annotate sites/all/modules/biblio/includes/Parser.php @ 4:ce11bbd8f642

added modules
author danieleb <danielebarchiesi@me.com>
date Thu, 19 Sep 2013 10:38:44 +0100
parents
children
rev   line source
danielebarchiesi@4 1 <?php
danielebarchiesi@4 2 /**
danielebarchiesi@4 3 * Works with a Name object to parse out the parts of a name.
danielebarchiesi@4 4 *
danielebarchiesi@4 5 * Example usage:
danielebarchiesi@4 6 * $parser = new Parser("John Q. Smith");
danielebarchiesi@4 7 * echo $parser->getLast() . ", " . $parser->getFirst();
danielebarchiesi@4 8 * //returns "Smith, John"
danielebarchiesi@4 9 *
danielebarchiesi@4 10 *
danielebarchiesi@4 11 */
danielebarchiesi@4 12 class HumanNameParser_Parser {
danielebarchiesi@4 13 private $name;
danielebarchiesi@4 14 private $nameParts = array();
danielebarchiesi@4 15 private $leadingInit;
danielebarchiesi@4 16 private $first;
danielebarchiesi@4 17 private $nicknames;
danielebarchiesi@4 18 private $middle;
danielebarchiesi@4 19 private $last;
danielebarchiesi@4 20 private $suffix;
danielebarchiesi@4 21 private $category;
danielebarchiesi@4 22 private $type;
danielebarchiesi@4 23 private $literal;
danielebarchiesi@4 24
danielebarchiesi@4 25 private $suffixes;
danielebarchiesi@4 26 private $prefixes;
danielebarchiesi@4 27
danielebarchiesi@4 28 /*
danielebarchiesi@4 29 * Constructor
danielebarchiesi@4 30 *
danielebarchiesi@4 31 * @param mixed $name Either a name as a string or as a Name object.
danielebarchiesi@4 32 */
danielebarchiesi@4 33 public function __construct($name = NULL)
danielebarchiesi@4 34 {
danielebarchiesi@4 35 $this->suffixes = array('esq','esquire','jr','sr','2','ii','iii','iv');
danielebarchiesi@4 36 $this->prefixes = array('bar','ben','bin','da','dal','de la', 'de', 'del','der','di',
danielebarchiesi@4 37 'ibn','la','le','san','st','ste','van', 'van der', 'van den', 'vel','von');
danielebarchiesi@4 38 $this->setName($name);
danielebarchiesi@4 39 }
danielebarchiesi@4 40
danielebarchiesi@4 41 public function parseName($name = NULL, $category = NULL) {
danielebarchiesi@4 42 $this->literal = 0;
danielebarchiesi@4 43 $this->category = 1;
danielebarchiesi@4 44 $this->type = 1;
danielebarchiesi@4 45 if (is_array($name) && isset($name['name'])) {
danielebarchiesi@4 46 if (isset($name['auth_category']) && !empty($name['auth_category']) && empty($category)) {
danielebarchiesi@4 47 $this->category = $name['auth_category'];
danielebarchiesi@4 48 }
danielebarchiesi@4 49 elseif (!empty($category)) {
danielebarchiesi@4 50 $this->category = $category;
danielebarchiesi@4 51 }
danielebarchiesi@4 52 if (isset($name['auth_type']) && !empty($name['auth_type'])) {
danielebarchiesi@4 53 $this->type = $name['auth_type'];
danielebarchiesi@4 54 }
danielebarchiesi@4 55 $this->nameParts = $name;
danielebarchiesi@4 56 $this->setName($name['name'], $category);
danielebarchiesi@4 57 }
danielebarchiesi@4 58 else {
danielebarchiesi@4 59 $this->nameParts['name'] = $name;
danielebarchiesi@4 60 $this->setName($name, $category);
danielebarchiesi@4 61 }
danielebarchiesi@4 62
danielebarchiesi@4 63 return $this->getArray();
danielebarchiesi@4 64 }
danielebarchiesi@4 65 /**
danielebarchiesi@4 66 * Sets name string and parses it.
danielebarchiesi@4 67 * Takes Name object or a simple string (converts the string into a Name obj),
danielebarchiesi@4 68 * parses and loads its constituant parts.
danielebarchiesi@4 69 *
danielebarchiesi@4 70 * @param mixed $name Either a name as a string or as a Name object.
danielebarchiesi@4 71 */
danielebarchiesi@4 72 public function setName($name = NULL, $category = NULL){
danielebarchiesi@4 73 if ($name) {
danielebarchiesi@4 74 $this->category == $category;
danielebarchiesi@4 75
danielebarchiesi@4 76 if (is_object($name) && get_class($name) == "HumanNameParser_Name") { // this is mostly for testing
danielebarchiesi@4 77 $this->name = $name;
danielebarchiesi@4 78 }
danielebarchiesi@4 79 elseif (is_array($name) && isset($name['name'])) {
danielebarchiesi@4 80 $this->name = new HumanNameParser_Name($name['name']);
danielebarchiesi@4 81 $this->nameParts = $name;
danielebarchiesi@4 82 }
danielebarchiesi@4 83 else {
danielebarchiesi@4 84 $this->name = new HumanNameParser_Name($name);
danielebarchiesi@4 85 }
danielebarchiesi@4 86
danielebarchiesi@4 87 $this->leadingInit = "";
danielebarchiesi@4 88 $this->first = "";
danielebarchiesi@4 89 $this->nicknames = "";
danielebarchiesi@4 90 $this->middle = "";
danielebarchiesi@4 91 $this->last = "";
danielebarchiesi@4 92 $this->suffix = "";
danielebarchiesi@4 93
danielebarchiesi@4 94 if ($this->category == 5 || $this->type == 5) {
danielebarchiesi@4 95 $this->last = $name;
danielebarchiesi@4 96 $this->literal = TRUE;
danielebarchiesi@4 97 }
danielebarchiesi@4 98 else {
danielebarchiesi@4 99 $this->parse();
danielebarchiesi@4 100 }
danielebarchiesi@4 101
danielebarchiesi@4 102 }
danielebarchiesi@4 103 }
danielebarchiesi@4 104
danielebarchiesi@4 105 public function getleadingInit() {
danielebarchiesi@4 106 return $this->leadingInit;
danielebarchiesi@4 107 }
danielebarchiesi@4 108 public function getFirst() {
danielebarchiesi@4 109 return $this->first;
danielebarchiesi@4 110 }
danielebarchiesi@4 111 public function getNicknames() {
danielebarchiesi@4 112 return $this->nicknames;
danielebarchiesi@4 113 }
danielebarchiesi@4 114
danielebarchiesi@4 115 public function getMiddle() {
danielebarchiesi@4 116 return $this->middle;
danielebarchiesi@4 117 }
danielebarchiesi@4 118
danielebarchiesi@4 119 public function getLast() {
danielebarchiesi@4 120 return $this->last;
danielebarchiesi@4 121 }
danielebarchiesi@4 122
danielebarchiesi@4 123 public function getSuffix() {
danielebarchiesi@4 124 return $this->suffix;
danielebarchiesi@4 125 }
danielebarchiesi@4 126 public function getName(){
danielebarchiesi@4 127 return $this->name;
danielebarchiesi@4 128 }
danielebarchiesi@4 129
danielebarchiesi@4 130 /**
danielebarchiesi@4 131 * returns all the parts of the name as an array
danielebarchiesi@4 132 *
danielebarchiesi@4 133 * @param String $arrType pass 'int' to get an integer-indexed array (default is associative)
danielebarchiesi@4 134 * @return array An array of the name-parts
danielebarchiesi@4 135 */
danielebarchiesi@4 136 public function getArray($arrType = 'assoc') {
danielebarchiesi@4 137 $arr = array();
danielebarchiesi@4 138 $arr['prefix'] = $this->leadingInit;
danielebarchiesi@4 139 $arr['firstname'] = $this->first;
danielebarchiesi@4 140 $arr['nicknames'] = $this->nicknames;
danielebarchiesi@4 141 $arr['initials'] = substr($this->middle, 0, 10);
danielebarchiesi@4 142 $arr['lastname'] = $this->last;
danielebarchiesi@4 143 $arr['suffix'] = $this->suffix;
danielebarchiesi@4 144 $arr['md5'] = md5(json_encode($arr));
danielebarchiesi@4 145 $arr['literal'] = $this->literal;
danielebarchiesi@4 146
danielebarchiesi@4 147 if ($arrType == 'assoc') {
danielebarchiesi@4 148 return array_merge($this->nameParts, $arr);
danielebarchiesi@4 149 }
danielebarchiesi@4 150 else if ($arrType == 'int'){
danielebarchiesi@4 151 return array_values($arr);
danielebarchiesi@4 152 }
danielebarchiesi@4 153 else {
danielebarchiesi@4 154 throw new Exception("Array must be associative ('assoc') or numeric ('num').");
danielebarchiesi@4 155 }
danielebarchiesi@4 156 }
danielebarchiesi@4 157
danielebarchiesi@4 158 /*
danielebarchiesi@4 159 * Parse the name into its constituent parts.
danielebarchiesi@4 160 *
danielebarchiesi@4 161 * Sequentially captures each name-part, working in from the ends and
danielebarchiesi@4 162 * trimming the namestring as it goes.
danielebarchiesi@4 163 *
danielebarchiesi@4 164 * @return boolean true on success
danielebarchiesi@4 165 */
danielebarchiesi@4 166 private function parse()
danielebarchiesi@4 167 {
danielebarchiesi@4 168 $suffixes = implode("\.*|", $this->suffixes) . "\.*"; // each suffix gets a "\.*" behind it.
danielebarchiesi@4 169 $prefixes = implode(" |", $this->prefixes) . " "; // each prefix gets a " " behind it.
danielebarchiesi@4 170
danielebarchiesi@4 171 // The regex use is a bit tricky. *Everything* matched by the regex will be replaced,
danielebarchiesi@4 172 // but you can select a particular parenthesized submatch to be returned.
danielebarchiesi@4 173 // Also, note that each regex requres that the preceding ones have been run, and matches chopped out.
danielebarchiesi@4 174 $nicknamesRegex = "/ ('|\"|\(\"*'*)(.+?)('|\"|\"*'*\)) /"; // names that starts or end w/ an apostrophe break this
danielebarchiesi@4 175 $suffixRegex = "/,* *($suffixes)$/";
danielebarchiesi@4 176 $lastRegex = "/(?!^)\b([^ ]+ y |$prefixes)*[^ ]+$/";
danielebarchiesi@4 177 $leadingInitRegex = "/^(.\.*)(?= \p{L}{2})/"; // note the lookahead, which isn't returned or replaced
danielebarchiesi@4 178 $firstRegex = "/^[^ ]+/"; //
danielebarchiesi@4 179
danielebarchiesi@4 180 // get nickname, if there is one
danielebarchiesi@4 181 $this->nicknames = $this->name->chopWithRegex($nicknamesRegex, 2);
danielebarchiesi@4 182
danielebarchiesi@4 183 // get suffix, if there is one
danielebarchiesi@4 184 $this->suffix = $this->name->chopWithRegex($suffixRegex, 1);
danielebarchiesi@4 185
danielebarchiesi@4 186 // flip the before-comma and after-comma parts of the name
danielebarchiesi@4 187 $this->name->flip(",");
danielebarchiesi@4 188
danielebarchiesi@4 189 // get the last name
danielebarchiesi@4 190 $this->last = $this->name->chopWithRegex($lastRegex, 0);
danielebarchiesi@4 191 if (!$this->last){
danielebarchiesi@4 192 throw new Exception("Couldn't find a last name in '{$this->name->getStr()}'.");
danielebarchiesi@4 193 }
danielebarchiesi@4 194
danielebarchiesi@4 195 // get the first initial, if there is one
danielebarchiesi@4 196 $this->leadingInit = $this->name->chopWithRegex($leadingInitRegex, 1);
danielebarchiesi@4 197
danielebarchiesi@4 198 // get the first name
danielebarchiesi@4 199 $this->first = $this->name->chopWithRegex($firstRegex, 0);
danielebarchiesi@4 200 if (!$this->first && $this->category != 5){
danielebarchiesi@4 201 throw new Exception("Couldn't find a first name in '{$this->name->getStr()}'");
danielebarchiesi@4 202 }
danielebarchiesi@4 203
danielebarchiesi@4 204 // if anything's left, that's the middle name
danielebarchiesi@4 205 $this->middle = $this->name->getStr();
danielebarchiesi@4 206 return true;
danielebarchiesi@4 207 }
danielebarchiesi@4 208
danielebarchiesi@4 209
danielebarchiesi@4 210
danielebarchiesi@4 211
danielebarchiesi@4 212
danielebarchiesi@4 213 }
danielebarchiesi@4 214 ?>