A python wrapper for streamtools: http://nytlabs.github.io/streamtools

pip install python-streamtools==0.1.1



A python wrapper for streamtools.


pip install python-streamtools


NOTE requires streamtools to be running on localhost:7070.



Low-level API Access

import streamtools 

st = streamtools.Api()

ticker_id = st.create_block(type='ticker', rule={'Interval': '1s'})
log_id = st.create_block(type='tolog')

conn_id = st.create_connection(from_id=ticker_id, to_id=log_id)
print st.get_connection(conn_id)

for msg in st.stream(ticker_id):
  print msg

Block, Connection, Pattern Construction:

from streamtools import Api, Block, Connection, Pattern 

# init api
st = Api()

# init empty Pattern
p = Pattern()

# construct blocks
b1 = Block(type='ticker', rule={'Interval':'1s'})
b2 = Block(type='tolog')

print b1.raw
print b2.raw

# construct a Connection explicity / implicitly
c = b1 + b2
print c.raw

# add connection to pattern
p += c
print p.raw

# pattern exists on the api too
print st.get_pattern()

# stream block output
for line in b1.listen():
  print line

Creating Custom Plugins

NOTE requires rabbitmq. python-streamtools includes a class for generating blocks which enable the use of a customizable python functions. Plugin's work be creating two built-in streamtools blocks: toampq and fromampq and spawns a reader which listens to toampq passes input to a custom function and emits output to the fromampq block. In order to initialize the connection you must call plugin.attach() which will run forever.

from streamtools import Plugin, Block 

def my_function(body):
  body['im'] = 'here'
  yield body

plugin = Plugin('my-plugin')
plugin.main = my_function 

ticker = Block(type='ticker', rule={'Interval':'1s'})
log = Block(type='tolog')

cin = ticker + plugin.in_block
cout = plugin.out_block + log 



  • Documentation is basic for now, refer to the streamtools docs, and the source code for full usage.
  • st.stream seems to have significant latency.


  • Websocket support
  • Custom Blocks which allow you to insert a python script, Would work via creating a series of blocks to query an internal api, pass input to a function, and pass output to another block.
  • Figure out how to more elegantly specify custom to_route when adding two blocks together, eg:
# right now 
c = b1 + b2 
c.to_route = 'rule'

c = b1 + b2('rule')