doc-check.sh

#!/bin/bash

# Example use:
# ./doc-check.sh https://docs.redhat.com/en/documentation/red_hat_satellite/6.18 html-single

DOC_URL="$1"
VARIANT="$2"

# Example URL: https://docs.redhat.com/en/documentation/red_hat_satellite/6.18/\$VARIANT/administering_red_hat_satellite/index#Using_the_Project_Content_Dashboard_admin
# Guide: administering_red_hat_satellite
# Section: index
# Anchor: Using_the_Project_Content_Dashboard_admin

# Assumptions:
# - if VARIANT is html-single, the section does not matter
# - if the guide is wrong, docs site will return 404 which will count as a a failure
# - if the anchor is wrong, there won't be any element on the page which would have it as it as an id attribute

# DOC_BASE=https://docs.redhat.com/en/documentation/red_hat_satellite/6.18/$VARIANT
DOC_BASE=${DOC_URL}/$VARIANT

report_failure() {
    kind="$1"
    element="$2"
    path="$3"

    echo
    echo "Failed to locate $kind '$element' in $path"
}

TOTAL=0
SUCCESS=0
FAILURE=0

while read DOC_PATH; do
    TMPFILE="$(mktemp)"
    if curl -L "${DOC_BASE}${DOC_PATH}" -q --fail -s >"$TMPFILE"; then
        if echo "$DOC_PATH" | grep -q '#'  ; then
            ANCHOR="$(echo "$DOC_PATH" | cut -d '#' -f 2)"
            if grep -q "id=\"$ANCHOR\"" "$TMPFILE"; then
                echo -n '.'
                SUCCESS=$((SUCCESS + 1))
            else
                report_failure anchor "$ANCHOR" "${DOC_BASE}${DOC_PATH}"
                FAILURE=$((FAILURE + 1))
            fi
        else
            echo -n '.'
            SUCCESS=$((SUCCESS + 1))
        fi
    else
        report_failure guide "$(echo "$DOC_PATH" | awk -F / '{ print $2 "/" $3 }')" "${DOC_BASE}${DOC_PATH}"
        FAILURE=$((FAILURE + 1))
    fi
    TOTAL=$((TOTAL + 1))
    rm "$TMPFILE"
done < <(grep -Po '#{ForemanThemeSatellite.documentation_root}[^"]+' lib/foreman_theme_satellite/documentation.rb | sed 's|#{ForemanThemeSatellite.documentation_root}||')

echo
echo "Checked $TOTAL links, $SUCCESS ok, $FAILURE broken"