tgsai/segy

The Ultimate Python SEG-Y I/O with Cloud Support and Schemas

segy's Issues

Handle Pandas and Struct header returns

Currently, we only implemented Pandas return. Downstream applications are more efficient with struct returns (MDIO), so we should handle this.

Unit tests for schemas

Let's make sure we have 100% test coverage for the schemas.

Implement parsing float headers

This is out of spec, but there exists a lot of SEG-Y files with IEEE float and IBM float 4-byte headers.

Allow aliases in header fields

People name things many different ways so it would be nice to have aliases for header fields.

Like the seismic inlines can be named

• inline
• Inline
• iline
• Iline
• inline_no

etc. We need to have a single point of "truth" for each field name though.

This would make the querying a lot nicer. I think it would encompass modifying StructuredFieldDescriptor to understand aliases.

Benchmarks

Benchmark against popular libraries:

• segyio
• obspy
• segysak
• segfast

Remove the option to disable transforms completely

And remove conditional checks from everywhere else. Settings come from SegySettings

Refactor endianness

Right now we got too much flexibility for endianness. Look into having only one at file level; or maybe one for binary one for trace.

Unit tests for indexers

Decouple Post-Post Processing

The post-processing after decoding is in indexers; we should separate it out for future enhancements.

Streamline SegyFactory interface

Some ideas

• getitem setitem
• hold traces in interval tree
• encapsulate low level logic
• flush methods (write trace chunks)
• finalize method (consolidate)

Implement SEG-Y Rev0 rules

The validation and post-processing (i.e. coordinate scaling) should be applied as post-processors.

We can design a composable way to do this for different SEG-Y versions.

High Priority:

• Sample axis from binary header (maybe can be a method of SegyFile?)

Backlog:

• Coordinates for src/rec/cdp

Implement Writing Functionality

Right now reading and parsing is within the indexers and we use the __getitem__ methods. We need to implement reciprocal __setitem__ methods.

Set up project linters

Move SegyFactory header and data pipelines to class attributes

And init it with the class. This will avoid repetitively building them every time we make traces.

Factory not writing SEG-Y sample format in binary file header

Need to implement functionality to write the sample format to the proper location in binary header.

Write custom exceptions

For common exceptions, we should make nice error messages for users.

Set up CI/CD

Set up all the github actions and push to pypi etc

Simplify SegySettings

The initial version is a bit confusing. Rethink and redo the organization and keys.

Evaluate if `SegyFile.from_spec` is Still Useful

It used to be, before we made SegyFile constructor more generic. Now seems a bit redundant?

Write tests for CLI

The new CLI needs tests, dragging coverage down

Decouple Decoding

The decoding of values from raw bytes in indexers need to be separated.