Is there a real-world example LAS file with comma-delimited ~ASCII section?

Try las_sample_3.0.las in the repo, I think all data sections in there are comma delimited.

@kinverarity1, How should Lasio identify to parse the ~ASCII section with comma-delimiter? I think these are the options. Is one of these (or something else) preferred?:

• Check for the delimiter in the first few lines and have some logic to make the decision. Then parse the data section with this 'found' delimiter.
• A command line parameter is passed in. Does this already exist?
• Check if the file is LAS-3 and if it is then parse the ~ASCII section assuming the delimiter is a comma?

Thanks,
DC

I'd prefer the second approach mixed with the "provisional" approach we use for WRAP and NULL (see the DLM item in the version section):

Then we could just provide a keyword argument to force a choice? Not sure what to default to - probably whitespace.

My preference is mainly driven by laziness 🙃 First option would be great and we can do it if you want!

Reading through 1.2/2.0 and 3.0 specs only 3.0 has DLM. The 3.0 spec says that if DLM is not stated then the default delimiter is space. So we could set the default to space and if DLM is stated then adopt its value. That would roughly be the First option and I think is pretty do-able. I'll attempt a draft branch of it.

Dev Notes (may change over time...) Edited 2021-08-10:

• Reread the 3.0 spec

• Review Kent's manual method to read LAS3 data https://gist.github.com/kinverarity1/92f00b781472512349a9312d75fd4c33

• Create a central a las.delimiter property and make it available to both parsing engines: normal and Numpy.

• In the ~Version parsing section, check for the DLM keyword, if it exists override the default delimiter with the DLM setting. The default delimiter will be a .

• Focus on parsing just the LAS3 '~ASCII' data section. Skip the other data sections for now because the code and internal data-structure isn't there yet to handle multiple data sections. In LAS3 the '~ASCII' section can be also called ~Log_Data. So add the ability to find either one. Optionally implement comma parsing with this file standards/examples/3.0/sample_las3.0_spec.las. It uses the LAS2 standard section names. Using it as the test file would enable separating the 'adding of alternate 'Log' names to a separate pull-request or separate/subsequent commit.

• The curve names may be in ~Log_Definition instead of ~Curves. Add the ability to check both, and to add the Data to las.curves

• Add a default.py::READ_POLICIES key-value for when the delimiter is a comma (this will exclude the regex_sub for converting commas to decimal). Add provisional_delimiter. Move the call in las.py to reader.py::get_substitutions() till after the provisional_delimiter has been processed so we can decide which default.py::READ_POLICIES to use.

• reader.py::read_data_section_iterative_normal_engine()'s sub-routine items() needs to be changed to enable parsing a line with different delimiters not just SPACE but also COMMA and TAB. Same for reader.py::inspect_data_section().

What is the current status of this issue? I see a pull request have been merged. Is it done?

I believe the PR only implemented it for the normal (slow) parser, not the numpy (fast) parser. And also didn't allow for user specification of the delimiter.