Given my lack of bandwidth for this project and the presence of a number of great alternatives popping up in recent years, I currently recommend against using Glide in long term / production scenarios.
Check out pypeln for an alternative that is more focused on flow & parallelization features that are useful in Glide.
Glide is an easy-to-use data pipelining tool inspired by Consecution and Apache Storm Topologies.
Like those libraries, Glide is:
- A simple, reusable approach to building robust ETL pipelines
- A system for wiring together processing nodes to form a directed acyclic graph (DAG)
Glide also has:
- An expanding suite of built-in nodes and pipelines that extract, transform, and load data from/to any combination of:
- SQL databases (SQLite, DBAPI, and SQLAlchemy support)
- Local or remote files (CSVs, Excel, and raw file support)
- URLs (JSON endpoints, file downloads, APIs, etc.)
- HTML Tables
- Emails
- Extensions for Pandas, Dask, Celery, Redis Queue and more
- A variety of node and DAG parallel/concurrent/distributed processing strategies
- A simple decorator to generate a command line interface from a pipeline in ~one line of code
- Flexible pipeline templating
Glide is not a task orchestration and/or dependency management tool like Airflow. Use Glide to define your easily developed/contained/reusable/testable data processing pipelines and then rely on a tool like Airflow to do what it's good at, namely scheduling and complex task dependency management.
- Installation
- Primer
- Basic Examples
- Flow Control Examples
- Parallelization & Concurrency
- Utility Examples
- CLI Generation
- Extensions
- Docs
- How to Contribute
โ ๏ธ Warning: This project is in an alpha state and is maintained on an as-needed basis. Please test carefully for production usage and report any issues.
$ pip install glide
You are encouraged to take a deeper look at the docs, but the short of it is the following:
1. A Node
is a part of a pipeline which has a run
method that typically
accepts data from upstream nodes, and pushes data to downstream nodes. For example:
class MyNode(Node):
def run(self, data):
# Some node-specific code here
self.push(data)
2. A Glider
is a pipeline of Node
objects wired together in a DAG. It
accepts input data in its consume
method. For example:
glider = Glider(
MyExtractNode("extract")
| MyTransformNode("transform")
| MyLoadNode("load")
)
glider.consume(data)
If a node's run
method has additional parameters, they are populated from
the node's context
. More info on creating nodes and populating runtime context
can be found here.
The following examples serve to quickly illustrate some core features and
built-in nodes. There is much more Glide
can do that is not shown
here. Everything below assumes you have used the following shortcut to import
all necessary node and pipeline classes:
from glide import *
Apply a transformation to data from a CSV, use a function to lowercase all strings, and load into an output CSV:
def lower_rows(data):
for row in data:
for k, v in row.items():
row[k] = v.lower() if type(v) == str else v
return data
glider = Glider(
CSVExtract("extract")
| Func("transform", func=lower_rows)
| CSVLoad("load")
)
glider.consume(
["/path/to/infile.csv"],
extract=dict(chunksize=100),
load=dict(outfile="/path/to/outfile.csv"),
)
Read from one table, write to another:
conn = get_my_sqlalchemy_conn()
sql = "select * from in_table limit 10"
glider = Glider(
SQLExtract("extract")
| SQLLoad("load"),
global_state=dict(conn=conn) # conn is automagically passed to any nodes that accept a "conn" argument
)
glider.consume(
[sql],
load=dict(table="out_table")
)
Start a transaction before writing to a database, rollback on failure:
glider = Glider(
SQLExtract("extract")
| SQLTransaction("tx")
| SQLLoad("load", rollback=True),
global_state=dict(conn=conn)
)
glider.consume(...)
Extract data from each URL in the list of requests and load to a URL endpoint:
glider = Glider(URLExtract("extract") | URLLoad("load"))
reqs = [
"https://jsonplaceholder.typicode.com/todos/1",
"https://jsonplaceholder.typicode.com/todos/2",
]
glider.consume(
reqs,
extract=dict(data_type="json"),
load=dict(
url="https://jsonplaceholder.typicode.com/todos",
data_param="json",
headers={"Content-type": "application/json; charset=UTF-8"},
),
)
Filter the propagation of data based on the result of a function:
def data_check(node, data):
# do some check on data, return True/False to control filtering
return True
glider = Glider(
MyExtract("extract")
| Filter("filter", func=data_check)
| MyLoad("load")
)
Push each row of an input iterable individually:
glider = Glider(
CSVExtract("extract", nrows=20)
| IterPush("iter")
| Print("load")
)
Split an iterable before pushing:
glider = Glider(SplitPush("push", split_count=2) | Print("print"))
glider.consume([range(4)])
Split an iterable evenly among downstream nodes:
glider = Glider(SplitByNode("push") | [Print("print1"), Print("print2")])
glider.consume([range(4)])
Collect all upstream node data before pushing:
glider = Glider(
CSVExtract("extract")
| Reduce("reduce")
| Print("load")
)
glider.consume(["/path/to/infile1.csv", "/path/to/infile2.csv"])
This will read both input CSVs and push them in a single iterable to the
downstream nodes. You can also use the flatten
option of Reduce
to
flatten the depth of the iterable before pushing (effectively a concat
operation).
Join data on one or more columns before pushing:
glider = Glider(
Reduce("reduce")
| Join("join")
| Print("load")
)
d1 = <list of dicts or DataFrame>
d2 = <list of dicts or DataFrame>
glider.consume([d1, d2], join=dict(on="common_key", how="inner"))
Route data to a particular downstream node using a router function:
def parity_router(row):
if int(row["mycolumn"]) % 2 == 0:
return "even"
return "odd"
glider = Glider(
CSVExtract("extract", nrows=20)
| IterPush("iter")
| [parity_zip_router, Print("even"), Print("odd")]
)
glider.consume(...)
This will push rows with even mycolumn
values to the "even" Print
node,
and rows with odd mycolumn
values to the "odd" Print
node.
Push a sliding window of the data:
glider = Glider(
CSVExtract("extract", nrows=5)
| WindowPush("window", size=3)
| MyWindowCalcNode("calc")
)
Generate a set of datetime windows and push them downstream:
import datetime
today = datetime.date.today()
glider = Glider(DateTimeWindowPush("windows") | PrettyPrint("print"))
glider.consume(
windows=dict(
start_date=today - datetime.timedelta(days=3),
end_date=today,
num_windows=2
)
)
Or use DateWindowPush
for date objects. Note that the data arg to consume
can be ignored because the top node (DateTimeWindowPush
) is a subclass of
NoInputNode
which takes no input data and generates data to push
on its own.
By default consume
does not return any values and assumes you will be
outputting your results to one or more endpoints in your terminating
nodes (files, databases, etc.). The Return
node will collect the data
from its parent node(s) and set it as a return value for consume
.
glider = Glider(
CSVExtract("extract")
| MyTransformer("transform")
| Return("return")
)
data = glider.consume(...)
Call a function in parallel processes on equal splits of data from a CSV:
glider = Glider(
CSVExtract("extract")
| ProcessPoolSubmit("transform", push_type=PushTypes.Result)
| CSVLoad("load")
)
glider.consume(
["infile.csv"],
transform=dict(func=lower_rows),
load=dict(outfile="outfile.csv"),
)
We passed push_type=PushTypes.Result
to force ProcessPoolSubmit
to fetch
and combine the asynchronous results before pushing to the downstream
node. The default is to just pass the asynchronous task/futures objects
forward, so the following would be equivalent:
glider = Glider(
CSVExtract("extract")
| ProcessPoolSubmit("transform")
| FuturesReduce("reduce")
| Flatten("flatten")
| CSVLoad("load")
)
The FuturesReduce
node waits for the results from each futures object, and
then Flatten
will combine each subresult back together into a single result
to be loaded in the final CSVLoad
node.
Completely parallelize a pipeline using a ParaGlider
(who said ETL isn't
fun?!?). Split processing of the inputs (two files in this case) over the
pool, with each process running the entire pipeline on part of the consumed
data:
glider = ProcessPoolParaGlider(
CSVExtract('extract')
| Print('load')
)
glider.consume(
["/path/to/infile1.csv", "/path/to/infile2.csv"],
extract=dict(nrows=50)
)
Branch into parallel execution in the middle of the DAG utilizing a parallel push node:
glider = Glider(
CSVExtract("extract", nrows=60)
| ProcessPoolPush("push", split=True)
| [Print("load1"), Print("load2"), Print("load3")]
)
glider.consume(["/path/to/infile.csv"])
The above example will extract 60 rows from a CSV and then push equal slices
to the logging nodes in parallel processes. Using split=False
(default)
would have passed the entire 60 rows to each logging node in parallel
processes.
Once you branch off into processes with a parallel push node there is no way to reduce/join the pipeline back into the original process and resume single-process operation. The entire remainder of the pipeline is executed in each subprocess. However, that is possible with threads as shown in the next example.
glider = Glider(
CSVExtract("extract", nrows=60)
| ThreadPoolPush("push", split=True)
| [Print("load1"), Print("load2"), Print("load3")]
| ThreadReduce("reduce")
| Print("loadall")
)
glider.consume(["/path/to/infile.csv"])
The above code will split the data and push to the first 3 logging nodes in
multiple threads. The ThreadReduce
node won't push until all of the previous
nodes have finished, and then the final logging node will print all of the
results.
Limited, experimental support is also available for concurrency via asyncio
in Python >= 3.7:
import asyncio
async def async_sleep(data):
# Dummy example. Await some real async work in here.
await asyncio.sleep(0.5)
return data
glider = Glider(
CSVExtract("extract", nrows=5)
| AsyncIOSubmit("transform", push_type=PushTypes.Result)
| Print("load")
)
glider.consume(
["/path/to/infile.csv"],
transform=dict(func=async_sleep)
)
The above example will split the input data into items to be processed on an
asyncio
event loop and synchronously wait for the results before pushing.
AsyncIOSubmit
supports specifying a split_count
as well as a timeout
when waiting for results. Alternatively, one can push asyncio
futures and
later reduce their results as follows:
glider = Glider(
CSVExtract("extract", nrows=5)
| AsyncIOSubmit("transform", push_type=PushTypes.Async)
| AsyncIOFuturesReduce("reduce", flatten=True)
| Print("load")
)
Note that the asyncio
nodes will create and start an event loop for you if
necessary. It's also perfectly fine to manage the event loop on your own, in
which case glide
will run tasks on the current thread's event loop.
Drop replacement nodes into an existing pipeline. Any node can be replaced by name:
glider = Glider(
PlaceholderNode("extract")
| CSVLoad("load")
)
glider["extract"] = CSVExtract("extract")
glider.consume(...)
Or reuse an existing structure of nodes with a NodeTemplate
:
nodes = NodeTemplate(
CSVExtract("extract")
| CSVLoad("load")
)
glider = Glider(nodes()) # Copy of nodes created with each call
Or reuse an existing pipeline structure with GliderTemplate
:
template = GliderTemplate(
CSVExtract("extract")
| CSVLoad("load")
)
glider = template() # Copy of pipeline created with each call
You can use the AssertFunc
node to assert that some condition of the data is
met:
glider = Glider(
CSVExtract("extract", chunksize=10, nrows=20)
| AssertFunc("length_check", func=lambda node, data: len(data) == 10)
| CSVLoad("load")
)
The func
callable must accept two parameters, a reference to the node object
and the data passed into that node. Any truthy value returned will pass the
assertion test.
Similarly, you can do a sql-based check with AssertSQL
, in this case simply
verifying the number of rows inserted:
glider = Glider(
SQLExtract("extract")
| SQLLoad("load")
| AssertSQL("sql_check")
)
sql = "select * from in_table limit 10"
assert_sql = "select (select count(*) as x from out_table) == 10 as assert"
glider.consume(
[sql],
extract=dict(conn=in_conn),
load=dict(conn=out_conn, table="out_table"),
sql_check=dict(conn=out_conn, sql=assert_sql)
)
This looks for a truthy value in the assert
column of the result to pass the
assertion. You can also use the data_check
option of AssertSQL
to instead
have it do a comparison to the result of some function of the data:
glider = ...
sql = "select * from in_table limit 10"
assert_sql = "select count(*) as assert from out_table"
glider.consume(
[sql],
extract=dict(conn=in_conn),
load=dict(conn=out_conn, table="out_table", push_data=True),
sql_check=dict(
conn=out_conn,
sql=assert_sql,
data_check=lambda node, data: len(data)
)
)
Note that we also added push_data=True
to the SQLLoad
node to have it push
the data instead of a table name.
To enable debug logging for Glide change the log level of the "glide" logger:
import logging
logging.getLogger("glide").setLevel(logging.DEBUG)
Glide will then print debug information about data passed through your pipeline.
You can also pass _log=True
to the init method of any node to enable logging
of processed data:
glider = Glider(
SQLExtract("extract", _log=True)
...
)
Additionaly, you can pass _debug=True
to the init method of any node to
cause the node to drop into PDB right before calling run
, assuming you
aren't executing the pipeline in a subprocess:
glider = Glider(
SQLExtract("extract", _debug=True)
...
)
Finally, there are a variety of print nodes you can place in your pipeline for
general logging or debugging, such as Print
, PrettyPrint
, LenPrint
,
ReprPrint
, and FormatPrint
. See the node documentation for more info.
Insert a Profile
node somewhere in your pipeline to get profiler information
for all downstream nodes:
glider = Glider(
Profile("profile")
...
)
If the hierarchy of nodes you want to form is not achievable with the |
operator, you can use the add_downstream
Node
method to form more complex
graphs. More info can be found
here.
If you have the Graphviz package installed, you can generate a plot of your pipelines by simply doing the following:
glider = Glider(...)
glider.plot("/path/to/filename.png")
With Glide you can create parameterized command line scripts from any pipeline with a simple decorator:
glider = Glider(
SQLLoad("extract")
| SQLExtract("load")
)
@glider.cli()
def main(glide_data, node_contexts):
glider.consume(glide_data, **node_contexts)
if __name__ == "__main__":
main()
The script arguments, their types, and whether they are required or not is all
inferred by inspecting the run
arguments on the nodes of the pipeline and
prefixing the node name. Please see the full documentation
here
for more details.
To install all extensions and dev dependencies:
$ pip install glide[complete]
You can also just install Glide plus a specific extension:
$ pip install glide[dask]
$ pip install glide[celery]
$ pip install glide[rq]
$ pip install glide[swifter]
To access installed extensions import from the glide.extensions
submodules
as necessary. Review the documentation and tests for current extensions for
help getting started.
The Pandas extension is actually supported by default with all glide
installs.
Below is a simple example that extracts from a CSV, lowercases all strings,
and then loads to another CSV using Pandas under the hood:
def lower(s):
return s.lower() if type(s) == str else s
glider = Glider(
DataFrameCSVExtract("extract")
| DataFrameApplyMap("transform", func=lower)
| DataFrameCSVLoad("load", index=False, mode="a")
)
glider.consume(...)
There are a variety of other helpful nodes built in, including ToDataFrame
,
FromDataFrame
, nodes to read/write other datasources, and nodes to deal with
rolling
calculations. There is also a generic DataFrameMethod
node that
passes through to any DataFrame method.
See the extension docs here for node/pipeline reference information. See the tests here for some additional examples.
See the extension docs here for node/pipeline reference information. See the tests here for some additional examples.
See the extension docs here for node/pipeline reference information. See the tests here for some additional examples.
See the extension docs here for node/pipeline reference information. See the tests here for some additional examples.
See the extension docs here for node/pipeline reference information. See the tests here for some additional examples.
More thorough documentation can be found here. You can supplement your knowledge by perusing the tests directory or the module reference.
See the CONTRIBUTING guide.