reverse-proxy/lib/parse-sites.awk

# =============================================================================
# parse-sites.awk
# -----------------------------------------------------------------------------
# Liest config/sites.yaml (fester, dokumentierter Formatstil - siehe Kopf der
# sites.yaml) und gibt eine einfache, tab-getrennte Zwischenform aus, die
# generate.sh mit reinem Bash weiterverarbeitet:
#
#   GLOBAL<TAB>acme_email
#   DOMAIN<TAB>name<TAB>backend<TAB>backend_scheme<TAB>backend_tls_insecure<TAB>tls_mode<TAB>allowed_ips(komma-getrennt)
#   TCP<TAB>name<TAB>listen_port<TAB>backend<TAB>allowed_ips(komma-getrennt)
#
# Bewusst ohne gawk-spezifische Features (kein 3-arg match(), kein gensub())
# geschrieben, damit es auch mit mawk oder busybox awk (Alpine) läuft.
# =============================================================================

function trim(s) {
    sub(/^[ \t\r]+/, "", s)
    sub(/[ \t\r]+$/, "", s)
    return s
}

function stripq(s) {
    gsub(/"/, "", s)
    return s
}

function flush_domain() {
    if (in_domain) {
        ips = ""
        for (i = 1; i <= n_ips; i++) { ips = ips (i > 1 ? "," : "") ips_arr[i] }
        printf "DOMAIN\t%s\t%s\t%s\t%s\t%s\t%s\n", d_name, d_backend, d_scheme, d_tls_insecure, d_tls_mode, ips
        in_domain = 0
    }
}

function flush_tcp() {
    if (in_tcp) {
        ips = ""
        for (i = 1; i <= n_ips; i++) { ips = ips (i > 1 ? "," : "") ips_arr[i] }
        printf "TCP\t%s\t%s\t%s\t%s\n", t_name, t_port, t_backend, ips
        in_tcp = 0
    }
}

BEGIN {
    section = ""
    in_domain = 0
    in_tcp = 0
    in_allowed_ips = 0
    acme_email = ""
}

{
    line = $0
    sub(/\r$/, "", line)

    # volle Kommentarzeilen und Leerzeilen überspringen
    if (line ~ /^[ \t]*#/ || line ~ /^[ \t]*$/) next

    # Einrückung ermitteln (Anzahl führender Leerzeichen)
    stripped = line
    gsub(/^ */, "", stripped)
    indent = length(line) - length(stripped)
    content = stripped

    # Top-Level-Abschnitt
    if (indent == 0) {
        flush_domain()
        flush_tcp()
        if (content ~ /^global:/) { section = "global" }
        else if (content ~ /^domains:/) { section = "domains" }
        else if (content ~ /^tcp_services:/) { section = "tcp_services" }
        else { section = "" }
        next
    }

    if (section == "global" && indent == 2) {
        if (content ~ /^acme_email:/) {
            val = content
            sub(/^acme_email:[ \t]*/, "", val)
            acme_email = stripq(trim(val))
        }
        next
    }

    if (section == "domains") {
        if (indent == 2 && content ~ /^- name:/) {
            flush_domain()
            in_domain = 1
            in_allowed_ips = 0
            n_ips = 0
            d_backend = ""
            d_scheme = ""
            d_tls_insecure = "false"
            d_tls_mode = "auto"
            val = content
            sub(/^- name:[ \t]*/, "", val)
            d_name = stripq(trim(val))
            next
        }
        if (in_domain && indent == 4) {
            if (content ~ /^backend:/) {
                val = content; sub(/^backend:[ \t]*/, "", val); d_backend = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^backend_scheme:/) {
                val = content; sub(/^backend_scheme:[ \t]*/, "", val); d_scheme = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^backend_tls_insecure:/) {
                val = content; sub(/^backend_tls_insecure:[ \t]*/, "", val); d_tls_insecure = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^tls_mode:/) {
                val = content; sub(/^tls_mode:[ \t]*/, "", val); d_tls_mode = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^allowed_ips:/) { in_allowed_ips = 1; next }
        }
        if (in_domain && in_allowed_ips && indent == 6 && content ~ /^- /) {
            val = content; sub(/^- [ \t]*/, "", val)
            n_ips++; ips_arr[n_ips] = stripq(trim(val))
            next
        }
    }

    if (section == "tcp_services") {
        if (indent == 2 && content ~ /^- name:/) {
            flush_tcp()
            in_tcp = 1
            in_allowed_ips = 0
            n_ips = 0
            t_port = ""
            t_backend = ""
            val = content
            sub(/^- name:[ \t]*/, "", val)
            t_name = stripq(trim(val))
            next
        }
        if (in_tcp && indent == 4) {
            if (content ~ /^listen_port:/) {
                val = content; sub(/^listen_port:[ \t]*/, "", val); t_port = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^backend:/) {
                val = content; sub(/^backend:[ \t]*/, "", val); t_backend = stripq(trim(val))
                in_allowed_ips = 0; next
            }
            if (content ~ /^allowed_ips:/) { in_allowed_ips = 1; next }
        }
        if (in_tcp && in_allowed_ips && indent == 6 && content ~ /^- /) {
            val = content; sub(/^- [ \t]*/, "", val)
            n_ips++; ips_arr[n_ips] = stripq(trim(val))
            next
        }
    }
}

END {
    flush_domain()
    flush_tcp()
    printf "GLOBAL\t%s\n", acme_email
}