Non-standard headers loading #6
Conversation
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
Co-authored-by: Alexander Kuvaev <[email protected]>
segfast/loader.py
Outdated
| @@ -5,7 +5,7 @@ | |||
|
|
|||
|
|
|||
| def Loader(path, engine='memmap', endian='big', strict=False, ignore_geometry=True): | |||
| """ !!. """ | |||
| """ SEG-Y loader. """ | |||
There was a problem hiding this comment.
Selector class for loading SEG-Y with either segyio-based loader or memmap-based one.
Make class references links, so they would be cool (and useful) in the doc
There was a problem hiding this comment.
It will be the next PR (see docs branch here https://github.com/AlexeyKozhevin/segfast/tree/docs)
| elif unused_len < 0: | ||
| raise ValueError(f'{header.name} header overlap') |
There was a problem hiding this comment.
Actually, does not violates anything (the loading of overlapping headers should still work just fine), but we probably don't want to even give the option
There was a problem hiding this comment.
If you want to make overlaped headers, we will have to rewrite all the part with 'unused_*' in dtype. I don't see any reasons to do it.
segfast/utils.py
Outdated
| @@ -35,6 +37,112 @@ def update(self, n=1): | |||
| notifier = Notifier | |||
|
|
|||
|
|
|||
| class TraceHeaderSpec: | |||
There was a problem hiding this comment.
Please, add to_tuple / to_dict method: motivation is to keep track of actually used headers in, for example, Geometry class of seismiqb.
In the same vain, I would think about returning list of TraceHeaderSpec instances out of Loader.load_headers methods, so that you have a detailed info on what was actually loaded.
There was a problem hiding this comment.
There is a make_headers_specs method
There was a problem hiding this comment.
The returned specs may differ from those returned by make_headers_specs, I'll add such an option in my PR later today
load_headersloads non-standard header by its start byte position and dtype.'>f4')TraceHeaderSpecclassExample: