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

/*
 * energysi-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/energysi/energysi-import.php,v 1.1 2007-11-13 23:22:18 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($row) // {{{
{
	global $db;
	$query = sprintf('SELECT id FROM publicpi_energy2.ezin_articles WHERE access_key LIKE "esi-%d-%s%%"',
			$row['id_art'], $row['lang']);
	$id = $db->getOne($query);
	if (DB::isError($id)) {
		error($id);
	}
	$id = intval($id);
	return $id;
} // }}}

function get_section($id_art) // {{{
{
	global $db;
	static $map_ar = array(
			'100' /* Energy News                   */ => 22,
			'108' /* Slovak Energy Forum           */ => 36,
			 '10' /* About ESI                     */ => 23,
			'105' /* ESI Board                     */ => 31,
			'107' /* Energy Forums                 */ => 34,
			 '20' /* Governance & Expertise        */ => 35,
			'106' /* Energy Studies                */ => 33,
			'104' /* Energy Events                 */ => 24,
			 '30' /* ESI Initiatives               */ => 25,
			 '50' /* ESI Insight                   */ => 26,
			 '60' /* Media & Publications          */ => 27,
			 '70' /* Partners                      */ => 28,
			 '80' /* Annual Report                 */ => 29,
			 '90' /* Contact                       */ => 30,
			'103' /* ESI Mission                   */ => 32,
			 '31' /* Energy Security               */ => 37,
			 '32' /* Liberalization & Consequences */ => 38,
			 '33' /* Transatlantic Energy Dialogue */ => 39,
			 '34' /* Central Asia Stability        */ => 40
			);
	$id_sec = $db->getOne('SELECT id_sec FROM sec_art WHERE id_art = '.intval($id_art));
	if (isset($map_ar[$id_sec])) {
		return intval($map_ar[$id_sec]);
	}
	echo('WARNING - cannot find section:'.$id_sec);
	return 22; /* default */
} // }}}

function get_access_key($row) // {{{
{
	return 'esi-'.intval($row['id_art']).'-'.$row['lang'].'-'.$row['show'];
} // }}}

// Data source name configuration
$dsn = 'mysql://publicpi_energy:kofola@localhost/publicpi_energy';

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

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

/*
 * Cleanup
 */

$query = 'DELETE FROM publicpi_energy2.ezin_cache';
if (DB::isError($res = $db->query($query))) {
	error($res);
}

/*
 * Articles migration
 */
$ids = array();
$idc = 0;
$max = $db->getOne('SELECT MAX(id) FROM publicpi_energy2.ezin_articles');
//echo "max = $max\n";exit;

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

while ($row = $res->fetchRow()) {
	echo "Processing ID $row[id_art]... ";
	$out = array(
			'id'                 => lookup_id($row),
			'author_id'          => 1, /* admin */
			'section_id'         => get_section($row['id_art']),
			'bind_lang'          => null,
			'bind_theme'         => null,
			'accesses_count'     => 0,    
			'sessions_count'     => 0,
			'message_count'      => 0, 
			'status'             => 'showed',
			'datetime'           => $row['date'],
			'hometext_wordcount' => str_word_count($row['sh_text']),
			'bodytext_wordcount' => str_word_count($row['full_text']),
			'hometext_flags'     => 'iER',
			'bodytext_flags'     => 'iER',
			'access_key'         => get_access_key($row),
			'template_key'       => null,
			'title'              => trim($row['subject']),
			'atitle'             => null,
			'hometext'           => trim($row['sh_text']),
			'bodytext'           => trim($row['full_text']),
			'created_by'         => 'convertor',
			'created_dt'         => '2007-11-17'
			);

	foreach (array('hometext', 'bodytext') as $key) {
		$out[$key] = preg_replace('|<img src="([^"]*)"|mi', '<image $1', $out[$key]);
		$out[$key] = preg_replace('|\s+|mi', ' ', $out[$key]);
		$out[$key] = preg_replace('|&nbsp;|', '~', $out[$key]);
		$out[$key] = preg_replace('|~+<|', '<', $out[$key]);
		$out[$key] = preg_replace('|\\\\\'|', "'", $out[$key]);
		$out[$key] = preg_replace('|\\\'|', "'", $out[$key]);
		$out[$key] = preg_replace('|<!--[^<>]*-->|mi', '', $out[$key]); // comments removal
		$out[$key] = preg_replace('|</p>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|</div>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|<div[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|</font>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|<font[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|</o:[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|<o:[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|</st1:[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|<st1:[^<>]*>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|<span[^<>]*>\s*</span>|mi', '', $out[$key]);
		$out[$key] = preg_replace('|\s*<br[^<>]*>\s*|mi', "<br>", $out[$key]);
		$out[$key] = preg_replace('|\s*<br>(\s*<br>\s*)+|mi', "<p>", $out[$key]);
		$out[$key] = preg_replace('|\s*<p[^<>]*>\s*|mi', "<p>", $out[$key]);
		$out[$key] = preg_replace('|(<p>)+|', '<p>', $out[$key]);
		$out[$key] = preg_replace('/((<br>|<p>|~)?\s*)+$/mi', '', $out[$key]);
		$out[$key] = preg_replace('/^((<br>|<p>|~)?\s*)+/mi', '', $out[$key]);
	}

	if (strlen($out['bodytext']) <= 0) {
		$out['bodytext'] = $out['hometext'];
		$out['hometext'] = '';
	}

	$out['hometext_wordcount'] = str_word_count($out['hometext']);
	$out['bodytext_wordcount'] = str_word_count($out['bodytext']);

	echo "\n";

	$query = createQuery('publicpi_energy2.ezin_articles', $out);
	if (DB::isError($result = $db->query($query))) {
			error($result);
	}
}

$query = 'UPDATE publicpi_energy2.ezin_articles SET
hometext = REPLACE(hometext, "<p>", "\r\n\r\n"),
bodytext = REPLACE(bodytext, "<p>", "\r\n\r\n")
WHERE created_by = "convertor" AND created_dt >= "2007-11-17"';
if (DB::isError($res = $db->query($query))) {
	error($res);
}

$db->disconnect();

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

?>