hi

i need to use unix to extract data from several rows of a table coded in html. I know that rows within a table have the tags <tr> </tr> and so i thought that my first step should be to to delete all of the other html code which is not contained within these tags. i could then use this method again but remove everything not in <td> </td> tags. but the big question is how can i do this? i think i need sed but at the moment it is just confusing me too much any help?

In principle you are right. The following script will extract everything between a "<tr>" and "</tr>" tag. It will assume that there are no multiple "<tr>-</tr>"-pairs on a single line and the tags themselves are all lowercase (no "<TR>").

The result might not be what you need, though, so you might consider giving us a sample of what you have and what you will need to get from it. This would help us to help you better.

I hope this helps.

bakunin

Hi.

See HTML table to CSV for another approach using lynx -dump.

In general, links to threads similar to yours are posted at the bottom of the thread ... cheers, drl

thanks bakunin that is really helpful. i cant post a sample of the html page for various reasons. the only problem with your solution is that most of the <tr> tags are across multiple lines in my html page. ie the tag may be opened on line 7 and then closed on line 20. hence is it possible with sed to delete everything on a line (including the line) BUT stop when it gets to a <tr> tag and start again when it gets to a </tr>? alternatively is there a way to make sed believe that the whole html page is on a single line?

as i am not familiar with the capabilities of sed, it makes it hard for me to know what the best way of completing this task is.

There's no reason you can't mock up an HTML page which looks like the one you're working with but which does not contain any sensitive information. Nobody is interested in throwing darts into a dark room.

If you post something, someone will post code. Otherwise, you're going to have to do it yourself. Try something like replacing all newlines in the file with spaces, splitting the file before each < or after each >, and going from there. If you may have a < or > within the data, then you're going to do a little extra work. That's the best I can do for you at the moment.

ShawnMilo

trying to answer my own question here, but im still struggling if this doesnt work then i will mock up an example, i just thought that my description may have been good enough withough having to waste time making an example table.

i found on this site Sed - An Introduction and Tutorial that you can create ranges by patterns. the example code is:

i tried making <tr> my start and </tr> my stop but i just kept getting errors. furthermore i would have to NOT (!) this so instead of deleting everything in the tags, it deletes everything outside the tags.
could someone please help me get this sed command working?

thanks

Well, i told you that - in absence of any example - i had to make some assumptions. Here is a new version which will work on tags ranging over several lines. It will still not catch the case of several "<tr>...</tr>" pairs on one line, though.

How this works: the "-n" clause will stop sed from printing every line it has read, so if you delete the script it would print just nothing. This is to (implicitly) throw out all the lines which are NOT in the specified range.

Everything between the curly braces is executed only when inside the range specified on line 1. As you can see the last command inside the curly braces is a "p", which will print everything inside this range. If you delete the two "s/...."-commands it would print something this:

As you can see the bold parts should be deleted as they are not part of what you want. The two "s/..."-commands (s=substitute) take care of that along with the tags themselves. At last the p(rint)-command outputs the result of all the trimming.

One more word, though: You got a second answer from me because i appreciated that you were doing genuine research on your own. You almost forfeited this answer because of this:

You might notice i have "wasted time" not only writing a script but even wasted more time explaining how it works in the hope of not only solving the problem at hand but enhancing your understanding at the same time. On top of that i "wasted some more time" to write a script in my first post which nobody is going to need because it was based on faulty assumptions. Assumptions which might not have been faulty at all would i have been able to work from an example created by "wasting time".

I am even now "wasting some more time" to explain to you why you might sometimes get no answer at all or some answer you can't use. Go figure.

I hope this helps.

bakunin