Parent Directory
| 
Revision Log
| 
Revision Graph
Fully functional links_admin.module. Some new
APIs in links.inc. Removed an extraneous diagnostic
message from links_weblink.module.
This is the initial beta release of links_admin.
| 1 | <?php |
| 2 | /** @file |
| 3 | * $Id: links.inc,v 1.27 2008/12/26 03:53:36 syscrusher Exp $ |
| 4 | * |
| 5 | * links.inc Utility functions for the links modules |
| 6 | * |
| 7 | * Author: Scott Courtney (Drupal user "syscrusher") scott@4th.com |
| 8 | * |
| 9 | * NOTE: This file is co-dependent with links.module. |
| 10 | */ |
| 11 | |
| 12 | /*! @addtogroup links_url URL editing and parsing |
| 13 | * @{ |
| 14 | */ |
| 15 | |
| 16 | /** |
| 17 | * Certain common GET parameters, such as known session tags (e.g., |
| 18 | * PHPSESSID) don't really belong in a URL that is being permanently |
| 19 | * stored. This function strips out these known-undesirable parms. |
| 20 | * |
| 21 | * WARNING: Changing this function, or others that it calls, may cause |
| 22 | * the MD5 sums of URLs in the database to need to be recalculated. If |
| 23 | * you make such changes, be sure to make a call to the function |
| 24 | * _links_recalc_all_url_md5() in the upgrade script. |
| 25 | */ |
| 26 | function links_normalize_url($url) { |
| 27 | // $url = links_remove_session_id($url); |
| 28 | // $url = links_remove_empty_get_parms($url); |
| 29 | // $url = links_check_trailing_slash($url); |
| 30 | // $url = links_sort_get_parms($url); |
| 31 | $parsed = links_parse_url($url); |
| 32 | $url_n = $parsed['normalized']; |
| 33 | return $url_n; |
| 34 | } |
| 35 | |
| 36 | /** |
| 37 | * Remove empty GET parameters from a URL or bare query string. |
| 38 | * For example: |
| 39 | * http://example.com/abc.php?this=&that=55&other= |
| 40 | * becomes |
| 41 | * http://example.com/abc.php?that=55 |
| 42 | * |
| 43 | * TODO: Correctly handle links of the form |
| 44 | * http://example.com/abc.xyz?parm=xyz&parm2=123 |
| 45 | * (character entity encoding of ampersand delimiter) |
| 46 | */ |
| 47 | function links_remove_empty_get_parms($url) { |
| 48 | $url = trim($url); |
| 49 | // The four consecutive calls are ugly as hell, but the |
| 50 | // pregexp misses consecutive empth parms otherwise. For |
| 51 | // some reason, substitutions like \1\2 in the replacement |
| 52 | // expression don't work as one would expect. |
| 53 | $url = preg_replace('/([&?])[^=]+=(&|$)/','\1',$url); |
| 54 | $url = preg_replace('/([&?])[^=]+=(&|$)/','\1',$url); |
| 55 | $url = preg_replace('/([&?])[^=]+=(&|$)/','\1',$url); |
| 56 | $url = preg_replace('/([&?])[^=]+=(&|$)/','\1',$url); |
| 57 | // Special handling for empty parm at beginning of string |
| 58 | $url = preg_replace('/^([^=]+=)(&.*|$)/','\2',$url); |
| 59 | return links_cleanup_parms($url); |
| 60 | } |
| 61 | |
| 62 | /** |
| 63 | * Order the GET parameters for a URL. This helps to detect |
| 64 | * URLs that are the same except for the parameter order. |
| 65 | * For exammple: |
| 66 | * http://example.com/abc.php?xyz=98&def=yes&abc=49 |
| 67 | * becomes: |
| 68 | * http://example.com/abc.php?abc=49&def=yes&xyz=98 |
| 69 | * |
| 70 | * Compare to links_sort_query_string() (which this function calls). |
| 71 | */ |
| 72 | function links_sort_get_parms($url) { |
| 73 | $url = trim($url); |
| 74 | $qmark = strpos($url, '?'); |
| 75 | if ($qmark === FALSE || $qmark == (strlen($url)-1)) { |
| 76 | // No GET parms present |
| 77 | return $url; |
| 78 | } |
| 79 | // Intentionally keep the question mark with this part |
| 80 | $page = substr($url, 0, $qmark+1); |
| 81 | $query = substr($url, $qmark+1); |
| 82 | $url = $page . links_sort_query_string($query); |
| 83 | return $url; |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * Orders the parameters in an HTTP GET query string. This |
| 88 | * helps to detect URLs that are the same except for the |
| 89 | * parameter order. |
| 90 | * For exammple: |
| 91 | * xyz=98&def=yes&abc=49 |
| 92 | * becomes: |
| 93 | * abc=49&def=yes&xyz=98 |
| 94 | * |
| 95 | * Accepts either "&" or the W3C-preferred "&" as a delimiter |
| 96 | * between parameters. |
| 97 | * |
| 98 | * Compare to links_sort_get_parms() (which calls this function). |
| 99 | */ |
| 100 | function links_sort_query_string($query) { |
| 101 | // Handles either "&" or "&" as delimiters, or a mixture. |
| 102 | $parms = preg_split('/(?:&|&)/i', $query); |
| 103 | $p_sort = array(); |
| 104 | foreach ($parms as $parm) { |
| 105 | $eq = strpos($parm, '='); |
| 106 | $key = ($eq === FALSE) ? $parm : substr($parm, 0, $eq); |
| 107 | $val = $parm; |
| 108 | $p_sort[$key] = $val; |
| 109 | } |
| 110 | ksort($p_sort); |
| 111 | // See which delimiter they preferred...believe it or not, |
| 112 | // & is actually the W3C preference, but most don't |
| 113 | // use it. |
| 114 | $delim = strpos($query, '&') === FALSE ? '&' : '&'; |
| 115 | $query2 = implode($delim, $p_sort); |
| 116 | return $query2; |
| 117 | } |
| 118 | |
| 119 | /** |
| 120 | * This function removes known session ID strings from a URL. These are |
| 121 | * transient data from a particular browser on a particular day, and |
| 122 | * definitely do not belong in a links database. The search is not |
| 123 | * case sensitive. This also works on a bare query string rather than |
| 124 | * a full URL, and can be used that way. |
| 125 | */ |
| 126 | function links_remove_session_id($url) { |
| 127 | $url = trim($url); |
| 128 | $url = preg_replace('/(^|[&?]|&)[0-9A-Z_]*SESS(ION|_ID|ID)*=[0-9A-Z_]*/i','\1',$url); |
| 129 | return links_cleanup_parms($url); |
| 130 | } |
| 131 | |
| 132 | /** |
| 133 | * If a link consists only of a protocol and hostname, add a trailing |
| 134 | * slash to explicitly request the default or index document. |
| 135 | * For example, "http://www.example.com" becomes "http://www.example.com/". |
| 136 | * This slightly speeds up page retrieval for the visitor, and also keeps |
| 137 | * from having these non-significant differences reflected in the database. |
| 138 | */ |
| 139 | function links_check_trailing_slash($url) { |
| 140 | $url = trim($url); |
| 141 | if (preg_match('!^[A-Z0-9]+://[^/?]+$!i', $url)) { |
| 142 | $url .= '/'; |
| 143 | } |
| 144 | return $url; |
| 145 | } |
| 146 | |
| 147 | /** |
| 148 | * The preg_replace() calls in the links_remove_...() functions |
| 149 | * may leave some cruft in the URL, such as "&&", "&?", or "?" or "&" as |
| 150 | * the first or last character. This function clears that problem up. It is |
| 151 | * called by the links_remove_...() functions and may also be |
| 152 | * of use to external applications. |
| 153 | */ |
| 154 | function links_cleanup_parms($url) { |
| 155 | $patterns = array('/(\&|\&)(\&|\&)/', '/^(\&|\&)/', '/\?(\&|\&)/', '/(\&|\&|\?)$/', '/(\&|\&)\?/'); |
| 156 | $replaces = array('\1', '', '?', '', '?'); |
| 157 | $url = preg_replace($patterns, $replaces, $url); |
| 158 | return $url; |
| 159 | } |
| 160 | |
| 161 | /** |
| 162 | * Given a link ID or URL, query the database to suggest a title for this link |
| 163 | * as it is being associated with a node. Basically, this will obtain the |
| 164 | * first assigned title (ordered by weight) of any existing node associations |
| 165 | * for this same link. A default value can be specified in case there's no |
| 166 | * good hint in the database. Failing all else, this function will return a |
| 167 | * very stripped down version of the URL itself as a suggested title. |
| 168 | */ |
| 169 | function links_suggest_link_title($link_spec, $default="") { |
| 170 | // We order the query by weight and then link title, excluding rows where title is empty. |
| 171 | $sql = links_get_link_node_query_sql($link_spec, "weight,link_title", FALSE); |
| 172 | $result = db_query_range($sql, 1, 1); |
| 173 | if (db_error()) { |
| 174 | watchdog("error", "links database error on query: $sql"); |
| 175 | return $default; |
| 176 | } |
| 177 | $row = db_fetch_array($result); |
| 178 | if (is_array($row) && !empty($row["link_title"])) { |
| 179 | return $row["link_title"]; |
| 180 | } |
| 181 | // At this point, we didn't find anything really good in the database. |
| 182 | // If the caller specified a default, we'll go with that -- but if not, |
| 183 | // we still have a few tricks up our sleeve (see below). |
| 184 | if (! empty($default)) { |
| 185 | return $default; |
| 186 | } |
| 187 | // Try mangling the URL to make something halfway friendly... |
| 188 | $linkrec = links_get_link($link_spec); |
| 189 | $url = is_array($linkrec) ? $linkrec["url"] : ''; |
| 190 | if (empty($url) && ! is_int($link_spec)) { |
| 191 | // Sigh...couldn't look it up...but can use the link spec itself |
| 192 | $url = $link_spec; |
| 193 | } |
| 194 | if (! empty($url)) { |
| 195 | $url_split = explode('?', links_normalize_url($url)); |
| 196 | $pre_parms = $url_split[0]; |
| 197 | $title = empty($url_split[1]) ? $pre_parms : $pre_parms . "?..."; |
| 198 | // Trim that thing to a reasonable size |
| 199 | if (strlen($title) > 80) { |
| 200 | $title = substr($title, 0, 75) . '...'; |
| 201 | } |
| 202 | } else { |
| 203 | // Method of last resort |
| 204 | $title = t("Web Link: ").$link_spec; |
| 205 | } |
| 206 | return $title; |
| 207 | } |
| 208 | |
| 209 | /** |
| 210 | * Checks the provided URL for valid syntax, and returns an associative array |
| 211 | * to report the findings. The array contains the following elements: |
| 212 | * |
| 213 | * 'url' => The original URL as passed to the function |
| 214 | * 'normalized' => The normalized version of the URL, which is also |
| 215 | * validated after normalization. This will be filled |
| 216 | * even if the source URL is not valid, though in that |
| 217 | * situation its results are undefined. |
| 218 | * 'valid' => Boolean TRUE or FALSE to indicate overall status |
| 219 | * 'errors' => A sub-array containing one or more error messages |
| 220 | * which the caller may optionally provide to the user. |
| 221 | * These have NOT been translated with t() before |
| 222 | * being returned; that is up to the calling application. |
| 223 | * This element will be UNSET if there were no errors. |
| 224 | * 'debug' => A sub-array containing diagnostics about the validation |
| 225 | * process. Used for testing; each element will be a string, |
| 226 | * though these are for programmers and not end users. |
| 227 | * 'scheme' => The URL scheme, if present or implied by type (based |
| 228 | * on the result URL, not the original), without trailing colon |
| 229 | * 'host' => The server name (host and/or domain), also set for |
| 230 | * bare email addresses if appropriate options enabled |
| 231 | * 'port' => Optional port number, if specified |
| 232 | * 'user' => Optional username, if specified, for regular URLS, or |
| 233 | * the local part of email addresses |
| 234 | * 'pass' => Optional password, if specified |
| 235 | * 'path' => An absolute or relative path depending on the URL format |
| 236 | * 'query' => GET parameters from the URL as a single string |
| 237 | * 'fragment' => Anything after a hashmark (#) character from the |
| 238 | * local part of the URL |
| 239 | * 'type' => One of the following, depending on the URL category: |
| 240 | * 'REMOTE' A traditional URL with sheme and hostname |
| 241 | * 'ABSOLUTE' A local absolute path |
| 242 | * 'RELATIVE' A local relative (or Drupal) path |
| 243 | * 'EMAIL' A bare email address or "mailto:" URL |
| 244 | * unset Unknown or invalid URL |
| 245 | * 'md5' => The MD5 hash of the lowercased, normalized URL |
| 246 | * |
| 247 | * The parameters to the function are: |
| 248 | * |
| 249 | * $url The URL to validate |
| 250 | * $accept_bare If TRUE, the function will also accept URLs with no |
| 251 | * scheme (that is, without "http://" etc.) as long as |
| 252 | * they begin with a valid hostname or match an email |
| 253 | * address. In this situation, the returned normalized |
| 254 | * URL will add the scheme as appropriate, including |
| 255 | * "mailto:" preceding an email address. The default scheme |
| 256 | * for anything not beginning with "ftp." is "http", except |
| 257 | * that "mailto" will be the scheme for email addresses. |
| 258 | * $accept_relative If TRUE, the function will accept relative paths (e.g., |
| 259 | * "node/1234"). In this situation, the returned normalized |
| 260 | * URL will not add a scheme or server name. If this is TRUE, |
| 261 | * then $accept_absolute is forced TRUE as well. |
| 262 | * $accept_absolute If TRUE, the function will accept local absolute paths |
| 263 | * (like a Drupal path, only beginning with a slash). in |
| 264 | * this situation, the returned normalized URL will not |
| 265 | * add a scheme or server name. |
| 266 | * |
| 267 | * The PHP parse_url() function is deliberately not used here, because its |
| 268 | * validation is not extensive since it is intended only as a parser and |
| 269 | * not as a validator. The resemblence of this function's return values to |
| 270 | * those of parse_url() is not, however, an accident. |
| 271 | */ |
| 272 | function links_parse_url($url, $accept_bare=TRUE, $accept_relative=TRUE, $accept_absolute=TRUE) { |
| 273 | $result = array( |
| 274 | 'url' => $url, |
| 275 | 'valid' => FALSE, |
| 276 | 'debug' => array(), |
| 277 | ); |
| 278 | $url_n = trim($url); |
| 279 | |
| 280 | // Trivial rejection |
| 281 | if (empty($url_n)) { |
| 282 | $result['errors'] = array('Empty URL'); |
| 283 | return $result; |
| 284 | } |
| 285 | |
| 286 | // URI scheme (protocol spec) |
| 287 | $p_scheme = "(?:([A-Za-z0-9]+)://|(mailto:))"; |
| 288 | // Fully qualified domain name (hostname or server name) |
| 289 | $p_fqdn = "([a-zA-Z0-9]+\.[a-zA-Z0-9-.]*[a-zA-Z0-9])"; |
| 290 | // Port specifier |
| 291 | $p_port = "(?::([0-9]+))"; |
| 292 | // Userid or local_part for email address |
| 293 | $p_user = "([a-zA-Z0-9_\-.%$]+)"; |
| 294 | // Password for URLs of the form "scheme://username:password@server/path" |
| 295 | $p_passwd = "([a-zA-Z0-9%_~#?&=.,;-]+)"; |
| 296 | // Combined userid and password authentication spec (see above) |
| 297 | $p_auth = "(?:(".$p_user."(?::".$p_passwd.")?)@)"; |
| 298 | // Combined specification of authentication credentials, hostname, and port |
| 299 | $p_server = "(".$p_auth."?".$p_fqdn.$p_port."?)"; |
| 300 | // Parameters for GET query string |
| 301 | // $p_query = "(?:\?([a-zA-Z0-9{}@:%_~&=.,/;-]+))"; |
| 302 | $p_query = "(?:\?([^#]*))"; |
| 303 | // Fragment indicator (after "#") |
| 304 | $p_frag = "(?:#([a-zA-Z0-9@:%_?~&=.,/;-]*))"; |
| 305 | // The path catches the leading slash, if present, separately |
| 306 | // so that we can later distinguish between relative and absolute |
| 307 | // $p_path = "(/?)([a-zA-Z0-9@:%_~&=.,/;-]*)"; |
| 308 | $p_path = "(/?)([^?#]*)"; |
| 309 | $p_local = $p_path . $p_query . "?" . $p_frag . "?"; |
| 310 | |
| 311 | // Set some path components based on optional parameters |
| 312 | if ($accept_bare) { |
| 313 | $p_scheme .= "?"; |
| 314 | } |
| 315 | $p_inet = "(".$p_scheme.$p_server.")"; |
| 316 | if ($accept_relative) { |
| 317 | $accept_absolute = TRUE; |
| 318 | } |
| 319 | if ($accept_absolute) { |
| 320 | // The entire remote host spec is optional |
| 321 | $p_inet .= "?"; |
| 322 | } |
| 323 | |
| 324 | // Now try to find a match |
| 325 | $pattern = "!".$p_inet.$p_local."$!i"; |
| 326 | $result['debug'][] = "Matching for regular URL using pattern: ".$pattern; |
| 327 | $matches = array(); |
| 328 | preg_match($pattern, $url_n, $matches); |
| 329 | if (count($matches) > 0) { |
| 330 | if ($matches[3] == "mailto:" |
| 331 | || (empty($matches[2]) && empty($matches[3]) |
| 332 | && !empty($matches[6]) && !empty($matches[8]) |
| 333 | && empty($matches[10]) && empty($matches[11]))) { |
| 334 | $result['scheme'] = "mailto"; |
| 335 | $result['type'] = 'EMAIL'; |
| 336 | } |
| 337 | if (! isset($result['type'])) { |
| 338 | $result['scheme'] = $matches[2]; |
| 339 | if (! empty($result['scheme'])) { |
| 340 | $result['type'] = 'REMOTE'; |
| 341 | $result['path'] = "/" . $matches[11]; |
| 342 | $result['port'] = $matches[9]; |
| 343 | } else if (! empty($matches[8])) { |
| 344 | if (preg_match("!^ftp.!", $matches[8])) { |
| 345 | $result['debug'][] = "Assuming remote FTP protocol for bare URL based on host name"; |
| 346 | $result['scheme'] = 'ftp'; |
| 347 | } else { |
| 348 | $result['debug'][] = "Assuming remote HTTP protocol for bare URL"; |
| 349 | $result['scheme'] = 'http'; |
| 350 | } |
| 351 | $result['type'] = 'REMOTE'; |
| 352 | $result['path'] = "/" . $matches[11]; |
| 353 | $result['port'] = $matches[9]; |
| 354 | } else { |
| 355 | if (!empty($matches[11])) { |
| 356 | if (empty($matches[10])) { |
| 357 | // Local relative path |
| 358 | $result['type'] = 'RELATIVE'; |
| 359 | $result['path'] = $matches[11]; |
| 360 | } else { |
| 361 | // Local absolute path |
| 362 | $result['type'] = 'ABSOLUTE'; |
| 363 | $result['path'] = '/' . $matches[11]; |
| 364 | } |
| 365 | } |
| 366 | } |
| 367 | if (isset($result['type'])) { |
| 368 | $result['pass'] = $matches[7]; |
| 369 | $result['query'] = $matches[12]; |
| 370 | $result['fragment'] = $matches[13]; |
| 371 | } |
| 372 | } |
| 373 | if (isset($result['type'])) { |
| 374 | $result['host'] = $matches[8]; |
| 375 | $result['user'] = $matches[6]; |
| 376 | } |
| 377 | } |
| 378 | if (isset($result['type'])) { |
| 379 | $result['valid'] = TRUE; |
| 380 | // Build the normalized URL |
| 381 | $norm = ''; |
| 382 | switch ($result['scheme']) { |
| 383 | case 'mailto': |
| 384 | $norm .= 'mailto:'; |
| 385 | break; |
| 386 | case '': |
| 387 | break; |
| 388 | default: |
| 389 | $norm .= $result['scheme'] . '://'; |
| 390 | } |
| 391 | $type = $result['type']; |
| 392 | if (! empty($result['user'])) { |
| 393 | $norm .= $result['user']; |
| 394 | if (! empty($result['pass'])) { |
| 395 | $norm .= ':' . $result['pass']; |
| 396 | } |
| 397 | $norm .= '@'; |
| 398 | } |
| 399 | $norm .= $result['host']; |
| 400 | if (! empty($result['port'])) { |
| 401 | $norm .= ':' . $result['port']; |
| 402 | } |
| 403 | $norm .= $result['path']; |
| 404 | if (! empty($result['query'])) { |
| 405 | $result['debug'][] = "Raw query as extracted: ".$result['query']; |
| 406 | $result['query'] = links_remove_session_id($result['query']); |
| 407 | $result['debug'][] = "Query after removing session IDs: ".$result['query']; |
| 408 | $result['query'] = links_remove_empty_get_parms($result['query']); |
| 409 | $result['debug'][] = "Query after removing empty parms: ".$result['query']; |
| 410 | $result['query'] = links_sort_query_string($result['query']); |
| 411 | $result['debug'][] = "Sorted query string: ".$result['query']; |
| 412 | } |
| 413 | if (! empty($result['query'])) { |
| 414 | $norm .= '?' . $result['query']; |
| 415 | } |
| 416 | if (! empty($result['fragment'])) { |
| 417 | $norm .= '#' . $result['fragment']; |
| 418 | } |
| 419 | $result['normalized'] = $norm; |
| 420 | $result['md5'] = md5(strtolower($norm)); |
| 421 | } |
| 422 | return $result; |
| 423 | } |
| 424 | |
| 425 | /*! |
| 426 | * @} |
| 427 | */ |
| 428 | |
| 429 | /*! @addtogroup links_text Text search and edit functions |
| 430 | * @{ |
| 431 | */ |
| 432 | |
| 433 | /** |
| 434 | * Search a text string, potentially multiline, and return an array |
| 435 | * containing all of the URLs found in that string. The array is |
| 436 | * an integer-subscripted array with the subscript reflecting the |
| 437 | * order in which the links were found. |
| 438 | * |
| 439 | * Each element of the main array is an inner array per link. These are |
| 440 | * associative, with the keys being: |
| 441 | * "matched" => the actual text matched in the search, which is |
| 442 | * later to be replaced with a "goto" link. |
| 443 | * "url" => the normalized URL to which this link should go; |
| 444 | * it may not be identical to the URL in the text. |
| 445 | * "http://" is added to bare URLs that don't specify |
| 446 | * a protocol. |
| 447 | * "title" => a nominal title string for the link. For links that |
| 448 | * had an HTML <A> tag around them, this is the text |
| 449 | * inside the tag. For bare links, this is the same |
| 450 | * as the url originally found in the text (the assumption |
| 451 | * being that this is what the author wanted to present). |
| 452 | * |
| 453 | * Some of the pregexps are adapted from the weblinks module, and others |
| 454 | * are adapted from examples on php.net. In all cases, they are complex |
| 455 | * and should be changed with great care and tested thoroughly, as small |
| 456 | * changes here could have large and unintended consequences! |
| 457 | */ |
| 458 | function links_find_links($text) { |
| 459 | $links = array(); |
| 460 | $text = ' ' . $text . ' '; |
| 461 | $patterns = array(); |
| 462 | // Finds links with a protocol spec (adapted from urlfilter.module) |
| 463 | $patterns['bare_url'] = "!(?:<p>|<br>|<br />|[ \n\r\t\(])((?:(?:[A-Za-z0-9]+://)|mailto:)(?:[a-zA-Z0-9@:%_~#?&=.,/;-]*[a-zA-Z0-9@:%_~#&=/;-]))([.,?]?)(?=(</p>|[ \n\r\t\)]))!i"; |
| 464 | // Finds WWW or FTP links without a protocol spec (adapted from urlfilter.module) |
| 465 | $patterns['bare_no_protocol'] = "!(?:<p>|<br>|<br />|[ \n\r\t\(])((?:www|ftp)\.[a-zA-Z0-9@:%_~#?&=.,/;-]*[a-zA-Z0-9@:%_~#\&=/;-])(?:[.,?]?)(?=(</p>|[ \n\r\t\)]))!i"; |
| 466 | // Finds pre-tagged links (pattern adapted from one posted to php.net by "martin at vertikal dot dk") |
| 467 | $patterns['tag'] = "!<a href=(?:\"([^\">]+)\"[^>]*|([^\" >]+?)[^>]*)>(.*)</a>!Uis"; |
| 468 | reset($patterns); |
| 469 | while (list($type, $pattern) = each($patterns)) { |
| 470 | $found = array(); |
| 471 | $count = preg_match_all($pattern, $text, $found); |
| 472 | $howmany = count($found[0]); |
| 473 | switch ($type) { |
| 474 | case 'bare_url': |
| 475 | $found[0] = $found[1]; |
| 476 | $found[2] = $found[1]; |
| 477 | unset($found[3]); |
| 478 | break; |
| 479 | case 'bare_no_protocol': |
| 480 | $found[0] = $found[1]; |
| 481 | $found[2] = $found[1]; |
| 482 | for ($i=0; $i<$howmany; $i++) { |
| 483 | $found[1][$i] = "http://" . $found[1][$i]; |
| 484 | } |
| 485 | break; |
| 486 | case 'tag': |
| 487 | // $found[1] = $found[2]; |
| 488 | $found[2] = $found[3]; |
| 489 | break; |
| 490 | } |
| 491 | // At this point, $found is a nested array where outer subscripts |
| 492 | // are as follows: |
| 493 | // 0 => array of all the text matched in the original item |
| 494 | // 1 => array of all the physical URLs that should be linked |
| 495 | // 2 => array of nominal display text for the links |
| 496 | for ($i=0; $i<$howmany; $i++) { |
| 497 | // Create an associative array for one link |
| 498 | $matched = $found[0][$i]; |
| 499 | $url = links_normalize_url($found[1][$i]); |
| 500 | $title = $found[2][$i]; |
| 501 | // We may have line breaks in the title -- remove them |
| 502 | $title = preg_replace("![\n\r\t]!",' ',$title,-1); |
| 503 | $link = array("matched"=>$matched, "url"=>$url, "title"=>$title); |
| 504 | $links[] = $link; |
| 505 | } |
| 506 | } |
| 507 | return $links; |
| 508 | } |
| 509 | |
| 510 | /*! |
| 511 | * @} |
| 512 | */ |
| 513 | |
| 514 | // ********* FUNCTIONS FOR MANAGING THE URL HASH *************** |
| 515 | |
| 516 | /*! |
| 517 | * @addtogroup links_hash URL hash management |
| 518 | * @{ |
| 519 | */ |
| 520 | |
| 521 | /** |
| 522 | * Given a URL, this function returns an MD5 hash of a case-insensitive, |
| 523 | * normalized version of that URL. The idea of this function is that two |
| 524 | * URLs that differ only in ways that are not significant to a web |
| 525 | * browser will have the same MD5 in the database. See the warning for |
| 526 | * function links_normalize_url() for information concerning |
| 527 | * changes made to this code in new versions. |
| 528 | */ |
| 529 | function links_hash_url($url) { |
| 530 | $url = strtolower(links_normalize_url($url)); |
| 531 | return md5($url); |
| 532 | } |
| 533 | |
| 534 | /*! |
| 535 | * @} |
| 536 | */ |
| 537 | |
| 538 | // ************** FUNCTIONS FOR DATABASE QUERIES AND UPDATES **************** |
| 539 | |
| 540 | /*! |
| 541 | * @addtogroup links_db Link storage and retrieval |
| 542 | * @{ |
| 543 | */ |
| 544 | |
| 545 | /** |
| 546 | * This internal function creates an SQL WHERE clause to select a |
| 547 | * record from the {links} table based on an integer link ID |
| 548 | * or a URL (which is normalized and hashed before the query). The |
| 549 | * returned SQL will be a WHERE clause with leading and trailing blanks |
| 550 | * for convenient concatenation. Note that no query is executed here. |
| 551 | * |
| 552 | * Accepts an MD5 string for the $link_spec, to match directly against |
| 553 | * the url_md5 field in the database. The assumption is that a raw |
| 554 | * hex number exactly 32 characters in length won't be a URL. The hex |
| 555 | * characters A-F are matched case-insensitively. |
| 556 | * |
| 557 | * The $tablename optional parameter defaults to {links} but can |
| 558 | * specify a table alias if this WHERE is to be used in a join query. |
| 559 | * Remember to use the curly-brackets for actual table names, but not |
| 560 | * for alias names, so that database prefixes will work right. |
| 561 | */ |
| 562 | function links_get_link_where_sql($link_spec, $tablename="{links}") { |
| 563 | $where = " WHERE ".$tablename; |
| 564 | // @TODO fix this - it just fails if the hash happens to start with a numeric (DOH) |
| 565 | if (intval($link_spec)) { |
| 566 | $where = $where . ".lid=" . $link_spec . " "; |
| 567 | } else { |
| 568 | if (preg_match('/^[0-9a-f]{32}$/i', $link_spec)) { |
| 569 | $hash = $link_spec; |
| 570 | } else { |
| 571 | $hash = links_hash_url($link_spec); |
| 572 | } |
| 573 | $where = $where . ".url_md5='" . $hash . "' "; |
| 574 | } |
| 575 | return $where; |
| 576 | } |
| 577 | |
| 578 | /** |
| 579 | * Given a URL, URL hash, or link ID (lid), retrieves any existing link record |
| 580 | * from the database as an array. |
| 581 | * |
| 582 | * @param string $link_spec The URL or hash of the URL to be fetched |
| 583 | */ |
| 584 | function links_get_link($link_spec) { |
| 585 | $where = links_get_link_where_sql($link_spec); |
| 586 | $sql = "SELECT * FROM {links}" . $where; |
| 587 | $result = db_query($sql); |
| 588 | if (db_error()) { |
| 589 | watchdog("links","links could not retrieve link matching '".$link_spec."'", array(), WATCHDOG_ERROR); |
| 590 | return NULL; |
| 591 | } |
| 592 | if ($link = db_fetch_array($result)){ |
| 593 | $link['link_title'] = trim($link['link_title']); |
| 594 | $link['url'] = trim($link['url']); |
| 595 | } |
| 596 | return $link; |
| 597 | } |
| 598 | |
| 599 | /** |
| 600 | * This function deletes a link record specified by its lid. |
| 601 | * Before doing this, it invokes the hook 'links_delete_link_references' |
| 602 | * with the specified link ID so that other modules can do what they |
| 603 | * need to do in order to adjust. In some cases, this may actually |
| 604 | * mean the other modules need to delete nodes. |
| 605 | */ |
| 606 | function links_delete_link($lid) { |
| 607 | module_invoke_all('links_delete_link_reference', $lid); |
| 608 | // This query exists in case a module implementor failed to |
| 609 | // implement the hook. It makes sure the table is tidy, |
| 610 | // but provides no opportunity for the module to add |
| 611 | // customized behavior. |
| 612 | $sql = "DELETE FROM {links_node} WHERE lid=%d"; |
| 613 | db_query($sql, $lid); |
| 614 | if (db_error()) { |
| 615 | watchdog("links","Links could not delete link !lid from {links_node} table", array('!lid'=>$lid), WATCHDOG_ERROR); |
| 616 | // Treat this one as non-fatal, so no return here |
| 617 | } else { |
| 618 | watchdog("links","Executed backstop clearing of {links_node} for lid=!lid", array('!lid'=>$lid), WATCHDOG_INFO); |
| 619 | } |
| 620 | // Now delete the actual link. |
| 621 | $sql = "DELETE FROM {links} WHERE lid=%d"; |
| 622 | db_query($sql, $lid); |
| 623 | if (db_error()) { |
| 624 | watchdog("links", "Failed to delete link !lid", array('!lid'=>$lid), WATCHDOG_ERROR); |
| 625 | return FALSE; |
| 626 | } else { |
| 627 | watchdog("links", "Deleted link !lid", array('!lid'=>$lid), WATCHDOG_INFO); |
| 628 | return TRUE; |
| 629 | } |
| 630 | } |
| 631 | |
| 632 | /** |
| 633 | * This low-level function updates an existing link record in place. If the |
| 634 | * URL has changed, it checks for another existing record that already matches |
| 635 | * the *new* URL, and returns that link ID if found. Otherwise, updates the |
| 636 | * URL and/or title of the existing record. |
| 637 | * |
| 638 | * If the existing record is not found, an insert will be attempted. |
| 639 | * |
| 640 | * In any case, if things are successful, the *new* $lid (which may or |
| 641 | * may not be the same as the old one) is returned. The caller should |
| 642 | * then call module_invoke_all('links_update_link_reference', $old_lid, $new_lid) |
| 643 | * to let other modules update any referring tables if needed. |
| 644 | * |
| 645 | * @see links_update_link |
| 646 | */ |
| 647 | function links_update_link_basic($old_lid, $url, $title='') { |
| 648 | if ($old_lid) { |
| 649 | $old = links_get_link($old_lid); |
| 650 | } else { |
| 651 | $old = links_get_link($url); |
| 652 | } |
| 653 | if (! is_array($old)) { |
| 654 | // No extant record. Attempt insert. |
| 655 | return links_put_link($url, $title); |
| 656 | } else { |
| 657 | $title = check_plain(trim($title)); |
| 658 | $old_lid = $old['lid']; |
| 659 | $new_hash = links_hash_url($url); |
| 660 | if ($new_hash == $old['url_md5']) { |
| 661 | if ($title == $old['link_title']) { |
| 662 | // Nothing to do! |
| 663 | return $old_lid; |
| 664 | } |
| 665 | // Safe to update the title in place |
| 666 | $sql = "UPDATE {links} SET link_title='%s' WHERE lid=%d"; |
| 667 | $result = db_query($sql, $title, $old_lid); |
| 668 | if (db_error()) { |
| 669 | drupal_set_message(t('Update of title in existing link record failed.'), 'error'); |
| 670 | watchdog("links","links update for existing link !lid failed", array('!lid'=>$old_lid), WATCHDOG_ERROR); |
| 671 | return 0; |
| 672 | } |
| 673 | return $old_lid; |
| 674 | } else { |
| 675 | // The URL has changed |
| 676 | $extant = links_get_link($url); |
| 677 | if (is_array($extant) && $extant['lid'] != $old_lid) { |
| 678 | // An attempt was made to change this link's URL to match one already in the |
| 679 | // database. What we need to do, therefore, is to return the existing record's |
| 680 | // key. Higher-level code should update references and then delete the old |
| 681 | // record ($old_lid) and keep the new one ($extant['lid']) instead. |
| 682 | return $extant['lid']; |
| 683 | } else { |
| 684 | // It's actually safe to update the URL and the title in place |
| 685 | $sql = "UPDATE {links} SET link_title='%s', url='%s', url_md5='%s' WHERE lid=%d"; |
| 686 | $title = trim($title); |
| 687 | $result = db_query($sql, $title, $url, $new_hash, $old_lid); |
| 688 | if (db_error()) { |
| 689 | drupal_set_message(t('Update of URL and title in existing link record failed.'), 'error'); |
| 690 | watchdog("links","Links update for existing link !lid failed", array('!lid'=>$old_lid), WATCHDOG_ERROR); |
| 691 | return 0; |
| 692 | } else { |
| 693 | watchdog("links","Changed URL for existing link !lid to !url", array('!lid'=>$old_lid, '!url'=>$url), WATCHDOG_INFO); |
| 694 | } |
| 695 | return $old_lid; |
| 696 | } |
| 697 | } |
| 698 | } |
| 699 | } |
| 700 | |
| 701 | /** |
| 702 | * For a specified link ID, checks all existing node references to be sure |
| 703 | * the title is really needed in the local record. If appropriate, the |
| 704 | * {links_node} record is set to use the master record's title. This is |
| 705 | * typically invoked after changing the title in the master record, because |
| 706 | * one or more of the node records may no longer need its title. |
| 707 | */ |
| 708 | function links_check_node_titles($lid) { |
| 709 | $master = links_get_link($lid); |
| 710 | if (! $master['lid']) { |
| 711 | // Don't do anything if no master record. |
| 712 | return; |
| 713 | } |
| 714 | $result = db_query("SELECT * FROM {links_node} WHERE lid=%d", $lid); |
| 715 | while ($row = db_fetch_array($result)) { |
| 716 | // In most cases this become a no-op |
| 717 | links_set_local_title($lid, $row['nid'], $row['module'], $row['link_title']); |
| 718 | } |
| 719 | } |
| 720 | |
| 721 | /** |
| 722 | * This is a higher-level wrapper around links_update_link_basic() that is |
| 723 | * well-behaved with respect to updating references to tables. Anything in |
| 724 | * {links_node} will be taken care of by our own code. Other modules only |
| 725 | * need to implement the hook if they use custom tables. |
| 726 | */ |
| 727 | function links_update_link($old_lid, $url, $title='') { |
| 728 | $new_lid = links_update_link_basic($old_lid, $url, $title); |
| 729 | links_check_node_titles($new_lid); |
| 730 | if ($new_lid && $old_lid && ($new_lid != $old_lid)) { |
| 731 | watchdog("links", "Updated link ID from !old to !new. Calling hook to update references, then old link will be deleted.", array('!old'=>$old_lid, '!new'=>$new_lid), WATCHDOG_INFO); |
| 732 | module_invoke_all('links_update_link_reference', $old_lid, $new_lid); |
| 733 | // Now delete the old link that is no longer used anywhere |
| 734 | links_delete_link($old_lid); |
| 735 | } |
| 736 | } |
| 737 | |
| 738 | /** |
| 739 | * Forces all of the references in {links_node} for the |
| 740 | * specified $lid to use the master title rather than their |
| 741 | * node-local title, if any. |
| 742 | * |
| 743 | * $lid is required; the other two parameters are optional. |
| 744 | */ |
| 745 | function links_force_master_title($lid, $nid=0, $module='') { |
| 746 | if ($nid==0) { |
| 747 | if (empty($module)) { |
| 748 | db_query("UPDATE {links_node} SET link_title='' WHERE lid=%d", $lid); |
| 749 | } else { |
| 750 | db_query("UPDATE {links_node} SET link_title='' WHERE lid=%d AND module='%s'", $lid, $module); |
| 751 | } |
| 752 | } else { |
| 753 | if (empty($module)) { |
| 754 | db_query("UPDATE {links_node} SET link_title='' WHERE lid=%d AND nid=%d", $lid, $nid); |
| 755 | } else { |
| 756 | db_query("UPDATE {links_node} SET link_title='' WHERE lid=%d AND nid=%d AND module='%s'", $lid, $nid, $module); |
| 757 | } |
| 758 | } |
| 759 | if (db_error()) { |
| 760 | watchdog("links", "Failed to force master title for link !lid", array('!lid'=>$lid), WATCHDOG_ERROR); |
| 761 | } else { |
| 762 | watchdog("links", "Forcing master title for link !lid", array('!lid'=>$lid), WATCHDOG_INFO); |
| 763 | } |
| 764 | } |
| 765 | |
| 766 | /** |
| 767 | * Returns the count of how many references exist for a given link in |
| 768 | * {links_node}. If $module is specified, the query is limited to that |
| 769 | * module only. |
| 770 | * |
| 771 | * TODO: Add more robust error handling. |
| 772 | */ |
| 773 | function links_count_nodes($link_spec, $module=NULL) { |
| 774 | $link = links_get_link($link_spec); |
| 775 | if (! is_array($link)) { |
| 776 | return 0; |
| 777 | } |
| 778 | $lid = intval($link['lid']); |
| 779 | if (empty($module)) { |
| 780 | $result = db_query("SELECT COUNT(*) AS n FROM {links_node} WHERE lid=%d", $lid); |
| 781 | } else { |
| 782 | $result = db_query("SELECT COUNT(*) AS n FROM {links_node} WHERE lid=%d AND module='%s'", $lid, $module); |
| 783 | } |
| 784 | $row = db_fetch_array($result); |
| 785 | if (is_array($row)) { |
| 786 | return intval($row['n']); |
| 787 | } else { |
| 788 | return 0; |
| 789 | } |
| 790 | } |
| 791 | |
| 792 | /** |
| 793 | * Implements hook_links_update_link_reference to change $old_lid to |
| 794 | * $new_lid in our own tables. Here we update only {links_node}. |
| 795 | */ |
| 796 | function links_links_update_link_reference($old_lid, $new_lid) { |
| 797 | db_query("UPDATE {links_node} SET lid=%d WHERE lid=%d", $new_lid, $old_lid); |
| 798 | if (db_error()) { |
| 799 | watchdog("links","links update for existing link record failed", array(), WATCHDOG_ERROR); |
| 800 | } |
| 801 | } |
| 802 | |
| 803 | /* |
| 804 | * Gets a specific record from {links_node} |
| 805 | */ |
| 806 | function links_get_node_reference($lid, $nid, $module) { |
| 807 | $result = db_query("SELECT * FROM {links_node} WHERE lid=%d AND nid=%d AND module='%s'", $lid, $nid, $module); |
| 808 | if (db_error()) { |
| 809 | watchdog("links", "database error while retrieving link reference lid=!lid, nid=!nid, module=!module", array('!lid'=>$lid, '!nid'=>$nid, '!module'=>$module), WATCHDOG_ERROR); |
| 810 | return NULL; |
| 811 | } |
| 812 | if ($row = db_fetch_array($result)) { |
| 813 | return $row; |
| 814 | } else { |
| 815 | return array(); |
| 816 | } |
| 817 | } |
| 818 | |
| 819 | /* |
| 820 | * Sets the local title for a specified link reference. If the affected |
| 821 | * node is the ONLY node using the link, this will also update the master |
| 822 | * link if the user has access to do so. Note that this function presumes |
| 823 | * a pre-existing {links_node} record. |
| 824 | */ |
| 825 | function links_set_local_title($lid, $nid, $module, $new_title) { |
| 826 | $link = links_get_link($lid); |
| 827 | if (! is_array($link)) { |
| 828 | watchdog("links", "cannot find master link for !lid while updating title in node reference", array('!lid'=>$lid), WATCHDOG_ERROR); |
| 829 | return; |
| 830 | } |
| 831 | $master_title = $link['link_title']; |
| 832 | $new_title = check_plain(trim($new_title)); |
| 833 | if (empty($new_title) || $new_title == $master_title) { |
| 834 | // An empty local title, or one matching the master title, says to use the master title |
| 835 | links_force_master_title($lid, $nid, $module); |
| 836 | return; |
| 837 | } |
| 838 | // Get the existing local title |
| 839 | $old = links_get_node_reference($lid, $nid, $module); |
| 840 | $old_title = $old['link_title']; |
| 841 | |
| 842 | // If the new and old titles are equal, nothing to do. Otherwise, update |
| 843 | // and set the master title if this is the only referring node. |
| 844 | if ($new_title != $old_title) { |
| 845 | $refcount = links_count_nodes($lid); |
| 846 | if ($refcount > 1 || ! user_access('administer links')) { |
| 847 | // Conventional local update |
| 848 | db_query("UPDATE {links_node} SET link_title='%s' WHERE lid=%d AND nid=%d AND module='%s'", $new_title, $lid, $nid, $module); |
| 849 | } else { |
| 850 | // Update the master record |
| 851 | db_query("UPDATE {links} SET link_title='%s' WHERE lid=%d", $new_title, $lid); |
| 852 | if (! db_error()) { |
| 853 | db_query("UPDATE {links_node} SET link_title='' WHERE lid=%d AND nid=%d AND module='%s'", $lid, $nid, $module); |
| 854 | } |
| 855 | } |
| 856 | if (db_error()) { |
| 857 | watchdog("links", "links update for link !lid, node !nid failed", array('!lid'=>$lid, '!nid'=>$nid), WATCHDOG_ERROR); |
| 858 | } |
| 859 | } |
| 860 | } |
| 861 | |
| 862 | /** |
| 863 | * Implements hook_links_delete_link_reference to remove $lid from |
| 864 | * the {links_node} table for cases where the 'module' colum is |
| 865 | * empty or null (which it shouldn't be, but...). Any module that |
| 866 | * uses {links_node} should implement this hook as well. |
| 867 | */ |
| 868 | function links_links_delete_link_reference($lid) { |
| 869 | // Here we just delete anything for null or empty module, |
| 870 | // which at least in theory shouldn't happen. If you implement |
| 871 | // this hook in your module, put your module's name in the |
| 872 | // appropriate field as a constant. |
| 873 | db_query("DELETE FROM {links_node} WHERE lid=%d AND module=''", $lid); |
| 874 | if (db_error()) { |
| 875 | watchdog("links","links delete references for link !lid failed for null module", array('!lid'=>$lid), WATCHDOG_ERROR); |
| 876 | } else { |
| 877 | watchdog("links","Deleted references for link !lid for null module", array('!lid'=>$lid), WATCHDOG_INFO); |
| 878 | } |
| 879 | } |
| 880 | |
| 881 | /** |
| 882 | * This low-level function simply adds a bare URL record, with no associated |
| 883 | * node, to the database. It returns the new link ID number |
| 884 | * if successful, or 0 if it fails. |
| 885 | * |
| 886 | * The function checks for an existing matching link before adding a new |
| 887 | * record, and if one is found, that links ID is returned instead of a new |
| 888 | * one being assigned. |
| 889 | * |
| 890 | * The optional title field is only used if a new link is being inserted. |
| 891 | */ |
| 892 | function links_put_link($url, $title='') { |
| 893 | $hash = links_hash_url($url); |
| 894 | $sql = "SELECT * FROM {links} WHERE url_md5='%s'"; |
| 895 | $result = db_query($sql,$hash); |
| 896 | if (db_error()) { |
| 897 | // Log an error message and return failure |
| 898 | drupal_set_message(t('Query for existing link record failed.'), 'error'); |
| 899 | watchdog("links","links query for existing link record failed", array(), WATCHDOG_ERROR); |
| 900 | return 0; |
| 901 | } |
| 902 | if ($result) { |
| 903 | $row = db_fetch_array($result); |
| 904 | if ($row["lid"]) { |
| 905 | // Existing record found |
| 906 | return $row["lid"]; |
| 907 | } |
| 908 | } |
| 909 | // Need to create a new record |
| 910 | $url = links_normalize_url($url); |
| 911 | if (empty($title)) { |
| 912 | $title = links_suggest_link_title($url); |
| 913 | } |
| 914 | $sql = "INSERT INTO {links} (lid, url, url_md5, link_title) VALUES (%d, '%s', '%s', '%s')"; |
| 915 | $result = db_query($sql, $lid, $url, $hash, $title); |
| 916 | $lid = db_last_insert_id('{links}','lid'); |
| 917 | if (db_error()) { |
| 918 | // Failed to add record -- log and fail |
| 919 | watchdog("error","Unable to insert URL '".$url."' into links as ID $lid"); |
| 920 | return 0; |
| 921 | } |
| 922 | // Success! |
| 923 | return $lid; |
| 924 | } |
| 925 | |
| 926 | /*! |
| 927 | * @} |
| 928 | * |
| 929 | * @addtogroup links_node Link-Node associations |
| 930 | * @{ |
| 931 | */ |
| 932 | |
| 933 | /** |
| 934 | * Given a link ID, URL, or URL hash, creates an SQL query to obtain brief node data for |
| 935 | * nodes that share that link's record. Columns returned from this query include: |
| 936 | * lid The link ID number (integer) |
| 937 | * url The actual (full) URL from the link record |
| 938 | * nid The node ID number (integer) |
| 939 | * node_title The node title (string) |
| 940 | * weight The weight of this link-node pairing, which determines the |
| 941 | * order when links are listed together. |
| 942 | * clicks The number of times a browser has followed this particular |
| 943 | * instance of this particular link-node association. |
| 944 | * link_title The title given to the link (string); this is the text that |
| 945 | * is displayed for the user as the title of the link. How and |
| 946 | * where it appears (on the page, mouse over, not at all...) |
| 947 | * depends on the settings of user and admin preferences. |
| 948 | * module If a specific module wants to own the link-node association, |
| 949 | * insert that module's name here. The default is 'links', which |
| 950 | * represents the entire links package collectively. |
| 951 | * |
| 952 | * Note that lid and url fields are the same on all returned rows; they are |
| 953 | * here for convenience. |
| 954 | * |
| 955 | * The optional parameter $order determines whether the outer array is |
| 956 | * ordered by the link's weight, by link_title, or by node_title. If |
| 957 | * $order is an empty string (not the default!), no ORDER BY will be added; this |
| 958 | * allows the caller more control, such as appending a WHERE clause or doing |
| 959 | * a compound ordering on multiple columns. The default orders by weight and |
| 960 | * then by link_title. |
| 961 | * |
| 962 | * Left joins are intentionally used in this query, with the deliberate possibility |
| 963 | * to return rows with NULL values in cases of (1) a link with no associated nodes, |
| 964 | * or (2) a link associated with node IDs that no longer exist. This is useful |
| 965 | * to find orphan records. To suppress this feature, set the optional parameter |
| 966 | * $orphans to FALSE (it defaults to TRUE). |
| 967 | * |
| 968 | * Note that this function returns an SQL statement but does not actually |
| 969 | * execute the query. |
| 970 | */ |
| 971 | function links_get_link_node_query_sql($link_spec, $order="weight,link_title", $orphans=TRUE, $module='links') { |
| 972 | $sql = "SELECT l.lid, l.url, ln.nid, n.title AS node_title, ln.weight, ln.clicks, ln.link_title AS link_title "; |
| 973 | $sql .= "FROM {links} AS l LEFT JOIN {links_node} AS ln ON l.lid=ln.nid "; |
| 974 | $sql .= "LEFT JOIN {node} AS n ON ln.nid=n.nid "; |
| 975 | $sql .= links_get_link_where_sql($link_spec, "l"); |
| 976 | if (! $orphans) { |
| 977 | $sql .= " AND ln.nid IS NOT NULL AND n.nid IS NOT NULL "; |
| 978 | } |
| 979 | if (! empty($module)) { |
| 980 | $sql .= " AND ln.module='" . $module . "' "; |
| 981 | } |
| 982 | switch ($order) { |
| 983 | case "": |
| 984 | break; |
| 985 | case "node_title": |
| 986 | case "link_title": |
| 987 | $sql .= " ORDER BY " . $order; |
| 988 | break; |
| 989 | default: |
| 990 | $sql .= " ORDER BY weight, link_title"; |
| 991 | } |
| 992 | $sql .= " "; |
| 993 | return $sql; |
| 994 | } |
| 995 | |
| 996 | /** |
| 997 | * Delete all the links for the specified node (object) |
| 998 | * or nid (integer) |
| 999 | */ |
| 1000 | function links_delete_links_for_node($node, $module='links_links') { |
| 1001 | $sql = "DELETE FROM {links_node} WHERE nid=%d"; |
| 1002 | if (! empty($module)) { |
| 1003 | $sql .= " AND module='" . $module . "' "; |
| 1004 | } |
| 1005 | if (is_object($node)) { |
| 1006 | return db_query($sql, $node->nid); |
| 1007 | } else { |
| 1008 | return db_query($sql, $node); |
| 1009 | } |
| 1010 | } |
| 1011 | |
| 1012 | /** |
| 1013 | * Save all the links for the specified node (object). The links to be |
| 1014 | * saved are assumed to exist as sub-arrays inside the main integer- |
| 1015 | * subscripted array named $node->$module, where $module is the |
| 1016 | * parameter that specifies the context for these links. |
| 1017 | * |
| 1018 | * Each link is an associative sub-array containing the fields 'url', |
| 1019 | * 'weight', and 'link_title'. The 'url' field is self-explanatory. |
| 1020 | * 'weight' will determine, when multiple links exist, their display |
| 1021 | * order when the node is rendered. 'link_title' is stored as either |
| 1022 | * the node-local title for the link, or as the global link title if |
| 1023 | * the link is new to the catalog. 'url' is the only required field; |
| 1024 | * the others have defaults. |
| 1025 | * |
| 1026 | * There is no equivalent function to save a single link for a node. If |
| 1027 | * modules require that functionality, they should simply pass an array |
| 1028 | * with only one link sub-array. |
| 1029 | * |
| 1030 | * @param &$node The node for which links are being saved |
| 1031 | * @param $module The module controlling the link(s) |
| 1032 | * @param $global If TRUE, modification of a URL for any of these |
| 1033 | * links will change that link's URL globally. If |
| 1034 | * FALSE (the default), a changed URL results in a |
| 1035 | * separate catalog entry, leaving other nodes' URLs |
| 1036 | * untouched. $global is ignored if the current user |
| 1037 | * lacks the 'change url globally' permission (set in |
| 1038 | * links.module). |
| 1039 | */ |
| 1040 | function links_save_links_for_node(&$node, $module='links_links', $global=FALSE) { |
| 1041 | $oldlinks =& links_load_links_for_node($node->nid, $module, NULL, NULL, 'lid'); |
| 1042 | $links =& $node->$module; |
| 1043 | $all_ok = TRUE; |
| 1044 | // This will track the link IDs that should be kept |
| 1045 | $lids = array(); |
| 1046 | foreach ($links as $i => $link) { |
| 1047 | $url = links_normalize_url($link['url']); |
| 1048 | $link['url'] = $url; |
| 1049 | $title = trim($link['link_title']); |
| 1050 | $link['link_title'] = $title; |
| 1051 | if (empty($url) || $link['delete']) { |
| 1052 | // Ignore empty URL, either from unused form fields or from |
| 1053 | // blanked-out existing URL field. In the latter case, this |
| 1054 | // link will be removed from the node later in this function. |
| 1055 | // Same action if the "delete" flag (boolean) is set in the |
| 1056 | // array (this normally would happen from a UI form). |
| 1057 | continue; |
| 1058 | } |
| 1059 | // Make sure the base link record exists, one way |
| 1060 | // or another. |
| 1061 | $lid = links_put_link($url, $link['link_title']); |
| 1062 | // $base_rec gets the full record from {links} whether |
| 1063 | // it was inserted or just retrieved |
| 1064 | $base_rec = links_get_link($lid); |
| 1065 | if (isset($oldlinks[$lid])) { |
| 1066 | // Old record from this node |
| 1067 | $old_rec = $oldlinks[$lid]; |
| 1068 | } else { |
| 1069 | // Defaults for some fields |
| 1070 | $old_rec = $base_rec; |
| 1071 | } |
| 1072 | if (is_array($base_rec) && $base_rec['lid']) { |
| 1073 | $link['lid'] = $old_rec['lid']; |
| 1074 | $lids[] = $old_rec['lid']; |
| 1075 | $new_title = trim($link['link_title']); |
| 1076 | $master_title = trim($base_rec['link_title']); |
| 1077 | if (empty($new_title)) { |
| 1078 | $suggested = TRUE; |
| 1079 | $new_title = links_suggest_link_title($base_rec['lid'], ''); |
| 1080 | } else { |
| 1081 | $suggested = FALSE; |
| 1082 | } |
| 1083 | // If we are first to title an existing link, make that |
| 1084 | // update directly before anything else |
| 1085 | if (empty($master_title)) { |
| 1086 | $sql = "UPDATE {links} SET link_title='%s' WHERE lid=%d"; |
| 1087 | $result = db_query($sql, $new_title, $lid); |
| 1088 | $master_title = $new_title; |
| 1089 | } |
| 1090 | // Now do the rest |
| 1091 | if ($new_title == $master_title || $suggested) { |
| 1092 | // Leave the node-specific title empty if we |
| 1093 | // already have it in the master record. Also, |
| 1094 | // an API-suggested title goes *only* in the master |
| 1095 | // record, and if applicable, we did that above. |
| 1096 | $new_title = ''; |
| 1097 | } else { |
| 1098 | if (empty($new_title)) { |
| 1099 | $new_title = $master_title; |
| 1100 | } |
| 1101 | } |
| 1102 | } else { |
| 1103 | drupal_set_message(t('Cannot find or create link record for %url',array('%url'=>$url)),'error'); |
| 1104 | $all_ok = FALSE; |
| 1105 | } |
| 1106 | $weight = intval($link['weight']); |
| 1107 | if (isset($oldlinks[$lid])) { |
| 1108 | $sql = "UPDATE {links_node} SET link_title='%s', weight=%d WHERE lid=%d AND nid=%d AND module='%s'"; |
| 1109 | $op = t('update'); |
| 1110 | } else { |
| 1111 | $sql = "INSERT INTO {links_node} (link_title, weight, lid, nid, clicks, module) VALUES ('%s',%d,%d,%d,0,'%s')"; |
| 1112 | $op = t('insert'); |
| 1113 | } |
| 1114 | $result = db_query($sql, $new_title, $weight, $lid, $node->nid, $module); |
| 1115 | if (! $result) { |
| 1116 | drupal_set_message(t('Unable to %op link "%t" in database.', array('%op'=>$op, '%t'=>$link['link_title'])),'error'); |
| 1117 | $all_ok = FALSE; |
| 1118 | } |
| 1119 | } |
| 1120 | // Now delete any leftover links from before, that weren't in the group we |
| 1121 | // just confirmed should stay |
| 1122 | if (count($lids)) { |
| 1123 | $inlist = implode(',',$lids); |
| 1124 | $sql = "DELETE FROM {links_node} WHERE nid=%d AND lid NOT IN (%s) AND module='%s'"; |
| 1125 | db_query($sql,$node->nid,$inlist, $module); |
| 1126 | } else { |
| 1127 | // There now are NO links for this node |
| 1128 | links_delete_links_for_node($node, $module); |
| 1129 | } |
| 1130 | return $all_ok; |
| 1131 | } |
| 1132 | |
| 1133 | /** |
| 1134 | * Given a node ID, returns all of the links currently defined for that node in |
| 1135 | * the database. The return value is a compound array, with the outer subscript |
| 1136 | * being an integer index and the elements associative arrays with the fields |
| 1137 | * lid, url, url_md5, title, clicks, and weight. |
| 1138 | * |
| 1139 | * For each record returned, there are actually up to *three* title fields: |
| 1140 | * link_link_title the title from the main {links} record |
| 1141 | * node_link_title the title, if any, from the {links_node} record |
| 1142 | * link_title the node_link_title if present, or the link_link_title |
| 1143 | * otherwise. This is normally the field that applications |
| 1144 | * will want to use for display. |
| 1145 | * |
| 1146 | * If a specific link ID is provided, returns ONLY that record. If $firstonly |
| 1147 | * is TRUE, returns only one record. $firstonly is provided as a convenience for |
| 1148 | * handling a "goto" URL pattern from the browser that specifies nid but not lid. |
| 1149 | * This is used, among other things, for legacy compatibility with weblink.module. |
| 1150 | */ |
| 1151 | function links_load_links_for_node($nid, $module='links', $lid=0, $firstonly=FALSE, $indexedby = NULL) { |
| 1152 | $links = array(); |
| 1153 | if ($lid) { |
| 1154 | $sql = "SELECT l.lid, ln.nid, url, url_md5, weight, clicks, module, l.link_title AS link_link_title, ln.link_title AS node_link_title FROM {links_node} ln LEFT JOIN {links} l ON l.lid=ln.lid WHERE ln.nid=%d AND l.lid=%d"; |
| 1155 | if (! empty($module)) { |
| 1156 | $sql .= " AND module='" . $module . "' "; |
| 1157 | } |
| 1158 | $sql .= " ORDER BY weight, l.link_title"; |
| 1159 | $result = db_query($sql, array($nid, $lid)); |
| 1160 | } else { |
| 1161 | $sql = "SELECT l.lid, ln.nid, url, url_md5, weight, clicks, module, l.link_title AS link_link_title, ln.link_title AS node_link_title FROM {links_node} ln LEFT JOIN {links} l ON l.lid=ln.lid WHERE ln.nid=%d"; |
| 1162 | if (! empty($module)) { |
| 1163 | $sql .= " AND module='" . $module . "' "; |
| 1164 | } |
| 1165 | $sql .= " ORDER BY weight, l.link_title"; |
| 1166 | $result = db_query($sql, array($nid)); |
| 1167 | } |
| 1168 | if (db_error($result)) { |
| 1169 | watchdog("error",t("links failed on query ").$sql); |
| 1170 | } else { |
| 1171 | while ($row = db_fetch_array($result)) { |
| 1172 | $row['link_title'] = trim($row['link_title']); |
| 1173 | $row['link_title'] = trim(empty($row['node_link_title']) ? $row['link_link_title'] : $row['node_link_title']); |
| 1174 | if($indexedby){ |
| 1175 | $links[$row[$indexedby]] = $row; |
| 1176 | } |
| 1177 | else { |
| 1178 | $links[] = $row; |
| 1179 | } |
| 1180 | if ($firstonly) { |
| 1181 | break; |
| 1182 | } |
| 1183 | } |
| 1184 | } |
| 1185 | return $links; |
| 1186 | } |
| 1187 | |
| 1188 | /** |
| 1189 | * Given a node ID and an optional link ID, goes to either the specified |
| 1190 | * link ID (with click credit to the appropriate {links_node} record) or |
| 1191 | * to the first [or only] link for the specified node, again with appropriate |
| 1192 | * click credit. |
| 1193 | * |
| 1194 | * It is valid for $nid to be zero, if $lid is nonzero. In that case, the |
| 1195 | * function will redirect to the URL specified by the link, but will not |
| 1196 | * tally the link for a particular node. |
| 1197 | * |
| 1198 | * Unlike other instances of $module, here we default to an empty string to |
| 1199 | * fetch any link regardless of context. Callers can override that behavior, |
| 1200 | * and indeed almost always will want to do so. |
| 1201 | * |
| 1202 | * Returns TRUE if a redirect was actually sent to the browser, FALSE if not. |
| 1203 | */ |
| 1204 | function links_goto_bynode($nid, $lid=0, $module='') { |
| 1205 | if ($nid == 0 && $lid > 0) { |
| 1206 | // Goto a link without crediting any specific node |
| 1207 | $link = array(links_get_link($lid)); |
| 1208 | } else { |
| 1209 | // Look it up by node ID |
| 1210 | $link =& links_load_links_for_node($nid, $module, $lid, TRUE); |
| 1211 | } |
| 1212 | if (! is_array($link)) { |
| 1213 | return FALSE; |
| 1214 | } |
| 1215 | return links_goto_link($link[0], $module); |
| 1216 | } |
| 1217 | |
| 1218 | /** |
| 1219 | * Given a link ID or the URL or its MD5 hash, redirects to that location and tallies one |
| 1220 | * click in the links_node table. |
| 1221 | * |
| 1222 | * If $module is not specified, the redirection will still work as long as |
| 1223 | * the link spec is valid, but the outbound click won't be tallied in the |
| 1224 | * database. $module is needed for tallying because it is legal for the |
| 1225 | * same node to refer to the same link from multiple module contexts. |
| 1226 | * |
| 1227 | * Returns TRUE if a redirect was actually sent to the browser, FALSE if not. |
| 1228 | */ |
| 1229 | function links_goto($link_spec, $module='') { |
| 1230 | $link = links_get_link($link_spec); |
| 1231 | if (! is_array($link)) { |
| 1232 | return FALSE; |
| 1233 | } |
| 1234 | return links_goto_link($link, $module); |
| 1235 | } |
| 1236 | |
| 1237 | /** |
| 1238 | * Given a link record array, redirects to the URL contained in the array. |
| 1239 | * |
| 1240 | * Returns TRUE if a redirect was actually sent to the browser, FALSE if not. |
| 1241 | */ |
| 1242 | function links_goto_link($link, $module='') { |
| 1243 | if ($link['lid']) { |
| 1244 | # We can only tally the click if we know the associated module, because the |
| 1245 | # same link could be re |