PyOphidia: Python bindings for Ophidia

PyOphidia is a GPLv3-licensed Python package for interacting with the Ophidia framework.

It aims at providing a user-friendly and programmatic interface for large-scale data analytics and a convenient way to submit SOAP HTTPS requests to an Ophidia server or to develop your own application using Python.

PyOphidia provides features for handling scientific data in the form of datacubes, managing workflow execution, enabling parallel processing on HPC/Cloud systems and supporting integration with well-known modules from the Python scientific ecosystem.

It runs on Python 2.7, 3.7, 3.8, 3.9, 3.10 and 3.11 it is pure-Python code and has some (optional) dependencies on Xarray, Pandas and Numpy. It requires a running Ophidia instance for client-server interactions. The latest PyOphidia version (v1.12) is compatible with Ophidia v1.8.

It provides 2 main modules:

• client.py: generic low level class to submit any type of requests (simple tasks and workflows), using SSL and SOAP with the client ophsubmit.py;
• cube.py: high level cube-oriented class to interact directly with cubes, with several methods wrapping the operators.

Documentation

https://pyophidia.readthedocs.io/en/latest/

Installation

To install PyOphidia package run the following command:

pip install pyophidia

Install with conda

To install PyOphidia with conda run the following command:

conda install -c conda-forge pyophidia

Installation from sources

To install the latest developement version run the following commands:

git clone https://github.com/OphidiaBigData/PyOphidia
cd PyOphidia
python setup.py install

Examples

Import PyOphidia

Import client module from PyOphidia package:

from PyOphidia import client

Instantiate a client

Create a new Client() using the login parameters username, password, host and port. It will also try to resume the last session the user was connected to, as well as the last working directory and the last produced cube.

ophclient = client.Client(username="oph-user",password="oph-passwd",server="127.0.0.1",port="11732")

In case of authentication token is used:

ophclient = client.Client(token="token",server="127.0.0.1",port="11732")

If OPH_USER, OPH_PASSWD (or OPH_TOKEN), OPH_SERVER_HOST and OPH_SERVER_PORT variables have been set in the environment (see the documentation for more details), a client can be also created reading directly the values from the environment without the need to specify any parameter.

ophclient = client.Client(read_env=True)

Client attributes

• username: Ophidia username
• password: Ophidia password
• server: Ophidia server address
• port: Ophidia server port (default is 11732)
• session: ID of the current session
• base_src_path: Server-side instance base source path
• cwd: Current Working Directory
• cdd: Current Data Directory
• cube: Last produced cube PID
• host_partition: Name of host partition being used
• exec_mode: Execution mode, 'sync' for synchronous mode (default), 'async' for asynchronous mode
• ncores: Number of cores for each operation (default is 1)
• last_request: Last submitted query
• last_response: Last response received from the server (JSON string)
• last_response_status: Status of last response received from the server (string)
• last_jobid: Job ID associated to the last request
• last_return_value: Last return value associated to response
• last_error: Last error value associated to response
• last_exec_time: Last execution time value associated to response
• project: Project to be used for the resource manager (if required)

Client methods

• submit(query, display) -> self: Submit a query like 'operator=myoperator;param1=value1;' or 'myoperator param1=value1;' to the Ophidia server according to all login parameters of the Client and its state.
• get_progress(id) -> dict : Get progress of a workflow, either by specifying the id or from the last submitted one.
• deserialize_response() -> dict: Return the last_response JSON string attribute as a Python dictionary.
• get_base_path(display) -> self : Get base path for data from the Ophidia server.
• resume_session(display) -> self: Resume the last session the user was connected to.
• resume_cwd(display) -> self: Resume the last cwd (current working directory) the user was located into.
• resume_cdd(display) -> self: Resume the last cdd (current working data directory) the user was located into.
• resume_cube(display) -> self: Resume the last cube produced by the user.
• wsubmit(workflow, *params) -> self: Submit an entire workflow passing a JSON string or the path of a JSON file and an optional series of parameters that will replace $1, $2 etc. in the workflow. The workflow will be validated against the Ophidia Workflow JSON Schema.
• wisvalid(workflow) -> bool: Return True if the workflow (a JSON string or a Python dict) is valid against the Ophidia Workflow JSON Schema or False and the related validation/error message.
• pretty_print(response, response_i) -> self: Prints the last_response JSON string attribute as a formatted response.

To display the command output set "display=True"

Submit a request

Execute the request oph_list level=2:

ophclient.submit("oph_list level=2", display=True)

Set a Client for the Cube class

Instantiate a new Client common to all Cube instances:

from PyOphidia import cube
cube.Cube.setclient(username="oph-user",password="oph-passwd",server="127.0.0.1",port="11732")

Cube attributes

Instance attributes:

• pid: Cube PID
• creation_date: Creation date of the cube
• measure: Name of the variable imported into the cube
• measure_type: Measure data type
• level: Number of operations between the original imported cube and the actual cube
• nfragments: Total number of fragments
• source_file: Parent of the actual cube
• hostxcube: Number of hosts on which the cube is stored
• fragxdb: Number of fragments for each database
• rowsxfrag: Number of rows for each fragment
• elementsxrow: Number of elements for each row
• compressed: If the cube is compressed or not
• size: Size of the cube
• nelements: Total number of elements
• dim_info: List of dict with information on each cube dimension

Class attributes:

• client: instance of class Client through which it is possible to submit all requests

Create a new container

Create a new container to contain our cubes called test, with 3 double dimensions (lat, lon and time):

cube.Cube.createcontainer(container='test',dim='lat|lon|time',dim_type='double|double|double',hierarchy='oph_base|oph_base|oph_time')

Import a new cube

Import the variable T2M from the NetCDF file /path/to/file.nc into a new cube inside the test container. Use lat and lon as explicit dimensions and time as implicit dimension expressed in days:

mycube = cube.Cube(container='test',exp_dim='lat|lon',imp_dim='time',measure='T2M',src_path='/path/to/file.nc',exp_concept_level='c|c',imp_concept_level='d')

Create a Cube object from an existing cube identifier

Instantiate a new Cube using the PID of an existing cube:

mycube2 = cube.Cube(pid='http://127.0.0.1/1/2')

Show a Cube structure and info

To shows metadata information about a data cube, its size and the dimensions related to it:

mycube2.info()

For the operators such as "cubeschema", "cubesize", "cubeelements", "explore", "hierarchy", "info", "list", "loggingbk", "operators", "search", "showgrid", "man", "metadata", "primitives", "provenance", "search", "showgrid", "tasks" and other operators that provide verbose output, the display parameter by default is "True". For the rest of operators, to display the result, "dispay=True" should be set.

Subset a Cube

To perform a subsetting operation along dimensions of a data cube (dimension values are used as input filters):

mycube3 = mycube2.subset(subset_dims='lat|lon',subset_filter='1:10|20:30',subset_type='coord')

Explore Cube

To explore a data cube filtering the data along its dimensions:

mycube2.explore(subset_dims='lat|lon',subset_filter='1:10|20:30',subset_type='coord')

Export to NetCDF file

To export data into a single NetCDF file:

mycube3.exportnc2(output_path='/home/user')

Export to Python array

To exports data in a python-friendly format:

data = mycube3.export_array(show_time='yes')

Run a Python script with Ophidia

To run a Python script through Ophidia load or define the Python function in the script where PyOphidia is used (works starting with Python 3+), e.g.:

def myScript(arg1):
        import subprocess
        return subprocess.call('ls -la ' + arg1, shell=True)

cube.Cube.script(python_code=True,script=myScript,args="/home/ophidia",display=True)