source: trunk/scripts/migrate-newsletters-to-html @ 1184

#! /usr/bin/env php
<?php
require_once(dirname($argv[0]) . "/utilities.php");
require_once(IA_ROOT_DIR . "common/textblock.php");
require_once(IA_ROOT_DIR . "common/db/textblock.php");

// Timestamp since all newsletters are HTML.
define("IA_NEWSLETTER_HTML_EPOCH", "2009-01-10 00:00:00");
log_assert(is_db_date(IA_NEWSLETTER_HTML_EPOCH));
// Basic HTTP URL Regexp;
define("IA_RE_HTTP_URL", 'https?:\/\/[^\s\(\)]*');

function newsletter_is_plaintext($textblock) {
    log_assert_valid(textblock_validate($textblock));
    $timestamp = $textblock['timestamp'];
    return ($timestamp && is_db_date($timestamp) &&
        db_date_parse($timestamp) < db_date_parse(IA_NEWSLETTER_HTML_EPOCH));
}

// Convert plaintext newsletter body to textile format.
function newsletter_plaintext_to_textile($plaintext) {
    // Indent paragraphs with leading whitespace.

    $paragraphs = preg_split("/(\n\r?){2,}/", $plaintext);
    $textile = '';
    foreach ($paragraphs as $paragraph) {
        $p_stripped = ltrim($paragraph);
        $num_spaces = strlen($paragraph) - strlen($p_stripped);
        if ($num_spaces > 0) {
            $indent = (double)$num_spaces / 2;
            $textile .= "p{padding-left: {$indent}em;}. ";
        }
        $textile .= $p_stripped."\n\n";
    }

    // Insert headings.

    $textile = preg_replace('/^==(.+)==\s*$/m', "h2. \\1\n", $textile);

    // Convert placeholder %tags% to textile macros.

    $tags = array(
        '"%full_name%"' => '==UserInfo(user="%username%" info="fullname")==',
        '%full_name%' => '==UserInfo(user="%username%" info="fullname")==',
        '%url_unsubscribe%' => '\'account/%username%\':account/%username%',
        '%your_rating_is%' => '==NewsletterRating(username="%username%")==',
    );
    foreach ($tags as $tag_name => $tag_value) {
        $textile = str_replace($tag_name, $tag_value, $textile);
    }

    // Convert links

    $textile = preg_replace('|%url_infoarena%/?utilizator/('.IA_RE_USER_NAME.')|',
            ' ==User(user="\1" type="link")== ', $textile);
    $textile = preg_replace('|%url_infoarena%/?('.IA_RE_PAGE_NAME.')\.|',
            '\'\1\':\1.', $textile);
    $textile = preg_replace('|%url_infoarena%/?('.IA_RE_PAGE_NAME.')|',
            '\'\1\':\1', $textile);
    $textile = preg_replace('|%url_infoarena%/?|', '\'infoarena\':home', $textile);
    $textile = preg_replace('|('.IA_RE_HTTP_URL.')|', '\'\1\':\1', $textile);
    return $textile;
}


db_connect();
foreach (textblock_get_by_prefix('newsletter/', true) as $textblock) {
    if (!newsletter_is_plaintext($textblock)) {
        log_print('Skipping '.$textblock['name']);
        continue;
    }
    log_print('Converting '.$textblock['name']);
    $new_text = newsletter_plaintext_to_textile($textblock['text']);
    textblock_add_revision($textblock['name'], $textblock['title'],
            $new_text, $textblock['user_id'], $textblock['security'],
            $textblock['forum_topic'], null, $textblock['creation_timestamp'],
            getattr($textblock, 'remote_ip_info'));
}

?>