#!/usr/bin/php -q
<?php

/*
 * amrophever-import.php
 *
 * Developed by Ondrej Jombik <nepto@platon.sk>
 * Copyright (c) 2007 Platon Group, http://platon.sk/
 * Licensed under terms of GNU General Public License.
 * All rights reserved.
 *
 */

/* $Platon: scripts/php/importers/amrophever/amrophever-import.php,v 1.6 2007-09-23 23:15:47 nepto Exp $ */

function error($err_obj) // {{{
{
	if (is_object($err_obj)) {
		$error_string = $err_obj->getMessage() . "\n" . $err_obj->getDebugInfo();
	} else {
		$error_string = $err_obj;
	}
	echo "\nFATAL ERROR: $error_string\n";
	die();
} // }}}

	function createQuery($table, $data) // {{{
	{
		$keys = '';
		$vals = '';
		foreach ($data as $key => $val) {
			if (is_null($val) || !strncmp($key, '__', 2)) {
				continue;
			}

			if (is_array($val)) {
				$val = array_unique($val);
				sort($val);
				$val = join("\n", $val);
			}

			if ($keys != '')
				$keys .= ', ';
			if ($vals != '')
				$vals .= ', ';

			$keys .= $key;
			//$val = DB::quoteString($val);
			$val = addslashes($val);
			$vals .= "'$val'";
		}
		if ($keys != '') {
			return sprintf('REPLACE INTO %s (%s) VALUES (%s)',
					$table, $keys, $vals);
		}
		return false;
	} // }}}

function lookup_id($ar, $row) // {{{
{
	global $db;
	$query = sprintf('SELECT id FROM ezin_articles WHERE access_key LIKE "ajg-%d-%d%%"',
			$row['id_clanok'], $row['language']);
	$id = $db->getOne($query);
	if (DB::isError($id)) {
		error($id);
	}
	$id = intval($id);
	if ($id > 0) {
		echo ' <simple> ';
		return $id;
	}

	echo ' <notfound> ';

	$query = sprintf('SELECT id FROM ezin_articles'
			.' WHERE datetime = %s AND section_id = %d'
			.' AND (hometext_wordcount = %d AND bodytext_wordcount = %d)',
			//$db->quote(substr($ar['title'], 0, 5).'%'),
			$db->quote($ar['datetime']),
			$db->quote($ar['section_id']),
			//$db->quote($ar['author_id']),
			$db->quote($ar['hometext_wordcount']),
			$db->quote($ar['bodytext_wordcount'])
			);
	$id = $db->getOne($query);
	if (DB::isError($id)) {
		error($id);
	}
	$id = intval($id);
	if ($id == 0 && $row['url']) {
		$query = sprintf('SELECT id FROM ezin_articles WHERE hometext LIKE %s',
				$db->quote('%'.$row['url'].'%'));
		$id = $db->getOne($query);
		if (DB::isError($id)) {
			error($id);
		}
		$id = intval($id);
		if ($id > 0) {
			echo " get-from-URL($row[url]): ";
		}
	}
	return $id;
} // }}}

function get_author($ar) // {{{
{
	global $db;
	if (strlen(trim($ar['autor'])) <= 0) {
		return 1;
	}
	while (($id = $db->getOne(
					sprintf('SELECT id FROM ezin_authors WHERE name LIKE %s',
						$db->quote($ar['autor'])))) <= 0)
	{
		$query = sprintf('INSERT INTO ezin_authors (name) VALUES (%s)',
				$db->quote($ar['autor']));
		$db->query($query);
	}
	return $id;
} // }}}

function get_section($ar) // {{{
{
	$ret = -1;
	if ($ar['language'] == 1) {
		if (intval($ar['sekcia2']) > 0) {
			switch ($ar['sekcia2']) {
				case 1: $ret = 99; break;
				case 2: $ret = 102; break;
				case 3: $ret = 103; break;
				case 4: $ret = 106; break;
				default: die('wrong sekcia2: '.$ar['sekcia2']);
			}
		} else {
			switch ($ar['sekcia1']) {
				case 1: $ret = 20; break;
				case 2: $ret = 21; break;
				case 3: $ret = 98; break;
				default: die('wrong sekcia1: '.$ar['sekcia1']);
			}
		}
	} else if ($ar['language'] == 2) {
		if (intval($ar['sekcia2']) > 0) {
			switch ($ar['sekcia2']) {
				case 1: $ret = 100; break;
				case 2: $ret = 101; break;
				case 3: $ret = 104; break;
				case 4: $ret = 105; break;
				default: die('wrong sekcia2: '.$ar['sekcia2']);
			}
		} else {
			switch ($ar['sekcia1']) {
				case 1: $ret = 63; break;
				case 2: $ret = 64; break;
				case 3: $ret = 65; break;
				default: die('wrong sekcia1: '.$ar['sekcia1']);
			}
		}
	} else {
		die('wrong language');
	}
	return $ret;
} // }}}

function get_access_key($row) // {{{
{
	return 'ajg-'.intval($row['id_clanok'])
		.'-'.intval($row['language'])
		.'-'.intval($row['sekcia1'])
		.'-'.intval($row['sekcia2'])
		.'-'.intval($row['sektor']);
} // }}}

// Data source name configuration
$dsn = 'mysql://nepto_amrophever:x@localhost/nepto_amrophever';

// PEAR DB class
require_once 'DB.php';

if (DB::isError($db = DB::connect($dsn))) {
	error($db);
}
$db->setFetchMode(DB_FETCHMODE_ASSOC);

/*
 * Cleanup
 */

$query = 'DELETE FROM ezin_cache';
if (DB::isError($res = $db->query($query))) {
	error($res);
}

/*
$query = 'DELETE FROM ezin_authors WHERE id > 1';
if (DB::isError($res = $db->query($query))) {
	error($res);
}
*/

/*
 * Articles migration
 */
$ids = array();
$idc = 0;
$max = $db->getOne('SELECT MAX(id) FROM ezin_articles');
//echo "max = $max\n";exit;

$query = 'SELECT * FROM clanky WHERE sekcia1 > 0';
if (DB::isError($res = $db->query($query))) {
	error($res);
}

while ($row = $res->fetchRow()) {
	echo "Processing ID $row[id_clanok]... ";
	if ($row['zobrazit'] != 1) {
		echo "\nWARNING: zobrazit != 1 - skipping\n";
		continue;
	}
	$dual_lang = false;
	if ($row['language'] == 3) { // dual language article
		$row['language'] = 1;
		$dual_lang = true;
	}
	$ar['autor'] = trim($ar['autor']);
	$ar['autor'] = str_replace("\n", ' ', $ar['autor']);
	$ar['autor'] = str_replace("\r", ' ', $ar['autor']);
	$ar['autor'] = preg_replace('|\s*|mi', ' ', $ar['autor']);
	$out = array(
			//'id'                 => 1000+$row['id_clanok'],
			'id'                 => null, // auto-increment
			'author_id'          => get_author($row),
			'section_id'         => get_section($row),
			'bind_lang'          => null,
			'bind_theme'         => null,
			'accesses_count'     => 0,    
			'sessions_count'     => 0,
			'message_count'      => 0, 
			'status'             => 'showed',
			'datetime'           => $row['datum'],
			'hometext_wordcount' => str_word_count($row['anotacia']),
			'bodytext_wordcount' => str_word_count($row['text']),
			'hometext_flags'     => 'iER',
			'bodytext_flags'     => 'iER',
			'access_key'         => get_access_key($row),
			'template_key'       => null,
			'title'              => $row['nadpis'],
			'atitle'             => null,
			'hometext'           => trim($row['anotacia']),
			'bodytext'           => trim($row['text'])
			);

	if (strlen($row['url']) > 0) {
		//$out['bodytext'] = $out['hometext'];
		if (strlen(trim($out['bodytext'])) > 0) {
			$out['bodytext'] .= "\n\r\n\r$row[url]";
		} else {
			$out['hometext'] .= "<br>\n\r";
			$out['hometext'] .= sprintf('<extern-link %s%s>%s</extern-link> ',
					str_replace(' ', '+', trim($row['url'])),
					intval($row['pdf_velkost']) > 0 ? sprintf(' title="PDF~%d~kB"', $row['pdf_velkost']) : '',
					($row['language'] == 1 ? 'čítať ďalej...' : 'read more...'));
		}
	}

	$out['hometext'] = preg_replace('|<img src="([^"]*)".*>|mi', '', $out['hometext']);
	$out['bodytext'] = preg_replace('|<img src="([^"]*)"|mi', '<image $1', $out['bodytext']);
	foreach (array('hometext', 'bodytext') as $key) {
		$out[$key] = preg_replace('|\s*<br>\s*<br>\s*|', "\n\r\n\r", $out[$key]);
		$out[$key] = preg_replace('|href="/|', 'target="_blank" href="http://www.jenewein.sk/', $out[$key]);
		$out[$key] = preg_replace('|<div align="center">|', '', $out[$key]);
	}

	$out['hometext_wordcount'] = str_word_count($row['anotacia']);
	$out['bodytext_wordcount'] = str_word_count($row['text']);

	echo 'LOOKUP for ID ...';
	$id = lookup_id($out, $row);
	echo "$id\n";
	if ($id > 0) {
		if (in_array($id, $ids)) {
			echo " - already IN ARRAY!\n";
		} else {
			array_push($ids, $id);
			$out['id'] = $id;
		}
	} elseif ($id == 0) { $idc++; }

	$query = createQuery('ezin_articles', $out);
	if (1 && DB::isError($result = $db->query($query))) {
			error($result);
	}

	if ($dual_lang) {
		$row['language']   = 2;
		$out['section_id'] = get_section($row);
		$out['access_key'] = get_access_key($row);

		printf("  dualang, reins:%03s ", $out['section_id']);
		echo 'LOOKUP for ID ...';
		$id = lookup_id($out, $row);
		echo "$id\n";
		if ($id > 0) {
			if (in_array($id, $ids)) {
				echo " - already IN ARRAY!\n";
			} else {
				array_push($ids, $id);
				$out['id'] = $id;
			}
		} elseif ($id == 0) { $idc++; }

		$query = createQuery('ezin_articles', $out);
		if (1 && DB::isError($result = $db->query($query))) {
				error($result);
		}
	}
}


$query = 'DELETE FROM ezin_articles'
	.' WHERE section_id IN (20,21,63,64,100,99,102,101,104,103,106,105, 98, 65)'
	.' AND datetime IS NOT NULL AND datetime > "0000-00-00"'
	.' AND id NOT IN ('.join(',', $ids).')'
	.' AND id <= '.$max;
// protected 97,22,98,65
			echo "----------\nRUN THIS:\n$query\nidc = $idc\n";
if (0 && DB::isError($res = $db->query($query))) {
	error($res);
}


$db->disconnect();

/* Modeline for ViM {{{
 * vim: set ts=4:
 * vim600: fdm=marker fdl=0 fdc=0:
 * }}} */

?>