Interactive recursive diff with extra hacks for some pipeline specifics

diff_runs

#!/usr/bin/env bash

# works much better if runs have the same name on the same machine (i.e. rename first directory after run)
# much easier to view diffs with colordiff installed

# some files will differ only by dates/times/run times/references to unique job names, just gotta check 'em

# we ignore whitespace in contents
# we try to match filenames that have SGE pid/job names (but not in contents)
# some files will get processsed out-of-order and fail this matching
# maybe do something with find before diff to get a defined name-based ordering?


RUN_ROOT=/data2/processed
saved=()

function main() {
    local runa=$1 && shift
    local runb=$1 && shift

    if [[ ! -d "${RUN_ROOT}/$runa" || ! -d "${RUN_ROOT}/$runb" ]]
    then
        echo supply two valid run names
        exit 1
    else
        compare "${RUN_ROOT}/$runa" "${RUN_ROOT}/$runb"
    fi
}

function compare() {
    local dira=$1 && shift
    local dirb=$1 && shift

    local diff=$(diff -r --brief "$dira" "$dirb")
    local prev_line
    IFS=$'\n'
    for line in $diff
    do
        # echo
        # echo LINE
        # echo $line
        # echo LINE
        # echo
        if [[ ! "$line" =~ ^Only ]]
        then
            if [[ "$prev_line" =~ ^Only ]]
            then
                # previous line was unpaired Only, check it
                check_line "$dira" "$dirb" "$prev_line" "$prev_prev_line" || break
            fi
            check_line "$dira" "$dirb" "$line" "$prev_line" || break
        elif [[ -n "$prev_line" && "$prev_line" =~ ^Only ]]
        then
            local filea=$(echo "$prev_line" | awk '{ print gensub("(.*):", "\\1/", "", $3) $4 }')
            local fileb=$(echo "$line" | awk '{ print gensub("(.*):", "\\1/", "", $3) $4 }')
            if pair "$dira" "$dirb" "$filea" "$fileb"
            then
                line=$(diff --brief "$filea" "$fileb")
                if [[ -n "$line" ]]
                then
                    prev_line=$prev_prev_line
                    check_line "$dira" "$dirb" "$line" "$prev_line" || break
                else
                    # echo pair no differ
                    prev_line=$prev_prev_line
                    continue
                fi
            else
                # echo no pair
                # Two Only lines but they do not match, danger here!
                check_line "$dira" "$dirb" "$prev_line" "$prev_prev_line" || break;
            fi
        else
            # echo no prev or prev not Only $prev_line
            # we have no previous line, or an Only line but previous was not, do nothing & update previous
            :
        fi
        prev_prev_line=$prev_line
        prev_line=$line
    done

    echo saved:
    echo
    for item in "${saved[@]}"
    do
        echo "$item"
    done
}

function pair() {
    # the horror...
    local dira=$1 && shift
    local dirb=$1 && shift
    local filea=$1 && shift
    local fileb=$1 && shift

    filea=$(pretty "$dira" "$dirb" "$filea")
    fileb=$(pretty "$dira" "$dirb" "$fileb")
    if [[ "$(dirname "$filea")" == "$(dirname "$filea")" && "${filea%_*}" == "${fileb%_*}" ]]
    then
        local exta="${filea##*.}"
        local extb="${fileb##*.}"
        [[ "$exta" == "$extb" ]] && return 0
        if [[ "$(basename "$(dirname "$filea")")" == "logs" ]]
        then
            [[ "${exta:0:1}" == "o" && "${exta:0:1}" == "${extb:0:1}" ]] && return 0
            [[ "${exta:0:1}" == "e" && "${exta:0:1}" == "${extb:0:1}" ]] && return 0
            [[ "${exta:0:2}" == "po" && "${exta:0:2}" == "${extb:0:2}" ]] && return 0
            [[ "${exta:0:2}" == "pe" && "${exta:0:2}" == "${extb:0:2}" ]] && return 0
        fi
    fi
    return 1
}

function pretty() {
    local dira=$1 && shift
    local dirb=$1 && shift
    local line=$1 && shift

    local pretty
    pretty=${line//$dira\//}
    pretty=${pretty//$dirb\//}
    echo "$pretty"
}

function check_line() {
    local dira=$1 && shift
    local dirb=$1 && shift
    local line=$1 && shift
    local prev_line=$1 && shift

    pretty "$dira" "$dirb" "$line"
    echo
    while true
    do
        local initial
        [[ "$line" =~ differ  ]] && initial="(d)iff/"
        read -r -n1 -p "${initial}(a)dd/(i)gnore/${prev_line:+(b)ack/}(q)uit? " action
        case $action in
            d)
                if [ -z "$initial" ]
                then
                    echo
                    echo "Cannot diff one file"
                else
                    local filea=$(echo "$line" | awk '{ print $2 }')
                    local fileb=$(echo "$line" | awk '{ print $4 }')
                    local formatter=cat
                    which colordiff >/dev/null 2>&1 && formatter=colordiff
                    diff -wu "$filea" "$fileb" | $formatter | less -FR
                fi
                ;;
            a)
                saved+=( $line )
                break
                ;;
            i)
                break
                ;;
            b)
                if [ -z "$prev_line" ]
                then
                    echo
                    echo "Cannot go back further"
                else
                    echo
                    check_line "$dira" "$dirb" "$prev_line" || return 1
                    check_line "$dira" "$dirb" "$line" "$prev_line" || return 1
                    break
                fi
                ;;
            q)
                echo
                return 1
                ;;
            *)
                echo
                echo "Not a valid action"
                echo
                echo LINE IS "$line"
                echo

                ;;
        esac
    done
    echo
    return 0
}

main "$@"