function extract_league($Html) { $Html = copy_be($Html, '<form name="hform"', '</form>'); $Html = implode("\r\n", extract_form_data($Html)); $Html = str_ireplace("showrosp=1\r\n", '', $Html); return $Html; }
function extract_league($Html) { $Html = copy_be($Html, '<form id="fbets"', '</form>'); $Html = replace_all($Html, 'долгосрочные ставки', '</form>', '</form>'); $Html = extract_tags($Html, '<input', '>', "\r\n", 'name=', 'value='); $Html = delete_all($Html, '<input', 'name='); $Html = str_ireplace(' value="', '=', $Html); $Html = str_ireplace('">', '', $Html); return $Html; }
function process_login($Html) { $PostHash = extract_form_hash($Html); $PostHash['username'] = '******'; $PostHash['passwd'] = 'grafdevalery'; $Html = copy_be($Html, '<form ', '>'); $Tags = extract_property_values($Html, 'action', "\r\n"); $Html = download($Tags, 'POST', 'http://www.buker.ru/dostaken.php', $PostHash); return $Html; }
function extract_league($Html, $Sport) { $Html = kill_space($Html); $Html = numbering_tag($Html, 'div'); $Html = extract_numbered_tags($Html, 'div', "\r\n", "navElem"); $TagNo = extract_tagno(copy_be($Html, "<div\t", "&SportID={$Sport}\">"), 'div'); $Html = extract_numbered_tag($Html, 'div', $TagNo); $Html = extract_tags($Html, 'EventGroupID=', "'", ','); $Html = str_ireplace('EventGroupID=', '', $Html); $Html = str_ireplace("'", '', $Html); return $Html; }
function extract_bet($Html) { $Html = copy_be($Html, '<form ', '</form>', 'betform'); $Html = kill_tag($Html, 'script|div'); $Html = kill_space($Html); $Html = kill_tag_bound($Html, 'b|span|br|i|a|input|font|img'); $Html = delete_all($Html, '<tr', '</tr>', 'line3'); $Html = kill_property_value_exclude($Html, 'class', 'line_table'); $Html = kill_property($Html, 'width|height|valign|align|bgcolor|style|border|cellspacing|cellpadding'); // Убираем "левые" данные $Html = replace_all_contain($Html, '<tr', '</tr>', '', 'Хозяева|' . 'Гости|' . 'Индив.тотал:|' . '-я четверть|' . '-я партия|' . ' тайм|' . '-й сет'); return $Html; }
function extract_league($Html) { $Html = copy_be($Html, '<form ', '</form>', '"frmTurnirs"'); $Html = extract_tags($Html, '<input', '>', "\r\n", 'id="sport_', 'name="turnir_'); $Html = delete_all($Html, '<input', '>', 'turnir_2_'); // Бейсбол $Html = delete_all($Html, '<input', '>', 'turnir_12_'); // Бокс $Html = delete_all($Html, '<input', '>', 'turnir_254_'); // Лотереи $Html = delete_all($Html, '<input', '>', 'turnir_255_'); // Другие виды спорта $Html = delete_all($Html, '<input', '>', 'turnir_6_'); // Хоккей $Html = extract_property_values($Html, 'name', ';'); return $Html; }
function extract_bet($Html) { // Разбираемся с концами строк $Html = str_ireplace("\n", '<br>', $Html); $Html = str_ireplace("\r", '<br>', $Html); $Html = copy_be($Html, '<form name=f1>', '</form>'); $Html = str_ireplace(' ', ' ', $Html); $Html = kill_tag_bound($Html, 'input|itemevent|a|b'); $Html = str_ireplace(' noshade', '', $Html); $Html = kill_property($Html, 'time|color'); $Html = replace_all($Html, '<sup>', '</sup>', ' '); $Html = kill_tag($Html, 'big'); $Html = str_ireplace('<endpreview/>', '', $Html); $Html = str_ireplace('<hr><br>', '<hr>', $Html); $Html = str_ireplace('<br>Оглавление<', '<', $Html); $Html = str_ireplace('<br>Оглавление<', '<', $Html); $Html = replace_all($Html, '<u>', '</form>', '</form>'); $Html = kill_tag_bound($Html, 'u|form'); $Html = delete_all($Html, '<span ', '</span>', 'maxcap'); return $Html; }
function extract_bet($Html) { $Html = copy_be($Html, '<form name=f1', '</form>', 'Дома'); $Html = kill_space($Html); $Html = kill_tag($Html, 'script'); $Html = kill_tag_bound($Html, 'b|a|tbody|small|colgroup'); // <!-- Выносим Экспресы и информацию о предыдущих встречах --> $Html = str_ireplace('<span class=tr>1</span>', '', $Html); $Html = delete_all($Html, '<tr>', '</tr>', '<span class=tr>'); $Html = delete_all($Html, '<span class=n>', '</span>'); // <!-- Убираем строки с кнопками --> $Html = delete_all($Html, '<table ', '</table>', '<input '); // <!-- Заменяем IMG с информацией на igif для последующего удаления --> $Html = replace_all($Html, '<img ', '>', "igif"); // <!-- Убираем левые коэффициенты в скобках --> $Html = replace_all($Html, '>(', ')', '>'); // <!-- Убираем оглавление --> $Html = delete_all($Html, "<div ", "</div", "Оглавление"); $Html = kill_property($Html, 'width|border'); $Html = replace_all_contain($Html, '<tr', '</tr>', '', '>Хозяева|' . '>Гости|' . '>Фолы|' . '>ж/к|' . '>угл.'); return $Html; }
private function extract_extra_bets(&$tournir_node, $html, $sport_sign, $tournir_id) { $html = str_ireplace('<li><h2>', '<li class="extra"><h2>', $html); $html = numbering_tag($html, 'li'); $table_rows = extract_all_numbered_tags($html, 'li', 'extra'); foreach ($table_rows as $row) { $row = kill_property($row, 'TagNo'); $event_id = extract_property_values(copy_be($html, '<ul', '>', 'rel'), 'rel', ''); $event_node = $this->event_find($tournir_node, $event_id); $phrase = copy_be($row, '<h2', '</h2>'); $phrase = copy_between($phrase, '>', '<'); $section_node = $this->findSection($sport_sign, $phrase); if (!(string) $section_node['Ignore']) { $bets = extract_all_tags($row, '<li', '</li>', 'rel'); foreach ($bets as $bet) { list($label, $koef) = $this->extract_label_koef($bet); $label = str_ireplace((string) $event_node['Gamer1_Name'], 'Gamer1', $label); $label = str_ireplace((string) $event_node['Gamer2_Name'], 'Gamer2', $label); $phrase_node = $this->findPhrase($sport_sign, $phrase, $label); if (!$phrase_node['Ignore']) { $this->addBet($event_node, (string) $phrase_node['BetKind'] . ';Koef=' . $koef); } } } } }
function extract_numbered_tag($Html, $TagName, $TagNo) { return copy_be($Html, "<{$TagName}\tTagNo=\"{$TagName}{$TagNo}\"", "</{$TagName}\tTagNo=\"{$TagName}{$TagNo}\">"); }
function extract_extra_bets(&$tournir_node, $html, $sport_sign, $tournir_id) { $filename_headers = "phrases/olymp/{$sport_sign}.headers.txt"; $filename_labels = "phrases/olymp/{$sport_sign}.labels.txt"; $phrases_headers = file_get_hash($filename_headers); $phrases_labels = file_get_hash($filename_labels); $html = str_ireplace('<li><h2>', '<li class="extra"><h2>', $html); $html = numbering_tag($html, 'li'); $table_rows = extract_all_numbered_tags($html, 'li', 'extra'); foreach ($table_rows as $row) { $row = kill_property($row, 'TagNo'); $event_id = extract_property_values(copy_be($html, '<ul', '>', 'rel'), 'rel', ''); $event_node = event_find($tournir_node, $event_id); $phrase = copy_be($row, '<h2', '</h2>'); $phrase = copy_between($phrase, '>', '<'); unset($bettype); $bettype_str = $phrases_headers[$phrase]; if ($bettype_str) { foreach (explode(';', $bettype_str) as $bet_pair) { list($key, $value) = explode('=', $bet_pair); $bettype[$key] = $value; } } else { $bettype['Modifier'] = 'Unknown'; $phrases_headers[$phrase] = 'Modifier=Unknown'; $phrases_headers_modified = true; } if (!in_array($bettype['Modifier'], array('Ignore', 'Unknown'))) { $bets = extract_all_tags($row, '<li', '</li>', 'rel'); foreach ($bets as $bet) { list($label, $koef) = extract_label_koef($bet); $label = str_ireplace((string) $event_node['Gamer1_Name'], 'Gamer1', $label); $label = str_ireplace((string) $event_node['Gamer2_Name'], 'Gamer2', $label); $modifier = $phrases_labels[$label]; if ($modifier) { foreach (explode(';', $modifier) as $bet_pair) { list($key, $value) = explode('=', $bet_pair); $bettype[$key] = $value; } } else { $bettype['Modifier'] = 'Unknown'; $phrases_labels[$label] = 'Modifier=Unknown'; $phrases_labels_modified = true; } if (!in_array($bettype['Modifier'], array('Ignore', 'Unknown'))) { $bet_node = $event_node->addChild('Bet'); foreach ($bettype as $key => $value) { $bet_node->addAttribute($key, $value); } $bet_node->addAttribute('Koef', $koef); } unset($bettype['Modifier']); } } $i++; } // отбираем новые и складываем их в новый файл if ($phrases_headers_modified) { $file_hash = file_get_hash($filename_headers); foreach ($file_hash as $key => $value) { unset($phrases_headers[$key]); } file_put_hash($filename_headers . '.new', $phrases_headers + file_get_hash($filename_headers . '.new')); } if ($phrases_labels_modified) { $file_hash = file_get_hash($filename_labels); foreach ($file_hash as $key => $value) { unset($phrases_labels[$key]); } file_put_hash($filename_labels . '.new', $phrases_labels + file_get_hash($filename_labels . '.new')); } }
private function extract_events(&$tournir_node, $html, $sport_sign, $tournir_id) { $html = copy_be($html, '<table', '</table>', 'class=date'); $html = kill_space($html); $tbodies = extract_all_tags($html, '<tbody', '</tbody>'); foreach ($tbodies as $tbody) { $tbody_class = copy_be($tbody, '<', '>'); $tbody_class = str_ireplace('id=', 'class=', $tbody_class); $tbody_class = extract_property_values($tbody_class, 'class', ''); if ($tbody_class == 'date') { list($this->day_no, $this->month_no, $this->year_no) = $this->decode_date($tbody); } elseif ($tbody_class == 'chead') { $this->extract_header($sport_sign, $tbody); } elseif ($tbody_class == 'line') { if ($sport_sign == 'tennis') { $this->extract_main_bets_tennis($tournir_node, $tbody, $sport_sign, $tournir_id); } elseif ($sport_sign == 'soccer') { $this->extract_main_bets_soccer($tournir_node, $tbody, $sport_sign, $tournir_id); } } } }
private function extract_main_bets(&$tournir_node, $html, $sport_sign, $tournir_id) { $html = kill_property($html, 'TagNo'); $html = numbering_tag($html, 'tr'); $events = extract_all_numbered_tags($html, 'tr', 'event-header'); foreach ($events as $event) { // получаем $event_info = take_be($event, '<table', '</table>'); file_put_contents("Event.html", $event_info); preg_match_all('/<div class="(.*)?member-name(.*)?">(.*)?<\\/div>/imsU', $event_info, $matches); $gamer1_name = $matches[3][0]; $gamer2_name = $matches[3][1]; $datetime_str = delete_all(copy_be($event_info, '<td', '</td>', '"date"'), '<', '>'); list($day_no, $month_no, $year_no, $hour, $minute) = $this->decode_datetime($datetime_str); $extrabet = copy_be($event_info, '<a', '</a>', 'treeid'); $event_id = extract_property_values($extrabet, 'treeid', ''); $extrabet = delete_all($extrabet, '<', '>'); $event_node = $this->event_create($tournir_node, $event_id, mktime($hour, $minute, 0, $month_no, $day_no, $year_no), $gamer1_name, $gamer2_name); $cells = extract_all_tags(replace_all($event, '<td', '>', '<td>'), '<td>', '</td>'); $i = 0; foreach ($cells as $cell) { $cell = trim(copy_between($cell, '<td>', '</td>')); if ($cell) { $cell = kill_tag_bound($cell, 'a'); unset($bettype); $bettype_str = $this->header[$i]; foreach (explode(';', $bettype_str) as $bet_pair) { list($key, $value) = explode('=', $bet_pair); $bettype[$key] = $value; } if ($bettype['Kind'] == 'Total') { preg_match('/\\((.+?)\\)<br\\/>(.+?)/iU', $cell, $matches); if ($matches[2]) { $matches[2] = delete_all($matches[2], '<', '>'); $this->addBet($event_node, $bettype_str . ';Value=' . $matches[1] . ';Koef=' . $matches[2]); } } elseif ($bettype['Kind'] == 'Fora') { preg_match('/\\(([\\+\\-]*?)(.+?)\\)<br\\/>(.+?)/iU', $cell, $matches); if ($matches[3]) { $matches[3] = delete_all($matches[3], '<', '>'); $this->addBet($event_node, $bettype_str . ';Value=' . $matches[1] . $matches[2] . ';Koef=' . $matches[3]); } } else { $cell = delete_all($cell, '<', '>'); $this->addBet($event_node, $bettype_str . ';Koef=' . $cell); } } $i++; } //if ($extrabet > 0) $this->extract_extra_bets($event_node, $tournir_id, $event_id); } }
function extract_bet($Html) { $Html = delete_all($Html, "(at ", ")"); $Html = delete_all($Html, "(score: ", ")"); $Html = str_ireplace('<wbr/>', '', $Html); $Html = str_replace("-.--", " ", $Html); $Html = kill_tag($Html, "h1|script"); $Html = kill_space($Html); $Html = kill_comment($Html); $Html = numbering_tag($Html, 'div'); $Html = extract_numbered_tags($Html, 'div', "", "bet-list"); $Html = extract_numbered_tags($Html, 'div', "", "dsBodyContent"); $Html = kill_property($Html, 'TagNo'); $Html = kill_property($Html, "onmouseover|onmouseout|onclick|bgcolor|border|cellpadding|leaguename|id|width|cellspacing|regionname"); $Html = kill_property_value($Html, "colspan", "1"); $Html = kill_tag_bound($Html, "form|input|a|nobr|img|tbody|span"); $Html = delete_all($Html, ' class="unselected ', '"'); $Html = delete_all($Html, ' class=\'unselected ', '\''); $Html = replace_all_contain($Html, '<div', '</div>', '', 'isFavourite|' . 'isNoFavourite'); $Html = replace_all_contain($Html, '<tr', '</tr>', '', 'sizer|' . 'bottomline'); $Html = kill_property_value_exclude($Html, "class", "controlHeaderNoShadow|topbar|normal|def|alt|ControlContent"); $Html = str_ireplace('<td></td>', '<td/>', $Html); $Html = kill_unclassed_tag($Html, 'div'); $Html = kill_apostrof($Html, 'colspan'); // !-- Приводим Totalы к нормальному виду --> $Html = replace_all($Html, "<h5>Сумма набранных очков ", '</h5>', '<h5>Total</h5>'); $Html = replace_all($Html, "<h5>Тотал ", '</h5>', '<h5>Total</h5>'); $Html = replace_all($Html, "<h5>Сколько будет забито голов?", '</h5>', '<h5>Total</h5>'); $Html = replace_all($Html, "<h5>Сколько голов будет забито?", '</h5>', '<h5>Total</h5>'); $Html = replace_all($Html, "<h5>Сколько геймов будет сыграно в матче?", '</h5>', '<h5>Total</h5>'); $Html = replace_all($Html, "<h5>Кто выиграет больше геймов в матче? (гандикап игрока)", '</h5>', '<h5>Fora</h5>'); $Html = replace_all($Html, "<h5>Двойной шанс", '</h5>', '<h5>1X_12_X2</h5>'); $Html = str_ireplace('<tr class="alt"', "\r\n" . '<tr class="alt"', $Html); $Html = str_ireplace('<tr class="def"', "\r\n" . '<tr class="alt"', $Html); $Tags = explode("\r\n", $Html); foreach ($Tags as $tag) { if (copy_be($tag, '>', '<', ' - ', ' - ', ':')) { $Html = str_ireplace("\r\n" . $tag, str_ireplace('="alt"', '="evn"', $tag), $Html); } } $Tags = explode("\r\n", $Html); foreach ($Tags as $tag) { if (strpos($tag, '"alt"')) { $newtag = numbering_tag($tag, 'tr'); $trAlt = extract_numbered_tags($newtag, 'tr', '', 'alt'); $h5 = delete_all($trAlt, '<', '>'); if (strpos($h5, ' ')) { $newtag = str_replace($trAlt, '', $newtag); $Normals = explode("\r\n", extract_numbered_tags($newtag, 'tr', "\r\n", 'normal')); foreach ($Normals as $normal) { $newtag = str_replace($normal, '', $newtag); } $newtag = kill_property($newtag, 'TagNo'); $Html = str_replace($tag, $newtag, $Html); } } } $Html = str_ireplace("\r\n", '', $Html); $Html = str_ireplace('<tr class="evn"', "\r\n" . '<tr class="evn"', $Html); $Tags = explode("\r\n", $Html); foreach ($Tags as $tag) { if (strpos($tag, '"evn"')) { $newtag = numbering_tag($tag, 'tr'); $trEvn = extract_numbered_tags($newtag, 'tr', '', 'evn'); if (strpos($newtag, '="normal"') == 0) { $newtag = str_replace($trEvn, '', $newtag); } $newtag = kill_property($newtag, 'TagNo'); $Html = str_replace($tag, $newtag, $Html); } } $Html = str_ireplace("\r\n", '', $Html); return $Html; }
function extract_tag_from_tag($html, $outer_tag, $inner_tag, $kill_bounds = null) { $ul_tagno = copy_be($html, "<{$outer_tag}", '>'); $ul_tagno = extract_tagno($ul_tagno, $outer_tag); $html = extract_numbered_tag($html, $outer_tag, $ul_tagno); $result = array(); while ($li_tagno = copy_be($html, "<{$inner_tag}", '>')) { $li_tagno = extract_tagno($li_tagno, $inner_tag); $li_html = extract_numbered_tag($html, $inner_tag, $li_tagno); $html = str_replace($li_html, '', $html); if ($kill_bounds) { $li_html = delete_all($li_html, "<{$inner_tag}", '>', "TagNo=\"{$inner_tag}{$li_tagno}\""); $li_html = delete_all($li_html, "</{$inner_tag}", '>', "TagNo=\"{$inner_tag}{$li_tagno}\""); } $result[] = $li_html; } return $result; }
$FileName = $Lines . "league.html"; if (!file_exists($FileName)) { $Html = download("{$Host}/bets.php?line_id[]={$Sport}", "GET", "{$Host}/center.php"); file_put_contents($FileName, $Html); } else { $Html = file_get_contents($FileName); } } else { $Html = download_page("{$Host}/bets.php?line_id[]={$Sport}", "GET", "{$Host}/center.php"); } function extract_league($Html) { $Html = replace_all($Html, '<table', '</table>', '', 'долгосрочные ставки'); return extract_form_hash($Html); } $Form = copy_be(win1251_to_utf8($Html), '<form id="fbets"', '</form>'); if ($debug) { file_put_contents($Lines . "form.txt", $Form); } $PostHash = extract_league($Form); $FormAction = extract_form_action($Form); if ($debug) { file_put_contents($Lines . "league.txt", implode_hash("\r\n", $PostHash)); } // Получаем перечень турниров $PostHash['simple'] = 'on'; if ($debug) { $FileName = $Lines . "bets.html"; if (!file_exists($FileName)) { $Html = download("{$Host}/{$FormAction}", "POST", "{$Host}/bets.php?line_id[]={$Sport}", $PostHash); file_put_contents($FileName, $Html);
function to_be_continue($html, $next_page) { $html = copy_be($html, '<a', '>', 'pagingGotoItem', "currentPage={$next_page}"); return $html != ''; }