class Dhis2Extractor(object):
"""This class fetches DHIS2 data using the dhis2 library and outputs it in various formats."""
FORMAT_CSV = "csv"
def __init__(self, url, *, username, password):
# Let's re-use an existing library instead of maintaining our own
self._api = Api(url, username, password)
def extract_organisation_units(self,
*,
fields=":all",
output_format,
output_path):
"""
Extract organisation units.
Methods should always expose parameters subject to changes, such as field names and output options.
:param fields:
:param output_format:
:param output_path:
:return:
"""
try:
response_data = self._api.get_paged("organisationUnits",
params={"fields": fields},
merge=True)
except RequestException as e: # Always handle exceptions
raise Dhis2ExtractorException(
# Prefer f-strings to concatenation / string interpolation
f"An error occurred while fetching DHIS2 data (URL: {e.url}, status code: {e.code})"
)
organisation_units = response_data["organisationUnits"]
dataframe = pd.DataFrame.from_dict(organisation_units)
return self._dump(dataframe,
output_format=output_format,
output_path=output_path)
def _dump(self, dataframe, *, output_format, output_path):
if output_format == self.FORMAT_CSV:
return dataframe.to_csv(output_path)
raise NotImplementedError(f'Unknown output format "{output_format}"')
def make_category_options_public(self, credentials):
api = Api(credentials.url, credentials.login, credentials.password)
for page in api.get_paged("categoryOptions", params={"fields": ":all"}, page_size=100):
for category_option in page["categoryOptions"]:
if category_option["name"] != "default":
try:
api.post(
"sharing?type=categoryOption&id=" + category_option["id"],
{
"meta": {"allowPublicAccess": True, "allowExternalAccess": False},
"object": {
"id": category_option["id"],
"name": category_option["name"],
"displayName": category_option["displayName"],
"publicAccess": "rwrw----",
"user": category_option["user"],
"externalAccess": False,
},
},
)
except Exception as e:
print("Failed to fix ", category_option["name"], e)
class Dhis2Client:
def __init__(self, *, url, username, password, verbose=False):
self._api = Api(url, username, password)
self.name = url
self.verbose = verbose
def fetch_info(self):
info = self._api.get_info()
self.name = info["systemName"]
return info
def fetch_data_elements(self):
results = []
for page in self._api.get_paged("dataElements",
params={"fields": ":all"},
page_size=100):
if self.verbose:
logger.info(
"sync_log %s: page from data_elements %s",
self.name,
page.get("pager"),
)
results.extend(
[DataElementResult(data) for data in page["dataElements"]])
return results
def fetch_datasets(self):
results = []
for page in self._api.get_paged("dataSets",
params={"fields": ":all"},
page_size=100):
if self.verbose:
logger.info("sync_log %s: page from datasets %s", self.name,
page.get("pager"))
results.extend([DataSetResult(data) for data in page["dataSets"]])
return results
def fetch_indicator_types(self):
results = []
for page in self._api.get_paged("indicatorTypes",
params={"fields": ":all"},
page_size=100):
if self.verbose:
logger.info(
"sync_log %s: page from indicator_types %s",
self.name,
page.get("pager"),
)
results.extend(
[IndicatorTypeResult(data) for data in page["indicatorTypes"]])
return results
def fetch_indicators(self):
results = []
for page in self._api.get_paged("indicators",
params={"fields": ":all"},
page_size=100):
if self.verbose:
logger.info("sync_log %s: page from indicators %s", self.name,
page.get("pager"))
results.extend(
[IndicatorResult(data) for data in page["indicators"]])
return results
def fetch_organisation_units(self):
results = []
for page in self._api.get_paged("organisationUnits",
params={"fields": ":all"},
page_size=100):
if self.verbose:
logger.info(
"sync_log %s: page from organisation_units %s",
self.name,
page.get("pager"),
)
# rewrite path -> replace "/" by "." for correct ltree path
# warning: in place edit, can side effect on tests
for element in page["organisationUnits"]:
if "path" in element:
element["path"] = element["path"].replace("/",
".").strip(".")
results.extend([
OrganisationUnitResult(data)
for data in page["organisationUnits"]
])
return results