parsing and replacing some strings in two files - linux

Question

I want to run a shell script with this usage:
./run A.txt B.xml
A.txt contain some statistic:
Accesses = 1
Hits = 2
Misses = 3
Evictions = 4
Retries = 5
B.xml looks like:
<stat name="total_accesses" value="0"/>
<stat name="total_misses" value="0"/>
<stat name="conflicts" value="0"/>
I want to replace some stats in B.xml from A.txt. For example, I want to
1- find "Accesses" in A.txt
2- find "total_accesses" in B.xml
3- replace 0 with 1
1- find "Misses" in A.txt
2- find "total_misses" in B.xml
3- replace 0 with 3
So B.xml will look like:
<stat name="total_accesses" value="1"/>
<stat name="total_misses" value="3"/>
<stat name="conflicts" value="0"/>
I want to do that with shell "sed" command. However I find it quite complex as the regexp is hard to understand.
Does "sed" help me in this problem or I have to find another way?

Answer 1

It might be a bit heavy-weight for such a simple case, but here's a Python script that does the job:
#!/usr/bin/env python
import sys
import xml.etree.ElementTree as etree
# read A.txt; fill stats
stats = {}
for line in open(sys.argv[1]):
if line.strip():
name, _, count = line.partition('=')
stats["total_"+name.lower().strip()] = count.strip()
# read B.xml; fix to make it a valid xml; replace stat[#value]
root = etree.fromstring("<root>%s</root>" % open(sys.argv[2]).read())
for s in root:
if s.get('name') in stats:
s.set('value', stats[s.get('name')])
print etree.tostring(s),
Example
$ python fill-xml-template.py A.txt B.xml
<stat name="total_accesses" value="1" />
<stat name="total_misses" value="3" />
<stat name="conflicts" value="0" />
To process input files incrementally or to makes changes inplace you could use the following:
#!/usr/bin/env python
import fileinput
import sys
import xml.etree.ElementTree as etree
try: sys.argv.remove('-i')
except ValueError:
inplace = False
else: inplace = True # make changes inplace if `-i` option is specified
# read A.txt; fill stats
stats = {}
for line in open(sys.argv.pop(1)):
if line.strip():
name, _, count = line.partition('=')
stats["total_"+name.lower().strip()] = count.strip()
# read input; replace stat[#value]
for line in fileinput.input(inplace=inplace):
s = etree.fromstring(line)
if s.get('name') in stats:
s.set('value', stats[s.get('name')])
print etree.tostring(s)
Example
$ python fill-xml-template.py A.txt B.xml -i
It can read from stdin or process several files:
$ cat B.xml | python fill-xml-template.py A.txt
<stat name="total_accesses" value="1" />
<stat name="total_misses" value="3" />
<stat name="conflicts" value="0" />

Answer 2

Here is a shell script that does what you want:
#!/bin/bash
while read line
do
key=`echo $line | cut -d' ' -f1`
value=`echo $line | cut -d' ' -f3`
xmlLine=`grep -i $key $2`
if [ -n "$xmlLine" ]; then
for num in `seq 5`
do
field[${num}]=`echo "$xmlLine" | cut -d'"' -f${num}`
done
echo ${field[1]}\"${field[2]}\"${field[3]}\"$value\"${field[5]}
fi
done
You can copy it to a file say A.sh , give run permissions to it (chmod +x A.sh) and then:
./A.sh A.txt B.xml
Please mind that this code is not suitable for production and regex is paramount for these scripts.

Answer 3

while you can hack this on the command line, I'd recommend not to do this.
XML is way too fragile to be handled this way - use a proper XML library and parse the XML before manipulating it. Otherwise you could easily end up with broken XML. e.g. write a script in Ruby, Python, or Perl and use an XML library.