I have a bash script that encapulates a perl script, which generates some report data, and a cron that pipes this data to mail. The problem is that the data being generated is raw HTML code, without any HTML or BODY tags, so the content sent to email includes these raw tags and looks kind of ugly.

Is there a quick way to add a script somewhere that either encapsulates the output in HTML tags, or strips out the tags from the output (<b>, </b>, &nbsp;, etc.)?

Just run tidy against it.



Cheers,
Tink

Thanks. But what if tidy is not installed (not on this server)?

Hi,

Install it

HTML Tidy Library Project

Either what druuna said, or edit the perl-script and stop
it from outputting the tags, or make it output the proper
header & body tags in the right spots, too? :}

It seems a bit silly to run something that produces extra
data and then use another tool to undo the efforts of the
first tool.



Cheers,
Tink

Here's the thing: the perl scripts are part of TWiki, one of which is spitting out the raw HTML. I'd rather not modify the TWiki code itself, and I've asked the owner of this particular bit of code to consider adding a flag for output format, but no luck so far.

In addition, I have limited access to installing new packages on the system, as the admins prefer prepackaged RPMs from their centralized repositories.

I thought a solution might be as "simple" as adding a step to the cron that threw the script output into a variable, and I could somehow add the necessary HTML tags "around" the variable, so the end result was a readable email.

However, if the viable solution is that tidy would be just an additional pipe in the cron command ("perl script | tidy | mail"), it makes sense to request an exception and get tidy installed on the server.

Well, you *could* brute-force it and (assuming that none of the pages
payload has "<" or ">" in it) use a pipe through sed to mangle the
pages twiki produces. The simple sed command below should get rid of
most HTML I'd think.

sed -r 's/<[^>]+>//g'



Cheers,
Tink

Or, if you really wanted to get frankenstein, you could have your shell script run a perl script that runs the aformentioned report generation. Thus, you could have the wrapper script parse the data however you want.

Still though I see the easiest and cleanest solution is to add several very minor alterations to the initial code

Things are already fairly Frankensteined in support of this application, so while I'd like to avoid further customization, another bolt here or there shouldn't make much of a difference.

Tink's sed command could be a good start. Since it strips out angle-bracketed tags, it also strips out line breaks (obviously), so the output is all on one line, and non-breaking space symbols (&nbsp; ) are left in.

I looked at Tidy, but couldn't find a binary for my Linux (RHEL4), and the source is C-based, which I've never compiled on Linux.

Finally, I'm looking at the code of TWiki's publishing script (Publish.pm) for the right place to insert: ... tags around the current output, and how. I thought the variable to be modified would be $tmpl, and the place would be right before this line that writes to HTML:
However, this would modify the actual file output, which seems to be working as expected. I just need to modify the standard output sent by the script itself (normally sent to the screen, but being piped to mail), not the file being produced by the script.

As this is a Perl app, you could also ask at www.perlmonks.org. There's some very sharp people there.