| Line 37... |
Line 37... |
| 37 |
*/
|
37 |
*/
|
| 38 |
public function set($name,$value) {
|
38 |
public function set($name,$value) {
|
| 39 |
$this->$name = $value;
|
39 |
$this->$name = $value;
|
| 40 |
}
|
40 |
}
|
| Line 41... |
Line -... |
| 41 |
|
- |
|
| 42 |
|
41 |
|
| 43 |
/**
|
42 |
/**
|
| 44 |
* Reduce Spaces
|
43 |
* Reduce Spaces
|
| 45 |
* This will reduce the string to only allow once space between characters
|
44 |
* This will reduce the string to only allow once space between characters
|
| 46 |
* @param string $str : string to reduce space
|
45 |
* @param string $str : string to reduce space
|
| Line 52... |
Line 51... |
| 52 |
$str = trim( $str );
|
51 |
$str = trim( $str );
|
| Line 53... |
Line 52... |
| 53 |
|
52 |
|
| 54 |
return( $str );
|
53 |
return( $str );
|
| Line 55... |
Line -... |
| 55 |
}
|
- |
|
| 56 |
|
54 |
}
|
| 57 |
|
55 |
|
| 58 |
/**
|
56 |
/**
|
| 59 |
* Function: parse_auth
|
57 |
* Function: parse_auth
|
| 60 |
* Purpose: Produce a parsed version of authority of a taxon name
|
58 |
* Purpose: Produce a parsed version of authority of a taxon name
|
| Line 142... |
Line 140... |
| 142 |
|
140 |
|
| Line 143... |
Line 141... |
| 143 |
$this->debug['parse_auth'][] = "6 (temp:$temp)";
|
141 |
$this->debug['parse_auth'][] = "6 (temp:$temp)";
|
| Line 144... |
Line 142... |
| 144 |
|
142 |
|
| 145 |
foreach( explode(' ', $temp) as $this_word ) {
|
- |
|
| - |
|
143 |
foreach( explode(' ', $temp) as $this_word ) {
|
| 146 |
|
144 |
|
| 147 |
$this->debug['parse_auth'][] = "7 (this_word:$this_word)";
|
145 |
//$this->debug['parse_auth'][] = "7 (this_word:$this_word)";
|
| 148 |
|
146 |
$elapsed_chars = '';
|
| 149 |
// like '(%'
|
147 |
// like '(%'
|
| 150 |
if ( preg_match('/^\(/', $this_word) ) {
|
148 |
if ( preg_match('/^\(/', $this_word) ) {
|
| 151 |
$elapsed_chars .= '(';
|
149 |
$elapsed_chars .= '(';
|
| Line 152... |
Line 150... |
| 152 |
$this_word = substr( $this_word, 1 );
|
150 |
$this_word = substr( $this_word, 1 );
|
| 153 |
$this->debug['parse_auth'][] = "7a (this_word:$this_word) (elapsed_chars:$elapsed_chars)";
|
151 |
//$this->debug['parse_auth'][] = "7a (this_word:$this_word) (elapsed_chars:$elapsed_chars)";
|
| 154 |
}
|
152 |
}
|
| 155 |
|
153 |
|
| Line 156... |
Line 154... |
| 156 |
// Add back the word to the final translation
|
154 |
// Add back the word to the final translation
|
| Line 157... |
Line 155... |
| 157 |
$elapsed_chars .= $this_word . ' ';
|
155 |
$elapsed_chars .= $this_word . ' ';
|
| Line 199... |
Line 197... |
| 199 |
*/
|
197 |
*/
|
| 200 |
public function parse( $str = NULL ) {
|
198 |
public function parse( $str = NULL ) {
|
| Line 201... |
Line 199... |
| 201 |
|
199 |
|
| Line 202... |
Line -... |
| 202 |
unset($this->debug['parse']);
|
- |
|
| 203 |
|
200 |
unset($this->debug['parse']);
|
| 204 |
|
201 |
|
| 205 |
$temp = '';
|
202 |
$temp = '';
|
| 206 |
$first_str_part = NULL;
|
203 |
$first_str_part = NULL;
|
| 207 |
$second_str_part = NULL;
|
204 |
$second_str_part = NULL;
|
| 208 |
$temp_genus = '';
|
205 |
$temp_genus = '';
|
| 209 |
$temp_species = '';
|
206 |
$temp_species = '';
|
| 210 |
$temp_genus_species = '';
|
207 |
$temp_genus_species = '';
|
| Line 211... |
Line 208... |
| 211 |
$temp_authority = '';
|
208 |
$temp_authority = '';
|
| Line 212... |
Line 209... |
| 212 |
$temp_infra = '';
|
209 |
$temp_infra = '';
|
| 213 |
|
210 |
|
| 214 |
$this->debug['parse'][] = "1";
|
211 |
//$this->debug['parse'][] = "1";
|
| 215 |
|
212 |
|
| 216 |
if ( ($str == NULL) || ( trim($str) == '') ) {
|
213 |
if ( ($str == NULL) || ( trim($str) == '') ) {
|
| 217 |
$this->debug[] = "N1a<br>";
|
214 |
//$this->debug[] = "N1a<br>";
|
| 218 |
return '';
|
215 |
return '';
|
| 219 |
} else {
|
216 |
} else {
|
| Line 220... |
Line 217... |
| 220 |
// trim any leading, trailing spaces or line feeds
|
217 |
// trim any leading, trailing spaces or line feeds
|
| 221 |
$temp = trim( $str );
|
218 |
$temp = trim( $str );
|
| 222 |
$this->debug['parse'][] = "1b";
|
219 |
//$this->debug['parse'][] = "1b";
|
| 223 |
}
|
220 |
}
|
| 224 |
|
221 |
|
| Line 225... |
Line 222... |
| 225 |
if ( $temp == NULL || $temp == '') {
|
222 |
if ( $temp == NULL || $temp == '') {
|
| 226 |
$this->debug['parse'][] = "2a";
|
223 |
//$this->debug['parse'][] = "2a";
|
| 227 |
return '';
|
224 |
return '';
|
| Line 228... |
Line 225... |
| 228 |
} else {
|
225 |
} else {
|
| Line 229... |
Line 226... |
| 229 |
$this->debug['parse'][] = "2b";
|
226 |
//$this->debug['parse'][] = "2b";
|
| 230 |
|
227 |
|
| 231 |
// replace any HTML ampersands
|
228 |
// replace any HTML ampersands
|
| 232 |
$set = array('%', '&', 'amp;%', 'AMP;%');
|
229 |
$set = array('%', '&', 'amp;%', 'AMP;%');
|
| 233 |
$temp = str_replace( $set, '&', $temp );
|
230 |
$temp = str_replace( $set, '&', $temp );
|
| Line 234... |
Line 231... |
| 234 |
|
231 |
|
| 235 |
$this->debug['parse'][] = "2b1 (temp:$temp)";
|
232 |
//$this->debug['parse'][] = "2b1 (temp:$temp)";
|
| 236 |
|
233 |
|
| 237 |
// remove any content in angle brackets (e.g. html tags - <i>, </i>, etc.)
|
234 |
// remove any content in angle brackets (e.g. html tags - <i>, </i>, etc.)
|
| 238 |
$html_pattern = "(\<(/?[^\>]+)\>)";
|
235 |
$html_pattern = "(\<(/?[^\>]+)\>)";
|
| 239 |
//? This should not just handle html tags but all <*>
|
236 |
//? This should not just handle html tags but all <*>
|
| 240 |
$temp = preg_replace( $html_pattern, '', $temp);
|
237 |
$temp = preg_replace( $html_pattern, '', $temp);
|
| 241 |
$this->debug['parse'][] = "2b2 (temp:$temp)";
|
238 |
//$this->debug['parse'][] = "2b2 (temp:$temp)";
|
| 242 |
|
239 |
|
| Line 243... |
Line 240... |
| 243 |
// if second term (only) is in round brackets, presume it is a subgenus or a comment and remove it
|
240 |
// if second term (only) is in round brackets, presume it is a subgenus or a comment and remove it
|
| 244 |
// examples: Barbatia (Mesocibota) bistrigata (Dunker, 1866) => Barbatia bistrigata (Dunker, 1866)
|
241 |
// examples: Barbatia (Mesocibota) bistrigata (Dunker, 1866) => Barbatia bistrigata (Dunker, 1866)
|
| 245 |
// Barbatia (?) bistrigata (Dunker, 1866) => Barbatia bistrigata (Dunker, 1866)
|
242 |
// Barbatia (?) bistrigata (Dunker, 1866) => Barbatia bistrigata (Dunker, 1866)
|
| 246 |
// (obviously this will not suit genus + author alone, where first part of authorname is in brackets,
|
243 |
// (obviously this will not suit genus + author alone, where first part of authorname is in brackets,
|
| 247 |
// however this is very rare?? and in any case we are not supporting genus+authority in this version)
|
244 |
// however this is very rare?? and in any case we are not supporting genus+authority in this version)
|
| 248 |
//if ( $temp like '% (%)%'
|
245 |
//if ( $temp like '% (%)%'
|
| Line 249... |
Line 246... |
| 249 |
$temp = preg_replace( "/ \(\w*\W*\)/", '', $temp, 1 );
|
246 |
$temp = preg_replace( "/ \(\w*\W*\)/", '', $temp, 1 );
|
| 250 |
//? Not sure if this will catch if
|
247 |
//? Not sure if this will catch if
|
| 251 |
$this->debug['parse'][] = "2b3 (temp:$temp)";
|
248 |
//$this->debug['parse'][] = "2b3 (temp:$temp)";
|
| 252 |
|
249 |
|
| 253 |
// if second term (only) is in square brackets, presume it is a comment and remove it
|
250 |
// if second term (only) is in square brackets, presume it is a comment and remove it
|
| 254 |
// example: Aphis [?] ficus Theobald, [1918] => Aphis ficus Theobald, [1918]
|
251 |
// example: Aphis [?] ficus Theobald, [1918] => Aphis ficus Theobald, [1918]
|
| 255 |
//if ( $temp like '% [%]%'
|
252 |
//if ( $temp like '% [%]%'
|
| 256 |
$temp = preg_replace( "/ \[\w*\W*\]/", '', $temp, 1 );
|
253 |
$temp = preg_replace( "/ \[\w*\W*\]/", '', $temp, 1 );
|
| Line 257... |
Line 254... |
| 257 |
//? Not sure if this will catch if
|
254 |
//? Not sure if this will catch if
|
| Line 258... |
Line 255... |
| 258 |
$this->debug['parse'][] = "2b4 (temp:$temp)";
|
255 |
//$this->debug['parse'][] = "2b4 (temp:$temp)";
|
| 259 |
|
256 |
|
| Line 260... |
Line 257... |
| 260 |
// drop indicators of questionable id's - presume all are lowercase for now (could extend as needed)
|
257 |
// drop indicators of questionable id's - presume all are lowercase for now (could extend as needed)
|
| Line 261... |
Line 258... |
| 261 |
$temp = preg_replace( "/ cf /", " ", $temp );
|
258 |
$temp = preg_replace( "/ cf /", " ", $temp );
|
| 262 |
$temp = preg_replace( "/ cf\. /", " ", $temp );
|
259 |
$temp = preg_replace( "/ cf\. /", " ", $temp );
|
| 263 |
$temp = preg_replace( "/ near /", " ", $temp );
|
260 |
$temp = preg_replace( "/ near /", " ", $temp );
|
| 264 |
$temp = preg_replace( "/ aff\. /", " ", $temp );
|
261 |
$temp = preg_replace( "/ aff\. /", " ", $temp );
|
| Line 283... |
Line 280... |
| 283 |
} else {
|
280 |
} else {
|
| 284 |
$temp_genus = $temp;
|
281 |
$temp_genus = $temp;
|
| 285 |
$temp = '';
|
282 |
$temp = '';
|
| 286 |
}
|
283 |
}
|
| Line 287... |
Line 284... |
| 287 |
|
284 |
|
| Line 288... |
Line 285... |
| 288 |
$this->debug['parse'][] = "2b7 (temp_genus:$temp_genus) (temp:$temp)";
|
285 |
//$this->debug['parse'][] = "2b7 (temp_genus:$temp_genus) (temp:$temp)";
|
| 289 |
|
286 |
|
| 290 |
// look for species epithet and authority
|
287 |
// look for species epithet and authority
|
| 291 |
$ar = explode( " ", $temp, 2);
|
288 |
$ar = explode( " ", $temp, 2);
|
| Line 303... |
Line 300... |
| 303 |
$temp_authority = preg_replace( "/ssp./", "subsp.", $temp_authority);
|
300 |
$temp_authority = preg_replace( "/ssp./", "subsp.", $temp_authority);
|
| 304 |
$temp_authority = preg_replace( "/ssp /", "subsp.", $temp_authority);
|
301 |
$temp_authority = preg_replace( "/ssp /", "subsp.", $temp_authority);
|
| 305 |
$temp_authority = preg_replace( "/subsp /", "subsp.", $temp_authority);
|
302 |
$temp_authority = preg_replace( "/subsp /", "subsp.", $temp_authority);
|
| 306 |
$temp_authority = preg_replace( "/var /", "var.", $temp_authority);
|
303 |
$temp_authority = preg_replace( "/var /", "var.", $temp_authority);
|
| Line -... |
Line 304... |
| - |
|
304 |
|
| - |
|
305 |
$temp_infra_authority = '';
|
| 307 |
|
306 |
$temp_infra_type = '';
|
| 308 |
foreach ($infras as $infra) {
|
307 |
foreach ($infras as $infra) {
|
| 309 |
$pos = strpos($temp_authority, $infra);
|
308 |
$pos = strpos($temp_authority, $infra);
|
| 310 |
if ($pos === false) {
|
309 |
if ($pos === false) {
|
| 311 |
continue;
|
- |
|
| 312 |
}
|
310 |
continue;
|
| 313 |
else {
|
311 |
} else {
|
| 314 |
$temp_infra=substr($temp_authority,$pos+strlen($infra));
|
312 |
$temp_infra=substr($temp_authority,$pos+strlen($infra));
|
| 315 |
$temp_authority=substr($temp_authority,0,$pos);
|
313 |
$temp_authority=substr($temp_authority,0,$pos);
|
| 316 |
$temp_infra=trim($temp_infra);
|
314 |
$temp_infra=trim($temp_infra);
|
| 317 |
$temp_infra_type=$infra;
|
315 |
$temp_infra_type=$infra;
|
| Line 323... |
Line 321... |
| 323 |
}
|
321 |
}
|
| 324 |
break; // on s'arrete au premier trouve
|
322 |
break; // on s'arrete au premier trouve
|
| 325 |
}
|
323 |
}
|
| 326 |
}
|
324 |
}
|
| Line 327... |
Line 325... |
| 327 |
|
325 |
|
| 328 |
$this->debug['parse'][] = "2b8 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
- |
|
| Line 329... |
Line 326... |
| 329 |
|
326 |
//$this->debug['parse'][] = "2b8 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
| 330 |
|
327 |
|
| 331 |
// replace selected ligatures here (Genus names can contain Æ, OE ligature)
|
328 |
// replace selected ligatures here (Genus names can contain Æ, OE ligature)
|
| 332 |
$temp_genus = str_replace( 'Æ', 'AE', $temp_genus);
|
329 |
$temp_genus = str_replace( 'Æ', 'AE', $temp_genus);
|
| Line 333... |
Line -... |
| 333 |
$temp_species = str_replace( 'Æ', 'AE', $temp_species);
|
- |
|
| 334 |
$temp_infra = str_replace( 'Æ', 'AE', $temp_infra );
|
330 |
$temp_species = str_replace( 'Æ', 'AE', $temp_species);
|
| Line 335... |
Line 331... |
| 335 |
|
331 |
$temp_infra = str_replace( 'Æ', 'AE', $temp_infra );
|
| 336 |
|
332 |
|
| 337 |
$this->debug['parse'][] = "2b9 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
333 |
//$this->debug['parse'][] = "2b9 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
| Line 338... |
Line 334... |
| 338 |
|
334 |
|
| 339 |
$temp_genus= trim($temp_genus);
|
335 |
$temp_genus= trim($temp_genus);
|
| 340 |
$temp_species= trim($temp_species);
|
336 |
$temp_species= trim($temp_species);
|
| 341 |
$temp_infra= trim($temp_infra );
|
337 |
$temp_infra= trim($temp_infra );
|
| Line 342... |
Line 338... |
| 342 |
|
338 |
|
| Line 343... |
Line 339... |
| 343 |
// reduce any new multiple internal spaces to single space, if present
|
339 |
// reduce any new multiple internal spaces to single space, if present
|
| 344 |
$temp_genus= $this->reduce_spaces( $temp_genus );
|
340 |
$temp_genus= $this->reduce_spaces( $temp_genus );
|
| 345 |
$temp_species= $this->reduce_spaces( $temp_species );
|
341 |
$temp_species= $this->reduce_spaces( $temp_species );
|
| Line 346... |
Line 342... |
| 346 |
$temp_infra= $this->reduce_spaces( $temp_infra );
|
342 |
$temp_infra= $this->reduce_spaces( $temp_infra );
|
| 347 |
|
343 |
|
| 348 |
$this->debug['parse'][] = "2b10 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
344 |
//$this->debug['parse'][] = "2b10 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
| 349 |
|
- |
|
| 350 |
if (isset($temp_authority) && ($temp_authority!='') ) {
|
- |
|
| 351 |
$temp_authority=$this->parse_auth($temp_authority);
|
345 |
|
| 352 |
}
|
- |
|
| 353 |
|
346 |
if (isset($temp_authority) && ($temp_authority!='') ) {
|
| 354 |
if (isset($temp_infra_authority) && ($temp_infra_authority!='') ) {
|
- |
|
| 355 |
$temp_infra_authority=$this->parse_auth($temp_infra_authority);
|
347 |
$temp_authority=$this->parse_auth($temp_authority);
|
| 356 |
}
|
- |
|
| 357 |
|
348 |
}
|
| 358 |
|
- |
|
| 359 |
$this->debug['parse'][] = "2b11 (temp_genus:$temp_genus) (temp_species:$temp_species) (temp_authority:$temp_authority) (temp_infra:$temp_infra) (temp_infra_authority:$temp_infra_authority) (temp:$temp)";
|
- |
|
| 360 |
|
- |
|
| 361 |
return array("genus"=>$temp_genus, "species"=>$temp_species, "authority"=>$temp_authority, "infra"=>$temp_infra, "infra_authority"=>$temp_infra_authority, "infra_type"=>$temp_infra_type);
|
349 |
|
| 362 |
|
- |
|
| 363 |
}
|
350 |
if (isset($temp_infra_authority) && ($temp_infra_authority!='') ) {
|