#!/usr/bin/env bash

#########################################################################################
#### seo-exporter.sh — Pull Google Search Console metrics into Prometheus            ####
#### Queries GSC API for clicks, impressions, CTR, position, top pages/queries       ####
#### Pushes Prometheus-format metrics to Pushgateway — runs on cron, no daemon       ####
####                                                                                 ####
#### Author: Phil Connor                                                             ####
#### Contact: contact@mylinux.work                                                   ####
#### License: MIT                                                                    ####
#### Version 1.00                                                                    ####
####                                                                                 ####
#### Usage:                                                                          ####
####   ./seo-exporter.sh                                                             ####
####   ./seo-exporter.sh --site https://mylinux.work/                                ####
####                                                                                 ####
#### See --help for all options.                                                     ####
#########################################################################################

set -euo pipefail

# ── Defaults ──────────────────────────────────────────────────────────
CREDENTIALS_FILE="${CREDENTIALS_FILE:-/etc/seo-exporter/credentials.json}"
SITE_URL="${SITE_URL:-https://mylinux.work/}"
PUSHGATEWAY_URL="${PUSHGATEWAY_URL:-http://localhost:9091}"
JOB_NAME="${JOB_NAME:-seo_exporter}"
DAYS_BACK="${DAYS_BACK:-7}"
TOP_N="${TOP_N:-50}"
VERBOSE="${VERBOSE:-false}"
COLOR="${COLOR:-auto}"

# ── State ─────────────────────────────────────────────────────────────
SCRIPT_NAME="$(basename "$0")"
readonly SCRIPT_NAME
ACCESS_TOKEN=""
METRICS=""

# ── Colors ────────────────────────────────────────────────────────────
setup_colors() {
    if [[ "$COLOR" == "never" ]]; then
        BLUE="" RED="" DIM="" RESET=""
        return
    fi
    if [[ "$COLOR" == "always" ]] || [[ -t 1 ]]; then
        BLUE='\033[0;34m'
        RED='\033[0;31m'
        DIM='\033[2m'
        RESET='\033[0m'
    else
        BLUE="" RED="" DIM="" RESET=""
    fi
}

# ── Logging ───────────────────────────────────────────────────────────
log()     { echo -e "${BLUE}[INFO]${RESET} $*"; }
err()     { echo -e "${RED}[ERROR]${RESET} $*" >&2; }
verbose() { if [[ "$VERBOSE" == "true" ]]; then echo -e "${DIM}[DEBUG]${RESET} $*"; fi; }

# ══════════════════════════════════════════════════════════════════════
#                       JWT / OAUTH2
# ══════════════════════════════════════════════════════════════════════

base64url() {
    openssl base64 -A | tr '+/' '-_' | tr -d '='
}

get_access_token() {
    local creds="$CREDENTIALS_FILE"

    if [[ ! -f "$creds" ]]; then
        err "Credentials file not found: $creds"
        exit 1
    fi

    local client_email token_uri private_key
    client_email=$(jq -r '.client_email' "$creds")
    token_uri=$(jq -r '.token_uri' "$creds")
    private_key=$(jq -r '.private_key' "$creds")

    local now exp
    now=$(date +%s)
    exp=$((now + 3600))

    # Build JWT header and claims
    local header claims
    header=$(printf '{"alg":"RS256","typ":"JWT"}' | base64url)
    claims=$(printf '{"iss":"%s","scope":"https://www.googleapis.com/auth/webmasters.readonly","aud":"%s","iat":%d,"exp":%d}' \
        "$client_email" "$token_uri" "$now" "$exp" | base64url)

    # Sign with private key
    local signature
    signature=$(printf '%s.%s' "$header" "$claims" \
        | openssl dgst -sha256 -sign <(printf '%s' "$private_key") \
        | base64url)

    local jwt="${header}.${claims}.${signature}"

    # Exchange JWT for access token
    local response
    response=$(curl -s -X POST "$token_uri" \
        -H "Content-Type: application/x-www-form-urlencoded" \
        -d "grant_type=urn:ietf:params:oauth:grant-type:jwt-bearer&assertion=${jwt}")

    ACCESS_TOKEN=$(echo "$response" | jq -r '.access_token // empty')

    if [[ -z "$ACCESS_TOKEN" ]]; then
        err "Failed to get access token"
        err "Response: $response"
        exit 1
    fi

    verbose "Access token acquired (${#ACCESS_TOKEN} chars)"
}

# ══════════════════════════════════════════════════════════════════════
#                       GSC API QUERIES
# ══════════════════════════════════════════════════════════════════════

gsc_query() {
    local body="$1"
    local encoded_site
    encoded_site=$(printf '%s' "$SITE_URL" | jq -sRr @uri)

    curl -s \
        -H "Authorization: Bearer ${ACCESS_TOKEN}" \
        -H "Content-Type: application/json" \
        -d "$body" \
        "https://searchconsole.googleapis.com/webmasters/v3/sites/${encoded_site}/searchAnalytics/query"
}

query_aggregate() {
    local end_date start_date
    end_date=$(date -d "-3 days" +%Y-%m-%d)
    start_date=$(date -d "-$((3 + DAYS_BACK)) days" +%Y-%m-%d)

    local body
    body=$(jq -n \
        --arg start "$start_date" \
        --arg end "$end_date" \
        '{"startDate": $start, "endDate": $end}')

    gsc_query "$body"
}

query_dimension() {
    local dimension="$1"
    local limit="${2:-$TOP_N}"
    local end_date start_date
    end_date=$(date -d "-3 days" +%Y-%m-%d)
    start_date=$(date -d "-$((3 + DAYS_BACK)) days" +%Y-%m-%d)

    local body
    body=$(jq -n \
        --arg start "$start_date" \
        --arg end "$end_date" \
        --arg dim "$dimension" \
        --argjson limit "$limit" \
        '{"startDate": $start, "endDate": $end, "dimensions": [$dim], "rowLimit": $limit}')

    gsc_query "$body"
}

# ══════════════════════════════════════════════════════════════════════
#                       METRICS BUILDER
# ══════════════════════════════════════════════════════════════════════

add_metric() {
    local name="$1" type="$2" help="$3" value="$4" labels="${5:-}"
    if [[ -z "$labels" ]]; then
        METRICS+="# HELP ${name} ${help}
# TYPE ${name} ${type}
${name} ${value}
"
    else
        # Only add HELP/TYPE if not already present
        if [[ "$METRICS" != *"# TYPE ${name}"* ]]; then
            METRICS+="# HELP ${name} ${help}
# TYPE ${name} ${type}
"
        fi
        METRICS+="${name}{${labels}} ${value}
"
    fi
}

# ══════════════════════════════════════════════════════════════════════
#                       COLLECT METRICS
# ══════════════════════════════════════════════════════════════════════

collect_metrics() {
    METRICS=""

    # Aggregate metrics
    log "Fetching aggregate metrics..."
    local agg
    agg=$(query_aggregate)

    local clicks impressions ctr position
    clicks=$(echo "$agg" | jq '.rows[0].clicks // 0')
    impressions=$(echo "$agg" | jq '.rows[0].impressions // 0')
    ctr=$(echo "$agg" | jq '(.rows[0].ctr // 0) * 100 | . * 100 | round / 100')
    position=$(echo "$agg" | jq '(.rows[0].position // 0) * 10 | round / 10')

    add_metric "seo_clicks_total" "gauge" "Total clicks" "$clicks"
    add_metric "seo_impressions_total" "gauge" "Total impressions" "$impressions"
    add_metric "seo_ctr_average" "gauge" "Average CTR percent" "$ctr"
    add_metric "seo_position_average" "gauge" "Average position" "$position"

    log "clicks=${clicks} impressions=${impressions} ctr=${ctr}% position=${position}"

    # Per-page metrics (top N)
    log "Fetching per-page metrics (top ${TOP_N})..."
    local pages
    pages=$(query_dimension "page")

    while IFS= read -r row; do
        [[ -z "$row" ]] && continue
        local page pg_clicks pg_imp pg_pos pg_ctr
        page=$(echo "$row" | jq -r '.keys[0]' | sed "s|${SITE_URL}|/|")
        pg_clicks=$(echo "$row" | jq '.clicks // 0')
        pg_imp=$(echo "$row" | jq '.impressions // 0')
        pg_pos=$(echo "$row" | jq '(.position // 0) * 10 | round / 10')
        pg_ctr=$(echo "$row" | jq '(.ctr // 0) * 100 | . * 100 | round / 100')

        add_metric "seo_page_clicks" "gauge" "Clicks per page" "$pg_clicks" "page=\"${page}\""
        add_metric "seo_page_impressions" "gauge" "Impressions per page" "$pg_imp" "page=\"${page}\""
        add_metric "seo_page_position" "gauge" "Position per page" "$pg_pos" "page=\"${page}\""
        add_metric "seo_page_ctr" "gauge" "CTR per page percent" "$pg_ctr" "page=\"${page}\""
    done < <(echo "$pages" | jq -c '.rows[]? // empty')

    # Per-query metrics (top N)
    log "Fetching per-query metrics (top ${TOP_N})..."
    local queries
    queries=$(query_dimension "query")

    while IFS= read -r row; do
        [[ -z "$row" ]] && continue
        local query q_clicks q_imp q_pos
        query=$(echo "$row" | jq -r '.keys[0]' | sed 's/"/\\"/g')
        q_clicks=$(echo "$row" | jq '.clicks // 0')
        q_imp=$(echo "$row" | jq '.impressions // 0')
        q_pos=$(echo "$row" | jq '(.position // 0) * 10 | round / 10')

        add_metric "seo_query_clicks" "gauge" "Clicks per query" "$q_clicks" "query=\"${query}\""
        add_metric "seo_query_impressions" "gauge" "Impressions per query" "$q_imp" "query=\"${query}\""
        add_metric "seo_query_position" "gauge" "Position per query" "$q_pos" "query=\"${query}\""
    done < <(echo "$queries" | jq -c '.rows[]? // empty')
}

# ══════════════════════════════════════════════════════════════════════
#                       PUSH TO PUSHGATEWAY
# ══════════════════════════════════════════════════════════════════════

push_metrics() {
    log "Pushing metrics to ${PUSHGATEWAY_URL}..."

    local http_code
    http_code=$(printf '%s' "$METRICS" | curl -s -o /dev/null -w "%{http_code}" \
        --data-binary @- \
        "${PUSHGATEWAY_URL}/metrics/job/${JOB_NAME}")

    if [[ "$http_code" == "200" ]] || [[ "$http_code" == "202" ]]; then
        log "Push successful (HTTP ${http_code})"
    else
        err "Push failed (HTTP ${http_code})"
        exit 1
    fi
}

# ══════════════════════════════════════════════════════════════════════
#                       USAGE
# ══════════════════════════════════════════════════════════════════════

usage() {
    cat <<EOF
${SCRIPT_NAME} — Pull Google Search Console metrics into Prometheus

USAGE:
  ${SCRIPT_NAME} [OPTIONS]

OPTIONS:
  --site URL            Site URL in GSC (default: ${SITE_URL})
  --credentials FILE    Path to service account JSON (default: ${CREDENTIALS_FILE})
  --pushgateway URL     Pushgateway URL (default: ${PUSHGATEWAY_URL})
  --days N              Days of data to fetch (default: ${DAYS_BACK})
  --top N               Top N pages/queries to track (default: ${TOP_N})
  --verbose             Enable debug output
  --no-color            Disable colored output
  --help                Show this help

ENVIRONMENT VARIABLES:
  CREDENTIALS_FILE      Service account JSON path
  SITE_URL              Site URL in GSC
  PUSHGATEWAY_URL       Pushgateway URL
  DAYS_BACK             Days of data to fetch
  TOP_N                 Number of top pages/queries
EOF
}

# ══════════════════════════════════════════════════════════════════════
#                       ARGUMENT PARSING
# ══════════════════════════════════════════════════════════════════════

parse_args() {
    while [[ $# -gt 0 ]]; do
        case "$1" in
            --site)          SITE_URL="$2"; shift 2 ;;
            --credentials)   CREDENTIALS_FILE="$2"; shift 2 ;;
            --pushgateway)   PUSHGATEWAY_URL="$2"; shift 2 ;;
            --days)          DAYS_BACK="$2"; shift 2 ;;
            --top)           TOP_N="$2"; shift 2 ;;
            --verbose)       VERBOSE="true"; shift ;;
            --no-color)      COLOR="never"; shift ;;
            --help|-h)       setup_colors; usage; exit 0 ;;
            *)
                err "Unknown option: $1"
                echo "Run ${SCRIPT_NAME} --help for usage" >&2
                exit 1 ;;
        esac
    done
}

# ══════════════════════════════════════════════════════════════════════
#                       MAIN
# ══════════════════════════════════════════════════════════════════════

main() {
    parse_args "$@"
    setup_colors

    log "SEO Exporter — $(date -u +%Y-%m-%dT%H:%M:%SZ)"
    log "Site: ${SITE_URL}"

    get_access_token
    collect_metrics
    push_metrics

    log "Done."
}

main "$@"