Skip to content

Latest commit

 

History

History
73 lines (58 loc) · 3.03 KB

README.md

File metadata and controls

73 lines (58 loc) · 3.03 KB
Raw

pyjseisio

This project attempts to enable JavaSeis IO capability in Python by providing SWIG wrappers for the C++ JSeisIO JavaSeis library. The current version is forked and modified from https://github.com/ericaddison/pyjseisio and requires the modified version JSeisIO on GitHub.

Status

Just enough of a SWIG interface has been written to allow basic reading of JavaSeis files. This includes opening a file and collecting metadata (extents, axis definitions, header words, etc), and reading single frames, both trace amplitudes and headers. Trace data and headers are returned as NumPy ndarray data.

Dependencies

  • NumPy
  • JSeisIO
  • cmake (for JSeisIO)
  • SWIG

Useful tips (do it first)

For some Linux distributions, this might be convenient (especially when you do not deal with any cross-platform compiling):

sudo ln -s /usr/local/lib /usr/local/lib64
sudo echo /usr/local/lib > /etc/ld.so.conf.d/libc.conf
sudo ldconfig

Build and install JSeisIO

See: JSeisIO on GitHub: build and install.

Installation

If you have JSeisIO and NumPy (and SWIG) installed, then it should be as easy as running (as adminstrator):

sudo python setup.py clean --all
python setup.py build
sudo pip install .

For per user installation:

python setup.py clean --all
python setup.py build
pip install --user .

If JSeisIO is installed in a non-path location, then you should set the library_dirs command in setup.cfg, and if you don't want to set LD_LIBRARY_PATH, then you should also set rpath.

API Examples

Jupyter notebooks are available in the demo/ directory which demonstrate use of pyjseisio:

Here is the super quick, condensed version of how to get started reading JavaSeis data:

import numpy as np
import pyjseisio as js

dataset = js.open("../test/synth.js")   # open the given JavaSeis file for reading
frame = dataset.readFrame(0)            # read trace data from frame 0

For convenience, calls to the wrapped C++ methods are made through the jsdataset class, which will hopefully hide all the interaction with SWIG/C++ types and let the user deal with purely python data.

Header words are packed into a hdrs dictionary for easier use:

dataset.hdrs.keys()                      # returns a list of all header names
dataset.hdrs['OFFSET']                   # returns a catalogedHdrEntry object for the header OFFSET
dataset.hdrs['OFFSET'].getDescription()  # returns the description string for the header OFFSET

hdrBuf = dataset.readFrameHeader(0)      # read the header buffer from frame 0
dataset.hdrs['OFFSET'].getVal(hdrBuf[0]) # returns the value of the OFFSET header for trace 0

TODO

  • Finish implementing methods from jsFileReader
  • Fix JS_BYTORDER
  • Start on jsFileWriter
  • geometry?
  • Can I use installed JSeisIO headers (in an include dir) instead of providing those files myself?
  • Should I use setuptools instead of distutils?