Shell Programming and Scripting

Sorry about that, I meant to post an abbreviated version of the ill formed XML which is only 65 lines. I have attached that here in case anyone want to have a look. The .zip also includes two .doc files. The first is the ill formed XML with comments describing the issues. The problem tags are in bold blue. The second is the revised version with comments describing the corrections. Additions are in bold red. I hope this will be helpful to anyone looking to correct a similar issue.

The script is working and I suppose isn't overly kludgey. I am hard coding the text format as ASCII. Is there an easy way in bash to id the encoding of the input file?

I have made a couple of changes. I am naming all the created attributes the same as the tag name, since this is simpler and now the replacements are done with a more general rule. The exception is the UgUn tag, since there is an extra space to be dealt with, I guess. I have also changed to perl based on the format of the earlier posts.

Here is the current script

Code:

#!/bin/bash

infile=$1
outfile=$1".mod.xml"

# add version and root begin tags to beginning of input file
echo '<?xml version="1.0" encoding="ASCII"?>' > temp1
echo "<net>" >> temp1
cat temp1 $1  > temp2

# format information tags (end with />)
perl -pe 's/<Fmt\ (\w+)>/<Fmt\ Fmt="\1"\/>/'         temp2 | \
perl -pe 's/<Name\ (\w+)>/<Name\ Name="\1"\/>/'            | \
perl -pe 's/<Epoch\ (\n+)>/<Epoch\ Epoch="\1"\/>/'         | \

# format UgUn tag arguments of different name than tag
sed 's/<UgUn  *\([^ ]\{1,9\}\) *>/<UgUn UgUn="\1">/g'      | \

# format tags where tag and arg have same value
# word | space | integer
perl -pe 's/<(\w+)\ (\n+)>/<\1 \1="\2">/'                   | \
# word | space | word
perl -pe 's/<(\w+)\ (\w+)>/<\1 \1="\2">/'                   | \

# format tags with multiple args
awk '{ if ( $0 ~ /<Cg 0 Fm:Input>/ ) {
     printf( "%s\n", "<Cg Cg=\"0\" Fm=\"Input\">" );
     } else {
          print $0;
     }
}'                                                         | \
awk '{ if ( $0 ~ /<Cg 0 Fm:Hidden>/ ) {
     printf( "%s\n", "<Cg Cg=\"0\" Fm=\"Hidden\">" );
     } else {
          print $0;
     }
}'                                                         | \

# format multi row, multi are data
perl -p0e 's/(<Un>\n)(.*)/\1<Bias>\2<\/Bias>/g'            | \
perl -pe 's/(\d+) ([\d.-]+)/<C\1>\2<\/C\1>/ if /<Cn/../<\/Cn/'   > $outfile

# add root close tag to end of file
echo "</net>" >> $outfile

# cleanup
rm temp1 temp2

The exception is the UnUg tag, which I can't get perl to find, and the tags with multiple args, which are still in awk. I am not sure about the current solution, since it doesn't take into account the fact that the args may have different values that are being searched for.

I could switch to using more temp files if you think that would help.

I really don't know perl all that well at all. On the whole, I am pretty poor at regex stuff. I use it some, since it's so d**n convenient, but what I know is just bits and pieces of various interpreters and stream editors. I can modify stuff in perl an awk, but that is about it. I wish I knew this material better, but that is just on the long list of such things.

LMHmedchem

Numeral characters class in Perl is \d, so I guess you want this: (\w+) (\d+)
Tomorrow I'll take a look at those .doc files.

I edited the post above and have changed most of the script to perl. I guessed and used (\w+) (\n+) and that seemed to work. I also added an excape (\w+)\ (\d+) since I wanted to match word | space | number, but maybe that's implicit. The only thing I can't manage is the UnUg tag where there is a space between the int and the >. I am also not sure what to do with the tags with multiple args, so those are still in awk.

Thanks very much for all your assistance.

LMHmedchem