<?php
/**
 * acb-diagnose-content.php — AI Content Bridge diagnostic
 *
 * AICOBR_CONTENT_DIAGNOSTIC_2026_08
 *
 * Answers one question with facts instead of theory: for each post type on this
 * site, where does the content actually live — post_content, _elementor_data,
 * or both — and would ACB's inbound-link scanner see it?
 *
 * WHY THIS EXISTS
 *
 * ACB's scanner reads post_content and SKIPS any post that has non-empty
 * _elementor_data (it will not write into a layout it cannot safely edit). The
 * open question is what that means on a real Elementor site, where content is
 * often stored in BOTH fields. This measures it directly rather than assuming.
 *
 * WHAT IT DOES NOT DO
 *
 * Nothing destructive. It reads. It writes no posts, changes no options, and
 * inserts no links. Safe to run on production.
 *
 * USAGE
 *
 *   wp eval-file acb-diagnose-content.php
 *   wp eval-file acb-diagnose-content.php --url=https://www.webuyvans.co.uk
 *
 * Optionally narrow to one type:
 *
 *   wp eval-file acb-diagnose-content.php -- van_model
 *
 * (Everything after `--` arrives in $args.)
 */

if (!defined('ABSPATH')) {
    fwrite(STDERR, "This is a WP-CLI script. Run it with: wp eval-file acb-diagnose-content.php\n");
    return;
}

global $wpdb;

$only_type = isset($args[0]) ? sanitize_key($args[0]) : '';

$types = get_post_types(['public' => true], 'names');
unset($types['attachment']);
if ($only_type) {
    $types = in_array($only_type, $types, true) ? [$only_type] : [];
    if (!$types) {
        WP_CLI::error("Post type '{$only_type}' is not a public post type on this site.");
        return;
    }
}

/**
 * A trimmed length that ignores empty-ish content: an Elementor page often has
 * a post_content of a few whitespace characters or a single wrapper, which is
 * not real editable prose. Anything under this many characters is treated as
 * "effectively empty" for the purpose of "can the scanner find text here".
 */
$MEANINGFUL = 40;

WP_CLI::log('');
WP_CLI::log('AI Content Bridge — content storage diagnostic');
WP_CLI::log('Where does each post type keep its content, and can the scanner read it?');
WP_CLI::log(str_repeat('-', 78));

$rows = [];

foreach ($types as $type) {
    $ids = get_posts([
        'post_type'      => $type,
        'post_status'    => 'publish',
        'posts_per_page' => 200,          // a sample is enough to characterise a type
        'fields'         => 'ids',
        'orderby'        => 'date',
        'order'          => 'DESC',
        'suppress_filters' => true,
    ]);

    $n = count($ids);
    if ($n === 0) {
        $rows[] = [$type, 0, '-', '-', '-', '-', 'no published posts'];
        continue;
    }

    $has_content   = 0;   // post_content holds meaningful text
    $has_elementor = 0;   // _elementor_data present and non-empty
    $both          = 0;   // both of the above
    $content_links = 0;   // at least one <a href in post_content
    $would_skip    = 0;   // scanner would skip as a SOURCE (elementor present)

    foreach ($ids as $id) {
        $content   = (string) get_post_field('post_content', $id);
        $elementor = (string) get_post_meta($id, '_elementor_data', true);

        $c = strlen(trim($content)) >= $MEANINGFUL;
        $e = ($elementor !== '' && $elementor !== '[]' && $elementor !== 'null');

        if ($c) $has_content++;
        if ($e) $has_elementor++;
        if ($c && $e) $both++;
        if (stripos($content, '<a ') !== false && stripos($content, 'href') !== false) $content_links++;

        // Mirrors is_page_builder(): _elementor_data is the trigger. If it is
        // present, ACB will not insert INTO this post as a link source.
        if ($e) $would_skip++;
    }

    $pct = fn($k) => $n ? round(($k / $n) * 100) : 0;

    $rows[] = [
        $type,
        $n,
        $pct($has_content) . '%',
        $pct($has_elementor) . '%',
        $pct($both) . '%',
        $pct($content_links) . '%',
        $pct($would_skip) . '% skipped',
    ];
}

if (class_exists('WP_CLI\\Utils\\format_items') || function_exists('WP_CLI\\Utils\\format_items')) {
    // preferred: a real table
    \WP_CLI\Utils\format_items(
        'table',
        array_map(function ($r) {
            return [
                'post type'       => $r[0],
                'sample'          => $r[1],
                'text in content' => $r[2],
                'has elementor'   => $r[3],
                'both fields'     => $r[4],
                'links in content'=> $r[5],
                'as a source'     => $r[6],
            ];
        }, $rows),
        ['post type', 'sample', 'text in content', 'has elementor', 'both fields', 'links in content', 'as a source']
    );
} else {
    foreach ($rows as $r) {
        WP_CLI::log(sprintf(
            "%-16s  n=%-4s  content:%-5s  elementor:%-5s  both:%-5s  links:%-5s  %s",
            $r[0], $r[1], $r[2], $r[3], $r[4], $r[5], $r[6]
        ));
    }
}

WP_CLI::log(str_repeat('-', 78));
WP_CLI::log('How to read this:');
WP_CLI::log('  text in content  — % of posts whose post_content holds real text.');
WP_CLI::log('                     This is the field ACB reads. High = ACB sees the content.');
WP_CLI::log('  has elementor    — % with _elementor_data. The front end renders from this.');
WP_CLI::log('  both fields      — % that keep text in BOTH. For these, ACB can read and');
WP_CLI::log('                     insert into post_content even though Elementor displays');
WP_CLI::log('                     its own copy. This is the case that explains "it works".');
WP_CLI::log('  as a source      — % ACB would SKIP when looking for pages to insert a link');
WP_CLI::log('                     INTO, because _elementor_data is present. Note a post can');
WP_CLI::log('                     be skipped as a SOURCE yet still work perfectly as a');
WP_CLI::log('                     TARGET (something else linking to it).');
WP_CLI::log('');
WP_CLI::log('The key comparison: if "text in content" is high but "as a source" also shows');
WP_CLI::log('most posts skipped, then ACB can SEE your content for counting and targeting');
WP_CLI::log('but is being conservative about writing INTO it. That is a setting worth');
WP_CLI::log('revisiting, not a hard limit.');
WP_CLI::log('');