Dear Maintainer,
please find a script below to reproduce a segmentation fault when
calling ledger convert:
```
#!/bin/sh
set -e
ledger --version
cd /tmp
rm -rf ledger-segfault && mkdir ledger-segfault
cd ledger-segfault
# create three example files (1 ledger, 2 csvs to be imported):
cat << EOF > test.dat
tag CSV
tag Imported
tag reference_id
assert value =~ /^[*]{8}[0-9]{19}/
2025/11/15 Order 12345
Assets:Accounts Receivable EUR 1025
Income:Sales
EOF
cat << EOF > final1.csv
date,description,total,amount,reference_id
2025-11-29,,EUR 25.0,EUR 25.0,********6521949711467142203
EOF
cat << EOF > final2.csv
date,description,total,amount,reference_id
2026-03-03,Order 12345,EUR 100.0,EUR 100.0,********9581361030270551453
EOF
set -x
# call ledger convert for the final1.csv (SUCCESS):
ledger convert final1.csv --input-date-format "%Y-%m-%d" --invert --account "Assets:Bank One" --rich-data --file test.dat --now 2026-03-09
# append ledger convert output of final1.csv to ledger (SUCCESS):
ledger convert final1.csv --input-date-format "%Y-%m-%d" --invert --account "Assets:Bank One" --rich-data --file test.dat --now 2026-03-09 >> test.dat
# call ledger convert for the final2.csv (SEGFAULT):
ledger convert final2.csv --input-date-format "%Y-%m-%d" --invert --account "Assets:Bank One" --rich-data --file test.dat --now 2026-03-09
```
Running the script through strace gives me this:
```
[...]
wait4(-1, [{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0, NULL) = 319633
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_EXITED, si_pid=319633, si_uid=1000, si_status=0, si_utime=0, si_stime=0} ---
rt_sigreturn({mask=[]}) = 319633
wait4(-1, 0x7ffdd799f06c, WNOHANG, NULL) = -1 ECHILD (No child processes)
dup2(11, 1) = 1
close(11) = 0
write(2, "+ ", 2+ ) = 2
write(2, "ledger", 6ledger) = 6
write(2, " convert", 8 convert) = 8
write(2, " final2.csv", 11 final2.csv) = 11
write(2, " --input-date-format", 20 --input-date-format) = 20
write(2, " %Y-%m-%d", 9 %Y-%m-%d) = 9
write(2, " --invert", 9 --invert) = 9
write(2, " --account", 10 --account) = 10
write(2, " Assets:Bank One", 16 Assets:Bank One) = 16
write(2, " --rich-data", 12 --rich-data) = 12
write(2, " --file", 7 --file) = 7
write(2, " test.dat", 9 test.dat) = 9
write(2, " --now", 6 --now) = 6
write(2, " 2026-03-09", 11 2026-03-09) = 11
write(2, "\n", 1
) = 1
rt_sigprocmask(SIG_SETMASK, ~[RTMIN RT_1], NULL, 8) = 0
vfork() = 319634
rt_sigprocmask(SIG_SETMASK, [], ~[KILL STOP RTMIN RT_1], 8) = 0
wait4(-1, [{WIFSIGNALED(s) && WTERMSIG(s) == SIGSEGV}], 0, NULL) = 319634
--- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_KILLED, si_pid=319634, si_uid=1000, si_status=SIGSEGV, si_utime=0, si_stime=0} ---
rt_sigreturn({mask=[]}) = 319634
write(2, "Segmentation fault\n", 19Segmentation fault
) = 19
0x7ffdd799f06c, WNOHANG, NULL) = -1 ECHILD (No child processes)
exit_group(139) = ?
+++ exited with 139 +++
```
which doesn't tell me much.
Commenting out the regex tag assertion for reference_id in test.dat
solves the problem, so I _assume_ it has something to do with it.
Might need forwarding upstream, if the issue persists on newer versions
of upstream. I haven't tested newer upstream versions.
Thank you for looking into this and your work on the package!