[tex4ht] [bug #625] Bad register code (65536) again on auto-generated Latex code

[Nasser M. Abbasi]

Thanks for the info Michal;

Yes, I changed my texmf.cnf first thing when installing texlive
to increase the counters, else will not be able to compile
most of my files.

>cat texmf.cnf 
buf_size=9000000
pool_size=9000000
main_memory=9000000
save_size=1000000
max_strings = 5000000
strings_free = 1000000
hash_extra=10000000
stack_size=500000


I've also just put the above file in the zip file:

<https://12000.org/tmp/bad_register_code_tex4ht.zip>

I think for now, I'll just delete these long equations and
not write them to the latex file as this seems simplest solution.

Since this is all done in code, I changed my code to check
for large Latex equations.

Computer algebra systems tend to generate huge Latex expressions.

Regards,
--Nasser



On 4/29/2024 5:32 AM, Michal Hoftich wrote:
> Follow-up Comment #2, bug #625 (project tex4ht):
> 
> Hi Nasser, this is not limited to TeX4ht but to l3regex, which is used to
> clean the MathJax code. See this question, where they found the same issue:
> https://tex.stackexchange.com/q/654633/2891
> 
> In theory, you could skip the regex cleaning using this configuration file:
> 
> %%%%%%%%%%%%%%%%
> \Preamble{xhtml}
> \ExplSyntaxOn
> \cs_set_protected:Npn \alteqtoks #1
> {
>    \HCode{\detokenize{#1}}
> }
> \ExplSyntaxOff
> \begin{document}
> \EndPreamble
> %%%%%%%%%%%%%%%%%%%%%
> 
> The downside is that you can end with incorrect characters in your HTML, in
> particular <, > and &. These can cause MathJax fail.
> 
> Also, to break your long strings, you can use this prefilter:
> 
> %%%%%%%%%%%%%%%
> local max_len = 255
> for line in io.lines() do
>    line = line:gsub("\r", "")
>    local str_len = string.len(line)
>    if str_len > max_len then
>      -- if the line is longer than maxlen, we will break it to a shorter
> segments
>      local curr_pos = max_len
>      local prev_pos = 1
>      while curr_pos < str_len do
>        -- find next command preceded by spaces starting at the current
> position
>        curr_pos, len = string.find(line, "%s+\\", curr_pos)
>        print(curr_pos, str_len,string.sub(line, prev_pos, curr_pos))
>        prev_pos = curr_pos
>        -- we must move the current position
>        curr_pos = curr_pos + max_len
>      end
>      -- print rest of the line
>      print(string.sub(line, prev_pos, str_len))
>    else
>      print(line)
>    end
> end
> 
> %%%%%%%%%%%%
> 
> Use it as:
> 
> $ texlua format.lua < original.tex > formated.tex
> 
> It will format your long lines into much shorter chunks, preventing the "!
> Unable to read an entire line---bufsize=200000" error I encountered. I suppose
> that you set the limit higher, but I didn't and couldn't compile your file
> originally.
> 
>      _______________________________________________________
> 
> Reply to this item at:
> 
>    <http://puszcza.gnu.org.ua/bugs/?625>
> 
> _______________________________________________
>    Message sent via/by Puszcza
>    http://puszcza.gnu.org.ua/
>