Can grep do something like this?

[PaulR]

Using BaCon with some shell commands (a total novice in the scripting department!)...

I'm trying to parse a dynamically generated (aspx) html file to get at data stored in a table.

First I'm using wget with a custom-built url to get the page to a local file (page.txt). Next I'm using grep like this to just get the lines containing data...

SYSTEM "grep '<tr class=\"aclass\">' page.txt > data.txt"


Then I'm stepping through the lines in data.txt and getting the information between the <td></td> tags.

Ir works great except that the first line contains a single record AND lots of stuff before it (the table tags, table headers etc).

My question therefore: is it possible for grep to return only the contents of the line after the match rather than the whole line? For example, if the whole line was like this;

<tr class=\"aclass\"><td>some data here</td></tr>

Can grep return:

<td>some data here</td></tr>

Ideally I'd like to do this in one line rather than execute a separate script

TIA

Paul

[seaside]

PaulR,

You can do this-

Code: Select all

 echo '<tr class=\"aclass\"><td>some data here</td></tr>'|grep -o '<td>.*'
<td>some data here</td></tr>

"Grep -o target.*" matches the target only and .* returns the rest of the line.

Regards,
s

[some1]

Hi
looks like an awk-job
awk is good for tag-parsing

-maybe as simple as this:

awk '/<td>/,/<\/td>/' "$filein">"$fileout"

get stuff between <td> and </td> : escape the /

Cheers

[PaulR]

They look promising! Thanks both, I'll try them out later

Paul

[Karl Godt]

Can grep return:

<td>some data here</td></tr>

grep has as many linux tools the --help or -h option .

This would show

-o, --only-matching show only the part of a line matching PATTERN

option .

Code: Select all

SYSTEM "grep -o  '<tr class=\"aclass\">' page.txt > data.txt"

DisClaimer > I am NO BaCon expert .

[seaside]

PaulR,

You could code it all in Bacon using just one system call to wget. Vovchik has coded Yweather which parses xml data all written in bacon basic for example.

http://murga-linux.com/puppy/viewtopic. ... h&id=63295

Regards,
s

[PaulR]

I couldn't get awk or grep to work - both kept returning the entire line (including the search target) rather than text from the target onwards.

In the end I made a work-around with two lines of BaCon code. It'll be fine as long as the format of the web page doesn't change radically!

Paul

[Ibidem]

I know it's late, but...
I would have done something like this:
sed -ne 's/.*<tr class="class">\(<td>.*</td></tr>\).*/\1/gp'

grep cannot modify matches; you print everything, the whole match, or nothing.

[PaulR]

Thanks for that, I don't pretend to understand the syntax at present but it's always useful to have an alternative!

Paul