<?php
/**
 * Plugin Name: Aliamin Auto Internal Linker
 * Plugin URI:  https://autointernallinker.com/
 * Description: Automatically inserts contextual internal links into post content based on keywords extracted from post titles. Includes a smart title-based mode in the free version.
 * Version:     1.4.5
 * Author:      Eugeniy Aliamin
 * License:     GPL-2.0+
 * Text Domain: aliamin-auto-internal-linker
 */

if ( ! defined( 'ABSPATH' ) ) {
    exit;
}

define( 'LFGAIL_VERSION', '1.4.5' );
define( 'LFGAIL_OPTION_KEY', 'lfgail_settings' );
define( 'LFGAIL_INDEX_KEY', 'lfgail_keyword_index' );
define( 'LFGAIL_LICENSE_KEY', 'lfgail_license' );
define( 'LFGAIL_FREE_VERSION', true );

define( 'LFGAIL_LS_STORE', 'internallinker' );
define( 'LFGAIL_PLAN_LIMITS', [
    'starter'    => 1,
    'small'      => 3,
    'business'   => 5,
    'agency'     => 10,
    'pro agency' => 50,
    'pro_agency' => 50,
] );

function lfgail_is_licensed() { return true; }
function lfgail_get_license() {
    return [ 'key' => '', 'status' => 'active', 'plan' => 'Free', 'site_limit' => 0, 'sites_used' => 0, 'expires' => '', 'checked_at' => 0 ];
}
function lfgail_activate_license( $key )   { return [ 'success' => false, 'message' => 'License activation is not available in the free version.' ]; }
function lfgail_deactivate_license()       { return [ 'success' => false, 'message' => 'Not applicable.' ]; }
function lfgail_plan_limit_from_name( $n ) { return 1; }

function lfgail_default_settings() {
    return [
        'max_links_per_post' => 6,
        'max_links_per_kw'   => 1,
        'link_target'        => '_self',
        'min_keyword_length' => 4,
        'min_words_between'  => 100,
        'skip_first_n_paras' => 1,
        'min_anchor_words'   => 1,
        'max_anchor_words'   => 3,
        'skip_headings'      => 1,
        'extraction_mode'    => 'smart',
        'claude_api_key'     => '',
        'blacklist'          => implode( "\n", [
            'the','and','but','for','nor','yet','so','this','that','these','those','your','their','there',
            'they','them','you','our','its','his','her','with','from','into','over','after','before','about',
            'above','below','under','between','through','during','have','make','know','take','come','look','will',
            'need','want','find','give','tell','keep','like','also','more','when','where','which','what','just',
            'guide','tips','ways','things','people','place','places','time','year','years','world','life','long','days',
            'area','part','parts','side','type','types','kind','site','sites','page','post','article','list',
            'most','best','first','last','every','each','only','some','even','than','then','such','well','right',
            'great','good','real','true','full','complete','ultimate','essential','perfect','top','main',
            'review','reviews','versus','compared','comparison','pros','cons','overview','introduction',
            'everything','something','nothing','anything',
        ] ),
    ];
}

function lfgail_get_settings() {
    $saved = get_option( LFGAIL_OPTION_KEY, [] );
    if ( ! is_array( $saved ) ) {
        $saved = [];
    }
    return wp_parse_args( $saved, lfgail_default_settings() );
}

function lfgail_extract_keywords_smart( $title, $blacklist, $min_len ) {
    $keywords = [];
    $colon_pos = strpos( $title, ':' );
    if ( $colon_pos !== false ) {
        $before = trim( substr( $title, 0, $colon_pos ) );
        $after  = trim( substr( $title, $colon_pos + 1 ) );
    } else {
        $before = $title;
        $after  = '';
    }

    $tokenise = function( $str ) {
        $str = preg_replace( '/[^a-z0-9\s]/i', ' ', $str );
        return array_values( array_filter( preg_split( '/\s+/', strtolower( trim( $str ) ) ) ) );
    };

    $is_ok = function( $word ) use ( $blacklist, $min_len ) {
        return strlen( $word ) >= $min_len && ! in_array( $word, $blacklist, true );
    };

    $bw = $tokenise( $before );
    $bw_clean = array_values( array_filter( $bw, $is_ok ) );

    if ( count( $bw_clean ) >= 2 && count( $bw_clean ) <= 5 ) {
        $keywords[] = implode( ' ', $bw_clean );
    }

    for ( $i = 0; $i < count( $bw ) - 2; $i++ ) {
        $w1 = $bw[ $i ]; $w2 = $bw[ $i + 1 ]; $w3 = $bw[ $i + 2 ];
        if ( $is_ok( $w1 ) && $is_ok( $w2 ) && $is_ok( $w3 ) ) {
            $keywords[] = "$w1 $w2 $w3";
        }
    }

    for ( $i = 0; $i < count( $bw ) - 1; $i++ ) {
        $w1 = $bw[ $i ]; $w2 = $bw[ $i + 1 ];
        if ( $is_ok( $w1 ) && $is_ok( $w2 ) ) {
            $keywords[] = "$w1 $w2";
        }
    }

    foreach ( $bw as $w ) {
        if ( $is_ok( $w ) ) {
            $keywords[] = $w;
        }
    }

    if ( $after !== '' ) {
        $aw = $tokenise( $after );
        for ( $i = 0; $i < count( $aw ) - 2; $i++ ) {
            $w1 = $aw[ $i ]; $w2 = $aw[ $i + 1 ]; $w3 = $aw[ $i + 2 ];
            if ( $is_ok( $w1 ) && $is_ok( $w2 ) && $is_ok( $w3 ) ) {
                $keywords[] = "$w1 $w2 $w3";
            }
        }
        for ( $i = 0; $i < count( $aw ) - 1; $i++ ) {
            $w1 = $aw[ $i ]; $w2 = $aw[ $i + 1 ];
            if ( $is_ok( $w1 ) && $is_ok( $w2 ) ) {
                $keywords[] = "$w1 $w2";
            }
        }
    }

    return array_unique( $keywords );
}

function lfgail_extract_keywords_claude( $posts_data, $api_key ) {
    $titles_json = wp_json_encode(
        array_map( fn( $p ) => [ 'id' => $p['post_id'], 'title' => $p['title'] ], $posts_data ),
        JSON_UNESCAPED_UNICODE
    );

    $prompt = "You are an SEO expert. For each blog post title below, extract 3-6 keyword phrases that would make excellent internal link anchor texts. Rules:\n"
        . "- Include the most specific and meaningful phrases from the title\n"
        . "- Include 1-word, 2-word, and 3-word variations\n"
        . "- Do NOT include stopwords, filler words, or generic adjectives\n"
        . "- Output ONLY valid JSON: an array of objects with \"id\" (number) and \"keywords\" (array of strings)\n"
        . "- All keywords must be lowercase\n\n"
        . "Titles:\n"
        . $titles_json;

    $response = wp_remote_post( 'https://api.anthropic.com/v1/messages', [
        'timeout' => 60,
        'headers' => [
            'Content-Type'      => 'application/json',
            'x-api-key'         => $api_key,
            'anthropic-version' => '2023-06-01',
        ],
        'body' => wp_json_encode( [
            'model'      => 'claude-haiku-4-5-20251001',
            'max_tokens' => 4096,
            'messages'   => [ [ 'role' => 'user', 'content' => $prompt ] ],
        ] ),
    ] );

    if ( is_wp_error( $response ) ) {
        return null;
    }

    $body = json_decode( wp_remote_retrieve_body( $response ), true );
    $text = $body['content'][0]['text'] ?? '';
    $text = preg_replace( '/^```(?:json)?\s*/m', '', $text );
    $text = preg_replace( '/\s*```$/m', '', $text );

    $parsed = json_decode( trim( $text ), true );
    if ( ! is_array( $parsed ) ) {
        return null;
    }

    $result = [];
    foreach ( $posts_data as $p ) {
        $result[ $p['post_id'] ] = $p;
    }
    foreach ( $parsed as $item ) {
        $id = (int) ( $item['id'] ?? 0 );
        if ( isset( $result[ $id ] ) && ! empty( $item['keywords'] ) ) {
            $result[ $id ]['claude_keywords'] = array_map( 'strtolower', (array) $item['keywords'] );
        }
    }

    return $result;
}

function lfgail_build_index( $force = false ) {
    $settings = lfgail_get_settings();

    if ( ! $force ) {
        $cached = get_option( LFGAIL_INDEX_KEY, null );
        if ( $cached !== null ) {
            return $cached;
        }
    }

    $blacklist = lfgail_get_blacklist_array( $settings['blacklist'] );
    $min_len   = (int) $settings['min_keyword_length'];
    $mode      = $settings['extraction_mode'];
    $api_key   = trim( (string) ( $settings['claude_api_key'] ?? '' ) );

    $post_ids = get_posts( [
        'post_type'      => 'post',
        'post_status'    => 'publish',
        'posts_per_page' => -1,
        'fields'         => 'ids',
    ] );

    $posts_data = [];
    foreach ( $post_ids as $id ) {
        $posts_data[] = [
            'post_id' => $id,
            'title'   => get_the_title( $id ),
            'url'     => get_permalink( $id ),
        ];
    }

    $index = [];

    if ( $mode === 'claude' && ! empty( $api_key ) ) {
        $batches = array_chunk( $posts_data, 50 );
        $claude_data = [];
        foreach ( $batches as $batch ) {
            $result = lfgail_extract_keywords_claude( $batch, $api_key );
            if ( $result ) {
                $claude_data = array_merge( $claude_data, $result );
            }
        }

        foreach ( $claude_data as $data ) {
            $keywords = $data['claude_keywords'] ?? lfgail_extract_keywords_smart( $data['title'], $blacklist, $min_len );
            foreach ( $keywords as $kw ) {
                $kw = trim( strtolower( $kw ) );
                if ( $kw !== '' && ! isset( $index[ $kw ] ) ) {
                    $index[ $kw ] = [
                        'post_id' => $data['post_id'],
                        'url'     => $data['url'],
                        'title'   => $data['title'],
                    ];
                }
            }
        }
    } else {
        foreach ( $posts_data as $data ) {
            $keywords = lfgail_extract_keywords_smart( $data['title'], $blacklist, $min_len );
            foreach ( $keywords as $kw ) {
                $kw = trim( strtolower( $kw ) );
                if ( $kw !== '' && ! isset( $index[ $kw ] ) ) {
                    $index[ $kw ] = [
                        'post_id' => $data['post_id'],
                        'url'     => $data['url'],
                        'title'   => $data['title'],
                    ];
                }
            }
        }
    }

    $min_w = max( 1, (int) $settings['min_anchor_words'] );
    $max_w = max( $min_w, (int) $settings['max_anchor_words'] );
    foreach ( array_keys( $index ) as $kw ) {
        $wc = count( explode( ' ', trim( $kw ) ) );
        if ( $wc < $min_w || $wc > $max_w ) {
            unset( $index[ $kw ] );
        }
    }

    uksort( $index, fn( $a, $b ) => strlen( $b ) - strlen( $a ) );
    update_option( LFGAIL_INDEX_KEY, $index, false );
    return $index;
}

function lfgail_get_blacklist_array( $raw ) {
    $lines = preg_split( '/[\r\n,]+/', strtolower( (string) $raw ) );
    return array_values( array_filter( array_map( 'trim', $lines ) ) );
}

add_action( 'save_post', 'lfgail_invalidate_index' );
add_action( 'delete_post', 'lfgail_invalidate_index' );

function lfgail_invalidate_index( $post_id ) {
    if ( wp_is_post_revision( $post_id ) || get_post_type( $post_id ) !== 'post' ) {
        return;
    }
    delete_option( LFGAIL_INDEX_KEY );
}

function lfgail_run_daily_rebuild() {
    lfgail_build_index( true );
}
add_action( 'lfgail_daily_rebuild', 'lfgail_run_daily_rebuild' );
if ( ! wp_next_scheduled( 'lfgail_daily_rebuild' ) ) {
    wp_schedule_event( time(), 'daily', 'lfgail_daily_rebuild' );
}

add_filter( 'the_content', 'lfgail_inject_links', 20 );

function lfgail_inject_links( $content ) {
    if ( ! is_singular( 'post' ) ) {
        return $content;
    }

    $current_id    = get_the_ID();
    $settings      = lfgail_get_settings();
    $index         = lfgail_build_index();

    if ( empty( $index ) ) {
        return $content;
    }

    $target        = $settings['link_target'] === '_blank' ? ' target="_blank" rel="noopener"' : '';
    $max_links     = (int) $settings['max_links_per_post'];
    $max_per_kw    = (int) $settings['max_links_per_kw'];
    $min_words_gap = (int) $settings['min_words_between'];
    $skip_paras    = (int) $settings['skip_first_n_paras'];
    $skip_headings = (bool) $settings['skip_headings'];

    $segments = preg_split( '/(<p[\s>].*?<\/p>)/is', $content, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY );

    $para_index = 0;
    $links_added = 0;
    $used_keywords = [];
    $used_urls = [];
    $words_since_last_link = $min_words_gap + 1;
    $result = [];
    $inside_heading = false;

    foreach ( $segments as $segment ) {
        $is_para = (bool) preg_match( '/^<p[\s>]/i', $segment );

        if ( $is_para ) {
            $para_index++;
            if ( $para_index <= $skip_paras || $links_added >= $max_links ) {
                $words_since_last_link += lfgail_count_words( $segment );
                $result[] = $segment;
                continue;
            }
        }

        $parts = preg_split( '/(<[^>]+>)/s', $segment, -1, PREG_SPLIT_DELIM_CAPTURE );
        $inside_a = false;
        $seg_out = [];

        foreach ( $parts as $part ) {
            if ( preg_match( '/^<[^>]+>$/s', $part ) ) {
                if ( preg_match( '/^<a[\s>]/i', $part ) ) $inside_a = true;
                if ( preg_match( '/^<\/a>/i', $part ) ) $inside_a = false;
                if ( $skip_headings ) {
                    if ( preg_match( '/^<h[1-6][\s>]/i', $part ) ) $inside_heading = true;
                    if ( preg_match( '/^<\/h[1-6]>/i', $part ) ) $inside_heading = false;
                }
                $seg_out[] = $part;
                continue;
            }

            if ( $inside_a || $inside_heading || $links_added >= $max_links ) {
                $words_since_last_link += lfgail_count_words( $part );
                $seg_out[] = $part;
                continue;
            }

            foreach ( $index as $keyword => $meta ) {
                if ( $links_added >= $max_links ) break;
                if ( (int) $meta['post_id'] === (int) $current_id ) continue;
                if ( ( $used_keywords[ $keyword ] ?? 0 ) >= $max_per_kw ) continue;
                if ( ( $used_urls[ $meta['url'] ] ?? 0 ) >= 2 ) continue;
                if ( $words_since_last_link < $min_words_gap ) continue;

                $pattern = '/(?<![a-zA-Z0-9\-])(' . preg_quote( $keyword, '/' ) . ')(?![a-zA-Z0-9\-])/iu';
                if ( preg_match( $pattern, $part ) ) {
                    $replacement = '<a href="' . esc_url( $meta['url'] ) . '" title="' . esc_attr( $meta['title'] ) . '"' . $target . '>$1</a>';
                    $new_part = preg_replace( $pattern, $replacement, $part, 1 );

                    if ( $new_part !== $part ) {
                        preg_match( $pattern, $part, $m, PREG_OFFSET_CAPTURE );
                        $words_since_last_link += lfgail_count_words( substr( $part, 0, $m[0][1] ) );
                        $part = wp_kses_post( $new_part );
                        $links_added++;
                        $used_keywords[ $keyword ] = ( $used_keywords[ $keyword ] ?? 0 ) + 1;
                        $used_urls[ $meta['url'] ] = ( $used_urls[ $meta['url'] ] ?? 0 ) + 1;
                        $words_since_last_link = 0;
                        preg_match( '/<\/a>(.*)/s', $part, $after );
                        if ( ! empty( $after[1] ) ) {
                            $words_since_last_link += lfgail_count_words( wp_strip_all_tags( $after[1] ) );
                        }
                        break;
                    }
                }
            }

            if ( strpos( $part, '</a>' ) === false ) {
                $words_since_last_link += lfgail_count_words( $part );
            }

            $seg_out[] = $part;
        }

        $result[] = implode( '', $seg_out );
    }

    return implode( '', $result );
}

function lfgail_count_words( $text ) {
    $clean = trim( preg_replace( '/\s+/', ' ', strip_tags( $text ) ) );
    return $clean === '' ? 0 : count( explode( ' ', $clean ) );
}