podman/hack/man-page-checker

#!/usr/bin/env bash
#
# man-page-checker - validate and cross-reference man page names
#

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


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

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

rc=0

for md in *.1.md;do
    # Read the first line after '# NAME' (or '## NAME'). (FIXME: # and ##
    #               are not the same; should we stick to one convention?)
    # There may be more than one name, e.g. podman-info.1.md has
    # podman-system-info then another line with podman-info. We
    # care only about the first.
    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 podman-foo.1.md matches the one
# in the table in podman.1.md. podman-remote is not a podman subcommand,
# so it is excluded here.
for md in $(ls -1 *-*.1.md | grep -v remote);do
    desc=$(egrep -A1 '^#* NAME' $md|tail -1|sed -e 's/^podman[^ ]\+ - //')

    # podman.1.md has a two-column table; podman-*.1.md all have three.
    parent=$(echo $md | sed -e 's/^\(.*\)-.*$/\1.1.md/')
    if [[ $parent =~ "podman-auto" ]]; then
        # podman-auto-update.1.md is special cased as it's structure differs
        # from that of other man pages where main and sub-commands split by
        # dashes.
        parent="podman.1.md"
    fi
    x=3
    if expr -- "$parent" : ".*-" >/dev/null; then
        x=4
    fi

    # Find the descriptive text in the parent man page.
    # Strip off the final period; let's not warn about such minutia.
    parent_desc=$(grep $md $parent | awk -F'|' "{print \$$x}" | sed -e 's/^ \+//' -e 's/ \+$//' -e 's/\.$//')

    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"

    # Sometimes in CI we run before podman gets built.
    test -x ../../../bin/podman || return

    # 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=$(sed -e "s/\*\*$cmd\*\*[[:space:]]*//" <<<"$from_man")
    from_help=$(sed -e "s/^[[:space:]]*$cmd[[:space:]]*//" <<<"$from_help")

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

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

    # 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 's/\*\([a-z-]\+\)\*/\U\1/g' <<<"$from_man")

    # FIXME: one of the common patterns is for --help to show 'POD [POD...]'
    # but man page show 'pod ...'. This conversion may help one day, but
    # not yet: there are too many inconsistencies such as '[pod ...]'
    # (brackets) and 'pod...' (no space between).
#    from_help=$(sed -e 's/\([A-Z]\+\)[[:space:]]\+\[\1[[:space:]]*\.\.\.\]/\1 .../' <<<"$from_help")

    # Compare man-page and --help usage strings. For now, do so only
    # when run with --verbose.
    if [[ "$from_man" != "$from_help" ]]; then
        if [ -n "$verbose" ]; then
            printf "%-25s man='%s' help='%s'\n" "$cmd:" "$from_man" "$from_help"
            # Yeah, we're not going to enable this as a blocker any time soon.
            # rc=1
        fi
    fi
}

# Pass 3: compare synopses.
#
# Make sure the SYNOPSIS line in podman-foo.1.md reads '**podman foo** ...'
for md in *.1.md;do
    # FIXME: several pages have a multi-line form of SYNOPSIS in which
    #        many or all flags are enumerated. Some of these are trivial
    #        and really should be made into one line (podman-container-exists,
    #        container-prune, others); some are more complicated and I
    #        would still like to see them one-lined (container-runlabel,
    #        image-trust) but I'm not 100% comfortable doing so myself.
    #        To view those:
    #           $ less $(for i in docs/*.1.md;do x=$(grep -A2 '^#* SYNOPSIS' $i|tail -1); if [ -n "$x" ]; then echo $i;fi;done)
    #
    synopsis=$(egrep -A1 '^#* SYNOPSIS' $md|tail -1)

    # Command name must be bracketed by double asterisks; options and
    # arguments are bracketed by single ones.
    #   E.g. '**podman volume inspect** [*options*] *volume*...'
    # Get the command name, and confirm that it matches the md file name.
    cmd=$(echo "$synopsis" | sed -e 's/\(.*\)\*\*.*/\1/' | tr -d \*)
    md_nodash=$(basename "$md" .1.md | tr '-' ' ')
    if [[ $md_nodash = 'podman auto update' ]]; then
        # special case: the command is "auto-update", with a hyphen
        md_nodash='podman auto-update'
    fi
    if [ "$cmd" != "$md_nodash" -a "$cmd" != "podman-remote" ]; 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 'podman 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/podman 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