def web_definition():
"""
Get dataset metadata from a URL
Loads the metadata for a file specified by the url and name parameters.
Response is the same as for `/dataset/definition`
---
parameters:
- name: name
in: query
description: The name of the dataset file that is to be loaded
required: false
type: string
defaultValue: "Mortality.monthly_MadrasIndia.1916_1921.tab"
- name: url
in: query
description:
The URL address of the file that is to be loaded
required: false
type: string
defaultValue: "http://example.com/interesting/file.csv"
tags:
- Dataverse
responses:
'200':
description: Dataset metadata retrieved
schema:
$ref: "#/definitions/DatasetSchema"
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_url = request.args.get('url', False)
dataset_name = request.args.get('name', False)
username = request.args.get('user', False)
log.debug("Web dataset: {}/{}".format(dataset_url, dataset_name))
# Check whether a file has been provided
if not (dataset_url and dataset_name and username):
raise (Exception("""You should provide a file id, name and user"""))
response = requests.get(dataset_url)
dataset_path = "{}/{}".format(username, dataset_name)
if response.status_code == 200:
gc.add_file(dataset_path, response.content)
dataset_definition = gc.load(dataset_path, dataset_path)
return jsonify(dataset_definition)
else:
raise (Exception(
"The dataset with URI {} could not be retrieved from the Web".
format(dataset_url)))
return jsonify(dataset_definition)
def web_definition():
"""
Get dataset metadata from a URL
Loads the metadata for a file specified by the url and name parameters.
Response is the same as for `/dataset/definition`
---
parameters:
- name: name
in: query
description: The name of the dataset file that is to be loaded
required: false
type: string
defaultValue: "Mortality.monthly_MadrasIndia.1916_1921.tab"
- name: url
in: query
description:
The URL address of the file that is to be loaded
required: false
type: string
defaultValue: "http://example.com/interesting/file.csv"
tags:
- Dataverse
responses:
'200':
description: Dataset metadata retrieved
schema:
$ref: "#/definitions/DatasetSchema"
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_url = request.args.get('url', False)
dataset_name = request.args.get('name', False)
username = request.args.get('user', False)
log.debug("Web dataset: {}/{}".format(dataset_url, dataset_name))
# Check whether a file has been provided
if not(dataset_url and dataset_name and username):
raise(Exception("""You should provide a file id, name and user"""))
response = requests.get(dataset_url)
dataset_path = "{}/{}".format(username, dataset_name)
if response.status_code == 200:
gc.add_file(dataset_path, response.content)
dataset_definition = gc.load(dataset_path, dataset_path)
return jsonify(dataset_definition)
else:
raise(Exception("The dataset with URI {} could not be retrieved from the Web".format(dataset_url)))
return jsonify(dataset_definition)
def dataverse_definition():
"""
Get dataset metadata from a dataverse file
Loads the metadata for a dataverse file specified by the id and name parameters.
Response is the same as for `/dataset/definition`
---
parameters:
- name: name
in: query
description: The name of the dataset file that is to be loaded
required: false
type: string
defaultValue: "Mortality.monthly_MadrasIndia.1916_1921.tab"
- name: id
in: query
description:
The id of a dataverse dataset file that is to be loaded
required: false
type: string
defaultValue: 2531997
- name: user
in: query
description: The email address of the Google user id
required: true
type: string
tags:
- Dataverse
responses:
'200':
description: Dataset metadata retrieved
schema:
$ref: "#/definitions/DatasetSchema"
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_id = request.args.get('id', False)
dataset_name = request.args.get('name', False)
username = request.args.get('user', False)
log.debug("Dataverse dataset: {}/{}".format(dataset_id, dataset_name))
# Check whether a file has been provided
if not (dataset_id and dataset_name and username):
raise (Exception("""You should provide a file id, name and user"""))
dataset_url = dc.Connection().get_access_url(dataset_id)
log.debug("Dataverse url: {}".format(dataset_url))
response = requests.get(dataset_url)
dataset_path = "{}/{}".format(username, dataset_name)
log.debug(dataset_path)
if response.status_code == 200:
gc.add_file(dataset_path, response.content)
dataset_definition = gc.load(dataset_path, dataset_path)
return jsonify(dataset_definition)
else:
raise (Exception(
"The dataset with URI {} could not be retrieved from Dataverse".
format(dataset_url)))
def get_dataset_definition():
"""
Get dataset metadata
Loads the metadata for a dataset specified by the 'path' relative path argument.
---
parameters:
- name: path
in: query
description: The relative path of the dataset file that is to be loaded
required: false
type: string
defaultValue: derived/utrecht_1829_clean_01.csv
tags:
- Dataset
responses:
'200':
description: Dataset metadata retrieved
schema:
id: DatasetSchema
type: object
properties:
name:
type: string
description: The name of the dataset
path:
type: string
description: The location of the dataset on disk (server side)
variables:
description: A dictionary of variable names and values occurring in the dataset
type: object
properties:
default:
description: The default uri for this variable name
type: string
uri:
description: The assigned uri for this variable name
type: string
label:
description: The label for this variable name (i.e. the name itself)
type: string
description:
description: The description of the variable
type: string
category:
description: The category of the variable (coded, identifier, other, community)
type: string
type:
description: The DataCube type of the variable (sdmx:DimensionProperty, ... etc.)
type: string
codelist:
description: If appliccable, the codelist for this variable
type: object
properties:
default:
description: The default URI for the codelist
type: string
uri:
description: The assigned URI for the codelist
type: string
label:
description: The label for the codelist
type: string
values:
description: An array with values and frequencies for this variable name
additionalProperties:
description: The values and frequencies for this variable
schema:
type: object
items:
type: object
default:
description: The default URI representation for this value
type: string
uri:
description: The assigned URI representation for this value
type: string
label:
description: The value as a label
type: string
literal:
description: The assigned Literal representation for this value
count:
type: integer
format: int32
required:
- name
- path
- mappings
- values
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_path = request.args.get('path', False)
# Check whether a file path has been provided
if not dataset_path:
raise (Exception("""You should provide a relative path to
the file you want to load, and specify its name"""))
dataset_name = os.path.basename(dataset_path)
# DEPRECATED: Create an absolute path
# absolute_dataset_path = os.path.join(config.TEMP_PATH, dataset_path)
log.debug('Dataset path: ' + dataset_path)
dataset_definition = gc.load(dataset_name, dataset_path)
return jsonify(dataset_definition)
def dataverse_definition():
"""
Get dataset metadata from a dataverse file
Loads the metadata for a dataverse file specified by the id and name parameters.
Response is the same as for `/dataset/definition`
---
parameters:
- name: name
in: query
description: The name of the dataset file that is to be loaded
required: false
type: string
defaultValue: "Mortality.monthly_MadrasIndia.1916_1921.tab"
- name: id
in: query
description:
The id of a dataverse dataset file that is to be loaded
required: false
type: string
defaultValue: 2531997
- name: user
in: query
description: The email address of the Google user id
required: true
type: string
tags:
- Dataverse
responses:
'200':
description: Dataset metadata retrieved
schema:
$ref: "#/definitions/DatasetSchema"
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_id = request.args.get('id', False)
dataset_name = request.args.get('name', False)
username = request.args.get('user', False)
log.debug("Dataverse dataset: {}/{}".format(dataset_id, dataset_name))
# Check whether a file has been provided
if not(dataset_id and dataset_name and username):
raise(Exception("""You should provide a file id, name and user"""))
dataset_url = dc.Connection().get_access_url(dataset_id)
log.debug("Dataverse url: {}".format(dataset_url))
response = requests.get(dataset_url)
dataset_path = "{}/{}".format(username, dataset_name)
log.debug(dataset_path)
if response.status_code == 200:
gc.add_file(dataset_path, response.content)
dataset_definition = gc.load(dataset_path, dataset_path)
return jsonify(dataset_definition)
else:
raise(Exception("The dataset with URI {} could not be retrieved from Dataverse".format(dataset_url)))
def get_dataset_definition():
"""
Get dataset metadata
Loads the metadata for a dataset specified by the 'path' relative path argument.
---
parameters:
- name: path
in: query
description: The relative path of the dataset file that is to be loaded
required: false
type: string
defaultValue: derived/utrecht_1829_clean_01.csv
tags:
- Dataset
responses:
'200':
description: Dataset metadata retrieved
schema:
id: DatasetSchema
type: object
properties:
name:
type: string
description: The name of the dataset
path:
type: string
description: The location of the dataset on disk (server side)
variables:
description: A dictionary of variable names and values occurring in the dataset
type: object
properties:
default:
description: The default uri for this variable name
type: string
uri:
description: The assigned uri for this variable name
type: string
label:
description: The label for this variable name (i.e. the name itself)
type: string
description:
description: The description of the variable
type: string
category:
description: The category of the variable (coded, identifier, other, community)
type: string
type:
description: The DataCube type of the variable (sdmx:DimensionProperty, ... etc.)
type: string
codelist:
description: If appliccable, the codelist for this variable
type: object
properties:
default:
description: The default URI for the codelist
type: string
uri:
description: The assigned URI for the codelist
type: string
label:
description: The label for the codelist
type: string
values:
description: An array with values and frequencies for this variable name
additionalProperties:
description: The values and frequencies for this variable
schema:
type: object
items:
type: object
default:
description: The default URI representation for this value
type: string
uri:
description: The assigned URI representation for this value
type: string
label:
description: The value as a label
type: string
literal:
description: The assigned Literal representation for this value
count:
type: integer
format: int32
required:
- name
- path
- mappings
- values
default:
description: Unexpected error
schema:
$ref: "#/definitions/Message"
"""
dataset_path = request.args.get('path', False)
# Check whether a file path has been provided
if not dataset_path:
raise(Exception("""You should provide a relative path to
the file you want to load, and specify its name"""))
dataset_name = os.path.basename(dataset_path)
# DEPRECATED: Create an absolute path
# absolute_dataset_path = os.path.join(config.TEMP_PATH, dataset_path)
log.debug('Dataset path: ' + dataset_path)
dataset_definition = gc.load(dataset_name, dataset_path)
return jsonify(dataset_definition)
