danielhrisca / asammdf

Fast Python reader and editor for ASAM MDF / MF4 (Measurement Data Format) files
GNU Lesser General Public License v3.0
655 stars 226 forks source link

##TX block unexpectedly converted to ##MD block upon saving file #992

Open markwaterbury opened 8 months ago

markwaterbury commented 8 months ago

Python version

('python=3.10.11 (tags/v3.10.11:7d4cc5a, Apr  5 2023, 00:38:17) [MSC v.1929 64 bit (AMD64)]')
'os=Windows-10-10.0.19045-SP0'
'numpy=1.26.4'
'asammdf=7.4.1'

Code

MDF version

'4.10'

Code snippet

m=MDF('<name>.mf4')
m.save('<name>_resave.mf4')
m.close()

Traceback

N/A

MDF Validator Error

Invalid XML: Element '{http://www.asam.net/mdf/v4}e' is unexpected according to content model of parent element '{http://www.asam.net/mdf/v4}CNcomment'. Expecting: {http://www.asam.net/mdf/v4}names, {http://www.asam.net/mdf/v4}linker_name, {http://www.asam.net/mdf/v4}linker_address, {http://www.asam.net/mdf/v4}axis_monotony, {http://www.asam.net/mdf/v4}raster, {http://www.asam.net/mdf/v4}formula, {http://www.asam.net/mdf/v4}address, {http://www.asam.net/mdf/v4}common_properties, {http://www.asam.net/mdf/v4}extensions.

Description

When saving a file, whether modifications were made to it or not, asammdf makes some changes which lead to issues in the MDF validator. This only happens when a channel has a valid cn_md_comment text block (id=="##TX"), and that block has a nonempty tx_data field.

What asammdf does:

danielhrisca commented 8 months ago

Please give a concrete example of the original block and the modified block

markwaterbury commented 8 months ago

Here is an event example. The same behavior can be seen with channel blocks.

Original Event Block (##EV) and ev_md_comment (##TX):

image image tx_data contents: <EVcomment><TX></TX> <e name="DateTime">2023-11-01 08:32:32 (local = UTC +01:00)</e> <e name="DateTimeUTC">2023-11-01 08:32:32 (UTC)</e> <e name="EventType">PreTrigger</e></EVcomment>

Resaved Event Block (##EV) and ev_md_comment (##MD):

image image md_data contents: <EVcomment><TX></TX> <e name="DateTime">2023-11-01 08:32:32 (local = UTC +01:00)</e> <e name="DateTimeUTC">2023-11-01 08:32:32 (UTC)</e> <e name="EventType">PreTrigger</e></EVcomment>

Not much difference in the actual event block, other than addresses (resaved file was not compressed, as the original was), but the ev_md_comment blocks show the difference. The data inside the tx/md data blocks remains unchanged, it is simply the reclassification as an XML text metadata block instead of a Text comment block that causes the XML validation to be run (and fail).

danielhrisca commented 8 months ago

If the comment block contains XML then it must be stored as ##MD

markwaterbury commented 8 months ago

When I read the MDF spec (5.7-5.8), it seems the inverse is true: if the block is ##MD, it must contain XML. But the ##TX block can have any UTF-8 text in it.

Additionally, it states:

if explicitly a TXBLOCK is required, it cannot be replaced by an MDBLOCK

As the tool wrote these as text blocks, I do not think they should have been replaced with meta data blocks, even if the text in them is XML.

danielhrisca commented 8 months ago

Which tool wrote the file originally? It looks like a dumb way to write the XML

markwaterbury commented 8 months ago

ATI VISION. I don't disagree with you, but it does appear to be valid.

danielhrisca commented 8 months ago

You should report an issue to ATI VISION. Custom e tags must be places inside the common_properties tag

markwaterbury commented 8 months ago

Again, I don't disagree with you, but if the text is not in an MDBLOCK it should not be interpreted as XML. So it seems like the TXBLOCK should be able to have anything in it, XML or not, regardless of XML schema.

Georgeeoo commented 6 months ago

please,could you tell me how to change METADATA?

Georgeeoo commented 6 months ago

vector validator only can watch METADATA, but now I dont know how to change it.