XSLT conversion and test strategy

convert.sh

#!/bin/bash

# Generalised conversion and validation script for XSLT conversions:
# 1. Convert all XMLs in an input directory to new files in an output directory
# 2. Validate all output XMLs against an XML Schema
# 3. Summarise the validation results
# (4. Fix XSLT and re-run)

# Input XML files are in `./inputxml`
# Output XML files go in `./outputxml`
# stderr output of the conversion go in `./conversionlogs` (only non-empty files are kept)
# stderr output of the validation go in `./validationlogs`
# Filtered validation logs go in `./filteredvalidationlogs`
# Summary of validation errors is written to `./validation-errors.txt`

for filename in `ls inputxml`; do
    
    # Convert XML
    echo "Converting $filename..."
    java -jar ~/Applications/saxon9he.jar -s:"inputxml/$filename" \
    -xsl:"in-out.xsl" \
    -o:"outputxml/$filename" 2> "conversionlogs/$filename.log"
    [[ ! -s "conversionlogs/$filename.log" ]] && rm "conversionlogs/$filename.log"
    
    # Validate XML
    echo "Checking $filename..."
    xmllint --noout --postvalid --schema "schema.xsd" \
    "outputxml/$filename" 2> "validationlogs/$filename.log" 
    
    # Filter validation results (if you want to ignore certain validation errors)
    sed -E -f "filtervalidation.txt" "validationlogs/$filename.log" > "filteredvalidationlogs/$filename.log"
done

# Filter all filtered validation logs: keep one line per error type, per file
sed -nE -e "s/^.*outputxml\/([0-9]+\.xml:)[0-9]+: /\1 /p" filteredvalidationlogs/*.xml.log | sort | uniq > validation-errors.txt

filtervalidation.txt

# Example set of sed commands to filter validation logs.

# Ignore the 'no DTD found' message; we're validating against an XML Schema
/no DTD found/d