(resource.descriptor['path'])
else:
return pd.read_csv(csv_source)
end = time.time()
print("Loaded data in {} seconds".format(end - start))
def meta_omniglot(data_root_folder=None,
std_num_classes=None,
std_num_examples=None,
one_hot_enc=True,
rand=0,
n_splits=None):
"""
Loads, and downloads if necessary, Omniglot meta-dataset
"""
data_folder_name = 'omniglot_resized'
if em is None:
return experiment_manager_not_available('meta_omniglot NOT AVAILABLE!')
if data_root_folder is None:
data_root_folder = os.path.join(os.getcwd(), 'DATA')
if not os.path.exists(data_root_folder):
os.mkdir(data_root_folder)
data_folder = os.path.join(data_root_folder, data_folder_name)
if os.path.exists(data_folder):
print('DATA FOLDER IS:', data_folder)
print('LOADING META-DATASET')
return em.load.meta_omniglot(data_folder,
std_num_classes=std_num_classes,
std_num_examples=std_num_examples,
one_hot_enc=one_hot_enc,
_rand=rand,
n_splits=n_splits)
else:
print('DOWNLOADING DATA')
package = datapackage.Package(
'https://datahub.io/lucfra/omniglot_resized/datapackage.json')
with open('tmp_omniglot_resized.zip', 'wb') as f:
f.write(package.get_resource('omniglot_resized').raw_read())
import zipfile
zip_ref = zipfile.ZipFile('tmp_omniglot_resized.zip', 'r')
print('EXTRACTING DATA')
zip_ref.extractall(data_root_folder)
zip_ref.close()
os.remove('tmp_omniglot_resized.zip')
print('DONE')
# os.tmpfile()
return meta_omniglot(data_root_folder, std_num_classes,
std_num_examples, one_hot_enc, rand, n_splits)
def extract_tickers():
data_url = 'https://datahub.io/core/nasdaq-listings/datapackage.json'
# to load Data Package into storage
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
for resource in resources:
if resource.tabular:
NASDAQ = pd.read_csv(resource.descriptor['path'])
NASDAQ.to_csv('../data/NASDAQ_Update.csv',index=False)
def validate_save_pkg(pkg_descriptor, pkg_dir):
"""
Validate a data package descriptor and save it to a json file.
Args:
pkg_descriptor (dict):
pkg_dir (path-like):
Returns:
report
"""
# Use that descriptor to instantiate a Package object
data_pkg = datapackage.Package(pkg_descriptor)
# Validate the data package descriptor before we go to
logger.info(
f"Validating JSON descriptor for {data_pkg.descriptor['name']} "
f"tabular data package...")
if not data_pkg.valid:
raise ValueError(
f"Invalid tabular data package: {data_pkg.descriptor['name']} "
f"Errors: {data_pkg.errors}")
logger.info('JSON descriptor appears valid!')
# pkg_json is the datapackage.json that we ultimately output:
pkg_json = os.path.join(pkg_dir, "datapackage.json")
data_pkg.save(pkg_json)
logger.info(
f"Validating a sample of data from {data_pkg.descriptor['name']} "
f"tabular data package using goodtables...")
# Validate the data within the package using goodtables:
report = goodtables.validate(
pkg_json,
# TODO: check which checks are applied... and uncomment out the line
# below when the checks are integrated
# checks=['structure', 'schema', 'foreign-key'],
row_limit=1000)
if not report["valid"]:
goodtables_errors = ""
for table in report["tables"]:
if not table["valid"]:
goodtables_errors += str(table["source"])
goodtables_errors += str(table["errors"])
raise ValueError(
f"Data package data validation failed with goodtables. "
f"Errors: {goodtables_errors}")
logger.info('Congrats! You made a valid data package!')
return report
def read_gold_prices():
data_url = 'https://datahub.io/core/gold-prices/datapackage.json'
package = datapackage.Package(data_url)
resources = package.resources
data = None
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
#print(data.head())
date_field = pd.to_datetime(data['Date'].astype(str), format='%Y-%m')
data['Date'] = date_field
#print(data)
for index, row in data.iterrows():
GoldPrice.objects.create(date=row['Date'], price=row['Price'])
def __init__(self, descriptor=None, title=None, **kwargs):
"""
You should not use the constructor directly but use the infer_from_file static method.
:param descriptor: An initial json data package descriptor. If none it will generate a blank
:param title: if a title is given it will be set in the descriptor along with the required name.
:param kwargs: kwargs that can de passed to the underlying frictionless datapackage.Package
"""
descriptor = descriptor or {}
if title:
descriptor['title'] = title
descriptor['name'] = slugify(title)
self.package = datapackage.Package(descriptor, **kwargs)
self.biosys_errors = []
# set the dataset type to be generic.
self.dataset_type = Dataset.TYPE_GENERIC
def __load_data(self, csv_source):
if 'http' in csv_source:
import datapackage
while 1:
try:
package = datapackage.Package(csv_source)
resources = package.resources
for resource in resources:
if resource.tabular:
return pd.read_csv(resource.descriptor['path'])
except:
print(
"Failed to load Data from {}. WIll reload. Tracebac: {}"
.format(csv_source, traceback.format_exc()))
else:
return pd.read_csv(csv_source)
def create_custom(base_fp, agreement_type="both"):
# Set "both" agreement type
if agreement_type == "both":
agreement_type = ["plain", "explicit"]
else:
agreement_type = [agreement_type]
# TODO: Add a plain/explicit question to the front of GUI and to this...
# ... this will choose the questions to ask and format to print
base_data = json.load(base_fp)
# TODO: The code exists to parse the correct datapackage data from the decisions list
# But, the current Jinja templates just take directly from the datapackage
# Eventually we want to just use the decisions template.
# Append Datapackages to base_data
try:
pkg = datapackage.Package(get_datapackage_path())
for name in pkg.resource_names:
log.debug("loading datapackage {0}".format(name))
resource = pkg.get_resource(name)
base_data[name] = resource.table.read(keyed=True)
except datapackage.exceptions.CastError as _e:
log.debug("Error while attempting to read datapackage resource")
log.error(_e.errors)
raise _e
if "plain" in agreement_type:
# Create plain
res = render("templates/plain.j2", base_data)
md_path = "outputs/plain_custom.md"
compose_agreement(md_path, res)
pdf_path = "outputs/plain_custom.pdf"
pandoc_command = [
"pandoc", "-V", "geometry:margin=1in", "-f", "markdown_github",
"-t", "latex", "-o", pdf_path, md_path
]
subprocess.check_call(pandoc_command)
if "explicit" in agreement_type:
# Create Explicit
res = render("templates/explicit.j2", base_data)
md_path = "outputs/explicit_custom.md"
compose_agreement(md_path, res)
pdf_path = "outputs/explicit_custom.pdf"
pandoc_command = [
"pandoc", "-V", "geometry:margin=1in", "-f", "markdown_github",
"-t", "latex", "-o", pdf_path, md_path
]
subprocess.check_call(pandoc_command)
print("DONE")
def cases_per_capita(df, df3):
data_url = "https://datahub.io/JohnSnowLabs/population-figures-by-country/datapackage.json"
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
for resource in resources:
if resource.tabular:
df2 = pd.read_csv(resource.descriptor['path'])
df3 = new_cases_modified(df)
dct2 = {}
list_of_countries_sorted = df3['Country/Region'].tolist()
new_cases_per_sorted_country = df3.iloc[:, -1].tolist()
dct = {}
result_dict = {}
for i in list_of_countries_sorted:
value = df2.loc[df2['Country'] == str(i)]['Year_2016'].tolist()
for j in value:
dct[str(i)] = j
for key, value in dct.items():
if key not in list_of_countries_sorted:
print(key)
for (country, value) in zip(list_of_countries_sorted,
new_cases_per_sorted_country):
dct2[str(country)] = value
country_list = []
cases_Capita = []
for keys in dct.keys():
country_list.append(keys)
for (key1, value1), (key2, value2) in zip(dct.items(), dct2.items()):
per_capita = value2 / value1
cases_Capita.append(per_capita)
for (i, j) in zip(country_list, cases_Capita):
result_dict[i] = str(j)
return result_dict
def objeto_del_gasto(config):
CT = COLUMN_MAPPING
CN = dict((k, v.replace(':', '-')) for k, v in CT.items())
lookup = {}
codes = datapackage.Package(
os.path.join(os.path.dirname(__file__),
'objeto_del_gasto.datapackage.zip'))
for resource in codes.resources:
kind = resource.name
lookup[kind] = {}
for row in resource.iter(keyed=True):
key = row[kind.upper().replace('Í', 'I')]
value = row['DESCRIPCION']
lookup[kind][key] = value
def process(row):
year = int(row['date-fiscal-year'])
# Skip the LAST year of the dataset (currently 2016) it has split columns already
if year < 2018:
objeto = row[CN['ID_CONCEPTO']]
if objeto:
row[CN['ID_CAPITULO']] = objeto[0] + '000'
row[CN['ID_CONCEPTO']] = objeto[:2] + '00'
row[CN['DESC_CAPITULO']] = lookup['capitulo'].get(
row[CN['ID_CAPITULO']])
row[CN['DESC_CONCEPTO']] = lookup['concepto'].get(
row[CN['ID_CONCEPTO']])
nb_generica_digits = 4 if year in (2008, 2009, 2010) else 3
if objeto and len(objeto) >= 4:
row[CN['ID_PARTIDA_GENERICA']] = objeto[:nb_generica_digits]
row[CN['DESC_PARTIDA_GENERICA']] = lookup['partida_generica'].get(
row.get(CN['ID_PARTIDA_GENERICA']))
if year not in (2008, 2009, 2010):
if objeto and len(objeto) >= 5:
row[CN['ID_PARTIDA_ESPECIFICA']] = objeto
row[CN['DESC_PARTIDA_ESPECIFICA']] = \
lookup['partida_específica'].get(row.get(CN['ID_PARTIDA_ESPECIFICA']))
return process
def get_corr_assets(dataset):
data_url = 'https://datahub.io/core/finance-vix/datapackage.json'
# to load Data Package into storage
package = datapackage.Package(data_url)
# to load only tabular data
resources = package.resources
for resource in resources:
if resource.tabular:
data = pd.read_csv(resource.descriptor['path'])
break
data['Date'] = pd.to_datetime(data.Date, format='%Y-%m-%d')
start = data[data['Date'] == dataset['Date'][0]].index[0]
# Create VIX Close and Open
dataset['vixClose'] = data['VIX Close'][start:].reset_index().drop('index', axis=1)
dataset['vixOpen'] = data['VIX Open'][start:].reset_index().drop('index', axis=1)
return dataset
def init_datapackage(resource_paths):
"""Create tabular data package with resources.
It will also infer the tabular resources' schemas.
Args:
resource_paths (List[str]): Paths to the data package resources.
Returns:
datapackage.Package: The data package.
"""
dp = datapackage.Package({
'name': 'change-me',
'schema': 'tabular-data-package',
})
for path in resource_paths:
dp.infer(path)
return dp
