#!/bin/bash

function die
{
   echo "$1"
   exit 1
}

if [[ $# -ne 1 ]]; then
   die "$0 <script.pl>"
fi
PROG=$1

perl -c -w $PROG > /dev/null 2>&1 || die "$PROG contains contains syntax errors"

# Check byte encoding
OUTPUT=$(perl $PROG testdata/test_0?.txt | awk '{print $2}' | uniq | wc -l)
if [[ "$OUTPUT" != "1" ]]; then
   die "$LINENO: Unexpected variation in high order nibble: $OUTPUT"
fi

OUTPUT=$(perl $PROG testdata/test_0?.txt testdata/test_0116a0.txt | awk '{print $3}' | sort -u | wc -l)
if [[ "$OUTPUT" != "16" ]]; then
   die "$LINENO: Insufficient variation in low order nibble: $OUTPUT"
fi

OUTPUT=$(perl $PROG testdata/test_?0.txt | awk '{print $2}' | sort -u | wc -l)
if [[ "$OUTPUT" != "16" ]]; then
   die "$LINENO: Insufficient variation in high order nibble: $OUTPUT"
fi

OUTPUT=$(perl $PROG testdata/test_?0.txt | awk '{print $3}' | uniq | wc -l)
if [[ "$OUTPUT" != "1" ]]; then
   die "$LINENO: Unexpected variation in low order nibble: $OUTPUT"
fi

# Check that reading from stdin produces same output as reading from file
OUTPUT1=$(perl $PROG testdata/test_00.txt | sed -e 's/\*.*//')
OUTPUT2=$(perl $PROG < testdata/test_00.txt | sed -e 's/\*.*//')
if [[ "$OUTPUT1" == "" ]]; then
   die "$LINENO: Unexpected empty digest"
fi
if [[ "$OUTPUT1" != "$OUTPUT2" ]]; then
   die "$LINENO: reading from stdin failed"
fi

# Check offset encoding
OUTPUT=$(perl $PROG testdata/test_??.txt | awk '{print $1}' | uniq | wc -l)
if [[ "$OUTPUT" != "1" ]]; then
   die "$LINENO: Unexpected variations in offsets: $OUTPUT"
fi
OUTPUT=$(perl $PROG testdata/test_01*.txt | awk '{print $1}' | sort -u | wc -l)
if [[ "$OUTPUT" != "2" ]]; then
   die "$LINENO: Unexpected variations in offsets: $OUTPUT"
fi

# Check that all files are given unique digests
OUTPUT1=$(perl $PROG testdata/test_*.txt | awk '{print $1 " " $2 " " $3}' | sort -u | wc -l)
OUTPUT2=$(ls testdata/test_*.txt | wc -l)
if [[ "$OUTPUT1" != "$OUTPUT2" ]]; then
   die "$LINENO: Digests are not unique: $OUTPUT1 vs $OUTPUT2"
fi

# Check that two byte encoding is used where needed
OUTPUT=$(perl $PROG testdata/collision_* | grep -F collision_base.txt)
if ! ( echo $OUTPUT | grep -qF " + " ); then
   die "$LINENO: Expected 2-tuple digest"
fi
OUTPUT=$(perl $PROG testdata/collision_base.txt testdata/collision_1* | grep -F collision_base.txt)
if ( echo $OUTPUT | grep -qF " + " ); then
   die "$LINENO: Unexpected 2-tuple digest"
fi

# Check that same abbreviated digest is used for two different files
# with the same MD5.
OUTPUT1=$(perl $PROG testdata/test_10.txt | sed -e 's/\*.*//')
OUTPUT2=$(perl $PROG testdata/collision_12.txt | sed -e 's/\*.*//')
if [[ "$OUTPUT1" == "" ]]; then
   die "$LINENO: Unexpected empty digest"
fi
if [[ "$OUTPUT1" != "$OUTPUT2" ]]; then
   die "$LINENO: Digest mismatched for same-MD5 files"
fi
if ( echo $OUTPUT1 | grep -qF " + " ); then
   die "$LINENO: Unexpected multi-tuple digest"
fi

# Generate and check digest for a set of files
TMPFILE=$(mktemp)
if [[ "$TMPFILE" == "" ]]; then
   die "$LINENO: Error creating temporary file"
fi
if ! ( perl $PROG testdata/test_*.txt > $TMPFILE ); then
   rm -f $TMPFILE
   die "$LINENO: Failed to generate digest"
fi
if [[ ! -s $TMPFILE ]]; then
   rm -f $TMPFILE
   die "$LINENO: Generated empty digest"
fi
OUTPUT=$(perl $PROG -c $TMPFILE 2>&1 | grep -vF "OK")
rm -f $TMPFILE
if [[ "$OUTPUT" != "" ]]; then
   die "$LINENO: Failed to verify digest from file"
fi

# Verify that digest can be read from stdin
OUTPUT=$(perl $PROG testdata/*.txt | perl $PROG -c 2>&1 | grep -vF "OK")
if [[ "$OUTPUT" != "" ]]; then
   die "$LINENO: Failed to verify digest from stdin"
fi
OUTPUT=$(perl $PROG testdata/*.txt | perl $PROG -c - 2>&1 | grep -vF "OK")
if [[ "$OUTPUT" != "" ]]; then
   die "$LINENO: Failed to verify digest from stdin"
fi

# Check that files with same digest are listed separately
OUTPUT=$(perl $PROG testdata/test_10.txt testdata/collision_12.txt | wc -l)
if [[ "$OUTPUT" != "2" ]]; then
   die "$LINENO: Expected two separate digest lines"
fi

# Check that file with the same digest uses only minimal encoding
OUTPUT=$(perl $PROG testdata/test_10.txt testdata/collision_12.txt | sed -e 's/\*.*//' | uniq | wc -l)
if [[ "$OUTPUT" != "1" ]]; then
   die "$LINENO: Expected a single digest variety"
fi
OUTPUT=$(perl $PROG testdata/test_10.txt testdata/collision_12.txt | sed -e 's/\*.*//' | uniq | wc | awk '{print $2}')
if [[ "$OUTPUT" != "3" ]]; then
   die "$LINENO: Expected 3-word digest"
fi

# Check that digests are specific to each file
OUTPUT1=$(perl $PROG testdata/test_0a.txt testdata/test_0b.txt | head -1 | sed -e 's/\*.*//')
OUTPUT2=$(perl $PROG testdata/test_0a.txt testdata/test_0b.txt | tail -1 | sed -e 's/\*.*//')
if [[ "$OUTPUT1" == "$OUTPUT2" ]]; then
   die "$LINENO: Expected different digests"
fi
OUTPUT=$(perl $PROG -c "$OUTPUT1" testdata/test_0a.txt 2>&1 | grep -F OK)
if [[ "$OUTPUT" == "" ]]; then
   die "$LINENO: Failed to verify digest for test_0a.txt"
fi
OUTPUT=$(perl $PROG -c "$OUTPUT1" testdata/test_0b.txt 2>&1 | grep -F FAIL)
if [[ "$OUTPUT" == "" ]]; then
   die "$LINENO: Failed to reject digest for test_0b.txt"
fi
OUTPUT=$(perl $PROG -c "$OUTPUT2" testdata/test_0a.txt 2>&1 | grep -F FAIL)
if [[ "$OUTPUT" == "" ]]; then
   die "$LINENO: Failed to reject digest for test_0a.txt"
fi
OUTPUT=$(perl $PROG -c "$OUTPUT2" testdata/test_0b.txt 2>&1 | grep -F OK)
if [[ "$OUTPUT" == "" ]]; then
   die "$LINENO: Failed to verify digest for test_0b.txt"
fi

# Check that directories are handled gracefully
OUTPUT=$(perl $PROG testdata 2>&1 | grep -F "Error reading testdata")
if [[ "$OUTPUT" == "" ]]; then
   die "$LINENO: Poor handling of directories"
fi

echo "$0: OK"
exit 0