I have this XML
<Result>
<Dataset name='ident1'>
<Row name='a1'>
<queryname>cat0</queryname>
<superfilename>cat1</superfilename>
<indexfilename>cat2</indexfilename>
</Row>
<Row name='a2'>
<queryname>cat3</queryname>
<superfilename>cat4</superfilename>
<indexfilename>cat5</indexfilename>
</Row>
<Row name='a3'>
<queryname>cat6</queryname>
<superfilename>cat7</superfilename>
<indexfilename>cat8</indexfilename>
</Row>
</Dataset>
<Dataset name='Result 2'>
</Dataset>
<Dataset name='Result 3'>
</Dataset>
<Dataset name='Result 4'>
</Dataset>
</Result>
I want to count the number of rows of DataSet named ident1. The xmlstarlet command I am using is:
xmlstarlet sel -t -v 'count(/Result/Dataset[@name='ident1']/Row)' oscar.xml
I think it should work but it is returning 0.
I have tried other variations but all of them return 0.
xmlstarlet sel -t -v 'count(/Result/Dataset[@name='ident1'])' oscar.xml
xmlstarlet sel -t -v 'count(/Result/Dataset[@name='ident1'][*]/Row)' oscar.xml
xmlstarlet sel -t -v 'count(/Result/Dataset[@name='ident1']/Row[*])' oscar.xml
What am I doing wrong?
NOTE
If I count other element like DataSet it returns correctly 4.
xmlstarlet sel -t -v 'count(/Result/Dataset)' oscar.xml
In this, the quotes are all shell syntax; in consequence, the shell strips the quotes before the query is given to XMLStarlet:
# bad: looks for @name=ident1, no quotes
# literal query is: count(/Result/Dataset[@name=ident1]/Row)
# ...which compares @name against the value of an element under Dataset named ident1
# ...since no such element exists, the result is a count of 0.
xmlstarlet sel -t -v 'count(/Result/Dataset[@name='ident1']/Row)' oscar.xml
Instead, make it:
# good: uses ""s on the outside, so ''s on the inside are literal
# literal query is: count(/Result/Dataset[@name='ident1']/Row)
xmlstarlet sel -t -v "count(/Result/Dataset[@name='ident1']/Row)" oscar.xml
...or, if your shell is bash:
# good (but nonportable): uses $'' syntax, which makes \' produce a single literal '
# literal query is: count(/Result/Dataset[@name='ident1']/Row)
xmlstarlet sel -t -v $'count(/Result/Dataset[@name=\'ident1\']/Row)' oscar.xml
All the above happens because quotes, in shell, are a per-character characteristic. You can, for instance, write:
echo "$foo"'$bar'$baz
...and $foo will be expanded per double-quote rules (literal replacement with comments), $bar will be treated as a literal string, and $baz will be expanded per unquoted-expansion rules (with string-splitting and globbing resulting in oft-unwanted behaviors).