pyo3_runtime.PanicException: no entry found for key

[raczben]

I tried to parse this file:
https://raw.githubusercontent.com/amrhas/PDRNoC/refs/heads/VCRouter/noctweak/Debug/wavform.vcd.vcd

When I get the following error:

thread '<unnamed>' (9314) panicked at wellen/src/vcd.rs:1244:39:
no entry found for key
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
 Parsing...Traceback (most recent call last):
  File "/****/main.py", line 105, in <module>
    vcd = pw.Waveform(file)
          ^^^^^^^^^^^^^^^^^
pyo3_runtime.PanicException: no entry found for key

System info:

• Python 3.12.3
• pywellen==0.20.2
• Ubuntu 24.04.3 LTS

[raczben]

The same issue with:
https://raw.githubusercontent.com/DarthSkipper/myHDL_Sigmoid/refs/heads/master/out/testbench/sigmoid_tb.vcd

[ekiwi]

Thanks for reporting. I was able to reproduce the crash on my machine. I will look into this over the next couple of weeks.

[oscargus]

For the second file, the problem is that the state variables are marked to be real: $var real 1 , state $end, but the value change is a string scount ,. Now, string is not part of the VCD standard, but it is clearly inconsistent to mark a string variable as real (although it used to work earlier in wellen IIRC as reals were not parsed but stored as strings).

If MyHDL emitted $var string 1 , state $end instead it will work.

[ekiwi]

Thanks for diagnosing, @oscargus. I have encountered the same problem with MyHDL before:

wellen/wellen/tests/diff_tests.rs

Line 553 in 4c40f7e

#[ignore] // TODO: this file declares a `real` signal and then emits strings ... :(

Back then, I decided not to fix it because MyHDL isn't super popular, and it is actually non-trivial to get wellen to play nice with signals that are declared one way and then get different data. There might be a shortcut, by detecting files from MyHDL and then pre-scan the whole VCD to find mismatches before doing the full parsing. I will keep this bug on my list, but it isn't super likely to get fixed anytime soon.

I will look at the SystemC file when I have time.

[ekiwi]

I made an issue to track the handling of the MyHDL bug: #91