skopeo/hack/man-page-checker

#!/usr/bin/env bash
#
# man-page-checker - validate and cross-reference man page names
#
# This is the script that cross-checks BETWEEN MAN PAGES. It is not the
# script that cross-checks that each option in skopeo foo --help is listed
# in skopeo-foo.1.md and vice-versa; that one is xref-helpmsgs-manpages.
#

verbose=
for i; do
    case "$i" in
        -v|--verbose)   verbose=verbose ;;
    esac
done


die() {
    echo "$(basename $0): $*" >&2
    exit 1
}

cd $(dirname $0)/../docs || die "Please run me from top-level skopeo dir"

rc=0

# Pass 1: cross-check file names with NAME section
#
# for a given skopeo-foo.1.md, the NAME should be 'skopeo-foo'
for md in *.1.md;do
    # Read the first line after '## NAME'
    name=$(egrep -A1 '^## NAME' $md|tail -1|awk '{print $1}' | tr -d \\\\)

    expect=$(basename $md .1.md)
    if [ "$name" != "$expect" ]; then
        echo
        printf "Inconsistent program NAME in %s:\n" $md
        printf "  NAME= %s  (expected: %s)\n" $name $expect
        rc=1
    fi
done

# Pass 2: compare descriptions.
#
# Make sure the descriptive text in skopeo-foo.1.md matches the one
# in the table in skopeo.1.md.
for md in $(ls -1 *-*.1.md);do
    desc=$(egrep -A1 '^## NAME' $md|tail -1|sed -E -e 's/^skopeo[^[:space:]]+ - //')

    # Find the descriptive text in the main skopeo man page.
    parent=skopeo.1.md
    parent_desc=$(grep $md $parent | awk -F'|' '{print $3}' | sed -E -e 's/^[[:space:]]+//' -e 's/[[:space:]]+$//')

    if [ "$desc" != "$parent_desc" ]; then
        echo
        printf "Inconsistent subcommand descriptions:\n"
        printf "  %-32s = '%s'\n" $md "$desc"
        printf "  %-32s = '%s'\n" $parent "$parent_desc"
        printf "Please ensure that the NAME section of $md\n"
        printf "matches the subcommand description in $parent\n"
        rc=1
    fi
done

# Helper function: compares man page synopsis vs --help usage message
function compare_usage() {
    local cmd="$1"
    local from_man="$2"

    # Run 'cmd --help', grab the line immediately after 'Usage:'
    local help_output=$(../bin/$cmd --help)
    local from_help=$(echo "$help_output" | grep -A1 '^Usage:' | tail -1)

    # strip off command name from both
    from_man=$(sed -E -e "s/\*\*$cmd\*\*[[:space:]]*//" <<<"$from_man")
    from_help=$(sed -E -e "s/^[[:space:]]*$cmd[[:space:]]*//" <<<"$from_help")

    # man page lists 'foo [*options*]', help msg shows 'foo [command options]'.
    # Make sure if one has it, the other does too.
    if expr "$from_man" : "\[\*options\*\]" >/dev/null; then
        if expr "$from_help" : "\[command options\]" >/dev/null; then
            :
        else
            echo "WARNING: $cmd: man page shows '[*options*]', help does not show [command options]"
            rc=1
       fi
    elif expr "$from_help" : "\[command options\]" >/dev/null; then
        echo "WARNING: $cmd: --help shows [command options], man page does not show [*options*]"
        rc=1
    fi

    # Strip off options and flags; start comparing arguments
    from_man=$(sed  -E -e 's/^\[\*options\*\][[:space:]]*//' <<<"$from_man")
    from_help=$(sed -E -e 's/^\[command options\][[:space:]]*//'      <<<"$from_help")

    # Constant strings in man page are '**foo**', in --help are 'foo'.
    from_man=$(sed -E -e 's/\*\*([^*]+)\*\*/\1/g' <<<"$from_man")

    # Args in man page are '_foo_', in --help are 'FOO'. Convert all to
    # UPCASE simply because it stands out better to the eye.
    from_man=$(sed -E -e 's/_([a-z-]+)_/\U\1/g' <<<"$from_man")

    # Compare man-page and --help usage strings. Skip 'skopeo' itself,
    # because the man page includes '[global options]' which we don't grok.
    if [[ "$from_man" != "$from_help" && "$cmd" != "skopeo" ]]; then
        printf "%-25s man='%s' help='%s'\n" "$cmd:" "$from_man" "$from_help"
        rc=1
    fi
}

# Pass 3: compare synopses.
#
# Make sure the SYNOPSIS line in skopeo-foo.1.md reads '**skopeo foo** ...'
for md in *.1.md;do
    synopsis=$(egrep -A1 '^#* SYNOPSIS' $md|tail -1)

    # Command name must be bracketed by double asterisks; options and
    # arguments are bracketed by single ones.
    #   E.g. '**skopeo copy** [*options*] _..._'
    # Get the command name, and confirm that it matches the md file name.
    cmd=$(echo "$synopsis" | sed -E -e 's/^\*\*([^*]+)\*\*.*/\1/' | tr -d \*)
    # Use sed, not tr, so we only replace the first dash: we want
    # skopeo-list-tags -> "skopeo list-tags", not "skopeo list tags"
    md_nodash=$(basename "$md" .1.md | sed -e 's/-/ /')
    if [ "$cmd" != "$md_nodash" ]; then
        echo
        printf "Inconsistent program name in SYNOPSIS in %s:\n" $md
        printf "  SYNOPSIS = %s (expected: '%s')\n" "$cmd" "$md_nodash"
        rc=1
    fi

    # The convention is to use UPPER CASE in 'skopeo foo --help',
    # but *lower case bracketed by asterisks* in the man page
    if expr "$synopsis" : ".*[A-Z]" >/dev/null; then
        echo
        printf "Inconsistent capitalization in SYNOPSIS in %s\n" $md
        printf "  '%s' should not contain upper-case characters\n" "$synopsis"
        rc=1
    fi

    # (for debugging, and getting a sense of standard conventions)
    #printf "  %-32s ------ '%s'\n" $md "$synopsis"

    # If bin/skopeo is available, run "cmd --help" and compare Usage
    # messages. This is complicated, so do it in a helper function.
    compare_usage "$md_nodash" "$synopsis"
done


exit $rc