Add EDF parser #581
Add EDF parser #581
Conversation
| file-extension: edf | ||
| endian: le | ||
| license: MIT | ||
|
|
There was a problem hiding this comment.
doc and doc-ref are mandatory. Also a link to some small samples is desireable.
| value: num_data_records.to_i | ||
|
|
||
| seq: | ||
| - id: "version" |
There was a problem hiding this comment.
There is need to quote ids here
| size: 8 | ||
| type: str | ||
| pad-right: 0x20 | ||
| encoding: ASCII |
There was a problem hiding this comment.
Should be moved to meta. Also should be lowercased. Also please check that UTF-8 is not allowed here by modern software.
There was a problem hiding this comment.
Also should be lowercased.
Why? Searching through the repo, both uppercase and lowercase are common (and uppercase is a bit more frequent).
There was a problem hiding this comment.
|
|
||
| signal: | ||
| params: | ||
| - id: "signal_index" |
There was a problem hiding this comment.
- Why not just
idx? - Should it really be signed?
There was a problem hiding this comment.
2. Should it really be signed?
Yes. Why not? Java only allows int for array subscription, IIRC, so if you put type: u4, the generated type will be long, which is wrong.
There should be a good reason to use unsigned types; if there isn't, signed types should be used by default.
There was a problem hiding this comment.
Thanks for the info on how it is going on in JVM-based langs.
There was a problem hiding this comment.
It is semantically incorrect to have negative indexes in an array.
Yes, but unsigned types are not a good solution to this.
There was a problem hiding this comment.
Unsigned types are only good solution for it and only they must be allowed there, any signed type used as an array index must be a compile-time error. In the case of subtraction - if a compiler can prove that the result is non-negative, it should be cast to unsigned automatically. If the compiler cannot - it must be a compile-time error and the programmer must code it so that the compiler can prove it.
It is JVM and Java that are flawed, not unsigned numbers. Wasting 1 bit for a sign that must always be 0 is waste.
| @@ -0,0 +1,182 @@ | |||
| meta: | |||
| id: edf | |||
There was a problem hiding this comment.
The id is too short, conflicts are possible. Please mind a more specific id.
| meta: | ||
| id: edf | ||
| file-extension: edf | ||
| endian: le |
There was a problem hiding this comment.
Here is a template for a metadata block you may find useful:
meta:
id:
title:
application:
file-extension: # todo: https://extension.nirsoft.net/
xref:
ansi:
din:
forensicswiki: # https://forensicswiki.xyz/
fileformat: # https://www.fileformat.info
incits:
iso:
justsolve: # http://fileformats.archiveteam.org/wiki/
loc: # https://www.loc.gov/preservation/digital/formats/
mime:
pronom: # https://www.nationalarchives.gov.uk/pronom/fmt/
- x-fmt/
rfc:
uti:
ffdg: # https://docs.fileformat.com
wikidata: Q
doc: |
doc-ref:
-
-
-
| seq: | ||
| - id: "signals" | ||
| repeat: expr | ||
| repeat-expr: _root.header.ns | ||
| type: signal(_index) |
There was a problem hiding this comment.
seq must come before types in the type specification - see https://doc.kaitai.io/ksy_style_guide.html#type:
2. Order of sections in a type spec
Use the following order of sections:
meta, if present, MUST go firstdocdoc-refseqinstances,types,enums— use one’s best judgement to order these 3 to maximize readability
This is what searching for a name of the type in which a seq structure is located looks like in practice if this rule is not followed (font/ttf.ksy:338-650):
|
Thanks for the comments. I will update the pr when I have time. |
Spec: https://www.edfplus.info/