Note 1: this repo is outdated. Please see this repo.

Note 2: use iod branch for older compatibility syntax.

Note 3: compatible with Python 2.X

Python Client Library for Haven OnDemand

Official Python client library to help with calling Haven OnDemand APIs http://havenondemand.com.

What is Haven OnDemand?

Haven OnDemand is a set of over 70 APIs for handling all sorts of unstructured data. Here are just some of our APIs' capabilities:

• Speech to text
• OCR
• Text extraction
• Indexing documents
• Smart search
• Language identification
• Concept extraction
• Sentiment analysis
• Web crawlers
• Machine learning

For a full list of all the APIs and to try them out, check out https://www.havenondemand.com/developer/apis

Installation

To install the latest version from this github repo:

pip install git+https://github.com/HP-Haven-OnDemand/havenondemand-python_deprecated

Importing

from havenondemand.hodindex import HODClient

Initializing the client

client = HODClient(apikey="myapikey", apiversiondefault=version_number)

You can find your API key here after signing up.

apiversiondefault is an optional parameter (defaults to 1) and can be either 1 or 2.

Proxies

http_proxy  = "ip:port"
proxyDict = {"http": http_proxy}
client = HODClient(apikey="myapikey", apiversiondefault=version_number, proxy=proxyDict)

The proxy parameter is optional and takes a dictionary of proxy urls. It will only use the one for the protocol chosen in the api url , http or https

Sending requests

r=client.post(handler,{'param1':'value1','param2':'value2'})
r=client.post('analyzesentiment',{'text':'I like cats'})

The client's post method takes the API path that you're sending your request to as well as an object containing the parameters you want to send to the api. You do not need to send your API key each time as the client will handle that automatically.

###Posting files

r=client.post('ocrdocument',files={'file':open('myimg.jpg','rb')})

Sending files is just as easy.

r=client.post('ocrdocument',{'mode':'photo'},files={'file':open('myimg.jpg','rb')})
r=client.post('ocrdocument',data={'mode':'photo'},files={'file':open('myimg.jpg','rb')})

Any extra parameters should be added in the same way as regular calls, or in the data parameter.

###Parsing the output

myjson=r.json()

The object returned is a response object from the python requests library and can easily be turned to json.

docs=myjson["documents"]
for doc in docs:
    #do stuff

###Indexing

Creating an index

client.createIndex('myindex')

An Index object can easily be created

Fetching indexes/an index

index = client.getIndex('myindex')

The getIndex call will return an hodindex Index object but will not check for existence.

indexes = client.listIndexes()
indexex.get('myindex',client.createIndex('myindex'))

Here we first check the list of our indexes and return a newly created index if the index does not already exist

Deleting an index

index.delete()
client.deleteIndex('myindex')

An index can be deleted in two equivalent ways

Indexing documents

doc1={'reference':'doc1','title':'title1','content':'this is my content'}
doc2={'reference':'doc2','title':'title2','content':'this is another content'}

Documents can be created as regular python objects

index.addDoc(doc1)
index.addDocs([doc1,doc2])

They can be added directly one at a time or in a batch.

for doc in docs:
  index.pushDoc(doc)
index.commit()

An alternative to addDocs and easy way to keep batch documents is to use the pushDoc method, the index will keep in memory a list of the documents it needs to index.

if index.countDocs()>10:
  index.commit()

It makes it easy to batch together groups of documents.

Asynchronous request

For each call the Async parameter can be set to true to send an asynchronous request.

r=client.post('analyzesentiment',{'text':'I like cats'},async=True)
print r.json()

r=index.commit(async=True)