experimental sdk for the socrata data-pipeline api
- Development
- Library Docs
This only supports python3.
Installation is available through pip. Using a virtualenv is advised. Install the package by running
pip3 install socrata-py
The only hard dependency is requests which will be installed via pip. Pandas is not required, but creating a dataset from a Pandas dataframe is supported. See below.
Try the command line example with
python -m examples.create 'Police Reports' ~/Desktop/catalog.data.gov/Seattle_Real_Time_Fire_911_Calls.csv 'pete-test.test-socrata.com' --username $SOCRATA_USERNAME --password $SOCRATA_PASSWORD# Import some stuff
from socrata.authorization import Authorization
from socrata import Socrata
import os
# Boilerplate...
# Make an auth object
auth = Authorization(
"pete-test.test-socrata.com",
os.environ['SOCRATA_USERNAME'],
os.environ['SOCRATA_PASSWORD']
)To create a dataset, you can do this:
with open('cool_dataset.csv', 'rb') as file:
# Upload + Transform step
# revision is the *change* to the view in the catalog, which has not yet been applied.
# output is the OutputSchema, which is a change to data which can be applied via the revision
(revision, output) = Socrata(auth).create(
name = "cool dataset",
description = "a description"
).csv(file)
# Transformation step
# We want to add some metadata to our column, drop another column, and add a new column which will
# be filled with values from another column and then transformed
(ok, output) = output\
.change_column_metadata('a_column', 'display_name').to('A Column!')\
.change_column_metadata('a_column', 'description').to('Here is a description of my A Column')\
.drop_column('b_column')\
.add_column('a_column_squared', 'A Column, but times itself', 'to_number(`a_column`) * to_number(`a_column`)', 'this is a column squared')\
.run()
# Validation of the results step
(ok, output) = output.wait_for_finish()
# The data has been validated now, and we can access errors that happened during validation. For example, if one of the cells in `a_column` couldn't be converted to a number in the call to `to_number`, that error would be reflected in this error_count
assert output.attributes['error_count'] == 0
# If you want, you can get a csv stream of all the errors
(ok, errors) = output_schema.schema_errors_csv()
for line in errors.iter_lines():
print(line)
# Update step
# Apply the revision - this will make it public and available to make
# visualizations from
(ok, job) = revision.apply(output_schema = output)
# This opens a browser window to your revision, and you will see the progress
# of the job
rev.open_in_browser()
# Application is async - this will block until all the data
# is in place and readable
job.wait_for_finish()Similar to the csv method are the xls, xlsx, and tsv methods, which upload
those files.
There is a blob method as well, which uploads blobby data to the source. This means the data will not be parsed, and will be displayed under "Files and Documents" in the catalog once the revision is applied.
Datasets can also be created from Pandas DataFrames
import pandas as pd
df = pd.read_csv('socrata-py/test/fixtures/simple.csv')
# Do various Pandas-y changes and modifications, then...
(revision, output) = Socrata(auth).create(
name = "Pandas Dataset",
description = "Dataset made from a Pandas Dataframe"
).df(df)
# Same code as above to apply the revision.A Socrata update is actually an upsert. Rows are updated or created based on the row identifier. If the row-identifer doesn't exist, all updates are just appends to the dataset.
A replace truncates the whole dataset and then inserts the new data.
# This is how we create our view initially
with open('cool_dataset.csv', 'rb') as file:
(revision, output) = Socrata(auth).create(
name = "cool dataset",
description = "a description"
).csv(file)
revision.apply(output_schema = output)
# This will build a configuration using the same settings (file parsing and
# data transformation rules) that we used to get our output. The action
# that we will take will be "update", though it could also be "replace"
(ok, config) = output.build_config("cool-dataset-config", "update")
# Now we need to save our configuration name and view id somewhere so we
# can update the view using our config
configuration_name = "cool-dataset-config"
view_id = revision.view_id()
# Now later, if we want to use that config to update our view, we just need the view and the configuration_name
socrata = Socrata(auth)
(ok, view) = socrata.views.lookup(view_id) # View will be the view we are updating with the new data
with open('updated-cool-dataset.csv', 'rb') as my_file:
(rev, job) = socrata.using_config(configuration_name, view).csv(my_file)
print(job) # Our update job is now running# This is our socrata object, using the auth variable from above
socrata = Socrata(auth)
# This will make our initial revision, on a view that doesn't yet exist
(ok, rev) = socrata.new({'name': 'cool dataset'})
assert ok
# rev is a Revision object, we can print it
print(rev)
Revision({'created_by': {'display_name': 'rozap',
'email': 'chris.duranti@socrata.com',
'user_id': 'tugg-ikce'},
'fourfour': 'ij46-xpxe',
'id': 346,
'inserted_at': '2017-02-27T23:05:08.522796',
'metadata': None,
'update_seq': 285,
'upsert_jobs': []})
# We can also access the attributes of the revision
print(rev.attributes['metadata']['name'])
'cool dataset'# Using that revision, we can create an upload
(ok, upload) = rev.create_upload('foo.csv')
assert ok
# And print it
print(upload)
Source({'content_type': None,
'created_by': {'display_name': 'rozap',
'email': 'chris.duranti@socrata.com',
'user_id': 'tugg-ikce'},
'source_type': {
'filename': 'foo.csv',
'type': 'upload'
},
'finished_at': None,
'id': 290,
'inserted_at': '2017-02-27T23:07:18.309676',
'schemas': []})# And using that upload we just created, we can put bytes into it
with open('test/fixtures/simple.csv', 'rb') as f:
(ok, source) = upload.csv(f)
assert okTransforming data consists of going from input data (data exactly as it appeared in the source) to output data (data as you want it to appear).
Transformation from input data to output data often has problems. You might, for example, have a column
full of numbers, but one row in that column is actually the value hehe! which cannot be transformed into
a number. Rather than failing at each datum which is dirty or wrong, transforming your data allows you to
reconcile these issues.
We might have a dataset called temps.csv that looks like
date, celsius
8-24-2017, 22
8-25-2017, 20
8-26-2017, 23
8-27-2017, hehe!
8-28-2017,
8-29-2017, 21
Suppose we uploaded it in our previous step, like this:
with open('temps.csv', 'rb') as f:
(ok, source) = upload.csv(f)
assert ok
input_schema = source.get_latest_input_schema()Our input_schema is the input data exactly as it appeared in the CSV, with all values of type string.
Our output_schema is the output data as it was guessed by Socrata. Guessing may not always be correct, which is why we have import configs to "lock in" a schema for automation. We can get the output_schema
like so:
(ok, output_schema) = input_schema.get_latest_output_schema()()
assert okWe can now make changes to the schema, like so
(ok, new_output_schema) = output
# Change the field_name of date to the_date
.change_column_metadata('date', 'field_name').to('the_date')\
# Change the description of the celsius column
.change_column_metadata('celsius', 'description').to('the temperature in celsius')\
# Change the display name of the celsius column
.change_column_metadata('celsius', 'display_name').to('Degrees (Celsius)')\
# Change the transform of the_date column to to_fixed_timestamp(`date`)
.change_column_transform('the_date').to('to_fixed_timestamp(`date`)')\
# Make the celsius column all numbers
.change_column_transform('celsius').to('to_number(`celsius`)')\
# Add a new column, which is computed from the `celsius` column
.add_column('fahrenheit', 'Degrees (Fahrenheit)', '(to_number(`celsius`) * (9 / 5)) + 32', 'the temperature in celsius')\
.run()change_column_metadata(column_name, column_attribute) takes the field name used to
identify the column and the column attribute to change (field_name, display_name, description, position)
add_column(field_name, display_name, transform_expression, description) will create a new column
We can also call drop_column(celsius) which will drop the column.
.run() will then make a request and return the new output_schema, or an error if something is invalid.
Transforms can be complex SoQL expressions. Available functions are listed here. You can do lots of stuff with them;
For example, you could change all null values into errors (which won't be imported) by doing
something like
(ok, new_output_schema) = output
.change_column_transform('celsius').to('coalesce(to_number(`celsius`), error("Celsius was null!"))')
.run()Or you could add a new column that says if the day was hot or not
(ok, new_output_schema) = output
.add_column('is_hot', 'Was the day hot?', 'to_number(`celsius`) >= 23')
.run()Or you could geocode a column, given the following CSV
address,city,zip,state
10028 Ravenna Ave NE, Seattle, 98128, WA
1600 Pennsylvania Avenue, Washington DC, 20500, DC
6511 32nd Ave NW, Seattle, 98155, WA
We could transform our first output_schema into a single column dataset, where that
single column is a Point of the address
(ok, output) = output\
.add_column('location', 'Incident Location', 'geocode(`address`, `city`, `state`, `zip`)')\
.drop_column('address')\
.drop_column('city')\
.drop_column('state')\
.drop_column('zip')\
.run()Composing these SoQL functions into expressions will allow you to validate, shape, clean and extend your data to make it more useful to the consumer.
Transformations are async, so if you want to wait for it to finish, you can do so
(ok, output_schema) = new_output_schema.wait_for_finish()
assert ok, output_schemaTransformations may have had errors, like in the previous example, we can't convert hehe! to a number. We can see the count of them like this:
print(output_schema.attributes['error_count'])We can view the detailed errors like this:
(ok, errors) = output_schema.schema_errors()We can get a CSV of the errors like this:
(ok, csv_stream) = output_schema.schema_errors_csv()We can look at the rows of our schema as well
(ok, rows) = output_schema.rows(offset = 0, limit = 20)
assert ok,
self.assertEqual(rows, [
{'b': {'ok': ' bfoo'}},
{'b': {'ok': ' bfoo'}},
{'b': {'ok': ' bfoo'}},
{'b': {'ok': ' bfoo'}}
])# Now we have transformed our data into the shape we want, let's do an upsert
(ok, job) = rev.apply(output_schema = output_schema)
# This will complete the upsert behind the scenes. If we want to
# re-fetch the current state of the upsert job, we can do so
(ok, job) = job.show()
# To get the progress
print(job.attributes['log'])
[
{'details': {'Errors': 0, 'Rows Created': 0, 'Rows Updated': 0, 'By RowIdentifier': 0, 'By SID': 0, 'Rows Deleted': 0}, 'time': '2017-02-28T20:20:59', 'stage': 'upsert_complete'},
{'details': {'created': 1}, 'time': '2017-02-28T20:20:59', 'stage': 'columns_created'},
{'details': {'created': 1}, 'time': '2017-02-28T20:20:59', 'stage': 'columns_created'},
{'details': None, 'time': '2017-02-28T20:20:59', 'stage': 'started'}
]
# So maybe we just want to wait here, printing the progress, until the job is done
job.wait_for_finish(progress = lambda job: print(job.attributes['log']))
# So now if we go look at our original four-four, our data will be thereWhen there is an existing Socrata view that you'd like to update the metadata of, you can do so by creating a Source which is the Socrata view.
(ok, view) = socrata.views.lookup('abba-cafe')
assert ok, view
(ok, rev) = view.revisions.create_replace_revision()
assert ok, rev
(ok, source) = revision.source_from_dataset()
assert ok, source
output_schema = source.get_latest_input_schema().get_latest_output_schema()
(ok, new_output_schema) = output_schema\
.change_column_metadata('a', 'description').to('meh')\
.change_column_metadata('b', 'display_name').to('bbbb')\
.change_column_metadata('c', 'field_name').to('ccc')\
.run()
revision.apply(output_schema = new_output_schema)Install test deps by running pip install -r requirements.txt. This will install pdoc and pandas which are required to run the tests.
Configuration is set in test/auth.py for tests. It reads the domain, username, and password from environment variables. If you want to run the tests, set those environment variables to something that will work.
If I wanted to run the tests against my local instance, I would run:
SOCRATA_DOMAIN=localhost SOCRATA_USERNAME=$SOCRATA_LOCAL_USER SOCRATA_PASSWORD=$SOCRATA_LOCAL_PASS bin/testmake the docs by running
make docsrelease to pypi by bumping the version to something reasonable and running
python setup.py sdist upload -r pypi
Note you'll need your .pypirc file in your home directory. For help, read this
ArgSpec
Args: auth
Top level publishing object.
All functions making HTTP calls return a result tuple, where the first element in the tuple is whether or not the call succeeded, and the second element is the returned object if it was a success, or a dictionary containing the error response if the call failed. 2xx responses are considered successes. 4xx and 5xx responses are considered failures. In the event of a socket hangup, an exception is raised.
Shortcut to create a dataset. Returns a Create object,
which contains functions which will create a view, upload
your file, and validate data quality in one step.
To actually place the validated data into a view, you can call .apply() on the revision
(revision, output_schema) Socrata(auth).create(
name = "cool dataset",
description = "a description"
).csv(file)
(ok, job) = revision.apply(output_schema = output_schema)
Args:
**kwargs: Arbitrary revision metadata values
Returns:
result (Revision, OutputSchema): Returns the revision that was created and the OutputSchema created from your uploaded file
Examples:
Socrata(auth).create(
name = "cool dataset",
description = "a description"
).csv(open('my-file.csv'))ArgSpec
Args: metadata
Create an empty revision, on a view that doesn't exist yet. The view will be created for you, and the initial revision will be returned.
Args:
metadata (dict): Metadata to apply to the revision
Returns:
result (bool, Revision | dict): Returns an API Result; the Revision if it was created or an API Error response
Examples:
(ok, rev) = Socrata(auth).new({
'name': 'hi',
'description': 'foo!',
'metadata': {
'view': 'metadata',
'anything': 'is allowed here'
}
})ArgSpec
Args: config_name, view
Update a dataset, using the configuration that you previously
created, and saved the name of. Takes the config_name parameter
which uniquely identifies the config, and the View object, which can
be obtained from socrata.views.lookup('view-id42')
Args:
config_name (str): The config name
view (View): The view to update
Returns:
result (Revision, Job): Returns the Revision and the Job, which is now running
Examples:
with open('my-file.csv', 'rb') as my_file:
(rev, job) = p.using_config(name, view).csv(my_file)
ArgSpec
Args: domain, username, password, request_id_prefix
Defaults: domain=
Manages basic authorization for accessing the socrata API.
This is passed into the Socrata object once, which is the entry
point for all operations.
auth = Authorization(
"data.seattle.gov",
os.environ['SOCRATA_USERNAME'],
os.environ['SOCRATA_PASSWORD']
)
publishing = Socrata(auth)
Disable SSL checking. Note that this should only be used while developing against a local Socrata instance.
ArgSpec
Args: fourfour, auth
ArgSpec
Args: metadata, permission
Defaults: metadata={}, permission=public
Create a revision on the view, which when applied, will replace the data.
Args:
metadata (dict): The metadata to change; these changes will be applied when the revision
is applied
permission (string): 'public' or 'private'
Returns:
result (bool, dict | Revision): The new revision, or an error
Examples:
>>> view.revisions.create_replace_revision(metadata = {'name': 'new dataset name', 'description': 'updated description'})
ArgSpec
Args: metadata, permission
Defaults: metadata={}, permission=public
Create a revision on the view, which when applied, will update the data rather than replacing it.
This is an upsert; if there is a rowId defined and you have duplicate ID values, those rows will be updated. Otherwise they will be appended.
Args:
metadata (dict): The metadata to change; these changes will be applied when the revision is applied
permission (string): 'public' or 'private'
Returns:
result (bool, dict | Revision): The new revision, or an error
Examples:
view.revisions.create_update_revision(metadata = {
'name': 'new dataset name',
'description': 'updated description'
})ArgSpec
Args: config
Create a revision for the given dataset.
List all the revisions on the view
Returns:
result (bool, dict | list[Revision])
ArgSpec
Args: revision_seq
Lookup a revision within the view based on the sequence number
Args:
revision_seq (int): The sequence number of the revision to lookup
Returns:
result (bool, dict | Revision): The Revision resulting from this API call, or an error
ArgSpec
Args: auth, response, parent
A revision is a change to a dataset
ArgSpec
Args: output_schema
Apply the Revision to the view that it was opened on
Args:
output_schema (OutputSchema): Optional output schema. If your revision includes
data changes, this should be included. If it is a metadata only revision,
then you will not have an output schema, and you do not need to pass anything
here
Returns:
result (bool, dict | Job): Returns the job that is being run to apply the revision
Examples:
(ok, job) = revision.apply(output_schema = my_output_schema)
ArgSpec
Args: filename
Create an upload within this revision
Args:
filename (str): The name of the file to upload
Returns:
result (bool, dict | Source): The Source created by this API call, or an error
Discard this open revision.
Returns:
result (bool, dict | Revision): The closed Revision or an error
Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
Open this revision in your browser, this will open a window
ArgSpec
Args: output_schema_id
Set the output schema id on the revision. This is what will get applied when the revision is applied if no ouput schema is explicitly supplied
Args:
output_schema_id (int): The output schema id
Returns:
result (bool, dict | Revision): The updated Revision as a result of this API call, or an error
Examples:
(ok, revision) = revision.set_output_schema(42)Create a dataset source within this revision
ArgSpec
Args: url
Create a URL source
Args:
url (str): The URL to create the dataset from
Returns:
result (bool, dict | Source): The Source created by this API call, or an error
This is the URL to the landing page in the UI for this revision
Returns:
url (str): URL you can paste into a browser to view the revision UI
ArgSpec
Args: body
Set the metadata to be applied to the view when this revision is applied
Args:
body (dict): The changes to make to this revision
Returns:
result (bool, dict | Revision): The updated Revision as a result of this API call, or an error
Examples:
(ok, revision) = revision.update({
'metadata': {
'name': 'new name',
'description': 'new description'
}
})ArgSpec
Args: auth
ArgSpec
Args: filename
Create a new source. Takes a body param, which must contain a filename
of the file.
Args:
filename (str): The name of the file you are uploading
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
(ok, upload) = revision.create_upload('foo.csv')ArgSpec
Args: auth, response, parent
Uploads bytes into the source. Requires content_type argument be set correctly for the file handle. It's advised you don't use this method directly, instead use one of the csv, xls, xlsx, or tsv methods which will correctly set the content_type for you.
ArgSpec
Args: revision
Associate this Source with the given revision.
ArgSpec
Args: file_handle
Uploads a Blob dataset. A blob is a file that will not be parsed as a data file, ie: an image, video, etc.
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-blob.jpg', 'rb') as f:
(ok, upload) = upload.blob(f)ArgSpec
Args: name
Change a parse option on the source.
If there are not yet bytes uploaded, these parse options will be used in order to parse the file.
If there are already bytes uploaded, this will trigger a re-parsing of
the file, and consequently a new InputSchema will be created. You can call
source.latest_input() to get the newest one.
Parse options are: header_count (int): the number of rows considered a header column_header (int): the one based index of row to use to generate the header encoding (string): defaults to guessing the encoding, but it can be explicitly set column_separator (string): For CSVs, this defaults to ",", and for TSVs " ", but you can use a custom separator quote_char (string): Character used to quote values that should be escaped. Defaults to """
Args:
name (string): One of the options above, ie: "column_separator" or "header_count"
Returns:
change (ParseOptionChange): implements a `.to(value)` function which you call to set the value
For our example, assume we have this dataset
This is my cool dataset
A, B, C
1, 2, 3
4, 5, 6
We want to say that the first 2 rows are headers, and the second of those 2 rows should be used to make the column header. We would do that like so:
Examples:
(ok, source) = source .change_parse_option('header_count').to(2) .change_parse_option('column_header').to(2) .run()ArgSpec
Args: file_handle
Upload a CSV, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-file.csv', 'rb') as f:
(ok, upload) = upload.csv(f)ArgSpec
Args: dataframe
Upload a pandas DataFrame, returns the new source.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
import pandas
df = pandas.read_csv('test/fixtures/simple.csv')
(ok, upload) = upload.df(df)ArgSpec
Args: file_handle
Upload a geojson file, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-geojson-file.geojson', 'rb') as f:
(ok, upload) = upload.geojson(f)ArgSpec
Args: file_handle
Upload a KML file, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-kml-file.kml', 'rb') as f:
(ok, upload) = upload.kml(f)Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
Open this source in your browser, this will open a window
ArgSpec
Args: file_handle
Upload a Shapefile, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-shapefile-archive.zip', 'rb') as f:
(ok, upload) = upload.shapefile(f)ArgSpec
Args: file_handle
Upload a TSV, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-file.tsv', 'rb') as f:
(ok, upload) = upload.tsv(f)This is the URL to the landing page in the UI for the sources
Returns:
url (str): URL you can paste into a browser to view the source UI
ArgSpec
Args: progress, timeout, sleeptime
Defaults: progress=<function noop at 0x7f6f42ffe7b8>, sleeptime=1
Wait for this dataset to finish transforming and validating. Accepts a progress function and a timeout.
ArgSpec
Args: file_handle
Upload an XLS, returns the new input schema
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-file.xls', 'rb') as f:
(ok, upload) = upload.xls(f)ArgSpec
Args: file_handle
Upload an XLSX, returns the new input schema.
Args:
file_handle: The file handle, as returned by the python function `open()`
Returns:
result (bool, Source | dict): Returns an API Result; the new Source or an error response
Examples:
with open('my-file.xlsx', 'rb') as f:
(ok, upload) = upload.xlsx(f)ArgSpec
Args: auth
ArgSpec
Args: name, data_action, parse_options, columns
Create a new ImportConfig. See http://docs.socratapublishing.apiary.io/
ImportConfig section for what is supported in data_action, parse_options,
and columns.
List all the ImportConfigs on this domain
ArgSpec
Args: name
Obtain a single ImportConfig by name
ArgSpec
Args: auth, response, parent
ArgSpec
Args: name
Change a parse option on the source.
If there are not yet bytes uploaded, these parse options will be used in order to parse the file.
If there are already bytes uploaded, this will trigger a re-parsing of
the file, and consequently a new InputSchema will be created. You can call
source.latest_input() to get the newest one.
Parse options are: header_count (int): the number of rows considered a header column_header (int): the one based index of row to use to generate the header encoding (string): defaults to guessing the encoding, but it can be explicitly set column_separator (string): For CSVs, this defaults to ",", and for TSVs " ", but you can use a custom separator quote_char (string): Character used to quote values that should be escaped. Defaults to """
Args:
name (string): One of the options above, ie: "column_separator" or "header_count"
Returns:
change (ParseOptionChange): implements a `.to(value)` function which you call to set the value
For our example, assume we have this dataset
This is my cool dataset
A, B, C
1, 2, 3
4, 5, 6
We want to say that the first 2 rows are headers, and the second of those 2 rows should be used to make the column header. We would do that like so:
Examples:
(ok, source) = source .change_parse_option('header_count').to(2) .change_parse_option('column_header').to(2) .run()ArgSpec
Args: fourfour
Create a new Revision in the context of this ImportConfig. Sources that happen in this Revision will take on the values in this Config.
Delete this ImportConfig. Note that this cannot be undone.
Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
ArgSpec
Args: body
Mutate this ImportConfig in place. Subsequent revisions opened against this ImportConfig will take on its new value.
ArgSpec
Args: auth, response, parent
This represents a schema exactly as it appeared in the source
Note that this does not make an API request
Returns: output_schema (OutputSchema): Returns the latest output schema
Get the latest (most recently created) OutputSchema which descends from this InputSchema
Returns: result (bool, OutputSchema | dict): Returns an API Result; the new OutputSchema or an error response
Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
ArgSpec
Args: body
Transform this InputSchema into an Output. Returns the new OutputSchema. Note that this call is async - the data may still be transforming even though the OutputSchema is returned. See OutputSchema.wait_for_finish to block until the
This is data as transformed from an InputSchema
ArgSpec
Args: field_name, display_name, transform_expr, description
Add a column
Args:
field_name (str): The column's field_name, must be unique
display_name (str): The columns display name
transform_expr (str): SoQL expression to evaluate and fill the column with data from
description (str): Optional column description
Returns:
output_schema (OutputSchema): Returns self for easy chaining
Examples:
(ok, new_output_schema) = output
# Add a new column, which is computed from the `celsius` column
.add_column('fahrenheit', 'Degrees (Fahrenheit)', '(to_number(`celsius`) * (9 / 5)) + 32', 'the temperature in celsius')
# Add a new column, which is computed from the `celsius` column
.add_column('kelvin', 'Degrees (Kelvin)', '(to_number(`celsius`) + 273.15')
.run()Whether or not any transform in this output schema has failed
ArgSpec
Args: name, data_action
Create a new ImportConfig from this OutputSchema. See the API docs for what an ImportConfig is and why they're useful
ArgSpec
Args: field_name, attribute
Change the column metadata. This returns a ColumnChange,
which implements a .to function, which takes the new value to change to
Args:
field_name (str): The column to change
attribute (str): The attribute of the column to change
Returns:
change (TransformChange): The transform change, which implements the `.to` function
Examples:
(ok, new_output_schema) = output
# Change the field_name of date to the_date
.change_column_metadata('date', 'field_name').to('the_date')
# Change the description of the celsius column
.change_column_metadata('celsius', 'description').to('the temperature in celsius')
# Change the display name of the celsius column
.change_column_metadata('celsius', 'display_name').to('Degrees (Celsius)')
.run()ArgSpec
Args: field_name
Change the column transform. This returns a TransformChange,
which implements a .to function, which takes a transform expression.
Args:
field_name (str): The column to change
Returns:
change (TransformChange): The transform change, which implements the `.to` function
Examples:
(ok, new_output_schema) = output
.change_column_transform('the_date').to('to_fixed_timestamp(`date`)')
# Make the celsius column all numbers
.change_column_transform('celsius').to('to_number(`celsius`)')
# Add a new column, which is computed from the `celsius` column
.add_column('fahrenheit', 'Degrees (Fahrenheit)', '(to_number(`celsius`) * (9 / 5)) + 32', 'the temperature in celsius')
.run()ArgSpec
Args: field_name
Drop the column
Args:
field_name (str): The column to drop
Returns:
output_schema (OutputSchema): Returns self for easy chaining
Examples:
(ok, new_output_schema) = output
.drop_column('foo')
.run()Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
ArgSpec
Args: offset, limit
Defaults: offset=0, limit=500
Get the rows for this OutputSchema. Acceps offset and limit params
for paging through the data.
Run all adds, drops, and column changes.
Returns:
result (bool, OutputSchema | dict): Returns an API Result; the new OutputSchema or an error response
Examples:
(ok, new_output_schema) = output
# Change the field_name of date to the_date
.change_column_metadata('date', 'field_name').to('the_date')
# Change the description of the celsius column
.change_column_metadata('celsius', 'description').to('the temperature in celsius')
# Change the display name of the celsius column
.change_column_metadata('celsius', 'display_name').to('Degrees (Celsius)')
# Change the transform of the_date column to to_fixed_timestamp(`date`)
.change_column_transform('the_date').to('to_fixed_timestamp(`date`)')
# Make the celsius column all numbers
.change_column_transform('celsius').to('to_number(`celsius`)')
# Add a new column, which is computed from the `celsius` column
.add_column('fahrenheit', 'Degrees (Fahrenheit)', '(to_number(`celsius`) * (9 / 5)) + 32', 'the temperature in celsius')
.run()ArgSpec
Args: offset, limit
Defaults: offset=0, limit=500
Get the errors that resulted in transforming into this output schema.
Accepts offset and limit params
Get the errors that results in transforming into this output schema as a CSV stream.
Note that this returns an (ok, Reponse) tuple, where Reponse is a python requests Reponse object
ArgSpec
Args: field_name
Set the row id. Note you must call validate_row_id before doing this.
Args:
field_name (str): The column to set as the row id
Returns:
result (bool, OutputSchema | dict): Returns an API Result; the new OutputSchema or an error response
Examples:
(ok, new_output_schema) = output.set_row_id('the_id_column')ArgSpec
Args: field_name
Set the row id. Note you must call validate_row_id before doing this.
Args:
field_name (str): The column to validate as the row id
Returns:
result (bool, dict): Returns an API Result; where the response says if it can be used as a row id
ArgSpec
Args: progress, timeout, sleeptime
Defaults: progress=<function noop at 0x7f6f42ffe7b8>, sleeptime=1
Wait for this dataset to finish transforming and validating. Accepts a progress function and a timeout.
ArgSpec
Args: auth, response, parent
Has this job finished or failed
Get a list of the operations that you can perform on this
object. These map directly onto what's returned from the API
in the links section of each resource
ArgSpec
Args: progress, timeout, sleeptime
Defaults: progress=<function noop at 0x7f6f42ffe7b8>, sleeptime=1
Wait for this dataset to finish transforming and validating. Accepts a progress function and a timeout.