Parsing the html based on tr and td tag values

I want to parse the html with bash where tr containing class as error like below in my whole html page.

<tr class="error">
<td>
<a href="https://exmple.com">Test failed for AAA</a>
</td>
<td>1</td>
<td>1</td>
<td>0</td>
<td>1</td>
<td>0</td>
<td>0.0%</td>
<td>3.640 seconds</td>
</tr>

Output like "Test failed for AAA"

I tried few things with sed but not working as expected & getting NULL values.

Any input could be helpful

As always, using line based regular-expression oriented tools for working with tree-based documents like HTML and XML is the wrong approach. Use tools aware of the format; much easier, less error prone and simpler to maintain to accommodate any potential future changes in the input data.

For example, using xmllint and an XPath query:

 $ xmllint --html --xpath '//tr[@class="error"]/td[1]/a/text()' input.html
Test failed for AAA

Or with W3C's HTML-XML Utils package and CSS selectors:

 $ hxselect -c 'tr.error td:first-child a' < input.html
Test failed for AAA

(These might not print a trailing newline at the end, which might be confusing if used interactively instead of capturing the result in a variable or whatever)

Look it isn't the prettiest but you can do this:

Find the lines in between tr class="error and </tr> and print only the lines that contain href :

awk '/tr class="error/{found=1};/<\/tr>/{found=0};found==1 && /href/' test.html

There is 3 scripts in here:

• Search for tr class="error and set variable found to 1
• Search for </tr> and set variable found to 0
• If found is 1 search for href and if that matches print $0 (which is current line)

Only print the stuff in between > and < :

grep -Po ">\K.*(?=<)"

Some things to note:

• -P perl regexes for more regex magic
• -o for only the matched parts and not the whole line
• \K in the regex is lookbehind (has to match but will not in the matched part) and all the stuff before the \K is lookbehind (its actually called reset capture group, but the difference is small)
• (?=<) The (?=...) construct is lookahead (has to match but will not be in matched part) and in this case it will look for <

Putting it together:

awk '/tr class="error/{found=1};/<\/tr>/{found=0};found==1 && /href/{print $0}' test.html | grep -Po ">\K.*(?=<)"

Another way would be to remove all newlines and use some regex in grep:

tr -d "\n" < test.html | grep -Po "class=\"error.*?href=.*?>\K.*?(?=<)"