| 1 |
1
|
wheeler
|
<?php
|
| 2 |
|
|
#############################################################################
|
| 3 |
|
|
# Driver file for taiper2flatClient, a simple script to
|
| 4 |
|
|
# consume taiper web services and save the results in a delimited file,
|
| 5 |
|
|
# one record per line. The current delimiter is backtick '`' but
|
| 6 |
|
|
# that can be changed by changing the $seperator variable located
|
| 7 |
|
|
# in configurableParams.php.
|
| 8 |
|
|
#
|
| 9 |
|
|
# Command line useage:
|
| 10 |
|
|
# php taiper2flat.php
|
| 11 |
|
|
#
|
| 12 |
|
|
# dependencies:
|
| 13 |
|
|
# php 5+
|
| 14 |
|
|
# pear
|
| 15 |
|
|
# HTTP_Request
|
| 16 |
|
|
# I had to install pear and HTTP_Request before it would work (on Ubuntu):
|
| 17 |
|
|
# sudo apt-get install php-http-request
|
| 18 |
|
|
# This should install all dependencies, but just in case, the full
|
| 19 |
|
|
# dependency list is:
|
| 20 |
|
|
# php-net-socket
|
| 21 |
|
|
# php-net-url
|
| 22 |
|
|
# php-pear
|
| 23 |
|
|
# php-http-request
|
| 24 |
|
|
#############################################################################
|
| 25 |
|
|
|
| 26 |
|
|
require_once('HTTP/Request.php'); // pear package
|
| 27 |
|
|
require_once('tapirRequestTemplate.php');
|
| 28 |
|
|
require_once('getAllConcepts.php');
|
| 29 |
|
|
|
| 30 |
|
|
global $url;
|
| 31 |
|
|
global $seperator;
|
| 32 |
|
|
|
| 33 |
|
|
$flat_filename = "specieslink.txt";
|
| 34 |
|
|
$record_num_filename = "recordnum.dat";
|
| 35 |
|
|
$error_log_filename = "error.log";
|
| 36 |
|
|
|
| 37 |
|
|
$supportedConcepts = getAllConcepts();
|
| 38 |
|
|
if(!is_array($supportedConcepts)){
|
| 39 |
|
|
$error_log = fopen($error_log_filename,"a");
|
| 40 |
|
|
fwrite($error_log, "Unable to retrieve supported concepts from service, failing.\n");
|
| 41 |
|
|
fclose($error_log);
|
| 42 |
|
|
exit;
|
| 43 |
|
|
}
|
| 44 |
|
|
|
| 45 |
|
|
$successive_failures = 0;
|
| 46 |
|
|
|
| 47 |
|
|
$start_record = 0; // Record to begin first request (ordinal position, not ID)
|
| 48 |
|
|
if(file_exists($record_num_filename)) {
|
| 49 |
|
|
$start_record = file_get_contents($record_num_filename);
|
| 50 |
|
|
}
|
| 51 |
|
|
|
| 52 |
|
|
$request_limit = 1000; // Number of records per request; cannot exceed 1000
|
| 53 |
|
|
|
| 54 |
|
|
//According to TAPIR docs, the estimate from the service
|
| 55 |
|
|
//is an upper bound on the number of records. Start out high.
|
| 56 |
|
|
$estimated_max_from_service = 4000000;
|
| 57 |
|
|
|
| 58 |
|
|
if(!file_exists($flat_filename)) {
|
| 59 |
|
|
$flat_file = fopen($flat_filename,"a");
|
| 60 |
|
|
$file_header = '';
|
| 61 |
|
|
foreach(array_keys($supportedConcepts) as $key){
|
| 62 |
|
|
$file_header .= $key . $seperator;
|
| 63 |
|
|
}
|
| 64 |
|
|
$file_header = substr_replace($file_header ,"",-1);
|
| 65 |
|
|
$file_header .= "\n";
|
| 66 |
|
|
fwrite($flat_file,$file_header);
|
| 67 |
|
|
fclose($flat_file);
|
| 68 |
|
|
}
|
| 69 |
|
|
|
| 70 |
|
|
$finished = false;
|
| 71 |
|
|
$itrNum = 0;
|
| 72 |
|
|
|
| 73 |
|
|
$http_request = new HTTP_Request();
|
| 74 |
|
|
|
| 75 |
|
|
while(!$finished && $start_record < $estimated_max_from_service) {
|
| 76 |
|
|
$itrNum++;
|
| 77 |
|
|
|
| 78 |
|
|
$body = buildRequest($start_record,$request_limit,$supportedConcepts);
|
| 79 |
|
|
|
| 80 |
|
|
$http_request->setMethod( 'POST' );
|
| 81 |
|
|
$http_request->addHeader('Content-Type', 'text/xml');
|
| 82 |
|
|
$http_request->addRawPostData( $body );
|
| 83 |
|
|
$http_request->setURL( $url );
|
| 84 |
|
|
$http_request->_timeout = 300;
|
| 85 |
|
|
$http_request->_readTimeout = 300;
|
| 86 |
|
|
|
| 87 |
|
|
// This can be used to see the entire request
|
| 88 |
|
|
#$raw_request = $http_request->_buildRequest();
|
| 89 |
|
|
#echo "\n\n" . $raw_request;
|
| 90 |
|
|
|
| 91 |
|
|
$http_request->sendRequest();
|
| 92 |
|
|
|
| 93 |
|
|
$response = $http_request->getResponseBody();
|
| 94 |
|
|
$code = $http_request->getResponseCode();
|
| 95 |
|
|
|
| 96 |
|
|
if ( $code != 200 ) // 200 = OK
|
| 97 |
|
|
{
|
| 98 |
|
|
$label = 'Unknown Error';
|
| 99 |
|
|
|
| 100 |
|
|
switch ( $code )
|
| 101 |
|
|
{
|
| 102 |
|
|
case 201: $label = 'Created'; break;
|
| 103 |
|
|
case 202: $label = 'Accepted'; break;
|
| 104 |
|
|
case 203: $label = 'Non-Authoritative Information'; break;
|
| 105 |
|
|
case 204: $label = 'No Content'; break;
|
| 106 |
|
|
case 205: $label = 'Reset Content'; break;
|
| 107 |
|
|
case 206: $label = 'Partial Content'; break;
|
| 108 |
|
|
case 300: $label = 'Multiple Choices'; break;
|
| 109 |
|
|
case 301: $label = 'Moved Permanently'; break;
|
| 110 |
|
|
case 302: $label = 'Found'; break;
|
| 111 |
|
|
case 303: $label = 'See Other'; break;
|
| 112 |
|
|
case 304: $label = 'Not Modified'; break;
|
| 113 |
|
|
case 305: $label = 'Use Proxy'; break;
|
| 114 |
|
|
case 307: $label = 'Temporary Redirect'; break;
|
| 115 |
|
|
case 400: $label = 'Bad Request'; break;
|
| 116 |
|
|
case 401: $label = 'Unauthorized'; break;
|
| 117 |
|
|
case 402: $label = 'Payment Required'; break;
|
| 118 |
|
|
case 403: $label = 'Forbidden'; break;
|
| 119 |
|
|
case 404: $label = 'Not Found'; break;
|
| 120 |
|
|
case 405: $label = 'Method Not Allowed'; break;
|
| 121 |
|
|
case 406: $label = 'Not Acceptable'; break;
|
| 122 |
|
|
case 407: $label = 'Proxy Authentication Required'; break;
|
| 123 |
|
|
case 408: $label = 'Request Timeout'; break;
|
| 124 |
|
|
case 409: $label = 'Conflict'; break;
|
| 125 |
|
|
case 410: $label = 'Gone'; break;
|
| 126 |
|
|
case 411: $label = 'Length Required'; break;
|
| 127 |
|
|
case 412: $label = 'Precondition Failed'; break;
|
| 128 |
|
|
case 413: $label = 'Request Entity Too Large'; break;
|
| 129 |
|
|
case 414: $label = 'Request-URI Too Long'; break;
|
| 130 |
|
|
case 415: $label = 'Unsupported Media Type'; break;
|
| 131 |
|
|
case 416: $label = 'Requested Range Not Satisfiable'; break;
|
| 132 |
|
|
case 417: $label = 'Expectation Failed'; break;
|
| 133 |
|
|
case 500: $label = 'Internal Server Error'; break;
|
| 134 |
|
|
case 501: $label = 'Not Implemented'; break;
|
| 135 |
|
|
case 502: $label = 'Bad Gateway'; break;
|
| 136 |
|
|
case 503: $label = 'Service Unavailable'; break;
|
| 137 |
|
|
case 504: $label = 'Gateway Timeout'; break;
|
| 138 |
|
|
case 505: $label = 'HTTP Version Not Supported'; break;
|
| 139 |
|
|
}
|
| 140 |
|
|
|
| 141 |
|
|
$error_log = fopen($error_log_filename,"a");
|
| 142 |
|
|
fwrite($error_log, "Service responded with HTTP ".$code." code: ".$label."\n");
|
| 143 |
|
|
|
| 144 |
|
|
$successive_failures += 1;
|
| 145 |
|
|
if($successive_failures >= 3) {
|
| 146 |
|
|
fwrite($error_log, "3 successive failures, quitting\n");
|
| 147 |
|
|
fclose($error_log);
|
| 148 |
|
|
exit;
|
| 149 |
|
|
}
|
| 150 |
|
|
|
| 151 |
|
|
fclose($error_log);
|
| 152 |
|
|
|
| 153 |
|
|
//Should repeat the same request
|
| 154 |
|
|
continue;
|
| 155 |
|
|
}
|
| 156 |
|
|
|
| 157 |
|
|
#Weird encoding bug - need to remove the ^F and ^L characters so
|
| 158 |
|
|
#that the xml parser won't choke.
|
| 159 |
|
|
$xmlDoc = new DOMDocument();
|
| 160 |
|
|
$response = preg_replace("//","6",$response);
|
| 161 |
|
|
$response = preg_replace("/ /","12",$response);
|
| 162 |
|
|
$xmlDoc->loadXML($response);
|
| 163 |
|
|
|
| 164 |
|
|
$errors = $xmlDoc->getElementsByTagName("error");
|
| 165 |
|
|
if($errors->length > 0) {
|
| 166 |
|
|
$error_log = fopen($error_log_filename,"a");
|
| 167 |
|
|
foreach($errors as $error) {
|
| 168 |
|
|
fwrite($error_log, $error->nodeValue . "\n");
|
| 169 |
|
|
}
|
| 170 |
|
|
|
| 171 |
|
|
$successive_failures += 1;
|
| 172 |
|
|
if($successive_failures >= 3) {
|
| 173 |
|
|
fwrite($error_log, "3 successive failures, quitting\n");
|
| 174 |
|
|
fclose($error_log);
|
| 175 |
|
|
exit;
|
| 176 |
|
|
}
|
| 177 |
|
|
|
| 178 |
|
|
fclose($error_log);
|
| 179 |
|
|
|
| 180 |
|
|
//Should repeat the same request
|
| 181 |
|
|
continue;
|
| 182 |
|
|
}
|
| 183 |
|
|
|
| 184 |
|
|
$summary = $xmlDoc->getElementsByTagName("summary");
|
| 185 |
|
|
if($summary->length == 0) {
|
| 186 |
|
|
$error_log = fopen($error_log_filename,"a");
|
| 187 |
|
|
fwrite($error_log, "No summary node, assuming there's a missed error.\n");
|
| 188 |
|
|
|
| 189 |
|
|
$successive_failures += 1;
|
| 190 |
|
|
if($successive_failures >= 3) {
|
| 191 |
|
|
fwrite($error_log, "3 successive failures, quitting\n");
|
| 192 |
|
|
fclose($error_log);
|
| 193 |
|
|
exit;
|
| 194 |
|
|
}
|
| 195 |
|
|
|
| 196 |
|
|
fclose($error_log);
|
| 197 |
|
|
continue;
|
| 198 |
|
|
} else {
|
| 199 |
|
|
$start_record_temp = $summary->item(0)->getAttribute("next");
|
| 200 |
|
|
$estimated_max_from_service = $summary->item(0)->getAttribute("totalMatched");
|
| 201 |
|
|
}
|
| 202 |
|
|
|
| 203 |
|
|
$records = $xmlDoc->getElementsByTagName("record");
|
| 204 |
|
|
$recordsStr = '';
|
| 205 |
|
|
foreach($records as $record) {
|
| 206 |
|
|
foreach(array_keys($supportedConcepts) as $key){
|
| 207 |
|
|
$element = $record->getElementsByTagName($key);
|
| 208 |
|
|
if($element->length > 0) {
|
| 209 |
|
|
$val = $element->item(0)->nodeValue;
|
| 210 |
|
|
$val = preg_replace("/$seperator/"," ",$val);
|
| 211 |
|
|
$val = preg_replace("/,/"," ",$val);
|
| 212 |
|
|
$val = preg_replace("/\"/"," ",$val);
|
| 213 |
|
|
$recordsStr .= $val;
|
| 214 |
|
|
}
|
| 215 |
|
|
$recordsStr .= $seperator;
|
| 216 |
|
|
}
|
| 217 |
|
|
$recordsStr = substr_replace($recordsStr ,"",-1);
|
| 218 |
|
|
$recordsStr .= "\n";
|
| 219 |
|
|
}
|
| 220 |
|
|
|
| 221 |
|
|
$flat_file = fopen($flat_filename,"a");
|
| 222 |
|
|
fwrite($flat_file,$recordsStr);
|
| 223 |
|
|
fclose($flat_file);
|
| 224 |
|
|
|
| 225 |
|
|
$successive_failures = 0;
|
| 226 |
|
|
|
| 227 |
|
|
$record_file = fopen($record_num_filename,"w");
|
| 228 |
|
|
fwrite($record_file,$start_record_temp);
|
| 229 |
|
|
fclose($record_file);
|
| 230 |
|
|
$start_record = $start_record_temp;
|
| 231 |
|
|
|
| 232 |
|
|
//If number of records is less than request_limit, it means
|
| 233 |
|
|
//the service is out of records.
|
| 234 |
|
|
$finished = ($request_limit > $records->length);
|
| 235 |
|
|
|
| 236 |
|
|
if($itrNum == 1) {
|
| 237 |
|
|
print "Extimated number of records: $estimated_max_from_service\n";
|
| 238 |
|
|
} elseif($itrNum % 10 == 0) {
|
| 239 |
|
|
print "Pulled $start_record records out of $estimated_max_from_service ".
|
| 240 |
|
|
"estimated total records.\n";
|
| 241 |
|
|
}
|
| 242 |
|
|
|
| 243 |
|
|
}
|
| 244 |
|
|
|
| 245 |
|
|
?>
|
