little-things/iasha1check

#!/bin/bash
# Fetch the SHA-1 hashes from an IA item and ensure that they match the local files (i.e. that the upload was successful)

identifier="$1"
escapedIdentifier="$(sed 's/[.[\*^$()+?{|]/\\&/g' <<<"${identifier}")"
readarray -t iasha1sums < <(curl -sL "https://archive.org/download/${identifier}/${identifier}_files.xml" | tr -d '\n' | grep -Po '<file .*?</file>' | grep 'source="original".*<sha1>' | sed 's,^.*name=",,; s,".*<sha1>, ,; s,</sha1>.*$,,' | grep -Pv "^${escapedIdentifier}"'(\.cdx\.(gz|idx)|_meta\.(sqlite|xml)) ' | sed 's,&amp;,\&,g' | awk '{ print $NF "  " substr($0, 1, length($0) - length($NF) - 1) }')

localFiles=()
while IFS= read -r -d $'\0' f; do localFiles+=("${f:2}"); done < <(find . -type f -print0)
readarray -t iaFiles < <(printf "%s\n" "${iasha1sums[@]}" | sed 's,^.\{40\}  ,,')
readarray -t localFilesSorted < <(printf "%s\n" "${localFiles[@]}" | sort)
readarray -t iaFilesSorted < <(printf "%s\n" "${iaFiles[@]}" | sort)
readarray -t localMissing < <(comm -13 <(printf "%s\n" "${localFilesSorted[@]}") <(printf "%s\n" "${iaFilesSorted[@]}"))
readarray -t iaMissing < <(comm -23 <(printf "%s\n" "${localFilesSorted[@]}") <(printf "%s\n" "${iaFilesSorted[@]}"))

status=0

RED=$'\x1b[1;31m'
GREEN=$'\x1b[0;32m'
RESET=$'\x1b[0m'

if [[ ${#localMissing[@]} -eq 0 && ${#iaMissing[@]} -eq 0 ]]
then
	echo "File list comparison: ${GREEN}OK${RESET}"
else
	echo "File list comparison: ${RED}FAIL${RESET}"
fi
if [[ ${#iaMissing[@]} -gt 0 ]]
then
	echo "Local files that are not in the IA item:"
	printf "  %s\n" "${iaMissing[@]}"
	status=1
fi
if [[ ${#localMissing[@]} -gt 0 ]]
then
	echo "IA item files that are not in the local directory:"
	printf "  %s\n" "${localMissing[@]}"
	status=1
fi

echo

echo "SHA-1 comparison:"
sha1sum -c < <(printf "%s\n" "${iasha1sums[@]}") > >(perl -pe 's,^(.*)(?<!: OK)$,'"${RED}"'\1'"${RESET}"',; s,: OK$,: '"${GREEN}OK${RESET}"',; s,^,  ,') 2>&1
shast=$?
wait  # Wait for Perl to finish; Bash 4.4+ only
if [[ ${shast} -eq 0 ]]
then
	echo "SHA-1 comparison: ${GREEN}OK${RESET}"
else
	echo "SHA-1 comparison: ${RED}FAIL${RESET}"
	status=1
fi

echo

if [[ ${status} -eq 0 ]]
then
	echo "${GREEN}OK${RESET}"
else
	echo "${RED}FAIL${RESET}"
fi
exit ${status}