@@ -140,1701 +140,1701 @@ |
||
140 | 140 | /// pack 64-bit signed |
141 | 141 | function sphPackI64($v) |
142 | 142 | { |
143 | - assert(is_numeric($v)); |
|
144 | - |
|
145 | - // x64 |
|
146 | - if (PHP_INT_SIZE>=8) |
|
147 | - { |
|
148 | - $v = (int)$v; |
|
149 | - return pack("NN", $v>>32, $v&0xFFFFFFFF); |
|
150 | - } |
|
151 | - |
|
152 | - // x32, int |
|
153 | - if (is_int($v)) |
|
154 | - return pack("NN", $v < 0 ? -1 : 0, $v); |
|
155 | - |
|
156 | - // x32, bcmath |
|
157 | - if (function_exists("bcmul")) |
|
158 | - { |
|
159 | - if (bccomp($v, 0) == -1) |
|
160 | - $v = bcadd("18446744073709551616", $v); |
|
161 | - $h = bcdiv($v, "4294967296", 0); |
|
162 | - $l = bcmod($v, "4294967296"); |
|
163 | - return pack("NN", (float)$h, (float)$l); // conversion to float is intentional; int would lose 31st bit |
|
164 | - } |
|
165 | - |
|
166 | - // x32, no-bcmath |
|
167 | - $p = max(0, strlen($v) - 13); |
|
168 | - $lo = abs((float)substr($v, $p)); |
|
169 | - $hi = abs((float)substr($v, 0, $p)); |
|
170 | - |
|
171 | - $m = $lo + $hi*1316134912.0; // (10 ^ 13) % (1 << 32) = 1316134912 |
|
172 | - $q = floor($m/4294967296.0); |
|
173 | - $l = $m - ($q*4294967296.0); |
|
174 | - $h = $hi*2328.0 + $q; // (10 ^ 13) / (1 << 32) = 2328 |
|
175 | - |
|
176 | - if ($v<0) |
|
177 | - { |
|
178 | - if ($l==0) |
|
179 | - $h = 4294967296.0 - $h; |
|
180 | - else |
|
181 | - { |
|
182 | - $h = 4294967295.0 - $h; |
|
183 | - $l = 4294967296.0 - $l; |
|
184 | - } |
|
185 | - } |
|
186 | - return pack("NN", $h, $l); |
|
143 | + assert(is_numeric($v)); |
|
144 | + |
|
145 | + // x64 |
|
146 | + if (PHP_INT_SIZE>=8) |
|
147 | + { |
|
148 | + $v = (int)$v; |
|
149 | + return pack("NN", $v>>32, $v&0xFFFFFFFF); |
|
150 | + } |
|
151 | + |
|
152 | + // x32, int |
|
153 | + if (is_int($v)) |
|
154 | + return pack("NN", $v < 0 ? -1 : 0, $v); |
|
155 | + |
|
156 | + // x32, bcmath |
|
157 | + if (function_exists("bcmul")) |
|
158 | + { |
|
159 | + if (bccomp($v, 0) == -1) |
|
160 | + $v = bcadd("18446744073709551616", $v); |
|
161 | + $h = bcdiv($v, "4294967296", 0); |
|
162 | + $l = bcmod($v, "4294967296"); |
|
163 | + return pack("NN", (float)$h, (float)$l); // conversion to float is intentional; int would lose 31st bit |
|
164 | + } |
|
165 | + |
|
166 | + // x32, no-bcmath |
|
167 | + $p = max(0, strlen($v) - 13); |
|
168 | + $lo = abs((float)substr($v, $p)); |
|
169 | + $hi = abs((float)substr($v, 0, $p)); |
|
170 | + |
|
171 | + $m = $lo + $hi*1316134912.0; // (10 ^ 13) % (1 << 32) = 1316134912 |
|
172 | + $q = floor($m/4294967296.0); |
|
173 | + $l = $m - ($q*4294967296.0); |
|
174 | + $h = $hi*2328.0 + $q; // (10 ^ 13) / (1 << 32) = 2328 |
|
175 | + |
|
176 | + if ($v<0) |
|
177 | + { |
|
178 | + if ($l==0) |
|
179 | + $h = 4294967296.0 - $h; |
|
180 | + else |
|
181 | + { |
|
182 | + $h = 4294967295.0 - $h; |
|
183 | + $l = 4294967296.0 - $l; |
|
184 | + } |
|
185 | + } |
|
186 | + return pack("NN", $h, $l); |
|
187 | 187 | } |
188 | 188 | |
189 | 189 | /// pack 64-bit unsigned |
190 | 190 | function sphPackU64($v) |
191 | 191 | { |
192 | - assert(is_numeric($v)); |
|
193 | - |
|
194 | - // x64 |
|
195 | - if (PHP_INT_SIZE>=8) |
|
196 | - { |
|
197 | - assert($v>=0); |
|
198 | - |
|
199 | - // x64, int |
|
200 | - if (is_int($v)) |
|
201 | - return pack("NN", $v>>32, $v&0xFFFFFFFF); |
|
202 | - |
|
203 | - // x64, bcmath |
|
204 | - if (function_exists("bcmul")) |
|
205 | - { |
|
206 | - $h = bcdiv($v, 4294967296, 0); |
|
207 | - $l = bcmod($v, 4294967296); |
|
208 | - return pack("NN", $h, $l); |
|
209 | - } |
|
210 | - |
|
211 | - // x64, no-bcmath |
|
212 | - $p = max(0, strlen($v) - 13); |
|
213 | - $lo = (int)substr($v, $p); |
|
214 | - $hi = (int)substr($v, 0, $p); |
|
215 | - |
|
216 | - $m = $lo + $hi*1316134912; |
|
217 | - $l = $m % 4294967296; |
|
218 | - $h = $hi*2328 + (int)($m/4294967296); |
|
219 | - |
|
220 | - return pack("NN", $h, $l); |
|
221 | - } |
|
222 | - |
|
223 | - // x32, int |
|
224 | - if (is_int($v)) |
|
225 | - return pack("NN", 0, $v); |
|
226 | - |
|
227 | - // x32, bcmath |
|
228 | - if (function_exists("bcmul")) |
|
229 | - { |
|
230 | - $h = bcdiv($v, "4294967296", 0); |
|
231 | - $l = bcmod($v, "4294967296"); |
|
232 | - return pack("NN", (float)$h, (float)$l); // conversion to float is intentional; int would lose 31st bit |
|
233 | - } |
|
234 | - |
|
235 | - // x32, no-bcmath |
|
236 | - $p = max(0, strlen($v) - 13); |
|
237 | - $lo = (float)substr($v, $p); |
|
238 | - $hi = (float)substr($v, 0, $p); |
|
239 | - |
|
240 | - $m = $lo + $hi*1316134912.0; |
|
241 | - $q = floor($m / 4294967296.0); |
|
242 | - $l = $m - ($q * 4294967296.0); |
|
243 | - $h = $hi*2328.0 + $q; |
|
244 | - |
|
245 | - return pack("NN", $h, $l); |
|
192 | + assert(is_numeric($v)); |
|
193 | + |
|
194 | + // x64 |
|
195 | + if (PHP_INT_SIZE>=8) |
|
196 | + { |
|
197 | + assert($v>=0); |
|
198 | + |
|
199 | + // x64, int |
|
200 | + if (is_int($v)) |
|
201 | + return pack("NN", $v>>32, $v&0xFFFFFFFF); |
|
202 | + |
|
203 | + // x64, bcmath |
|
204 | + if (function_exists("bcmul")) |
|
205 | + { |
|
206 | + $h = bcdiv($v, 4294967296, 0); |
|
207 | + $l = bcmod($v, 4294967296); |
|
208 | + return pack("NN", $h, $l); |
|
209 | + } |
|
210 | + |
|
211 | + // x64, no-bcmath |
|
212 | + $p = max(0, strlen($v) - 13); |
|
213 | + $lo = (int)substr($v, $p); |
|
214 | + $hi = (int)substr($v, 0, $p); |
|
215 | + |
|
216 | + $m = $lo + $hi*1316134912; |
|
217 | + $l = $m % 4294967296; |
|
218 | + $h = $hi*2328 + (int)($m/4294967296); |
|
219 | + |
|
220 | + return pack("NN", $h, $l); |
|
221 | + } |
|
222 | + |
|
223 | + // x32, int |
|
224 | + if (is_int($v)) |
|
225 | + return pack("NN", 0, $v); |
|
226 | + |
|
227 | + // x32, bcmath |
|
228 | + if (function_exists("bcmul")) |
|
229 | + { |
|
230 | + $h = bcdiv($v, "4294967296", 0); |
|
231 | + $l = bcmod($v, "4294967296"); |
|
232 | + return pack("NN", (float)$h, (float)$l); // conversion to float is intentional; int would lose 31st bit |
|
233 | + } |
|
234 | + |
|
235 | + // x32, no-bcmath |
|
236 | + $p = max(0, strlen($v) - 13); |
|
237 | + $lo = (float)substr($v, $p); |
|
238 | + $hi = (float)substr($v, 0, $p); |
|
239 | + |
|
240 | + $m = $lo + $hi*1316134912.0; |
|
241 | + $q = floor($m / 4294967296.0); |
|
242 | + $l = $m - ($q * 4294967296.0); |
|
243 | + $h = $hi*2328.0 + $q; |
|
244 | + |
|
245 | + return pack("NN", $h, $l); |
|
246 | 246 | } |
247 | 247 | |
248 | 248 | // unpack 64-bit unsigned |
249 | 249 | function sphUnpackU64($v) |
250 | 250 | { |
251 | - list($hi, $lo) = array_values(unpack("N*N*", $v)); |
|
252 | - |
|
253 | - if (PHP_INT_SIZE>=8) |
|
254 | - { |
|
255 | - if ($hi<0) $hi += (1<<32); // because php 5.2.2 to 5.2.5 is totally fucked up again |
|
256 | - if ($lo<0) $lo += (1<<32); |
|
257 | - |
|
258 | - // x64, int |
|
259 | - if ($hi<=2147483647) |
|
260 | - return ($hi<<32) + $lo; |
|
261 | - |
|
262 | - // x64, bcmath |
|
263 | - if (function_exists("bcmul")) |
|
264 | - return bcadd($lo, bcmul($hi, "4294967296")); |
|
265 | - |
|
266 | - // x64, no-bcmath |
|
267 | - $C = 100000; |
|
268 | - $h = ((int)($hi / $C) << 32) + (int)($lo / $C); |
|
269 | - $l = (($hi % $C) << 32) + ($lo % $C); |
|
270 | - if ($l>$C) |
|
271 | - { |
|
272 | - $h += (int)($l / $C); |
|
273 | - $l = $l % $C; |
|
274 | - } |
|
275 | - |
|
276 | - if ($h==0) |
|
277 | - return $l; |
|
278 | - return sprintf("%d%05d", $h, $l); |
|
279 | - } |
|
280 | - |
|
281 | - // x32, int |
|
282 | - if ($hi==0) |
|
283 | - { |
|
284 | - if ($lo>0) |
|
285 | - return $lo; |
|
286 | - return sprintf("%u", $lo); |
|
287 | - } |
|
288 | - |
|
289 | - $hi = sprintf("%u", $hi); |
|
290 | - $lo = sprintf("%u", $lo); |
|
291 | - |
|
292 | - // x32, bcmath |
|
293 | - if (function_exists("bcmul")) |
|
294 | - return bcadd($lo, bcmul($hi, "4294967296")); |
|
295 | - |
|
296 | - // x32, no-bcmath |
|
297 | - $hi = (float)$hi; |
|
298 | - $lo = (float)$lo; |
|
299 | - |
|
300 | - $q = floor($hi/10000000.0); |
|
301 | - $r = $hi - $q*10000000.0; |
|
302 | - $m = $lo + $r*4967296.0; |
|
303 | - $mq = floor($m/10000000.0); |
|
304 | - $l = $m - $mq*10000000.0; |
|
305 | - $h = $q*4294967296.0 + $r*429.0 + $mq; |
|
306 | - |
|
307 | - $h = sprintf("%.0f", $h); |
|
308 | - $l = sprintf("%07.0f", $l); |
|
309 | - if ($h=="0") |
|
310 | - return sprintf( "%.0f", (float)$l); |
|
311 | - return $h . $l; |
|
251 | + list($hi, $lo) = array_values(unpack("N*N*", $v)); |
|
252 | + |
|
253 | + if (PHP_INT_SIZE>=8) |
|
254 | + { |
|
255 | + if ($hi<0) $hi += (1<<32); // because php 5.2.2 to 5.2.5 is totally fucked up again |
|
256 | + if ($lo<0) $lo += (1<<32); |
|
257 | + |
|
258 | + // x64, int |
|
259 | + if ($hi<=2147483647) |
|
260 | + return ($hi<<32) + $lo; |
|
261 | + |
|
262 | + // x64, bcmath |
|
263 | + if (function_exists("bcmul")) |
|
264 | + return bcadd($lo, bcmul($hi, "4294967296")); |
|
265 | + |
|
266 | + // x64, no-bcmath |
|
267 | + $C = 100000; |
|
268 | + $h = ((int)($hi / $C) << 32) + (int)($lo / $C); |
|
269 | + $l = (($hi % $C) << 32) + ($lo % $C); |
|
270 | + if ($l>$C) |
|
271 | + { |
|
272 | + $h += (int)($l / $C); |
|
273 | + $l = $l % $C; |
|
274 | + } |
|
275 | + |
|
276 | + if ($h==0) |
|
277 | + return $l; |
|
278 | + return sprintf("%d%05d", $h, $l); |
|
279 | + } |
|
280 | + |
|
281 | + // x32, int |
|
282 | + if ($hi==0) |
|
283 | + { |
|
284 | + if ($lo>0) |
|
285 | + return $lo; |
|
286 | + return sprintf("%u", $lo); |
|
287 | + } |
|
288 | + |
|
289 | + $hi = sprintf("%u", $hi); |
|
290 | + $lo = sprintf("%u", $lo); |
|
291 | + |
|
292 | + // x32, bcmath |
|
293 | + if (function_exists("bcmul")) |
|
294 | + return bcadd($lo, bcmul($hi, "4294967296")); |
|
295 | + |
|
296 | + // x32, no-bcmath |
|
297 | + $hi = (float)$hi; |
|
298 | + $lo = (float)$lo; |
|
299 | + |
|
300 | + $q = floor($hi/10000000.0); |
|
301 | + $r = $hi - $q*10000000.0; |
|
302 | + $m = $lo + $r*4967296.0; |
|
303 | + $mq = floor($m/10000000.0); |
|
304 | + $l = $m - $mq*10000000.0; |
|
305 | + $h = $q*4294967296.0 + $r*429.0 + $mq; |
|
306 | + |
|
307 | + $h = sprintf("%.0f", $h); |
|
308 | + $l = sprintf("%07.0f", $l); |
|
309 | + if ($h=="0") |
|
310 | + return sprintf( "%.0f", (float)$l); |
|
311 | + return $h . $l; |
|
312 | 312 | } |
313 | 313 | |
314 | 314 | // unpack 64-bit signed |
315 | 315 | function sphUnpackI64($v) |
316 | 316 | { |
317 | - list($hi, $lo) = array_values(unpack("N*N*", $v)); |
|
318 | - |
|
319 | - // x64 |
|
320 | - if (PHP_INT_SIZE>=8) |
|
321 | - { |
|
322 | - if ($hi<0) $hi += (1<<32); // because php 5.2.2 to 5.2.5 is totally fucked up again |
|
323 | - if ($lo<0) $lo += (1<<32); |
|
324 | - |
|
325 | - return ($hi<<32) + $lo; |
|
326 | - } |
|
327 | - |
|
328 | - // x32, int |
|
329 | - if ($hi==0) |
|
330 | - { |
|
331 | - if ($lo>0) |
|
332 | - return $lo; |
|
333 | - return sprintf("%u", $lo); |
|
334 | - } |
|
335 | - // x32, int |
|
336 | - elseif ($hi==-1) |
|
337 | - { |
|
338 | - if ($lo<0) |
|
339 | - return $lo; |
|
340 | - return sprintf("%.0f", $lo - 4294967296.0); |
|
341 | - } |
|
342 | - |
|
343 | - $neg = ""; |
|
344 | - $c = 0; |
|
345 | - if ($hi<0) |
|
346 | - { |
|
347 | - $hi = ~$hi; |
|
348 | - $lo = ~$lo; |
|
349 | - $c = 1; |
|
350 | - $neg = "-"; |
|
351 | - } |
|
352 | - |
|
353 | - $hi = sprintf("%u", $hi); |
|
354 | - $lo = sprintf("%u", $lo); |
|
355 | - |
|
356 | - // x32, bcmath |
|
357 | - if (function_exists("bcmul")) |
|
358 | - return $neg . bcadd(bcadd($lo, bcmul($hi, "4294967296")), $c); |
|
359 | - |
|
360 | - // x32, no-bcmath |
|
361 | - $hi = (float)$hi; |
|
362 | - $lo = (float)$lo; |
|
363 | - |
|
364 | - $q = floor($hi/10000000.0); |
|
365 | - $r = $hi - $q*10000000.0; |
|
366 | - $m = $lo + $r*4967296.0; |
|
367 | - $mq = floor($m/10000000.0); |
|
368 | - $l = $m - $mq*10000000.0 + $c; |
|
369 | - $h = $q*4294967296.0 + $r*429.0 + $mq; |
|
370 | - if ($l==10000000) |
|
371 | - { |
|
372 | - $l = 0; |
|
373 | - $h += 1; |
|
374 | - } |
|
375 | - |
|
376 | - $h = sprintf("%.0f", $h); |
|
377 | - $l = sprintf("%07.0f", $l); |
|
378 | - if ($h=="0") |
|
379 | - return $neg . sprintf( "%.0f", (float)$l); |
|
380 | - return $neg . $h . $l; |
|
317 | + list($hi, $lo) = array_values(unpack("N*N*", $v)); |
|
318 | + |
|
319 | + // x64 |
|
320 | + if (PHP_INT_SIZE>=8) |
|
321 | + { |
|
322 | + if ($hi<0) $hi += (1<<32); // because php 5.2.2 to 5.2.5 is totally fucked up again |
|
323 | + if ($lo<0) $lo += (1<<32); |
|
324 | + |
|
325 | + return ($hi<<32) + $lo; |
|
326 | + } |
|
327 | + |
|
328 | + // x32, int |
|
329 | + if ($hi==0) |
|
330 | + { |
|
331 | + if ($lo>0) |
|
332 | + return $lo; |
|
333 | + return sprintf("%u", $lo); |
|
334 | + } |
|
335 | + // x32, int |
|
336 | + elseif ($hi==-1) |
|
337 | + { |
|
338 | + if ($lo<0) |
|
339 | + return $lo; |
|
340 | + return sprintf("%.0f", $lo - 4294967296.0); |
|
341 | + } |
|
342 | + |
|
343 | + $neg = ""; |
|
344 | + $c = 0; |
|
345 | + if ($hi<0) |
|
346 | + { |
|
347 | + $hi = ~$hi; |
|
348 | + $lo = ~$lo; |
|
349 | + $c = 1; |
|
350 | + $neg = "-"; |
|
351 | + } |
|
352 | + |
|
353 | + $hi = sprintf("%u", $hi); |
|
354 | + $lo = sprintf("%u", $lo); |
|
355 | + |
|
356 | + // x32, bcmath |
|
357 | + if (function_exists("bcmul")) |
|
358 | + return $neg . bcadd(bcadd($lo, bcmul($hi, "4294967296")), $c); |
|
359 | + |
|
360 | + // x32, no-bcmath |
|
361 | + $hi = (float)$hi; |
|
362 | + $lo = (float)$lo; |
|
363 | + |
|
364 | + $q = floor($hi/10000000.0); |
|
365 | + $r = $hi - $q*10000000.0; |
|
366 | + $m = $lo + $r*4967296.0; |
|
367 | + $mq = floor($m/10000000.0); |
|
368 | + $l = $m - $mq*10000000.0 + $c; |
|
369 | + $h = $q*4294967296.0 + $r*429.0 + $mq; |
|
370 | + if ($l==10000000) |
|
371 | + { |
|
372 | + $l = 0; |
|
373 | + $h += 1; |
|
374 | + } |
|
375 | + |
|
376 | + $h = sprintf("%.0f", $h); |
|
377 | + $l = sprintf("%07.0f", $l); |
|
378 | + if ($h=="0") |
|
379 | + return $neg . sprintf( "%.0f", (float)$l); |
|
380 | + return $neg . $h . $l; |
|
381 | 381 | } |
382 | 382 | |
383 | 383 | |
384 | 384 | function sphFixUint($value) |
385 | 385 | { |
386 | - if (PHP_INT_SIZE>=8) |
|
387 | - { |
|
388 | - // x64 route, workaround broken unpack() in 5.2.2+ |
|
389 | - if ($value<0) $value += (1<<32); |
|
390 | - return $value; |
|
391 | - } |
|
392 | - else |
|
393 | - { |
|
394 | - // x32 route, workaround php signed/unsigned braindamage |
|
395 | - return sprintf("%u", $value); |
|
396 | - } |
|
386 | + if (PHP_INT_SIZE>=8) |
|
387 | + { |
|
388 | + // x64 route, workaround broken unpack() in 5.2.2+ |
|
389 | + if ($value<0) $value += (1<<32); |
|
390 | + return $value; |
|
391 | + } |
|
392 | + else |
|
393 | + { |
|
394 | + // x32 route, workaround php signed/unsigned braindamage |
|
395 | + return sprintf("%u", $value); |
|
396 | + } |
|
397 | 397 | } |
398 | 398 | |
399 | 399 | function sphSetBit($flag, $bit, $on) |
400 | 400 | { |
401 | - if ($on) |
|
402 | - { |
|
403 | - $flag |=(1<<$bit); |
|
404 | - } else |
|
405 | - { |
|
406 | - $reset = 16777215 ^(1<<$bit); |
|
407 | - $flag = $flag & $reset; |
|
408 | - } |
|
409 | - return $flag; |
|
401 | + if ($on) |
|
402 | + { |
|
403 | + $flag |=(1<<$bit); |
|
404 | + } else |
|
405 | + { |
|
406 | + $reset = 16777215 ^(1<<$bit); |
|
407 | + $flag = $flag & $reset; |
|
408 | + } |
|
409 | + return $flag; |
|
410 | 410 | } |
411 | 411 | |
412 | 412 | |
413 | 413 | /// sphinx searchd client class |
414 | 414 | class SphinxClient |
415 | 415 | { |
416 | - var $_host; ///< searchd host (default is "localhost") |
|
417 | - var $_port; ///< searchd port (default is 9312) |
|
418 | - var $_offset; ///< how many records to seek from result-set start (default is 0) |
|
419 | - var $_limit; ///< how many records to return from result-set starting at offset (default is 20) |
|
420 | - var $_mode; ///< query matching mode (default is SPH_MATCH_EXTENDED2) |
|
421 | - var $_weights; ///< per-field weights (default is 1 for all fields) |
|
422 | - var $_sort; ///< match sorting mode (default is SPH_SORT_RELEVANCE) |
|
423 | - var $_sortby; ///< attribute to sort by (defualt is "") |
|
424 | - var $_min_id; ///< min ID to match (default is 0, which means no limit) |
|
425 | - var $_max_id; ///< max ID to match (default is 0, which means no limit) |
|
426 | - var $_filters; ///< search filters |
|
427 | - var $_groupby; ///< group-by attribute name |
|
428 | - var $_groupfunc; ///< group-by function (to pre-process group-by attribute value with) |
|
429 | - var $_groupsort; ///< group-by sorting clause (to sort groups in result set with) |
|
430 | - var $_groupdistinct; ///< group-by count-distinct attribute |
|
431 | - var $_maxmatches; ///< max matches to retrieve |
|
432 | - var $_cutoff; ///< cutoff to stop searching at (default is 0) |
|
433 | - var $_retrycount; ///< distributed retries count |
|
434 | - var $_retrydelay; ///< distributed retries delay |
|
435 | - var $_anchor; ///< geographical anchor point |
|
436 | - var $_indexweights; ///< per-index weights |
|
437 | - var $_ranker; ///< ranking mode (default is SPH_RANK_PROXIMITY_BM25) |
|
438 | - var $_rankexpr; ///< ranking mode expression (for SPH_RANK_EXPR) |
|
439 | - var $_maxquerytime; ///< max query time, milliseconds (default is 0, do not limit) |
|
440 | - var $_fieldweights; ///< per-field-name weights |
|
441 | - var $_overrides; ///< per-query attribute values overrides |
|
442 | - var $_select; ///< select-list (attributes or expressions, with optional aliases) |
|
443 | - var $_query_flags; ///< per-query various flags |
|
444 | - var $_predictedtime; ///< per-query max_predicted_time |
|
445 | - var $_outerorderby; ///< outer match sort by |
|
446 | - var $_outeroffset; ///< outer offset |
|
447 | - var $_outerlimit; ///< outer limit |
|
448 | - var $_hasouter; |
|
449 | - |
|
450 | - var $_error; ///< last error message |
|
451 | - var $_warning; ///< last warning message |
|
452 | - var $_connerror; ///< connection error vs remote error flag |
|
453 | - |
|
454 | - var $_reqs; ///< requests array for multi-query |
|
455 | - var $_mbenc; ///< stored mbstring encoding |
|
456 | - var $_arrayresult; ///< whether $result["matches"] should be a hash or an array |
|
457 | - var $_timeout; ///< connect timeout |
|
458 | - |
|
459 | - ///////////////////////////////////////////////////////////////////////////// |
|
460 | - // common stuff |
|
461 | - ///////////////////////////////////////////////////////////////////////////// |
|
462 | - |
|
463 | - /// create a new client object and fill defaults |
|
464 | - function SphinxClient () |
|
465 | - { |
|
466 | - // per-client-object settings |
|
467 | - $this->_host = "localhost"; |
|
468 | - $this->_port = 9312; |
|
469 | - $this->_path = false; |
|
470 | - $this->_socket = false; |
|
471 | - |
|
472 | - // per-query settings |
|
473 | - $this->_offset = 0; |
|
474 | - $this->_limit = 20; |
|
475 | - $this->_mode = SPH_MATCH_EXTENDED2; |
|
476 | - $this->_weights = array (); |
|
477 | - $this->_sort = SPH_SORT_RELEVANCE; |
|
478 | - $this->_sortby = ""; |
|
479 | - $this->_min_id = 0; |
|
480 | - $this->_max_id = 0; |
|
481 | - $this->_filters = array (); |
|
482 | - $this->_groupby = ""; |
|
483 | - $this->_groupfunc = SPH_GROUPBY_DAY; |
|
484 | - $this->_groupsort = "@group desc"; |
|
485 | - $this->_groupdistinct = ""; |
|
486 | - $this->_maxmatches = 1000; |
|
487 | - $this->_cutoff = 0; |
|
488 | - $this->_retrycount = 0; |
|
489 | - $this->_retrydelay = 0; |
|
490 | - $this->_anchor = array (); |
|
491 | - $this->_indexweights = array (); |
|
492 | - $this->_ranker = SPH_RANK_PROXIMITY_BM25; |
|
493 | - $this->_rankexpr = ""; |
|
494 | - $this->_maxquerytime = 0; |
|
495 | - $this->_fieldweights = array(); |
|
496 | - $this->_overrides = array(); |
|
497 | - $this->_select = "*"; |
|
498 | - $this->_query_flags = sphSetBit(0, 6, true); // default idf=tfidf_normalized |
|
499 | - $this->_predictedtime = 0; |
|
500 | - $this->_outerorderby = ""; |
|
501 | - $this->_outeroffset = 0; |
|
502 | - $this->_outerlimit = 0; |
|
503 | - $this->_hasouter = false; |
|
504 | - |
|
505 | - $this->_error = ""; // per-reply fields (for single-query case) |
|
506 | - $this->_warning = ""; |
|
507 | - $this->_connerror = false; |
|
508 | - |
|
509 | - $this->_reqs = array ();// requests storage (for multi-query case) |
|
510 | - $this->_mbenc = ""; |
|
511 | - $this->_arrayresult = false; |
|
512 | - $this->_timeout = 0; |
|
513 | - } |
|
514 | - |
|
515 | - function __destruct() |
|
516 | - { |
|
517 | - if ($this->_socket !== false) |
|
518 | - fclose($this->_socket); |
|
519 | - } |
|
520 | - |
|
521 | - /// get last error message (string) |
|
522 | - function GetLastError () |
|
523 | - { |
|
524 | - return $this->_error; |
|
525 | - } |
|
526 | - |
|
527 | - /// get last warning message (string) |
|
528 | - function GetLastWarning () |
|
529 | - { |
|
530 | - return $this->_warning; |
|
531 | - } |
|
532 | - |
|
533 | - /// get last error flag (to tell network connection errors from searchd errors or broken responses) |
|
534 | - function IsConnectError() |
|
535 | - { |
|
536 | - return $this->_connerror; |
|
537 | - } |
|
538 | - |
|
539 | - /// set searchd host name (string) and port (integer) |
|
540 | - function SetServer($host, $port = 0) |
|
541 | - { |
|
542 | - assert(is_string($host)); |
|
543 | - if ($host[0] == '/') |
|
544 | - { |
|
545 | - $this->_path = 'unix://' . $host; |
|
546 | - return; |
|
547 | - } |
|
548 | - if (substr($host, 0, 7)=="unix://") |
|
549 | - { |
|
550 | - $this->_path = $host; |
|
551 | - return; |
|
552 | - } |
|
553 | - |
|
554 | - $this->_host = $host; |
|
555 | - $port = intval($port); |
|
556 | - assert(0<=$port && $port<65536); |
|
557 | - $this->_port =($port==0) ? 9312 : $port; |
|
558 | - $this->_path = ''; |
|
559 | - } |
|
560 | - |
|
561 | - /// set server connection timeout (0 to remove) |
|
562 | - function SetConnectTimeout($timeout) |
|
563 | - { |
|
564 | - assert(is_numeric($timeout)); |
|
565 | - $this->_timeout = $timeout; |
|
566 | - } |
|
567 | - |
|
568 | - |
|
569 | - function _Send($handle, $data, $length) |
|
570 | - { |
|
571 | - if (feof($handle) || fwrite($handle, $data, $length) !== $length) |
|
572 | - { |
|
573 | - $this->_error = 'connection unexpectedly closed (timed out?)'; |
|
574 | - $this->_connerror = true; |
|
575 | - return false; |
|
576 | - } |
|
577 | - return true; |
|
578 | - } |
|
579 | - |
|
580 | - ///////////////////////////////////////////////////////////////////////////// |
|
581 | - |
|
582 | - /// enter mbstring workaround mode |
|
583 | - function _MBPush () |
|
584 | - { |
|
585 | - $this->_mbenc = ""; |
|
586 | - if (ini_get("mbstring.func_overload") & 2) |
|
587 | - { |
|
588 | - $this->_mbenc = mb_internal_encoding(); |
|
589 | - mb_internal_encoding("latin1"); |
|
590 | - } |
|
591 | - } |
|
592 | - |
|
593 | - /// leave mbstring workaround mode |
|
594 | - function _MBPop () |
|
595 | - { |
|
596 | - if ($this->_mbenc) |
|
597 | - mb_internal_encoding($this->_mbenc); |
|
598 | - } |
|
599 | - |
|
600 | - /// connect to searchd server |
|
601 | - function _Connect () |
|
602 | - { |
|
603 | - if ($this->_socket!==false) |
|
604 | - { |
|
605 | - // we are in persistent connection mode, so we have a socket |
|
606 | - // however, need to check whether it's still alive |
|
607 | - if (!@feof($this->_socket)) |
|
608 | - return $this->_socket; |
|
609 | - |
|
610 | - // force reopen |
|
611 | - $this->_socket = false; |
|
612 | - } |
|
613 | - |
|
614 | - $errno = 0; |
|
615 | - $errstr = ""; |
|
616 | - $this->_connerror = false; |
|
617 | - |
|
618 | - if ($this->_path) |
|
619 | - { |
|
620 | - $host = $this->_path; |
|
621 | - $port = 0; |
|
622 | - } |
|
623 | - else |
|
624 | - { |
|
625 | - $host = $this->_host; |
|
626 | - $port = $this->_port; |
|
627 | - } |
|
628 | - |
|
629 | - if ($this->_timeout<=0) |
|
630 | - $fp = @fsockopen($host, $port, $errno, $errstr); |
|
631 | - else |
|
632 | - $fp = @fsockopen($host, $port, $errno, $errstr, $this->_timeout); |
|
633 | - |
|
634 | - if (!$fp) |
|
635 | - { |
|
636 | - if ($this->_path) |
|
637 | - $location = $this->_path; |
|
638 | - else |
|
639 | - $location = "{$this->_host}:{$this->_port}"; |
|
640 | - |
|
641 | - $errstr = trim($errstr); |
|
642 | - $this->_error = "connection to $location failed (errno=$errno, msg=$errstr)"; |
|
643 | - $this->_connerror = true; |
|
644 | - return false; |
|
645 | - } |
|
646 | - |
|
647 | - // send my version |
|
648 | - // this is a subtle part. we must do it before (!) reading back from searchd. |
|
649 | - // because otherwise under some conditions (reported on FreeBSD for instance) |
|
650 | - // TCP stack could throttle write-write-read pattern because of Nagle. |
|
651 | - if (!$this->_Send($fp, pack("N", 1), 4)) |
|
652 | - { |
|
653 | - fclose($fp); |
|
654 | - $this->_error = "failed to send client protocol version"; |
|
655 | - return false; |
|
656 | - } |
|
657 | - |
|
658 | - // check version |
|
659 | - list(,$v) = unpack("N*", fread($fp, 4)); |
|
660 | - $v = (int)$v; |
|
661 | - if ($v<1) |
|
662 | - { |
|
663 | - fclose($fp); |
|
664 | - $this->_error = "expected searchd protocol version 1+, got version '$v'"; |
|
665 | - return false; |
|
666 | - } |
|
667 | - |
|
668 | - return $fp; |
|
669 | - } |
|
670 | - |
|
671 | - /// get and check response packet from searchd server |
|
672 | - function _GetResponse($fp, $client_ver) |
|
673 | - { |
|
674 | - $response = ""; |
|
675 | - $len = 0; |
|
676 | - |
|
677 | - $header = fread($fp, 8); |
|
678 | - if (strlen($header)==8) |
|
679 | - { |
|
680 | - list($status, $ver, $len) = array_values(unpack("n2a/Nb", $header)); |
|
681 | - $left = $len; |
|
682 | - while ($left>0 && !feof($fp)) |
|
683 | - { |
|
684 | - $chunk = fread($fp, min(8192, $left)); |
|
685 | - if ($chunk) |
|
686 | - { |
|
687 | - $response .= $chunk; |
|
688 | - $left -= strlen($chunk); |
|
689 | - } |
|
690 | - } |
|
691 | - } |
|
692 | - if ($this->_socket === false) |
|
693 | - fclose($fp); |
|
694 | - |
|
695 | - // check response |
|
696 | - $read = strlen($response); |
|
697 | - if (!$response || $read!=$len) |
|
698 | - { |
|
699 | - $this->_error = $len |
|
700 | - ? "failed to read searchd response (status=$status, ver=$ver, len=$len, read=$read)" |
|
701 | - : "received zero-sized searchd response"; |
|
702 | - return false; |
|
703 | - } |
|
704 | - |
|
705 | - // check status |
|
706 | - if ($status==SEARCHD_WARNING) |
|
707 | - { |
|
708 | - list(,$wlen) = unpack("N*", substr($response, 0, 4)); |
|
709 | - $this->_warning = substr($response, 4, $wlen); |
|
710 | - return substr($response, 4+$wlen); |
|
711 | - } |
|
712 | - if ($status==SEARCHD_ERROR) |
|
713 | - { |
|
714 | - $this->_error = "searchd error: " . substr($response, 4); |
|
715 | - return false; |
|
716 | - } |
|
717 | - if ($status==SEARCHD_RETRY) |
|
718 | - { |
|
719 | - $this->_error = "temporary searchd error: " . substr($response, 4); |
|
720 | - return false; |
|
721 | - } |
|
722 | - if ($status!=SEARCHD_OK) |
|
723 | - { |
|
724 | - $this->_error = "unknown status code '$status'"; |
|
725 | - return false; |
|
726 | - } |
|
727 | - |
|
728 | - // check version |
|
729 | - if ($ver<$client_ver) |
|
730 | - { |
|
731 | - $this->_warning = sprintf("searchd command v.%d.%d older than client's v.%d.%d, some options might not work", |
|
732 | - $ver>>8, $ver&0xff, $client_ver>>8, $client_ver&0xff); |
|
733 | - } |
|
734 | - |
|
735 | - return $response; |
|
736 | - } |
|
737 | - |
|
738 | - ///////////////////////////////////////////////////////////////////////////// |
|
739 | - // searching |
|
740 | - ///////////////////////////////////////////////////////////////////////////// |
|
741 | - |
|
742 | - /// set offset and count into result set, |
|
743 | - /// and optionally set max-matches and cutoff limits |
|
744 | - function SetLimits($offset, $limit, $max=0, $cutoff=0) |
|
745 | - { |
|
746 | - assert(is_int($offset)); |
|
747 | - assert(is_int($limit)); |
|
748 | - assert($offset>=0); |
|
749 | - assert($limit>0); |
|
750 | - assert($max>=0); |
|
751 | - $this->_offset = $offset; |
|
752 | - $this->_limit = $limit; |
|
753 | - if ($max>0) |
|
754 | - $this->_maxmatches = $max; |
|
755 | - if ($cutoff>0) |
|
756 | - $this->_cutoff = $cutoff; |
|
757 | - } |
|
758 | - |
|
759 | - /// set maximum query time, in milliseconds, per-index |
|
760 | - /// integer, 0 means "do not limit" |
|
761 | - function SetMaxQueryTime($max) |
|
762 | - { |
|
763 | - assert(is_int($max)); |
|
764 | - assert($max>=0); |
|
765 | - $this->_maxquerytime = $max; |
|
766 | - } |
|
767 | - |
|
768 | - /// set matching mode |
|
769 | - function SetMatchMode($mode) |
|
770 | - { |
|
771 | - trigger_error('DEPRECATED: Do not call this method or, even better, use SphinxQL instead of an API', E_USER_DEPRECATED); |
|
772 | - assert($mode==SPH_MATCH_ALL |
|
773 | - || $mode==SPH_MATCH_ANY |
|
774 | - || $mode==SPH_MATCH_PHRASE |
|
775 | - || $mode==SPH_MATCH_BOOLEAN |
|
776 | - || $mode==SPH_MATCH_EXTENDED |
|
777 | - || $mode==SPH_MATCH_FULLSCAN |
|
778 | - || $mode==SPH_MATCH_EXTENDED2); |
|
779 | - $this->_mode = $mode; |
|
780 | - } |
|
781 | - |
|
782 | - /// set ranking mode |
|
783 | - function SetRankingMode($ranker, $rankexpr="") |
|
784 | - { |
|
785 | - assert($ranker===0 || $ranker>=1 && $ranker<SPH_RANK_TOTAL); |
|
786 | - assert(is_string($rankexpr)); |
|
787 | - $this->_ranker = $ranker; |
|
788 | - $this->_rankexpr = $rankexpr; |
|
789 | - } |
|
790 | - |
|
791 | - /// set matches sorting mode |
|
792 | - function SetSortMode($mode, $sortby="") |
|
793 | - { |
|
794 | - assert ( |
|
795 | - $mode==SPH_SORT_RELEVANCE || |
|
796 | - $mode==SPH_SORT_ATTR_DESC || |
|
797 | - $mode==SPH_SORT_ATTR_ASC || |
|
798 | - $mode==SPH_SORT_TIME_SEGMENTS || |
|
799 | - $mode==SPH_SORT_EXTENDED || |
|
800 | - $mode==SPH_SORT_EXPR); |
|
801 | - assert(is_string($sortby)); |
|
802 | - assert($mode==SPH_SORT_RELEVANCE || strlen($sortby)>0); |
|
803 | - |
|
804 | - $this->_sort = $mode; |
|
805 | - $this->_sortby = $sortby; |
|
806 | - } |
|
807 | - |
|
808 | - /// bind per-field weights by order |
|
809 | - /// DEPRECATED; use SetFieldWeights() instead |
|
810 | - function SetWeights($weights) |
|
811 | - { |
|
812 | - die("This method is now deprecated; please use SetFieldWeights instead"); |
|
813 | - } |
|
814 | - |
|
815 | - /// bind per-field weights by name |
|
816 | - function SetFieldWeights($weights) |
|
817 | - { |
|
818 | - assert(is_array($weights)); |
|
819 | - foreach ($weights as $name=>$weight) |
|
820 | - { |
|
821 | - assert(is_string($name)); |
|
822 | - assert(is_int($weight)); |
|
823 | - } |
|
824 | - $this->_fieldweights = $weights; |
|
825 | - } |
|
826 | - |
|
827 | - /// bind per-index weights by name |
|
828 | - function SetIndexWeights($weights) |
|
829 | - { |
|
830 | - assert(is_array($weights)); |
|
831 | - foreach ($weights as $index=>$weight) |
|
832 | - { |
|
833 | - assert(is_string($index)); |
|
834 | - assert(is_int($weight)); |
|
835 | - } |
|
836 | - $this->_indexweights = $weights; |
|
837 | - } |
|
838 | - |
|
839 | - /// set IDs range to match |
|
840 | - /// only match records if document ID is beetwen $min and $max (inclusive) |
|
841 | - function SetIDRange($min, $max) |
|
842 | - { |
|
843 | - assert(is_numeric($min)); |
|
844 | - assert(is_numeric($max)); |
|
845 | - assert($min<=$max); |
|
846 | - $this->_min_id = $min; |
|
847 | - $this->_max_id = $max; |
|
848 | - } |
|
849 | - |
|
850 | - /// set values set filter |
|
851 | - /// only match records where $attribute value is in given set |
|
852 | - function SetFilter($attribute, $values, $exclude=false) |
|
853 | - { |
|
854 | - assert(is_string($attribute)); |
|
855 | - assert(is_array($values)); |
|
856 | - assert(count($values)); |
|
857 | - |
|
858 | - if (is_array($values) && count($values)) |
|
859 | - { |
|
860 | - foreach ($values as $value) |
|
861 | - assert(is_numeric($value)); |
|
862 | - |
|
863 | - $this->_filters[] = array("type"=>SPH_FILTER_VALUES, "attr"=>$attribute, "exclude"=>$exclude, "values"=>$values); |
|
864 | - } |
|
865 | - } |
|
866 | - |
|
867 | - /// set string filter |
|
868 | - /// only match records where $attribute value is equal |
|
869 | - function SetFilterString($attribute, $value, $exclude=false) |
|
870 | - { |
|
871 | - assert(is_string($attribute)); |
|
872 | - assert(is_string($value)); |
|
873 | - $this->_filters[] = array("type"=>SPH_FILTER_STRING, "attr"=>$attribute, "exclude"=>$exclude, "value"=>$value); |
|
874 | - } |
|
875 | - |
|
876 | - /// set range filter |
|
877 | - /// only match records if $attribute value is beetwen $min and $max (inclusive) |
|
878 | - function SetFilterRange($attribute, $min, $max, $exclude=false) |
|
879 | - { |
|
880 | - assert(is_string($attribute)); |
|
881 | - assert(is_numeric($min)); |
|
882 | - assert(is_numeric($max)); |
|
883 | - assert($min<=$max); |
|
884 | - |
|
885 | - $this->_filters[] = array("type"=>SPH_FILTER_RANGE, "attr"=>$attribute, "exclude"=>$exclude, "min"=>$min, "max"=>$max); |
|
886 | - } |
|
887 | - |
|
888 | - /// set float range filter |
|
889 | - /// only match records if $attribute value is beetwen $min and $max (inclusive) |
|
890 | - function SetFilterFloatRange($attribute, $min, $max, $exclude=false) |
|
891 | - { |
|
892 | - assert(is_string($attribute)); |
|
893 | - assert(is_float($min)); |
|
894 | - assert(is_float($max)); |
|
895 | - assert($min<=$max); |
|
896 | - |
|
897 | - $this->_filters[] = array("type"=>SPH_FILTER_FLOATRANGE, "attr"=>$attribute, "exclude"=>$exclude, "min"=>$min, "max"=>$max); |
|
898 | - } |
|
899 | - |
|
900 | - /// setup anchor point for geosphere distance calculations |
|
901 | - /// required to use @geodist in filters and sorting |
|
902 | - /// latitude and longitude must be in radians |
|
903 | - function SetGeoAnchor($attrlat, $attrlong, $lat, $long) |
|
904 | - { |
|
905 | - assert(is_string($attrlat)); |
|
906 | - assert(is_string($attrlong)); |
|
907 | - assert(is_float($lat)); |
|
908 | - assert(is_float($long)); |
|
909 | - |
|
910 | - $this->_anchor = array("attrlat"=>$attrlat, "attrlong"=>$attrlong, "lat"=>$lat, "long"=>$long); |
|
911 | - } |
|
912 | - |
|
913 | - /// set grouping attribute and function |
|
914 | - function SetGroupBy($attribute, $func, $groupsort="@group desc") |
|
915 | - { |
|
916 | - assert(is_string($attribute)); |
|
917 | - assert(is_string($groupsort)); |
|
918 | - assert($func==SPH_GROUPBY_DAY |
|
919 | - || $func==SPH_GROUPBY_WEEK |
|
920 | - || $func==SPH_GROUPBY_MONTH |
|
921 | - || $func==SPH_GROUPBY_YEAR |
|
922 | - || $func==SPH_GROUPBY_ATTR |
|
923 | - || $func==SPH_GROUPBY_ATTRPAIR); |
|
924 | - |
|
925 | - $this->_groupby = $attribute; |
|
926 | - $this->_groupfunc = $func; |
|
927 | - $this->_groupsort = $groupsort; |
|
928 | - } |
|
929 | - |
|
930 | - /// set count-distinct attribute for group-by queries |
|
931 | - function SetGroupDistinct($attribute) |
|
932 | - { |
|
933 | - assert(is_string($attribute)); |
|
934 | - $this->_groupdistinct = $attribute; |
|
935 | - } |
|
936 | - |
|
937 | - /// set distributed retries count and delay |
|
938 | - function SetRetries($count, $delay=0) |
|
939 | - { |
|
940 | - assert(is_int($count) && $count>=0); |
|
941 | - assert(is_int($delay) && $delay>=0); |
|
942 | - $this->_retrycount = $count; |
|
943 | - $this->_retrydelay = $delay; |
|
944 | - } |
|
945 | - |
|
946 | - /// set result set format (hash or array; hash by default) |
|
947 | - /// PHP specific; needed for group-by-MVA result sets that may contain duplicate IDs |
|
948 | - function SetArrayResult($arrayresult) |
|
949 | - { |
|
950 | - assert(is_bool($arrayresult)); |
|
951 | - $this->_arrayresult = $arrayresult; |
|
952 | - } |
|
953 | - |
|
954 | - /// set attribute values override |
|
955 | - /// there can be only one override per attribute |
|
956 | - /// $values must be a hash that maps document IDs to attribute values |
|
957 | - function SetOverride($attrname, $attrtype, $values) |
|
958 | - { |
|
959 | - trigger_error('DEPRECATED: Do not call this method. Use SphinxQL REMAP() function instead.', E_USER_DEPRECATED); |
|
960 | - assert(is_string($attrname)); |
|
961 | - assert(in_array($attrtype, array(SPH_ATTR_INTEGER, SPH_ATTR_TIMESTAMP, SPH_ATTR_BOOL, SPH_ATTR_FLOAT, SPH_ATTR_BIGINT))); |
|
962 | - assert(is_array($values)); |
|
963 | - |
|
964 | - $this->_overrides[$attrname] = array("attr"=>$attrname, "type"=>$attrtype, "values"=>$values); |
|
965 | - } |
|
966 | - |
|
967 | - /// set select-list (attributes or expressions), SQL-like syntax |
|
968 | - function SetSelect($select) |
|
969 | - { |
|
970 | - assert(is_string($select)); |
|
971 | - $this->_select = $select; |
|
972 | - } |
|
973 | - |
|
974 | - function SetQueryFlag($flag_name, $flag_value) |
|
975 | - { |
|
976 | - $known_names = array("reverse_scan", "sort_method", "max_predicted_time", "boolean_simplify", "idf", "global_idf", "low_priority"); |
|
977 | - $flags = array ( |
|
978 | - "reverse_scan" => array(0, 1), |
|
979 | - "sort_method" => array("pq", "kbuffer"), |
|
980 | - "max_predicted_time" => array(0), |
|
981 | - "boolean_simplify" => array(true, false), |
|
982 | - "idf" => array ("normalized", "plain", "tfidf_normalized", "tfidf_unnormalized"), |
|
983 | - "global_idf" => array(true, false), |
|
984 | - "low_priority" => array(true, false) |
|
985 | - ); |
|
986 | - |
|
987 | - assert(isset($flag_name, $known_names)); |
|
988 | - assert(in_array( $flag_value, $flags[$flag_name], true) ||($flag_name=="max_predicted_time" && is_int($flag_value) && $flag_value>=0)); |
|
989 | - |
|
990 | - if ($flag_name=="reverse_scan") $this->_query_flags = sphSetBit($this->_query_flags, 0, $flag_value==1); |
|
991 | - if ($flag_name=="sort_method") $this->_query_flags = sphSetBit($this->_query_flags, 1, $flag_value=="kbuffer"); |
|
992 | - if ($flag_name=="max_predicted_time") |
|
993 | - { |
|
994 | - $this->_query_flags = sphSetBit($this->_query_flags, 2, $flag_value>0); |
|
995 | - $this->_predictedtime = (int)$flag_value; |
|
996 | - } |
|
997 | - if ($flag_name=="boolean_simplify") $this->_query_flags = sphSetBit($this->_query_flags, 3, $flag_value); |
|
998 | - if ($flag_name=="idf" &&($flag_value=="normalized" || $flag_value=="plain")) $this->_query_flags = sphSetBit($this->_query_flags, 4, $flag_value=="plain"); |
|
999 | - if ($flag_name=="global_idf") $this->_query_flags = sphSetBit($this->_query_flags, 5, $flag_value); |
|
1000 | - if ($flag_name=="idf" &&($flag_value=="tfidf_normalized" || $flag_value=="tfidf_unnormalized")) $this->_query_flags = sphSetBit($this->_query_flags, 6, $flag_value=="tfidf_normalized"); |
|
1001 | - if ($flag_name=="low_priority") $this->_query_flags = sphSetBit($this->_query_flags, 8, $flag_value); |
|
1002 | - } |
|
1003 | - |
|
1004 | - /// set outer order by parameters |
|
1005 | - function SetOuterSelect($orderby, $offset, $limit) |
|
1006 | - { |
|
1007 | - assert(is_string($orderby)); |
|
1008 | - assert(is_int($offset)); |
|
1009 | - assert(is_int($limit)); |
|
1010 | - assert($offset>=0); |
|
1011 | - assert($limit>0); |
|
1012 | - |
|
1013 | - $this->_outerorderby = $orderby; |
|
1014 | - $this->_outeroffset = $offset; |
|
1015 | - $this->_outerlimit = $limit; |
|
1016 | - $this->_hasouter = true; |
|
1017 | - } |
|
1018 | - |
|
1019 | - |
|
1020 | - ////////////////////////////////////////////////////////////////////////////// |
|
1021 | - |
|
1022 | - /// clear all filters (for multi-queries) |
|
1023 | - function ResetFilters () |
|
1024 | - { |
|
1025 | - $this->_filters = array(); |
|
1026 | - $this->_anchor = array(); |
|
1027 | - } |
|
1028 | - |
|
1029 | - /// clear groupby settings (for multi-queries) |
|
1030 | - function ResetGroupBy () |
|
1031 | - { |
|
1032 | - $this->_groupby = ""; |
|
1033 | - $this->_groupfunc = SPH_GROUPBY_DAY; |
|
1034 | - $this->_groupsort = "@group desc"; |
|
1035 | - $this->_groupdistinct = ""; |
|
1036 | - } |
|
1037 | - |
|
1038 | - /// clear all attribute value overrides (for multi-queries) |
|
1039 | - function ResetOverrides () |
|
1040 | - { |
|
1041 | - $this->_overrides = array (); |
|
1042 | - } |
|
1043 | - |
|
1044 | - function ResetQueryFlag () |
|
1045 | - { |
|
1046 | - $this->_query_flags = sphSetBit(0, 6, true); // default idf=tfidf_normalized |
|
1047 | - $this->_predictedtime = 0; |
|
1048 | - } |
|
1049 | - |
|
1050 | - function ResetOuterSelect () |
|
1051 | - { |
|
1052 | - $this->_outerorderby = ''; |
|
1053 | - $this->_outeroffset = 0; |
|
1054 | - $this->_outerlimit = 0; |
|
1055 | - $this->_hasouter = false; |
|
1056 | - } |
|
1057 | - |
|
1058 | - ////////////////////////////////////////////////////////////////////////////// |
|
1059 | - |
|
1060 | - /// connect to searchd server, run given search query through given indexes, |
|
1061 | - /// and return the search results |
|
1062 | - function Query($query, $index="*", $comment="") |
|
1063 | - { |
|
1064 | - assert(empty($this->_reqs)); |
|
1065 | - |
|
1066 | - $this->AddQuery($query, $index, $comment); |
|
1067 | - $results = $this->RunQueries (); |
|
1068 | - $this->_reqs = array (); // just in case it failed too early |
|
1069 | - |
|
1070 | - if (!is_array($results)) |
|
1071 | - return false; // probably network error; error message should be already filled |
|
1072 | - |
|
1073 | - $this->_error = $results[0]["error"]; |
|
1074 | - $this->_warning = $results[0]["warning"]; |
|
1075 | - if ($results[0]["status"]==SEARCHD_ERROR) |
|
1076 | - return false; |
|
1077 | - else |
|
1078 | - return $results[0]; |
|
1079 | - } |
|
1080 | - |
|
1081 | - /// helper to pack floats in network byte order |
|
1082 | - function _PackFloat($f) |
|
1083 | - { |
|
1084 | - $t1 = pack("f", $f); // machine order |
|
1085 | - list(,$t2) = unpack("L*", $t1); // int in machine order |
|
1086 | - return pack("N", $t2); |
|
1087 | - } |
|
1088 | - |
|
1089 | - /// add query to multi-query batch |
|
1090 | - /// returns index into results array from RunQueries() call |
|
1091 | - function AddQuery($query, $index="*", $comment="") |
|
1092 | - { |
|
1093 | - // mbstring workaround |
|
1094 | - $this->_MBPush (); |
|
1095 | - |
|
1096 | - // build request |
|
1097 | - $req = pack("NNNNN", $this->_query_flags, $this->_offset, $this->_limit, $this->_mode, $this->_ranker); |
|
1098 | - if ($this->_ranker==SPH_RANK_EXPR) |
|
1099 | - $req .= pack("N", strlen($this->_rankexpr)) . $this->_rankexpr; |
|
1100 | - $req .= pack("N", $this->_sort); // (deprecated) sort mode |
|
1101 | - $req .= pack("N", strlen($this->_sortby)) . $this->_sortby; |
|
1102 | - $req .= pack("N", strlen($query)) . $query; // query itself |
|
1103 | - $req .= pack("N", count($this->_weights)); // weights |
|
1104 | - foreach ($this->_weights as $weight) |
|
1105 | - $req .= pack("N", (int)$weight); |
|
1106 | - $req .= pack("N", strlen($index)) . $index; // indexes |
|
1107 | - $req .= pack("N", 1); // id64 range marker |
|
1108 | - $req .= sphPackU64($this->_min_id) . sphPackU64($this->_max_id); // id64 range |
|
1109 | - |
|
1110 | - // filters |
|
1111 | - $req .= pack("N", count($this->_filters)); |
|
1112 | - foreach ($this->_filters as $filter) |
|
1113 | - { |
|
1114 | - $req .= pack("N", strlen($filter["attr"])) . $filter["attr"]; |
|
1115 | - $req .= pack("N", $filter["type"]); |
|
1116 | - switch ($filter["type"]) |
|
1117 | - { |
|
1118 | - case SPH_FILTER_VALUES: |
|
1119 | - $req .= pack("N", count($filter["values"])); |
|
1120 | - foreach ($filter["values"] as $value) |
|
1121 | - $req .= sphPackI64($value); |
|
1122 | - break; |
|
1123 | - |
|
1124 | - case SPH_FILTER_RANGE: |
|
1125 | - $req .= sphPackI64($filter["min"]) . sphPackI64($filter["max"]); |
|
1126 | - break; |
|
1127 | - |
|
1128 | - case SPH_FILTER_FLOATRANGE: |
|
1129 | - $req .= $this->_PackFloat($filter["min"]) . $this->_PackFloat($filter["max"]); |
|
1130 | - break; |
|
1131 | - |
|
1132 | - case SPH_FILTER_STRING: |
|
1133 | - $req .= pack("N", strlen($filter["value"])) . $filter["value"]; |
|
1134 | - break; |
|
1135 | - |
|
1136 | - default: |
|
1137 | - assert(0 && "internal error: unhandled filter type"); |
|
1138 | - } |
|
1139 | - $req .= pack("N", $filter["exclude"]); |
|
1140 | - } |
|
1141 | - |
|
1142 | - // group-by clause, max-matches count, group-sort clause, cutoff count |
|
1143 | - $req .= pack("NN", $this->_groupfunc, strlen($this->_groupby)) . $this->_groupby; |
|
1144 | - $req .= pack("N", $this->_maxmatches); |
|
1145 | - $req .= pack("N", strlen($this->_groupsort)) . $this->_groupsort; |
|
1146 | - $req .= pack("NNN", $this->_cutoff, $this->_retrycount, $this->_retrydelay); |
|
1147 | - $req .= pack("N", strlen($this->_groupdistinct)) . $this->_groupdistinct; |
|
1148 | - |
|
1149 | - // anchor point |
|
1150 | - if (empty($this->_anchor)) |
|
1151 | - { |
|
1152 | - $req .= pack("N", 0); |
|
1153 | - } else |
|
1154 | - { |
|
1155 | - $a =& $this->_anchor; |
|
1156 | - $req .= pack("N", 1); |
|
1157 | - $req .= pack("N", strlen($a["attrlat"])) . $a["attrlat"]; |
|
1158 | - $req .= pack("N", strlen($a["attrlong"])) . $a["attrlong"]; |
|
1159 | - $req .= $this->_PackFloat($a["lat"]) . $this->_PackFloat($a["long"]); |
|
1160 | - } |
|
1161 | - |
|
1162 | - // per-index weights |
|
1163 | - $req .= pack("N", count($this->_indexweights)); |
|
1164 | - foreach ($this->_indexweights as $idx=>$weight) |
|
1165 | - $req .= pack("N", strlen($idx)) . $idx . pack("N", $weight); |
|
1166 | - |
|
1167 | - // max query time |
|
1168 | - $req .= pack("N", $this->_maxquerytime); |
|
1169 | - |
|
1170 | - // per-field weights |
|
1171 | - $req .= pack("N", count($this->_fieldweights)); |
|
1172 | - foreach ($this->_fieldweights as $field=>$weight) |
|
1173 | - $req .= pack("N", strlen($field)) . $field . pack("N", $weight); |
|
1174 | - |
|
1175 | - // comment |
|
1176 | - $req .= pack("N", strlen($comment)) . $comment; |
|
1177 | - |
|
1178 | - // attribute overrides |
|
1179 | - $req .= pack("N", count($this->_overrides)); |
|
1180 | - foreach ($this->_overrides as $key => $entry) |
|
1181 | - { |
|
1182 | - $req .= pack("N", strlen($entry["attr"])) . $entry["attr"]; |
|
1183 | - $req .= pack("NN", $entry["type"], count($entry["values"])); |
|
1184 | - foreach ($entry["values"] as $id=>$val) |
|
1185 | - { |
|
1186 | - assert(is_numeric($id)); |
|
1187 | - assert(is_numeric($val)); |
|
1188 | - |
|
1189 | - $req .= sphPackU64($id); |
|
1190 | - switch ($entry["type"]) |
|
1191 | - { |
|
1192 | - case SPH_ATTR_FLOAT: $req .= $this->_PackFloat($val); break; |
|
1193 | - case SPH_ATTR_BIGINT: $req .= sphPackI64($val); break; |
|
1194 | - default: $req .= pack("N", $val); break; |
|
1195 | - } |
|
1196 | - } |
|
1197 | - } |
|
1198 | - |
|
1199 | - // select-list |
|
1200 | - $req .= pack("N", strlen($this->_select)) . $this->_select; |
|
1201 | - |
|
1202 | - // max_predicted_time |
|
1203 | - if ($this->_predictedtime>0) |
|
1204 | - $req .= pack("N", (int)$this->_predictedtime); |
|
1205 | - |
|
1206 | - $req .= pack("N", strlen($this->_outerorderby)) . $this->_outerorderby; |
|
1207 | - $req .= pack("NN", $this->_outeroffset, $this->_outerlimit); |
|
1208 | - if ($this->_hasouter) |
|
1209 | - $req .= pack("N", 1); |
|
1210 | - else |
|
1211 | - $req .= pack("N", 0); |
|
1212 | - |
|
1213 | - // mbstring workaround |
|
1214 | - $this->_MBPop (); |
|
1215 | - |
|
1216 | - // store request to requests array |
|
1217 | - $this->_reqs[] = $req; |
|
1218 | - return count($this->_reqs)-1; |
|
1219 | - } |
|
1220 | - |
|
1221 | - /// connect to searchd, run queries batch, and return an array of result sets |
|
1222 | - function RunQueries () |
|
1223 | - { |
|
1224 | - if (empty($this->_reqs)) |
|
1225 | - { |
|
1226 | - $this->_error = "no queries defined, issue AddQuery() first"; |
|
1227 | - return false; |
|
1228 | - } |
|
1229 | - |
|
1230 | - // mbstring workaround |
|
1231 | - $this->_MBPush (); |
|
1232 | - |
|
1233 | - if (!( $fp = $this->_Connect())) |
|
1234 | - { |
|
1235 | - $this->_MBPop (); |
|
1236 | - return false; |
|
1237 | - } |
|
1238 | - |
|
1239 | - // send query, get response |
|
1240 | - $nreqs = count($this->_reqs); |
|
1241 | - $req = join("", $this->_reqs); |
|
1242 | - $len = 8+strlen($req); |
|
1243 | - $req = pack("nnNNN", SEARCHD_COMMAND_SEARCH, VER_COMMAND_SEARCH, $len, 0, $nreqs) . $req; // add header |
|
1244 | - |
|
1245 | - if (!( $this->_Send($fp, $req, $len+8)) || |
|
1246 | - !( $response = $this->_GetResponse($fp, VER_COMMAND_SEARCH))) |
|
1247 | - { |
|
1248 | - $this->_MBPop (); |
|
1249 | - return false; |
|
1250 | - } |
|
1251 | - |
|
1252 | - // query sent ok; we can reset reqs now |
|
1253 | - $this->_reqs = array (); |
|
1254 | - |
|
1255 | - // parse and return response |
|
1256 | - return $this->_ParseSearchResponse($response, $nreqs); |
|
1257 | - } |
|
1258 | - |
|
1259 | - /// parse and return search query (or queries) response |
|
1260 | - function _ParseSearchResponse($response, $nreqs) |
|
1261 | - { |
|
1262 | - $p = 0; // current position |
|
1263 | - $max = strlen($response); // max position for checks, to protect against broken responses |
|
1264 | - |
|
1265 | - $results = array (); |
|
1266 | - for($ires=0; $ires<$nreqs && $p<$max; $ires++) |
|
1267 | - { |
|
1268 | - $results[] = array(); |
|
1269 | - $result =& $results[$ires]; |
|
1270 | - |
|
1271 | - $result["error"] = ""; |
|
1272 | - $result["warning"] = ""; |
|
1273 | - |
|
1274 | - // extract status |
|
1275 | - list(,$status) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1276 | - $result["status"] = $status; |
|
1277 | - if ($status!=SEARCHD_OK) |
|
1278 | - { |
|
1279 | - list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1280 | - $message = substr($response, $p, $len); $p += $len; |
|
1281 | - |
|
1282 | - if ($status==SEARCHD_WARNING) |
|
1283 | - { |
|
1284 | - $result["warning"] = $message; |
|
1285 | - } else |
|
1286 | - { |
|
1287 | - $result["error"] = $message; |
|
1288 | - continue; |
|
1289 | - } |
|
1290 | - } |
|
1291 | - |
|
1292 | - // read schema |
|
1293 | - $fields = array (); |
|
1294 | - $attrs = array (); |
|
1295 | - |
|
1296 | - list(,$nfields) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1297 | - while ($nfields-->0 && $p<$max) |
|
1298 | - { |
|
1299 | - list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1300 | - $fields[] = substr($response, $p, $len); $p += $len; |
|
1301 | - } |
|
1302 | - $result["fields"] = $fields; |
|
1303 | - |
|
1304 | - list(,$nattrs) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1305 | - while ($nattrs-->0 && $p<$max ) |
|
1306 | - { |
|
1307 | - list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1308 | - $attr = substr($response, $p, $len); $p += $len; |
|
1309 | - list(,$type) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1310 | - $attrs[$attr] = $type; |
|
1311 | - } |
|
1312 | - $result["attrs"] = $attrs; |
|
1313 | - |
|
1314 | - // read match count |
|
1315 | - list(,$count) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1316 | - list(,$id64) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1317 | - |
|
1318 | - // read matches |
|
1319 | - $idx = -1; |
|
1320 | - while ($count-->0 && $p<$max) |
|
1321 | - { |
|
1322 | - // index into result array |
|
1323 | - $idx++; |
|
1324 | - |
|
1325 | - // parse document id and weight |
|
1326 | - if ($id64) |
|
1327 | - { |
|
1328 | - $doc = sphUnpackU64(substr($response, $p, 8)); $p += 8; |
|
1329 | - list(,$weight) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1330 | - } |
|
1331 | - else |
|
1332 | - { |
|
1333 | - list($doc, $weight) = array_values(unpack("N*N*", |
|
1334 | - substr($response, $p, 8))); |
|
1335 | - $p += 8; |
|
1336 | - $doc = sphFixUint($doc); |
|
1337 | - } |
|
1338 | - $weight = sprintf("%u", $weight); |
|
1339 | - |
|
1340 | - // create match entry |
|
1341 | - if ($this->_arrayresult) |
|
1342 | - $result["matches"][$idx] = array("id"=>$doc, "weight"=>$weight); |
|
1343 | - else |
|
1344 | - $result["matches"][$doc]["weight"] = $weight; |
|
1345 | - |
|
1346 | - // parse and create attributes |
|
1347 | - $attrvals = array (); |
|
1348 | - foreach ($attrs as $attr=>$type) |
|
1349 | - { |
|
1350 | - // handle 64bit ints |
|
1351 | - if ($type==SPH_ATTR_BIGINT) |
|
1352 | - { |
|
1353 | - $attrvals[$attr] = sphUnpackI64(substr($response, $p, 8)); $p += 8; |
|
1354 | - continue; |
|
1355 | - } |
|
1356 | - |
|
1357 | - // handle floats |
|
1358 | - if ($type==SPH_ATTR_FLOAT) |
|
1359 | - { |
|
1360 | - list(,$uval) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1361 | - list(,$fval) = unpack("f*", pack("L", $uval)); |
|
1362 | - $attrvals[$attr] = $fval; |
|
1363 | - continue; |
|
1364 | - } |
|
1365 | - |
|
1366 | - // handle everything else as unsigned ints |
|
1367 | - list(,$val) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1368 | - if ($type==SPH_ATTR_MULTI) |
|
1369 | - { |
|
1370 | - $attrvals[$attr] = array (); |
|
1371 | - $nvalues = $val; |
|
1372 | - while ($nvalues-->0 && $p<$max) |
|
1373 | - { |
|
1374 | - list(,$val) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1375 | - $attrvals[$attr][] = sphFixUint($val); |
|
1376 | - } |
|
1377 | - } else if ($type==SPH_ATTR_MULTI64) |
|
1378 | - { |
|
1379 | - $attrvals[$attr] = array (); |
|
1380 | - $nvalues = $val; |
|
1381 | - while ($nvalues>0 && $p<$max) |
|
1382 | - { |
|
1383 | - $attrvals[$attr][] = sphUnpackI64(substr($response, $p, 8)); $p += 8; |
|
1384 | - $nvalues -= 2; |
|
1385 | - } |
|
1386 | - } else if ($type==SPH_ATTR_STRING) |
|
1387 | - { |
|
1388 | - $attrvals[$attr] = substr($response, $p, $val); |
|
1389 | - $p += $val; |
|
1390 | - } else if ($type==SPH_ATTR_FACTORS) |
|
1391 | - { |
|
1392 | - $attrvals[$attr] = substr($response, $p, $val-4); |
|
1393 | - $p += $val-4; |
|
1394 | - } else |
|
1395 | - { |
|
1396 | - $attrvals[$attr] = sphFixUint($val); |
|
1397 | - } |
|
1398 | - } |
|
1399 | - |
|
1400 | - if ($this->_arrayresult) |
|
1401 | - $result["matches"][$idx]["attrs"] = $attrvals; |
|
1402 | - else |
|
1403 | - $result["matches"][$doc]["attrs"] = $attrvals; |
|
1404 | - } |
|
1405 | - |
|
1406 | - list($total, $total_found, $msecs, $words) = |
|
1407 | - array_values(unpack("N*N*N*N*", substr($response, $p, 16))); |
|
1408 | - $result["total"] = sprintf("%u", $total); |
|
1409 | - $result["total_found"] = sprintf("%u", $total_found); |
|
1410 | - $result["time"] = sprintf("%.3f", $msecs/1000); |
|
1411 | - $p += 16; |
|
1412 | - |
|
1413 | - while ($words-->0 && $p<$max) |
|
1414 | - { |
|
1415 | - list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1416 | - $word = substr($response, $p, $len); $p += $len; |
|
1417 | - list($docs, $hits) = array_values(unpack("N*N*", substr($response, $p, 8))); $p += 8; |
|
1418 | - $result["words"][$word] = array ( |
|
1419 | - "docs"=>sprintf("%u", $docs), |
|
1420 | - "hits"=>sprintf("%u", $hits)); |
|
1421 | - } |
|
1422 | - } |
|
1423 | - |
|
1424 | - $this->_MBPop (); |
|
1425 | - return $results; |
|
1426 | - } |
|
1427 | - |
|
1428 | - ///////////////////////////////////////////////////////////////////////////// |
|
1429 | - // excerpts generation |
|
1430 | - ///////////////////////////////////////////////////////////////////////////// |
|
1431 | - |
|
1432 | - /// connect to searchd server, and generate exceprts (snippets) |
|
1433 | - /// of given documents for given query. returns false on failure, |
|
1434 | - /// an array of snippets on success |
|
1435 | - function BuildExcerpts($docs, $index, $words, $opts=array()) |
|
1436 | - { |
|
1437 | - assert(is_array($docs)); |
|
1438 | - assert(is_string($index)); |
|
1439 | - assert(is_string($words)); |
|
1440 | - assert(is_array($opts)); |
|
1441 | - |
|
1442 | - $this->_MBPush (); |
|
1443 | - |
|
1444 | - if (!( $fp = $this->_Connect())) |
|
1445 | - { |
|
1446 | - $this->_MBPop(); |
|
1447 | - return false; |
|
1448 | - } |
|
1449 | - |
|
1450 | - ///////////////// |
|
1451 | - // fixup options |
|
1452 | - ///////////////// |
|
1453 | - |
|
1454 | - if (!isset($opts["before_match"])) $opts["before_match"] = "<b>"; |
|
1455 | - if (!isset($opts["after_match"])) $opts["after_match"] = "</b>"; |
|
1456 | - if (!isset($opts["chunk_separator"])) $opts["chunk_separator"] = " ... "; |
|
1457 | - if (!isset($opts["limit"])) $opts["limit"] = 256; |
|
1458 | - if (!isset($opts["limit_passages"])) $opts["limit_passages"] = 0; |
|
1459 | - if (!isset($opts["limit_words"])) $opts["limit_words"] = 0; |
|
1460 | - if (!isset($opts["around"])) $opts["around"] = 5; |
|
1461 | - if (!isset($opts["exact_phrase"])) $opts["exact_phrase"] = false; |
|
1462 | - if (!isset($opts["single_passage"])) $opts["single_passage"] = false; |
|
1463 | - if (!isset($opts["use_boundaries"])) $opts["use_boundaries"] = false; |
|
1464 | - if (!isset($opts["weight_order"])) $opts["weight_order"] = false; |
|
1465 | - if (!isset($opts["query_mode"])) $opts["query_mode"] = false; |
|
1466 | - if (!isset($opts["force_all_words"])) $opts["force_all_words"] = false; |
|
1467 | - if (!isset($opts["start_passage_id"])) $opts["start_passage_id"] = 1; |
|
1468 | - if (!isset($opts["load_files"])) $opts["load_files"] = false; |
|
1469 | - if (!isset($opts["html_strip_mode"])) $opts["html_strip_mode"] = "index"; |
|
1470 | - if (!isset($opts["allow_empty"])) $opts["allow_empty"] = false; |
|
1471 | - if (!isset($opts["passage_boundary"])) $opts["passage_boundary"] = "none"; |
|
1472 | - if (!isset($opts["emit_zones"])) $opts["emit_zones"] = false; |
|
1473 | - if (!isset($opts["load_files_scattered"])) $opts["load_files_scattered"] = false; |
|
1474 | - |
|
1475 | - |
|
1476 | - ///////////////// |
|
1477 | - // build request |
|
1478 | - ///////////////// |
|
1479 | - |
|
1480 | - // v.1.2 req |
|
1481 | - $flags = 1; // remove spaces |
|
1482 | - if ($opts["exact_phrase"]) $flags |= 2; |
|
1483 | - if ($opts["single_passage"]) $flags |= 4; |
|
1484 | - if ($opts["use_boundaries"]) $flags |= 8; |
|
1485 | - if ($opts["weight_order"]) $flags |= 16; |
|
1486 | - if ($opts["query_mode"]) $flags |= 32; |
|
1487 | - if ($opts["force_all_words"]) $flags |= 64; |
|
1488 | - if ($opts["load_files"]) $flags |= 128; |
|
1489 | - if ($opts["allow_empty"]) $flags |= 256; |
|
1490 | - if ($opts["emit_zones"]) $flags |= 512; |
|
1491 | - if ($opts["load_files_scattered"]) $flags |= 1024; |
|
1492 | - $req = pack("NN", 0, $flags); // mode=0, flags=$flags |
|
1493 | - $req .= pack("N", strlen($index)) . $index; // req index |
|
1494 | - $req .= pack("N", strlen($words)) . $words; // req words |
|
1495 | - |
|
1496 | - // options |
|
1497 | - $req .= pack("N", strlen($opts["before_match"])) . $opts["before_match"]; |
|
1498 | - $req .= pack("N", strlen($opts["after_match"])) . $opts["after_match"]; |
|
1499 | - $req .= pack("N", strlen($opts["chunk_separator"])) . $opts["chunk_separator"]; |
|
1500 | - $req .= pack("NN", (int)$opts["limit"], (int)$opts["around"]); |
|
1501 | - $req .= pack("NNN", (int)$opts["limit_passages"], (int)$opts["limit_words"], (int)$opts["start_passage_id"]); // v.1.2 |
|
1502 | - $req .= pack("N", strlen($opts["html_strip_mode"])) . $opts["html_strip_mode"]; |
|
1503 | - $req .= pack("N", strlen($opts["passage_boundary"])) . $opts["passage_boundary"]; |
|
1504 | - |
|
1505 | - // documents |
|
1506 | - $req .= pack("N", count($docs)); |
|
1507 | - foreach ($docs as $doc) |
|
1508 | - { |
|
1509 | - assert(is_string($doc)); |
|
1510 | - $req .= pack("N", strlen($doc)) . $doc; |
|
1511 | - } |
|
1512 | - |
|
1513 | - //////////////////////////// |
|
1514 | - // send query, get response |
|
1515 | - //////////////////////////// |
|
1516 | - |
|
1517 | - $len = strlen($req); |
|
1518 | - $req = pack("nnN", SEARCHD_COMMAND_EXCERPT, VER_COMMAND_EXCERPT, $len) . $req; // add header |
|
1519 | - if (!( $this->_Send($fp, $req, $len+8)) || |
|
1520 | - !( $response = $this->_GetResponse($fp, VER_COMMAND_EXCERPT))) |
|
1521 | - { |
|
1522 | - $this->_MBPop (); |
|
1523 | - return false; |
|
1524 | - } |
|
1525 | - |
|
1526 | - ////////////////// |
|
1527 | - // parse response |
|
1528 | - ////////////////// |
|
1529 | - |
|
1530 | - $pos = 0; |
|
1531 | - $res = array (); |
|
1532 | - $rlen = strlen($response); |
|
1533 | - for($i=0; $i<count($docs); $i++) |
|
1534 | - { |
|
1535 | - list(,$len) = unpack("N*", substr($response, $pos, 4)); |
|
1536 | - $pos += 4; |
|
1537 | - |
|
1538 | - if ($pos+$len > $rlen) |
|
1539 | - { |
|
1540 | - $this->_error = "incomplete reply"; |
|
1541 | - $this->_MBPop (); |
|
1542 | - return false; |
|
1543 | - } |
|
1544 | - $res[] = $len ? substr($response, $pos, $len) : ""; |
|
1545 | - $pos += $len; |
|
1546 | - } |
|
1547 | - |
|
1548 | - $this->_MBPop (); |
|
1549 | - return $res; |
|
1550 | - } |
|
1551 | - |
|
1552 | - |
|
1553 | - ///////////////////////////////////////////////////////////////////////////// |
|
1554 | - // keyword generation |
|
1555 | - ///////////////////////////////////////////////////////////////////////////// |
|
1556 | - |
|
1557 | - /// connect to searchd server, and generate keyword list for a given query |
|
1558 | - /// returns false on failure, |
|
1559 | - /// an array of words on success |
|
1560 | - function BuildKeywords($query, $index, $hits) |
|
1561 | - { |
|
1562 | - assert(is_string($query)); |
|
1563 | - assert(is_string($index)); |
|
1564 | - assert(is_bool($hits)); |
|
1565 | - |
|
1566 | - $this->_MBPush (); |
|
1567 | - |
|
1568 | - if (!( $fp = $this->_Connect())) |
|
1569 | - { |
|
1570 | - $this->_MBPop(); |
|
1571 | - return false; |
|
1572 | - } |
|
1573 | - |
|
1574 | - ///////////////// |
|
1575 | - // build request |
|
1576 | - ///////////////// |
|
1577 | - |
|
1578 | - // v.1.0 req |
|
1579 | - $req = pack("N", strlen($query)) . $query; // req query |
|
1580 | - $req .= pack("N", strlen($index)) . $index; // req index |
|
1581 | - $req .= pack("N", (int)$hits); |
|
1582 | - |
|
1583 | - //////////////////////////// |
|
1584 | - // send query, get response |
|
1585 | - //////////////////////////// |
|
1586 | - |
|
1587 | - $len = strlen($req); |
|
1588 | - $req = pack("nnN", SEARCHD_COMMAND_KEYWORDS, VER_COMMAND_KEYWORDS, $len) . $req; // add header |
|
1589 | - if (!( $this->_Send($fp, $req, $len+8)) || |
|
1590 | - !( $response = $this->_GetResponse($fp, VER_COMMAND_KEYWORDS))) |
|
1591 | - { |
|
1592 | - $this->_MBPop (); |
|
1593 | - return false; |
|
1594 | - } |
|
1595 | - |
|
1596 | - ////////////////// |
|
1597 | - // parse response |
|
1598 | - ////////////////// |
|
1599 | - |
|
1600 | - $pos = 0; |
|
1601 | - $res = array (); |
|
1602 | - $rlen = strlen($response); |
|
1603 | - list(,$nwords) = unpack("N*", substr($response, $pos, 4)); |
|
1604 | - $pos += 4; |
|
1605 | - for($i=0; $i<$nwords; $i++) |
|
1606 | - { |
|
1607 | - list(,$len) = unpack("N*", substr($response, $pos, 4)); $pos += 4; |
|
1608 | - $tokenized = $len ? substr($response, $pos, $len) : ""; |
|
1609 | - $pos += $len; |
|
1610 | - |
|
1611 | - list(,$len) = unpack("N*", substr($response, $pos, 4)); $pos += 4; |
|
1612 | - $normalized = $len ? substr($response, $pos, $len) : ""; |
|
1613 | - $pos += $len; |
|
1614 | - |
|
1615 | - $res[] = array("tokenized"=>$tokenized, "normalized"=>$normalized); |
|
1616 | - |
|
1617 | - if ($hits) |
|
1618 | - { |
|
1619 | - list($ndocs,$nhits) = array_values(unpack("N*N*", substr($response, $pos, 8))); |
|
1620 | - $pos += 8; |
|
1621 | - $res [$i]["docs"] = $ndocs; |
|
1622 | - $res [$i]["hits"] = $nhits; |
|
1623 | - } |
|
1624 | - |
|
1625 | - if ($pos > $rlen) |
|
1626 | - { |
|
1627 | - $this->_error = "incomplete reply"; |
|
1628 | - $this->_MBPop (); |
|
1629 | - return false; |
|
1630 | - } |
|
1631 | - } |
|
1632 | - |
|
1633 | - $this->_MBPop (); |
|
1634 | - return $res; |
|
1635 | - } |
|
1636 | - |
|
1637 | - function EscapeString($string) |
|
1638 | - { |
|
1639 | - $from = array('\\', '(',')','|','-','!','@','~','"','&', '/', '^', '$', '=', '<'); |
|
1640 | - $to = array('\\\\', '\(','\)','\|','\-','\!','\@','\~','\"', '\&', '\/', '\^', '\$', '\=', '\<'); |
|
1641 | - |
|
1642 | - return str_replace($from, $to, $string); |
|
1643 | - } |
|
1644 | - |
|
1645 | - ///////////////////////////////////////////////////////////////////////////// |
|
1646 | - // attribute updates |
|
1647 | - ///////////////////////////////////////////////////////////////////////////// |
|
1648 | - |
|
1649 | - /// batch update given attributes in given rows in given indexes |
|
1650 | - /// returns amount of updated documents (0 or more) on success, or -1 on failure |
|
1651 | - function UpdateAttributes($index, $attrs, $values, $mva=false, $ignorenonexistent=false) |
|
1652 | - { |
|
1653 | - // verify everything |
|
1654 | - assert(is_string($index)); |
|
1655 | - assert(is_bool($mva)); |
|
1656 | - assert(is_bool($ignorenonexistent)); |
|
1657 | - |
|
1658 | - assert(is_array($attrs)); |
|
1659 | - foreach ($attrs as $attr) |
|
1660 | - assert(is_string($attr)); |
|
1661 | - |
|
1662 | - assert(is_array($values)); |
|
1663 | - foreach ($values as $id=>$entry) |
|
1664 | - { |
|
1665 | - assert(is_numeric($id)); |
|
1666 | - assert(is_array($entry)); |
|
1667 | - assert(count($entry)==count($attrs)); |
|
1668 | - foreach ($entry as $v) |
|
1669 | - { |
|
1670 | - if ($mva) |
|
1671 | - { |
|
1672 | - assert(is_array($v)); |
|
1673 | - foreach ($v as $vv) |
|
1674 | - assert(is_int($vv)); |
|
1675 | - } else |
|
1676 | - assert(is_int($v)); |
|
1677 | - } |
|
1678 | - } |
|
1679 | - |
|
1680 | - // build request |
|
1681 | - $this->_MBPush (); |
|
1682 | - $req = pack("N", strlen($index)) . $index; |
|
1683 | - |
|
1684 | - $req .= pack("N", count($attrs)); |
|
1685 | - $req .= pack("N", $ignorenonexistent ? 1 : 0); |
|
1686 | - foreach ($attrs as $attr) |
|
1687 | - { |
|
1688 | - $req .= pack("N", strlen($attr)) . $attr; |
|
1689 | - $req .= pack("N", $mva ? 1 : 0); |
|
1690 | - } |
|
1691 | - |
|
1692 | - $req .= pack("N", count($values)); |
|
1693 | - foreach ($values as $id=>$entry) |
|
1694 | - { |
|
1695 | - $req .= sphPackU64($id); |
|
1696 | - foreach ($entry as $v) |
|
1697 | - { |
|
1698 | - $req .= pack("N", $mva ? count($v) : $v); |
|
1699 | - if ($mva) |
|
1700 | - foreach ($v as $vv) |
|
1701 | - $req .= pack("N", $vv); |
|
1702 | - } |
|
1703 | - } |
|
1704 | - |
|
1705 | - // connect, send query, get response |
|
1706 | - if (!( $fp = $this->_Connect())) |
|
1707 | - { |
|
1708 | - $this->_MBPop (); |
|
1709 | - return -1; |
|
1710 | - } |
|
1711 | - |
|
1712 | - $len = strlen($req); |
|
1713 | - $req = pack("nnN", SEARCHD_COMMAND_UPDATE, VER_COMMAND_UPDATE, $len) . $req; // add header |
|
1714 | - if (!$this->_Send($fp, $req, $len+8)) |
|
1715 | - { |
|
1716 | - $this->_MBPop (); |
|
1717 | - return -1; |
|
1718 | - } |
|
1719 | - |
|
1720 | - if (!( $response = $this->_GetResponse($fp, VER_COMMAND_UPDATE))) |
|
1721 | - { |
|
1722 | - $this->_MBPop (); |
|
1723 | - return -1; |
|
1724 | - } |
|
1725 | - |
|
1726 | - // parse response |
|
1727 | - list(,$updated) = unpack("N*", substr($response, 0, 4)); |
|
1728 | - $this->_MBPop (); |
|
1729 | - return $updated; |
|
1730 | - } |
|
1731 | - |
|
1732 | - ///////////////////////////////////////////////////////////////////////////// |
|
1733 | - // persistent connections |
|
1734 | - ///////////////////////////////////////////////////////////////////////////// |
|
1735 | - |
|
1736 | - function Open() |
|
1737 | - { |
|
1738 | - if ($this->_socket !== false) |
|
1739 | - { |
|
1740 | - $this->_error = 'already connected'; |
|
1741 | - return false; |
|
1742 | - } |
|
1743 | - if (!$fp = $this->_Connect()) |
|
1744 | - return false; |
|
1745 | - |
|
1746 | - // command, command version = 0, body length = 4, body = 1 |
|
1747 | - $req = pack("nnNN", SEARCHD_COMMAND_PERSIST, 0, 4, 1); |
|
1748 | - if (!$this->_Send($fp, $req, 12)) |
|
1749 | - return false; |
|
1750 | - |
|
1751 | - $this->_socket = $fp; |
|
1752 | - return true; |
|
1753 | - } |
|
1754 | - |
|
1755 | - function Close() |
|
1756 | - { |
|
1757 | - if ($this->_socket === false) |
|
1758 | - { |
|
1759 | - $this->_error = 'not connected'; |
|
1760 | - return false; |
|
1761 | - } |
|
1762 | - |
|
1763 | - fclose($this->_socket); |
|
1764 | - $this->_socket = false; |
|
1765 | - |
|
1766 | - return true; |
|
1767 | - } |
|
1768 | - |
|
1769 | - ////////////////////////////////////////////////////////////////////////// |
|
1770 | - // status |
|
1771 | - ////////////////////////////////////////////////////////////////////////// |
|
1772 | - |
|
1773 | - function Status ($session=false) |
|
1774 | - { |
|
1775 | - assert(is_bool($session)); |
|
1776 | - |
|
1777 | - $this->_MBPush (); |
|
1778 | - if (!( $fp = $this->_Connect())) |
|
1779 | - { |
|
1780 | - $this->_MBPop(); |
|
1781 | - return false; |
|
1782 | - } |
|
1783 | - |
|
1784 | - $req = pack("nnNN", SEARCHD_COMMAND_STATUS, VER_COMMAND_STATUS, 4, $session?0:1); // len=4, body=1 |
|
1785 | - if (!( $this->_Send($fp, $req, 12)) || |
|
1786 | - !( $response = $this->_GetResponse($fp, VER_COMMAND_STATUS))) |
|
1787 | - { |
|
1788 | - $this->_MBPop (); |
|
1789 | - return false; |
|
1790 | - } |
|
1791 | - |
|
1792 | - $res = substr($response, 4); // just ignore length, error handling, etc |
|
1793 | - $p = 0; |
|
1794 | - list($rows, $cols) = array_values(unpack("N*N*", substr($response, $p, 8))); $p += 8; |
|
1795 | - |
|
1796 | - $res = array(); |
|
1797 | - for($i=0; $i<$rows; $i++) |
|
1798 | - for($j=0; $j<$cols; $j++) |
|
1799 | - { |
|
1800 | - list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1801 | - $res[$i][] = substr($response, $p, $len); $p += $len; |
|
1802 | - } |
|
1803 | - |
|
1804 | - $this->_MBPop (); |
|
1805 | - return $res; |
|
1806 | - } |
|
1807 | - |
|
1808 | - ////////////////////////////////////////////////////////////////////////// |
|
1809 | - // flush |
|
1810 | - ////////////////////////////////////////////////////////////////////////// |
|
1811 | - |
|
1812 | - function FlushAttributes () |
|
1813 | - { |
|
1814 | - $this->_MBPush (); |
|
1815 | - if (!( $fp = $this->_Connect())) |
|
1816 | - { |
|
1817 | - $this->_MBPop(); |
|
1818 | - return -1; |
|
1819 | - } |
|
1820 | - |
|
1821 | - $req = pack("nnN", SEARCHD_COMMAND_FLUSHATTRS, VER_COMMAND_FLUSHATTRS, 0); // len=0 |
|
1822 | - if (!( $this->_Send($fp, $req, 8)) || |
|
1823 | - !( $response = $this->_GetResponse($fp, VER_COMMAND_FLUSHATTRS))) |
|
1824 | - { |
|
1825 | - $this->_MBPop (); |
|
1826 | - return -1; |
|
1827 | - } |
|
1828 | - |
|
1829 | - $tag = -1; |
|
1830 | - if (strlen($response)==4) |
|
1831 | - list(,$tag) = unpack("N*", $response); |
|
1832 | - else |
|
1833 | - $this->_error = "unexpected response length"; |
|
1834 | - |
|
1835 | - $this->_MBPop (); |
|
1836 | - return $tag; |
|
1837 | - } |
|
416 | + var $_host; ///< searchd host (default is "localhost") |
|
417 | + var $_port; ///< searchd port (default is 9312) |
|
418 | + var $_offset; ///< how many records to seek from result-set start (default is 0) |
|
419 | + var $_limit; ///< how many records to return from result-set starting at offset (default is 20) |
|
420 | + var $_mode; ///< query matching mode (default is SPH_MATCH_EXTENDED2) |
|
421 | + var $_weights; ///< per-field weights (default is 1 for all fields) |
|
422 | + var $_sort; ///< match sorting mode (default is SPH_SORT_RELEVANCE) |
|
423 | + var $_sortby; ///< attribute to sort by (defualt is "") |
|
424 | + var $_min_id; ///< min ID to match (default is 0, which means no limit) |
|
425 | + var $_max_id; ///< max ID to match (default is 0, which means no limit) |
|
426 | + var $_filters; ///< search filters |
|
427 | + var $_groupby; ///< group-by attribute name |
|
428 | + var $_groupfunc; ///< group-by function (to pre-process group-by attribute value with) |
|
429 | + var $_groupsort; ///< group-by sorting clause (to sort groups in result set with) |
|
430 | + var $_groupdistinct; ///< group-by count-distinct attribute |
|
431 | + var $_maxmatches; ///< max matches to retrieve |
|
432 | + var $_cutoff; ///< cutoff to stop searching at (default is 0) |
|
433 | + var $_retrycount; ///< distributed retries count |
|
434 | + var $_retrydelay; ///< distributed retries delay |
|
435 | + var $_anchor; ///< geographical anchor point |
|
436 | + var $_indexweights; ///< per-index weights |
|
437 | + var $_ranker; ///< ranking mode (default is SPH_RANK_PROXIMITY_BM25) |
|
438 | + var $_rankexpr; ///< ranking mode expression (for SPH_RANK_EXPR) |
|
439 | + var $_maxquerytime; ///< max query time, milliseconds (default is 0, do not limit) |
|
440 | + var $_fieldweights; ///< per-field-name weights |
|
441 | + var $_overrides; ///< per-query attribute values overrides |
|
442 | + var $_select; ///< select-list (attributes or expressions, with optional aliases) |
|
443 | + var $_query_flags; ///< per-query various flags |
|
444 | + var $_predictedtime; ///< per-query max_predicted_time |
|
445 | + var $_outerorderby; ///< outer match sort by |
|
446 | + var $_outeroffset; ///< outer offset |
|
447 | + var $_outerlimit; ///< outer limit |
|
448 | + var $_hasouter; |
|
449 | + |
|
450 | + var $_error; ///< last error message |
|
451 | + var $_warning; ///< last warning message |
|
452 | + var $_connerror; ///< connection error vs remote error flag |
|
453 | + |
|
454 | + var $_reqs; ///< requests array for multi-query |
|
455 | + var $_mbenc; ///< stored mbstring encoding |
|
456 | + var $_arrayresult; ///< whether $result["matches"] should be a hash or an array |
|
457 | + var $_timeout; ///< connect timeout |
|
458 | + |
|
459 | + ///////////////////////////////////////////////////////////////////////////// |
|
460 | + // common stuff |
|
461 | + ///////////////////////////////////////////////////////////////////////////// |
|
462 | + |
|
463 | + /// create a new client object and fill defaults |
|
464 | + function SphinxClient () |
|
465 | + { |
|
466 | + // per-client-object settings |
|
467 | + $this->_host = "localhost"; |
|
468 | + $this->_port = 9312; |
|
469 | + $this->_path = false; |
|
470 | + $this->_socket = false; |
|
471 | + |
|
472 | + // per-query settings |
|
473 | + $this->_offset = 0; |
|
474 | + $this->_limit = 20; |
|
475 | + $this->_mode = SPH_MATCH_EXTENDED2; |
|
476 | + $this->_weights = array (); |
|
477 | + $this->_sort = SPH_SORT_RELEVANCE; |
|
478 | + $this->_sortby = ""; |
|
479 | + $this->_min_id = 0; |
|
480 | + $this->_max_id = 0; |
|
481 | + $this->_filters = array (); |
|
482 | + $this->_groupby = ""; |
|
483 | + $this->_groupfunc = SPH_GROUPBY_DAY; |
|
484 | + $this->_groupsort = "@group desc"; |
|
485 | + $this->_groupdistinct = ""; |
|
486 | + $this->_maxmatches = 1000; |
|
487 | + $this->_cutoff = 0; |
|
488 | + $this->_retrycount = 0; |
|
489 | + $this->_retrydelay = 0; |
|
490 | + $this->_anchor = array (); |
|
491 | + $this->_indexweights = array (); |
|
492 | + $this->_ranker = SPH_RANK_PROXIMITY_BM25; |
|
493 | + $this->_rankexpr = ""; |
|
494 | + $this->_maxquerytime = 0; |
|
495 | + $this->_fieldweights = array(); |
|
496 | + $this->_overrides = array(); |
|
497 | + $this->_select = "*"; |
|
498 | + $this->_query_flags = sphSetBit(0, 6, true); // default idf=tfidf_normalized |
|
499 | + $this->_predictedtime = 0; |
|
500 | + $this->_outerorderby = ""; |
|
501 | + $this->_outeroffset = 0; |
|
502 | + $this->_outerlimit = 0; |
|
503 | + $this->_hasouter = false; |
|
504 | + |
|
505 | + $this->_error = ""; // per-reply fields (for single-query case) |
|
506 | + $this->_warning = ""; |
|
507 | + $this->_connerror = false; |
|
508 | + |
|
509 | + $this->_reqs = array ();// requests storage (for multi-query case) |
|
510 | + $this->_mbenc = ""; |
|
511 | + $this->_arrayresult = false; |
|
512 | + $this->_timeout = 0; |
|
513 | + } |
|
514 | + |
|
515 | + function __destruct() |
|
516 | + { |
|
517 | + if ($this->_socket !== false) |
|
518 | + fclose($this->_socket); |
|
519 | + } |
|
520 | + |
|
521 | + /// get last error message (string) |
|
522 | + function GetLastError () |
|
523 | + { |
|
524 | + return $this->_error; |
|
525 | + } |
|
526 | + |
|
527 | + /// get last warning message (string) |
|
528 | + function GetLastWarning () |
|
529 | + { |
|
530 | + return $this->_warning; |
|
531 | + } |
|
532 | + |
|
533 | + /// get last error flag (to tell network connection errors from searchd errors or broken responses) |
|
534 | + function IsConnectError() |
|
535 | + { |
|
536 | + return $this->_connerror; |
|
537 | + } |
|
538 | + |
|
539 | + /// set searchd host name (string) and port (integer) |
|
540 | + function SetServer($host, $port = 0) |
|
541 | + { |
|
542 | + assert(is_string($host)); |
|
543 | + if ($host[0] == '/') |
|
544 | + { |
|
545 | + $this->_path = 'unix://' . $host; |
|
546 | + return; |
|
547 | + } |
|
548 | + if (substr($host, 0, 7)=="unix://") |
|
549 | + { |
|
550 | + $this->_path = $host; |
|
551 | + return; |
|
552 | + } |
|
553 | + |
|
554 | + $this->_host = $host; |
|
555 | + $port = intval($port); |
|
556 | + assert(0<=$port && $port<65536); |
|
557 | + $this->_port =($port==0) ? 9312 : $port; |
|
558 | + $this->_path = ''; |
|
559 | + } |
|
560 | + |
|
561 | + /// set server connection timeout (0 to remove) |
|
562 | + function SetConnectTimeout($timeout) |
|
563 | + { |
|
564 | + assert(is_numeric($timeout)); |
|
565 | + $this->_timeout = $timeout; |
|
566 | + } |
|
567 | + |
|
568 | + |
|
569 | + function _Send($handle, $data, $length) |
|
570 | + { |
|
571 | + if (feof($handle) || fwrite($handle, $data, $length) !== $length) |
|
572 | + { |
|
573 | + $this->_error = 'connection unexpectedly closed (timed out?)'; |
|
574 | + $this->_connerror = true; |
|
575 | + return false; |
|
576 | + } |
|
577 | + return true; |
|
578 | + } |
|
579 | + |
|
580 | + ///////////////////////////////////////////////////////////////////////////// |
|
581 | + |
|
582 | + /// enter mbstring workaround mode |
|
583 | + function _MBPush () |
|
584 | + { |
|
585 | + $this->_mbenc = ""; |
|
586 | + if (ini_get("mbstring.func_overload") & 2) |
|
587 | + { |
|
588 | + $this->_mbenc = mb_internal_encoding(); |
|
589 | + mb_internal_encoding("latin1"); |
|
590 | + } |
|
591 | + } |
|
592 | + |
|
593 | + /// leave mbstring workaround mode |
|
594 | + function _MBPop () |
|
595 | + { |
|
596 | + if ($this->_mbenc) |
|
597 | + mb_internal_encoding($this->_mbenc); |
|
598 | + } |
|
599 | + |
|
600 | + /// connect to searchd server |
|
601 | + function _Connect () |
|
602 | + { |
|
603 | + if ($this->_socket!==false) |
|
604 | + { |
|
605 | + // we are in persistent connection mode, so we have a socket |
|
606 | + // however, need to check whether it's still alive |
|
607 | + if (!@feof($this->_socket)) |
|
608 | + return $this->_socket; |
|
609 | + |
|
610 | + // force reopen |
|
611 | + $this->_socket = false; |
|
612 | + } |
|
613 | + |
|
614 | + $errno = 0; |
|
615 | + $errstr = ""; |
|
616 | + $this->_connerror = false; |
|
617 | + |
|
618 | + if ($this->_path) |
|
619 | + { |
|
620 | + $host = $this->_path; |
|
621 | + $port = 0; |
|
622 | + } |
|
623 | + else |
|
624 | + { |
|
625 | + $host = $this->_host; |
|
626 | + $port = $this->_port; |
|
627 | + } |
|
628 | + |
|
629 | + if ($this->_timeout<=0) |
|
630 | + $fp = @fsockopen($host, $port, $errno, $errstr); |
|
631 | + else |
|
632 | + $fp = @fsockopen($host, $port, $errno, $errstr, $this->_timeout); |
|
633 | + |
|
634 | + if (!$fp) |
|
635 | + { |
|
636 | + if ($this->_path) |
|
637 | + $location = $this->_path; |
|
638 | + else |
|
639 | + $location = "{$this->_host}:{$this->_port}"; |
|
640 | + |
|
641 | + $errstr = trim($errstr); |
|
642 | + $this->_error = "connection to $location failed (errno=$errno, msg=$errstr)"; |
|
643 | + $this->_connerror = true; |
|
644 | + return false; |
|
645 | + } |
|
646 | + |
|
647 | + // send my version |
|
648 | + // this is a subtle part. we must do it before (!) reading back from searchd. |
|
649 | + // because otherwise under some conditions (reported on FreeBSD for instance) |
|
650 | + // TCP stack could throttle write-write-read pattern because of Nagle. |
|
651 | + if (!$this->_Send($fp, pack("N", 1), 4)) |
|
652 | + { |
|
653 | + fclose($fp); |
|
654 | + $this->_error = "failed to send client protocol version"; |
|
655 | + return false; |
|
656 | + } |
|
657 | + |
|
658 | + // check version |
|
659 | + list(,$v) = unpack("N*", fread($fp, 4)); |
|
660 | + $v = (int)$v; |
|
661 | + if ($v<1) |
|
662 | + { |
|
663 | + fclose($fp); |
|
664 | + $this->_error = "expected searchd protocol version 1+, got version '$v'"; |
|
665 | + return false; |
|
666 | + } |
|
667 | + |
|
668 | + return $fp; |
|
669 | + } |
|
670 | + |
|
671 | + /// get and check response packet from searchd server |
|
672 | + function _GetResponse($fp, $client_ver) |
|
673 | + { |
|
674 | + $response = ""; |
|
675 | + $len = 0; |
|
676 | + |
|
677 | + $header = fread($fp, 8); |
|
678 | + if (strlen($header)==8) |
|
679 | + { |
|
680 | + list($status, $ver, $len) = array_values(unpack("n2a/Nb", $header)); |
|
681 | + $left = $len; |
|
682 | + while ($left>0 && !feof($fp)) |
|
683 | + { |
|
684 | + $chunk = fread($fp, min(8192, $left)); |
|
685 | + if ($chunk) |
|
686 | + { |
|
687 | + $response .= $chunk; |
|
688 | + $left -= strlen($chunk); |
|
689 | + } |
|
690 | + } |
|
691 | + } |
|
692 | + if ($this->_socket === false) |
|
693 | + fclose($fp); |
|
694 | + |
|
695 | + // check response |
|
696 | + $read = strlen($response); |
|
697 | + if (!$response || $read!=$len) |
|
698 | + { |
|
699 | + $this->_error = $len |
|
700 | + ? "failed to read searchd response (status=$status, ver=$ver, len=$len, read=$read)" |
|
701 | + : "received zero-sized searchd response"; |
|
702 | + return false; |
|
703 | + } |
|
704 | + |
|
705 | + // check status |
|
706 | + if ($status==SEARCHD_WARNING) |
|
707 | + { |
|
708 | + list(,$wlen) = unpack("N*", substr($response, 0, 4)); |
|
709 | + $this->_warning = substr($response, 4, $wlen); |
|
710 | + return substr($response, 4+$wlen); |
|
711 | + } |
|
712 | + if ($status==SEARCHD_ERROR) |
|
713 | + { |
|
714 | + $this->_error = "searchd error: " . substr($response, 4); |
|
715 | + return false; |
|
716 | + } |
|
717 | + if ($status==SEARCHD_RETRY) |
|
718 | + { |
|
719 | + $this->_error = "temporary searchd error: " . substr($response, 4); |
|
720 | + return false; |
|
721 | + } |
|
722 | + if ($status!=SEARCHD_OK) |
|
723 | + { |
|
724 | + $this->_error = "unknown status code '$status'"; |
|
725 | + return false; |
|
726 | + } |
|
727 | + |
|
728 | + // check version |
|
729 | + if ($ver<$client_ver) |
|
730 | + { |
|
731 | + $this->_warning = sprintf("searchd command v.%d.%d older than client's v.%d.%d, some options might not work", |
|
732 | + $ver>>8, $ver&0xff, $client_ver>>8, $client_ver&0xff); |
|
733 | + } |
|
734 | + |
|
735 | + return $response; |
|
736 | + } |
|
737 | + |
|
738 | + ///////////////////////////////////////////////////////////////////////////// |
|
739 | + // searching |
|
740 | + ///////////////////////////////////////////////////////////////////////////// |
|
741 | + |
|
742 | + /// set offset and count into result set, |
|
743 | + /// and optionally set max-matches and cutoff limits |
|
744 | + function SetLimits($offset, $limit, $max=0, $cutoff=0) |
|
745 | + { |
|
746 | + assert(is_int($offset)); |
|
747 | + assert(is_int($limit)); |
|
748 | + assert($offset>=0); |
|
749 | + assert($limit>0); |
|
750 | + assert($max>=0); |
|
751 | + $this->_offset = $offset; |
|
752 | + $this->_limit = $limit; |
|
753 | + if ($max>0) |
|
754 | + $this->_maxmatches = $max; |
|
755 | + if ($cutoff>0) |
|
756 | + $this->_cutoff = $cutoff; |
|
757 | + } |
|
758 | + |
|
759 | + /// set maximum query time, in milliseconds, per-index |
|
760 | + /// integer, 0 means "do not limit" |
|
761 | + function SetMaxQueryTime($max) |
|
762 | + { |
|
763 | + assert(is_int($max)); |
|
764 | + assert($max>=0); |
|
765 | + $this->_maxquerytime = $max; |
|
766 | + } |
|
767 | + |
|
768 | + /// set matching mode |
|
769 | + function SetMatchMode($mode) |
|
770 | + { |
|
771 | + trigger_error('DEPRECATED: Do not call this method or, even better, use SphinxQL instead of an API', E_USER_DEPRECATED); |
|
772 | + assert($mode==SPH_MATCH_ALL |
|
773 | + || $mode==SPH_MATCH_ANY |
|
774 | + || $mode==SPH_MATCH_PHRASE |
|
775 | + || $mode==SPH_MATCH_BOOLEAN |
|
776 | + || $mode==SPH_MATCH_EXTENDED |
|
777 | + || $mode==SPH_MATCH_FULLSCAN |
|
778 | + || $mode==SPH_MATCH_EXTENDED2); |
|
779 | + $this->_mode = $mode; |
|
780 | + } |
|
781 | + |
|
782 | + /// set ranking mode |
|
783 | + function SetRankingMode($ranker, $rankexpr="") |
|
784 | + { |
|
785 | + assert($ranker===0 || $ranker>=1 && $ranker<SPH_RANK_TOTAL); |
|
786 | + assert(is_string($rankexpr)); |
|
787 | + $this->_ranker = $ranker; |
|
788 | + $this->_rankexpr = $rankexpr; |
|
789 | + } |
|
790 | + |
|
791 | + /// set matches sorting mode |
|
792 | + function SetSortMode($mode, $sortby="") |
|
793 | + { |
|
794 | + assert ( |
|
795 | + $mode==SPH_SORT_RELEVANCE || |
|
796 | + $mode==SPH_SORT_ATTR_DESC || |
|
797 | + $mode==SPH_SORT_ATTR_ASC || |
|
798 | + $mode==SPH_SORT_TIME_SEGMENTS || |
|
799 | + $mode==SPH_SORT_EXTENDED || |
|
800 | + $mode==SPH_SORT_EXPR); |
|
801 | + assert(is_string($sortby)); |
|
802 | + assert($mode==SPH_SORT_RELEVANCE || strlen($sortby)>0); |
|
803 | + |
|
804 | + $this->_sort = $mode; |
|
805 | + $this->_sortby = $sortby; |
|
806 | + } |
|
807 | + |
|
808 | + /// bind per-field weights by order |
|
809 | + /// DEPRECATED; use SetFieldWeights() instead |
|
810 | + function SetWeights($weights) |
|
811 | + { |
|
812 | + die("This method is now deprecated; please use SetFieldWeights instead"); |
|
813 | + } |
|
814 | + |
|
815 | + /// bind per-field weights by name |
|
816 | + function SetFieldWeights($weights) |
|
817 | + { |
|
818 | + assert(is_array($weights)); |
|
819 | + foreach ($weights as $name=>$weight) |
|
820 | + { |
|
821 | + assert(is_string($name)); |
|
822 | + assert(is_int($weight)); |
|
823 | + } |
|
824 | + $this->_fieldweights = $weights; |
|
825 | + } |
|
826 | + |
|
827 | + /// bind per-index weights by name |
|
828 | + function SetIndexWeights($weights) |
|
829 | + { |
|
830 | + assert(is_array($weights)); |
|
831 | + foreach ($weights as $index=>$weight) |
|
832 | + { |
|
833 | + assert(is_string($index)); |
|
834 | + assert(is_int($weight)); |
|
835 | + } |
|
836 | + $this->_indexweights = $weights; |
|
837 | + } |
|
838 | + |
|
839 | + /// set IDs range to match |
|
840 | + /// only match records if document ID is beetwen $min and $max (inclusive) |
|
841 | + function SetIDRange($min, $max) |
|
842 | + { |
|
843 | + assert(is_numeric($min)); |
|
844 | + assert(is_numeric($max)); |
|
845 | + assert($min<=$max); |
|
846 | + $this->_min_id = $min; |
|
847 | + $this->_max_id = $max; |
|
848 | + } |
|
849 | + |
|
850 | + /// set values set filter |
|
851 | + /// only match records where $attribute value is in given set |
|
852 | + function SetFilter($attribute, $values, $exclude=false) |
|
853 | + { |
|
854 | + assert(is_string($attribute)); |
|
855 | + assert(is_array($values)); |
|
856 | + assert(count($values)); |
|
857 | + |
|
858 | + if (is_array($values) && count($values)) |
|
859 | + { |
|
860 | + foreach ($values as $value) |
|
861 | + assert(is_numeric($value)); |
|
862 | + |
|
863 | + $this->_filters[] = array("type"=>SPH_FILTER_VALUES, "attr"=>$attribute, "exclude"=>$exclude, "values"=>$values); |
|
864 | + } |
|
865 | + } |
|
866 | + |
|
867 | + /// set string filter |
|
868 | + /// only match records where $attribute value is equal |
|
869 | + function SetFilterString($attribute, $value, $exclude=false) |
|
870 | + { |
|
871 | + assert(is_string($attribute)); |
|
872 | + assert(is_string($value)); |
|
873 | + $this->_filters[] = array("type"=>SPH_FILTER_STRING, "attr"=>$attribute, "exclude"=>$exclude, "value"=>$value); |
|
874 | + } |
|
875 | + |
|
876 | + /// set range filter |
|
877 | + /// only match records if $attribute value is beetwen $min and $max (inclusive) |
|
878 | + function SetFilterRange($attribute, $min, $max, $exclude=false) |
|
879 | + { |
|
880 | + assert(is_string($attribute)); |
|
881 | + assert(is_numeric($min)); |
|
882 | + assert(is_numeric($max)); |
|
883 | + assert($min<=$max); |
|
884 | + |
|
885 | + $this->_filters[] = array("type"=>SPH_FILTER_RANGE, "attr"=>$attribute, "exclude"=>$exclude, "min"=>$min, "max"=>$max); |
|
886 | + } |
|
887 | + |
|
888 | + /// set float range filter |
|
889 | + /// only match records if $attribute value is beetwen $min and $max (inclusive) |
|
890 | + function SetFilterFloatRange($attribute, $min, $max, $exclude=false) |
|
891 | + { |
|
892 | + assert(is_string($attribute)); |
|
893 | + assert(is_float($min)); |
|
894 | + assert(is_float($max)); |
|
895 | + assert($min<=$max); |
|
896 | + |
|
897 | + $this->_filters[] = array("type"=>SPH_FILTER_FLOATRANGE, "attr"=>$attribute, "exclude"=>$exclude, "min"=>$min, "max"=>$max); |
|
898 | + } |
|
899 | + |
|
900 | + /// setup anchor point for geosphere distance calculations |
|
901 | + /// required to use @geodist in filters and sorting |
|
902 | + /// latitude and longitude must be in radians |
|
903 | + function SetGeoAnchor($attrlat, $attrlong, $lat, $long) |
|
904 | + { |
|
905 | + assert(is_string($attrlat)); |
|
906 | + assert(is_string($attrlong)); |
|
907 | + assert(is_float($lat)); |
|
908 | + assert(is_float($long)); |
|
909 | + |
|
910 | + $this->_anchor = array("attrlat"=>$attrlat, "attrlong"=>$attrlong, "lat"=>$lat, "long"=>$long); |
|
911 | + } |
|
912 | + |
|
913 | + /// set grouping attribute and function |
|
914 | + function SetGroupBy($attribute, $func, $groupsort="@group desc") |
|
915 | + { |
|
916 | + assert(is_string($attribute)); |
|
917 | + assert(is_string($groupsort)); |
|
918 | + assert($func==SPH_GROUPBY_DAY |
|
919 | + || $func==SPH_GROUPBY_WEEK |
|
920 | + || $func==SPH_GROUPBY_MONTH |
|
921 | + || $func==SPH_GROUPBY_YEAR |
|
922 | + || $func==SPH_GROUPBY_ATTR |
|
923 | + || $func==SPH_GROUPBY_ATTRPAIR); |
|
924 | + |
|
925 | + $this->_groupby = $attribute; |
|
926 | + $this->_groupfunc = $func; |
|
927 | + $this->_groupsort = $groupsort; |
|
928 | + } |
|
929 | + |
|
930 | + /// set count-distinct attribute for group-by queries |
|
931 | + function SetGroupDistinct($attribute) |
|
932 | + { |
|
933 | + assert(is_string($attribute)); |
|
934 | + $this->_groupdistinct = $attribute; |
|
935 | + } |
|
936 | + |
|
937 | + /// set distributed retries count and delay |
|
938 | + function SetRetries($count, $delay=0) |
|
939 | + { |
|
940 | + assert(is_int($count) && $count>=0); |
|
941 | + assert(is_int($delay) && $delay>=0); |
|
942 | + $this->_retrycount = $count; |
|
943 | + $this->_retrydelay = $delay; |
|
944 | + } |
|
945 | + |
|
946 | + /// set result set format (hash or array; hash by default) |
|
947 | + /// PHP specific; needed for group-by-MVA result sets that may contain duplicate IDs |
|
948 | + function SetArrayResult($arrayresult) |
|
949 | + { |
|
950 | + assert(is_bool($arrayresult)); |
|
951 | + $this->_arrayresult = $arrayresult; |
|
952 | + } |
|
953 | + |
|
954 | + /// set attribute values override |
|
955 | + /// there can be only one override per attribute |
|
956 | + /// $values must be a hash that maps document IDs to attribute values |
|
957 | + function SetOverride($attrname, $attrtype, $values) |
|
958 | + { |
|
959 | + trigger_error('DEPRECATED: Do not call this method. Use SphinxQL REMAP() function instead.', E_USER_DEPRECATED); |
|
960 | + assert(is_string($attrname)); |
|
961 | + assert(in_array($attrtype, array(SPH_ATTR_INTEGER, SPH_ATTR_TIMESTAMP, SPH_ATTR_BOOL, SPH_ATTR_FLOAT, SPH_ATTR_BIGINT))); |
|
962 | + assert(is_array($values)); |
|
963 | + |
|
964 | + $this->_overrides[$attrname] = array("attr"=>$attrname, "type"=>$attrtype, "values"=>$values); |
|
965 | + } |
|
966 | + |
|
967 | + /// set select-list (attributes or expressions), SQL-like syntax |
|
968 | + function SetSelect($select) |
|
969 | + { |
|
970 | + assert(is_string($select)); |
|
971 | + $this->_select = $select; |
|
972 | + } |
|
973 | + |
|
974 | + function SetQueryFlag($flag_name, $flag_value) |
|
975 | + { |
|
976 | + $known_names = array("reverse_scan", "sort_method", "max_predicted_time", "boolean_simplify", "idf", "global_idf", "low_priority"); |
|
977 | + $flags = array ( |
|
978 | + "reverse_scan" => array(0, 1), |
|
979 | + "sort_method" => array("pq", "kbuffer"), |
|
980 | + "max_predicted_time" => array(0), |
|
981 | + "boolean_simplify" => array(true, false), |
|
982 | + "idf" => array ("normalized", "plain", "tfidf_normalized", "tfidf_unnormalized"), |
|
983 | + "global_idf" => array(true, false), |
|
984 | + "low_priority" => array(true, false) |
|
985 | + ); |
|
986 | + |
|
987 | + assert(isset($flag_name, $known_names)); |
|
988 | + assert(in_array( $flag_value, $flags[$flag_name], true) ||($flag_name=="max_predicted_time" && is_int($flag_value) && $flag_value>=0)); |
|
989 | + |
|
990 | + if ($flag_name=="reverse_scan") $this->_query_flags = sphSetBit($this->_query_flags, 0, $flag_value==1); |
|
991 | + if ($flag_name=="sort_method") $this->_query_flags = sphSetBit($this->_query_flags, 1, $flag_value=="kbuffer"); |
|
992 | + if ($flag_name=="max_predicted_time") |
|
993 | + { |
|
994 | + $this->_query_flags = sphSetBit($this->_query_flags, 2, $flag_value>0); |
|
995 | + $this->_predictedtime = (int)$flag_value; |
|
996 | + } |
|
997 | + if ($flag_name=="boolean_simplify") $this->_query_flags = sphSetBit($this->_query_flags, 3, $flag_value); |
|
998 | + if ($flag_name=="idf" &&($flag_value=="normalized" || $flag_value=="plain")) $this->_query_flags = sphSetBit($this->_query_flags, 4, $flag_value=="plain"); |
|
999 | + if ($flag_name=="global_idf") $this->_query_flags = sphSetBit($this->_query_flags, 5, $flag_value); |
|
1000 | + if ($flag_name=="idf" &&($flag_value=="tfidf_normalized" || $flag_value=="tfidf_unnormalized")) $this->_query_flags = sphSetBit($this->_query_flags, 6, $flag_value=="tfidf_normalized"); |
|
1001 | + if ($flag_name=="low_priority") $this->_query_flags = sphSetBit($this->_query_flags, 8, $flag_value); |
|
1002 | + } |
|
1003 | + |
|
1004 | + /// set outer order by parameters |
|
1005 | + function SetOuterSelect($orderby, $offset, $limit) |
|
1006 | + { |
|
1007 | + assert(is_string($orderby)); |
|
1008 | + assert(is_int($offset)); |
|
1009 | + assert(is_int($limit)); |
|
1010 | + assert($offset>=0); |
|
1011 | + assert($limit>0); |
|
1012 | + |
|
1013 | + $this->_outerorderby = $orderby; |
|
1014 | + $this->_outeroffset = $offset; |
|
1015 | + $this->_outerlimit = $limit; |
|
1016 | + $this->_hasouter = true; |
|
1017 | + } |
|
1018 | + |
|
1019 | + |
|
1020 | + ////////////////////////////////////////////////////////////////////////////// |
|
1021 | + |
|
1022 | + /// clear all filters (for multi-queries) |
|
1023 | + function ResetFilters () |
|
1024 | + { |
|
1025 | + $this->_filters = array(); |
|
1026 | + $this->_anchor = array(); |
|
1027 | + } |
|
1028 | + |
|
1029 | + /// clear groupby settings (for multi-queries) |
|
1030 | + function ResetGroupBy () |
|
1031 | + { |
|
1032 | + $this->_groupby = ""; |
|
1033 | + $this->_groupfunc = SPH_GROUPBY_DAY; |
|
1034 | + $this->_groupsort = "@group desc"; |
|
1035 | + $this->_groupdistinct = ""; |
|
1036 | + } |
|
1037 | + |
|
1038 | + /// clear all attribute value overrides (for multi-queries) |
|
1039 | + function ResetOverrides () |
|
1040 | + { |
|
1041 | + $this->_overrides = array (); |
|
1042 | + } |
|
1043 | + |
|
1044 | + function ResetQueryFlag () |
|
1045 | + { |
|
1046 | + $this->_query_flags = sphSetBit(0, 6, true); // default idf=tfidf_normalized |
|
1047 | + $this->_predictedtime = 0; |
|
1048 | + } |
|
1049 | + |
|
1050 | + function ResetOuterSelect () |
|
1051 | + { |
|
1052 | + $this->_outerorderby = ''; |
|
1053 | + $this->_outeroffset = 0; |
|
1054 | + $this->_outerlimit = 0; |
|
1055 | + $this->_hasouter = false; |
|
1056 | + } |
|
1057 | + |
|
1058 | + ////////////////////////////////////////////////////////////////////////////// |
|
1059 | + |
|
1060 | + /// connect to searchd server, run given search query through given indexes, |
|
1061 | + /// and return the search results |
|
1062 | + function Query($query, $index="*", $comment="") |
|
1063 | + { |
|
1064 | + assert(empty($this->_reqs)); |
|
1065 | + |
|
1066 | + $this->AddQuery($query, $index, $comment); |
|
1067 | + $results = $this->RunQueries (); |
|
1068 | + $this->_reqs = array (); // just in case it failed too early |
|
1069 | + |
|
1070 | + if (!is_array($results)) |
|
1071 | + return false; // probably network error; error message should be already filled |
|
1072 | + |
|
1073 | + $this->_error = $results[0]["error"]; |
|
1074 | + $this->_warning = $results[0]["warning"]; |
|
1075 | + if ($results[0]["status"]==SEARCHD_ERROR) |
|
1076 | + return false; |
|
1077 | + else |
|
1078 | + return $results[0]; |
|
1079 | + } |
|
1080 | + |
|
1081 | + /// helper to pack floats in network byte order |
|
1082 | + function _PackFloat($f) |
|
1083 | + { |
|
1084 | + $t1 = pack("f", $f); // machine order |
|
1085 | + list(,$t2) = unpack("L*", $t1); // int in machine order |
|
1086 | + return pack("N", $t2); |
|
1087 | + } |
|
1088 | + |
|
1089 | + /// add query to multi-query batch |
|
1090 | + /// returns index into results array from RunQueries() call |
|
1091 | + function AddQuery($query, $index="*", $comment="") |
|
1092 | + { |
|
1093 | + // mbstring workaround |
|
1094 | + $this->_MBPush (); |
|
1095 | + |
|
1096 | + // build request |
|
1097 | + $req = pack("NNNNN", $this->_query_flags, $this->_offset, $this->_limit, $this->_mode, $this->_ranker); |
|
1098 | + if ($this->_ranker==SPH_RANK_EXPR) |
|
1099 | + $req .= pack("N", strlen($this->_rankexpr)) . $this->_rankexpr; |
|
1100 | + $req .= pack("N", $this->_sort); // (deprecated) sort mode |
|
1101 | + $req .= pack("N", strlen($this->_sortby)) . $this->_sortby; |
|
1102 | + $req .= pack("N", strlen($query)) . $query; // query itself |
|
1103 | + $req .= pack("N", count($this->_weights)); // weights |
|
1104 | + foreach ($this->_weights as $weight) |
|
1105 | + $req .= pack("N", (int)$weight); |
|
1106 | + $req .= pack("N", strlen($index)) . $index; // indexes |
|
1107 | + $req .= pack("N", 1); // id64 range marker |
|
1108 | + $req .= sphPackU64($this->_min_id) . sphPackU64($this->_max_id); // id64 range |
|
1109 | + |
|
1110 | + // filters |
|
1111 | + $req .= pack("N", count($this->_filters)); |
|
1112 | + foreach ($this->_filters as $filter) |
|
1113 | + { |
|
1114 | + $req .= pack("N", strlen($filter["attr"])) . $filter["attr"]; |
|
1115 | + $req .= pack("N", $filter["type"]); |
|
1116 | + switch ($filter["type"]) |
|
1117 | + { |
|
1118 | + case SPH_FILTER_VALUES: |
|
1119 | + $req .= pack("N", count($filter["values"])); |
|
1120 | + foreach ($filter["values"] as $value) |
|
1121 | + $req .= sphPackI64($value); |
|
1122 | + break; |
|
1123 | + |
|
1124 | + case SPH_FILTER_RANGE: |
|
1125 | + $req .= sphPackI64($filter["min"]) . sphPackI64($filter["max"]); |
|
1126 | + break; |
|
1127 | + |
|
1128 | + case SPH_FILTER_FLOATRANGE: |
|
1129 | + $req .= $this->_PackFloat($filter["min"]) . $this->_PackFloat($filter["max"]); |
|
1130 | + break; |
|
1131 | + |
|
1132 | + case SPH_FILTER_STRING: |
|
1133 | + $req .= pack("N", strlen($filter["value"])) . $filter["value"]; |
|
1134 | + break; |
|
1135 | + |
|
1136 | + default: |
|
1137 | + assert(0 && "internal error: unhandled filter type"); |
|
1138 | + } |
|
1139 | + $req .= pack("N", $filter["exclude"]); |
|
1140 | + } |
|
1141 | + |
|
1142 | + // group-by clause, max-matches count, group-sort clause, cutoff count |
|
1143 | + $req .= pack("NN", $this->_groupfunc, strlen($this->_groupby)) . $this->_groupby; |
|
1144 | + $req .= pack("N", $this->_maxmatches); |
|
1145 | + $req .= pack("N", strlen($this->_groupsort)) . $this->_groupsort; |
|
1146 | + $req .= pack("NNN", $this->_cutoff, $this->_retrycount, $this->_retrydelay); |
|
1147 | + $req .= pack("N", strlen($this->_groupdistinct)) . $this->_groupdistinct; |
|
1148 | + |
|
1149 | + // anchor point |
|
1150 | + if (empty($this->_anchor)) |
|
1151 | + { |
|
1152 | + $req .= pack("N", 0); |
|
1153 | + } else |
|
1154 | + { |
|
1155 | + $a =& $this->_anchor; |
|
1156 | + $req .= pack("N", 1); |
|
1157 | + $req .= pack("N", strlen($a["attrlat"])) . $a["attrlat"]; |
|
1158 | + $req .= pack("N", strlen($a["attrlong"])) . $a["attrlong"]; |
|
1159 | + $req .= $this->_PackFloat($a["lat"]) . $this->_PackFloat($a["long"]); |
|
1160 | + } |
|
1161 | + |
|
1162 | + // per-index weights |
|
1163 | + $req .= pack("N", count($this->_indexweights)); |
|
1164 | + foreach ($this->_indexweights as $idx=>$weight) |
|
1165 | + $req .= pack("N", strlen($idx)) . $idx . pack("N", $weight); |
|
1166 | + |
|
1167 | + // max query time |
|
1168 | + $req .= pack("N", $this->_maxquerytime); |
|
1169 | + |
|
1170 | + // per-field weights |
|
1171 | + $req .= pack("N", count($this->_fieldweights)); |
|
1172 | + foreach ($this->_fieldweights as $field=>$weight) |
|
1173 | + $req .= pack("N", strlen($field)) . $field . pack("N", $weight); |
|
1174 | + |
|
1175 | + // comment |
|
1176 | + $req .= pack("N", strlen($comment)) . $comment; |
|
1177 | + |
|
1178 | + // attribute overrides |
|
1179 | + $req .= pack("N", count($this->_overrides)); |
|
1180 | + foreach ($this->_overrides as $key => $entry) |
|
1181 | + { |
|
1182 | + $req .= pack("N", strlen($entry["attr"])) . $entry["attr"]; |
|
1183 | + $req .= pack("NN", $entry["type"], count($entry["values"])); |
|
1184 | + foreach ($entry["values"] as $id=>$val) |
|
1185 | + { |
|
1186 | + assert(is_numeric($id)); |
|
1187 | + assert(is_numeric($val)); |
|
1188 | + |
|
1189 | + $req .= sphPackU64($id); |
|
1190 | + switch ($entry["type"]) |
|
1191 | + { |
|
1192 | + case SPH_ATTR_FLOAT: $req .= $this->_PackFloat($val); break; |
|
1193 | + case SPH_ATTR_BIGINT: $req .= sphPackI64($val); break; |
|
1194 | + default: $req .= pack("N", $val); break; |
|
1195 | + } |
|
1196 | + } |
|
1197 | + } |
|
1198 | + |
|
1199 | + // select-list |
|
1200 | + $req .= pack("N", strlen($this->_select)) . $this->_select; |
|
1201 | + |
|
1202 | + // max_predicted_time |
|
1203 | + if ($this->_predictedtime>0) |
|
1204 | + $req .= pack("N", (int)$this->_predictedtime); |
|
1205 | + |
|
1206 | + $req .= pack("N", strlen($this->_outerorderby)) . $this->_outerorderby; |
|
1207 | + $req .= pack("NN", $this->_outeroffset, $this->_outerlimit); |
|
1208 | + if ($this->_hasouter) |
|
1209 | + $req .= pack("N", 1); |
|
1210 | + else |
|
1211 | + $req .= pack("N", 0); |
|
1212 | + |
|
1213 | + // mbstring workaround |
|
1214 | + $this->_MBPop (); |
|
1215 | + |
|
1216 | + // store request to requests array |
|
1217 | + $this->_reqs[] = $req; |
|
1218 | + return count($this->_reqs)-1; |
|
1219 | + } |
|
1220 | + |
|
1221 | + /// connect to searchd, run queries batch, and return an array of result sets |
|
1222 | + function RunQueries () |
|
1223 | + { |
|
1224 | + if (empty($this->_reqs)) |
|
1225 | + { |
|
1226 | + $this->_error = "no queries defined, issue AddQuery() first"; |
|
1227 | + return false; |
|
1228 | + } |
|
1229 | + |
|
1230 | + // mbstring workaround |
|
1231 | + $this->_MBPush (); |
|
1232 | + |
|
1233 | + if (!( $fp = $this->_Connect())) |
|
1234 | + { |
|
1235 | + $this->_MBPop (); |
|
1236 | + return false; |
|
1237 | + } |
|
1238 | + |
|
1239 | + // send query, get response |
|
1240 | + $nreqs = count($this->_reqs); |
|
1241 | + $req = join("", $this->_reqs); |
|
1242 | + $len = 8+strlen($req); |
|
1243 | + $req = pack("nnNNN", SEARCHD_COMMAND_SEARCH, VER_COMMAND_SEARCH, $len, 0, $nreqs) . $req; // add header |
|
1244 | + |
|
1245 | + if (!( $this->_Send($fp, $req, $len+8)) || |
|
1246 | + !( $response = $this->_GetResponse($fp, VER_COMMAND_SEARCH))) |
|
1247 | + { |
|
1248 | + $this->_MBPop (); |
|
1249 | + return false; |
|
1250 | + } |
|
1251 | + |
|
1252 | + // query sent ok; we can reset reqs now |
|
1253 | + $this->_reqs = array (); |
|
1254 | + |
|
1255 | + // parse and return response |
|
1256 | + return $this->_ParseSearchResponse($response, $nreqs); |
|
1257 | + } |
|
1258 | + |
|
1259 | + /// parse and return search query (or queries) response |
|
1260 | + function _ParseSearchResponse($response, $nreqs) |
|
1261 | + { |
|
1262 | + $p = 0; // current position |
|
1263 | + $max = strlen($response); // max position for checks, to protect against broken responses |
|
1264 | + |
|
1265 | + $results = array (); |
|
1266 | + for($ires=0; $ires<$nreqs && $p<$max; $ires++) |
|
1267 | + { |
|
1268 | + $results[] = array(); |
|
1269 | + $result =& $results[$ires]; |
|
1270 | + |
|
1271 | + $result["error"] = ""; |
|
1272 | + $result["warning"] = ""; |
|
1273 | + |
|
1274 | + // extract status |
|
1275 | + list(,$status) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1276 | + $result["status"] = $status; |
|
1277 | + if ($status!=SEARCHD_OK) |
|
1278 | + { |
|
1279 | + list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1280 | + $message = substr($response, $p, $len); $p += $len; |
|
1281 | + |
|
1282 | + if ($status==SEARCHD_WARNING) |
|
1283 | + { |
|
1284 | + $result["warning"] = $message; |
|
1285 | + } else |
|
1286 | + { |
|
1287 | + $result["error"] = $message; |
|
1288 | + continue; |
|
1289 | + } |
|
1290 | + } |
|
1291 | + |
|
1292 | + // read schema |
|
1293 | + $fields = array (); |
|
1294 | + $attrs = array (); |
|
1295 | + |
|
1296 | + list(,$nfields) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1297 | + while ($nfields-->0 && $p<$max) |
|
1298 | + { |
|
1299 | + list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1300 | + $fields[] = substr($response, $p, $len); $p += $len; |
|
1301 | + } |
|
1302 | + $result["fields"] = $fields; |
|
1303 | + |
|
1304 | + list(,$nattrs) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1305 | + while ($nattrs-->0 && $p<$max ) |
|
1306 | + { |
|
1307 | + list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1308 | + $attr = substr($response, $p, $len); $p += $len; |
|
1309 | + list(,$type) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1310 | + $attrs[$attr] = $type; |
|
1311 | + } |
|
1312 | + $result["attrs"] = $attrs; |
|
1313 | + |
|
1314 | + // read match count |
|
1315 | + list(,$count) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1316 | + list(,$id64) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1317 | + |
|
1318 | + // read matches |
|
1319 | + $idx = -1; |
|
1320 | + while ($count-->0 && $p<$max) |
|
1321 | + { |
|
1322 | + // index into result array |
|
1323 | + $idx++; |
|
1324 | + |
|
1325 | + // parse document id and weight |
|
1326 | + if ($id64) |
|
1327 | + { |
|
1328 | + $doc = sphUnpackU64(substr($response, $p, 8)); $p += 8; |
|
1329 | + list(,$weight) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1330 | + } |
|
1331 | + else |
|
1332 | + { |
|
1333 | + list($doc, $weight) = array_values(unpack("N*N*", |
|
1334 | + substr($response, $p, 8))); |
|
1335 | + $p += 8; |
|
1336 | + $doc = sphFixUint($doc); |
|
1337 | + } |
|
1338 | + $weight = sprintf("%u", $weight); |
|
1339 | + |
|
1340 | + // create match entry |
|
1341 | + if ($this->_arrayresult) |
|
1342 | + $result["matches"][$idx] = array("id"=>$doc, "weight"=>$weight); |
|
1343 | + else |
|
1344 | + $result["matches"][$doc]["weight"] = $weight; |
|
1345 | + |
|
1346 | + // parse and create attributes |
|
1347 | + $attrvals = array (); |
|
1348 | + foreach ($attrs as $attr=>$type) |
|
1349 | + { |
|
1350 | + // handle 64bit ints |
|
1351 | + if ($type==SPH_ATTR_BIGINT) |
|
1352 | + { |
|
1353 | + $attrvals[$attr] = sphUnpackI64(substr($response, $p, 8)); $p += 8; |
|
1354 | + continue; |
|
1355 | + } |
|
1356 | + |
|
1357 | + // handle floats |
|
1358 | + if ($type==SPH_ATTR_FLOAT) |
|
1359 | + { |
|
1360 | + list(,$uval) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1361 | + list(,$fval) = unpack("f*", pack("L", $uval)); |
|
1362 | + $attrvals[$attr] = $fval; |
|
1363 | + continue; |
|
1364 | + } |
|
1365 | + |
|
1366 | + // handle everything else as unsigned ints |
|
1367 | + list(,$val) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1368 | + if ($type==SPH_ATTR_MULTI) |
|
1369 | + { |
|
1370 | + $attrvals[$attr] = array (); |
|
1371 | + $nvalues = $val; |
|
1372 | + while ($nvalues-->0 && $p<$max) |
|
1373 | + { |
|
1374 | + list(,$val) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1375 | + $attrvals[$attr][] = sphFixUint($val); |
|
1376 | + } |
|
1377 | + } else if ($type==SPH_ATTR_MULTI64) |
|
1378 | + { |
|
1379 | + $attrvals[$attr] = array (); |
|
1380 | + $nvalues = $val; |
|
1381 | + while ($nvalues>0 && $p<$max) |
|
1382 | + { |
|
1383 | + $attrvals[$attr][] = sphUnpackI64(substr($response, $p, 8)); $p += 8; |
|
1384 | + $nvalues -= 2; |
|
1385 | + } |
|
1386 | + } else if ($type==SPH_ATTR_STRING) |
|
1387 | + { |
|
1388 | + $attrvals[$attr] = substr($response, $p, $val); |
|
1389 | + $p += $val; |
|
1390 | + } else if ($type==SPH_ATTR_FACTORS) |
|
1391 | + { |
|
1392 | + $attrvals[$attr] = substr($response, $p, $val-4); |
|
1393 | + $p += $val-4; |
|
1394 | + } else |
|
1395 | + { |
|
1396 | + $attrvals[$attr] = sphFixUint($val); |
|
1397 | + } |
|
1398 | + } |
|
1399 | + |
|
1400 | + if ($this->_arrayresult) |
|
1401 | + $result["matches"][$idx]["attrs"] = $attrvals; |
|
1402 | + else |
|
1403 | + $result["matches"][$doc]["attrs"] = $attrvals; |
|
1404 | + } |
|
1405 | + |
|
1406 | + list($total, $total_found, $msecs, $words) = |
|
1407 | + array_values(unpack("N*N*N*N*", substr($response, $p, 16))); |
|
1408 | + $result["total"] = sprintf("%u", $total); |
|
1409 | + $result["total_found"] = sprintf("%u", $total_found); |
|
1410 | + $result["time"] = sprintf("%.3f", $msecs/1000); |
|
1411 | + $p += 16; |
|
1412 | + |
|
1413 | + while ($words-->0 && $p<$max) |
|
1414 | + { |
|
1415 | + list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1416 | + $word = substr($response, $p, $len); $p += $len; |
|
1417 | + list($docs, $hits) = array_values(unpack("N*N*", substr($response, $p, 8))); $p += 8; |
|
1418 | + $result["words"][$word] = array ( |
|
1419 | + "docs"=>sprintf("%u", $docs), |
|
1420 | + "hits"=>sprintf("%u", $hits)); |
|
1421 | + } |
|
1422 | + } |
|
1423 | + |
|
1424 | + $this->_MBPop (); |
|
1425 | + return $results; |
|
1426 | + } |
|
1427 | + |
|
1428 | + ///////////////////////////////////////////////////////////////////////////// |
|
1429 | + // excerpts generation |
|
1430 | + ///////////////////////////////////////////////////////////////////////////// |
|
1431 | + |
|
1432 | + /// connect to searchd server, and generate exceprts (snippets) |
|
1433 | + /// of given documents for given query. returns false on failure, |
|
1434 | + /// an array of snippets on success |
|
1435 | + function BuildExcerpts($docs, $index, $words, $opts=array()) |
|
1436 | + { |
|
1437 | + assert(is_array($docs)); |
|
1438 | + assert(is_string($index)); |
|
1439 | + assert(is_string($words)); |
|
1440 | + assert(is_array($opts)); |
|
1441 | + |
|
1442 | + $this->_MBPush (); |
|
1443 | + |
|
1444 | + if (!( $fp = $this->_Connect())) |
|
1445 | + { |
|
1446 | + $this->_MBPop(); |
|
1447 | + return false; |
|
1448 | + } |
|
1449 | + |
|
1450 | + ///////////////// |
|
1451 | + // fixup options |
|
1452 | + ///////////////// |
|
1453 | + |
|
1454 | + if (!isset($opts["before_match"])) $opts["before_match"] = "<b>"; |
|
1455 | + if (!isset($opts["after_match"])) $opts["after_match"] = "</b>"; |
|
1456 | + if (!isset($opts["chunk_separator"])) $opts["chunk_separator"] = " ... "; |
|
1457 | + if (!isset($opts["limit"])) $opts["limit"] = 256; |
|
1458 | + if (!isset($opts["limit_passages"])) $opts["limit_passages"] = 0; |
|
1459 | + if (!isset($opts["limit_words"])) $opts["limit_words"] = 0; |
|
1460 | + if (!isset($opts["around"])) $opts["around"] = 5; |
|
1461 | + if (!isset($opts["exact_phrase"])) $opts["exact_phrase"] = false; |
|
1462 | + if (!isset($opts["single_passage"])) $opts["single_passage"] = false; |
|
1463 | + if (!isset($opts["use_boundaries"])) $opts["use_boundaries"] = false; |
|
1464 | + if (!isset($opts["weight_order"])) $opts["weight_order"] = false; |
|
1465 | + if (!isset($opts["query_mode"])) $opts["query_mode"] = false; |
|
1466 | + if (!isset($opts["force_all_words"])) $opts["force_all_words"] = false; |
|
1467 | + if (!isset($opts["start_passage_id"])) $opts["start_passage_id"] = 1; |
|
1468 | + if (!isset($opts["load_files"])) $opts["load_files"] = false; |
|
1469 | + if (!isset($opts["html_strip_mode"])) $opts["html_strip_mode"] = "index"; |
|
1470 | + if (!isset($opts["allow_empty"])) $opts["allow_empty"] = false; |
|
1471 | + if (!isset($opts["passage_boundary"])) $opts["passage_boundary"] = "none"; |
|
1472 | + if (!isset($opts["emit_zones"])) $opts["emit_zones"] = false; |
|
1473 | + if (!isset($opts["load_files_scattered"])) $opts["load_files_scattered"] = false; |
|
1474 | + |
|
1475 | + |
|
1476 | + ///////////////// |
|
1477 | + // build request |
|
1478 | + ///////////////// |
|
1479 | + |
|
1480 | + // v.1.2 req |
|
1481 | + $flags = 1; // remove spaces |
|
1482 | + if ($opts["exact_phrase"]) $flags |= 2; |
|
1483 | + if ($opts["single_passage"]) $flags |= 4; |
|
1484 | + if ($opts["use_boundaries"]) $flags |= 8; |
|
1485 | + if ($opts["weight_order"]) $flags |= 16; |
|
1486 | + if ($opts["query_mode"]) $flags |= 32; |
|
1487 | + if ($opts["force_all_words"]) $flags |= 64; |
|
1488 | + if ($opts["load_files"]) $flags |= 128; |
|
1489 | + if ($opts["allow_empty"]) $flags |= 256; |
|
1490 | + if ($opts["emit_zones"]) $flags |= 512; |
|
1491 | + if ($opts["load_files_scattered"]) $flags |= 1024; |
|
1492 | + $req = pack("NN", 0, $flags); // mode=0, flags=$flags |
|
1493 | + $req .= pack("N", strlen($index)) . $index; // req index |
|
1494 | + $req .= pack("N", strlen($words)) . $words; // req words |
|
1495 | + |
|
1496 | + // options |
|
1497 | + $req .= pack("N", strlen($opts["before_match"])) . $opts["before_match"]; |
|
1498 | + $req .= pack("N", strlen($opts["after_match"])) . $opts["after_match"]; |
|
1499 | + $req .= pack("N", strlen($opts["chunk_separator"])) . $opts["chunk_separator"]; |
|
1500 | + $req .= pack("NN", (int)$opts["limit"], (int)$opts["around"]); |
|
1501 | + $req .= pack("NNN", (int)$opts["limit_passages"], (int)$opts["limit_words"], (int)$opts["start_passage_id"]); // v.1.2 |
|
1502 | + $req .= pack("N", strlen($opts["html_strip_mode"])) . $opts["html_strip_mode"]; |
|
1503 | + $req .= pack("N", strlen($opts["passage_boundary"])) . $opts["passage_boundary"]; |
|
1504 | + |
|
1505 | + // documents |
|
1506 | + $req .= pack("N", count($docs)); |
|
1507 | + foreach ($docs as $doc) |
|
1508 | + { |
|
1509 | + assert(is_string($doc)); |
|
1510 | + $req .= pack("N", strlen($doc)) . $doc; |
|
1511 | + } |
|
1512 | + |
|
1513 | + //////////////////////////// |
|
1514 | + // send query, get response |
|
1515 | + //////////////////////////// |
|
1516 | + |
|
1517 | + $len = strlen($req); |
|
1518 | + $req = pack("nnN", SEARCHD_COMMAND_EXCERPT, VER_COMMAND_EXCERPT, $len) . $req; // add header |
|
1519 | + if (!( $this->_Send($fp, $req, $len+8)) || |
|
1520 | + !( $response = $this->_GetResponse($fp, VER_COMMAND_EXCERPT))) |
|
1521 | + { |
|
1522 | + $this->_MBPop (); |
|
1523 | + return false; |
|
1524 | + } |
|
1525 | + |
|
1526 | + ////////////////// |
|
1527 | + // parse response |
|
1528 | + ////////////////// |
|
1529 | + |
|
1530 | + $pos = 0; |
|
1531 | + $res = array (); |
|
1532 | + $rlen = strlen($response); |
|
1533 | + for($i=0; $i<count($docs); $i++) |
|
1534 | + { |
|
1535 | + list(,$len) = unpack("N*", substr($response, $pos, 4)); |
|
1536 | + $pos += 4; |
|
1537 | + |
|
1538 | + if ($pos+$len > $rlen) |
|
1539 | + { |
|
1540 | + $this->_error = "incomplete reply"; |
|
1541 | + $this->_MBPop (); |
|
1542 | + return false; |
|
1543 | + } |
|
1544 | + $res[] = $len ? substr($response, $pos, $len) : ""; |
|
1545 | + $pos += $len; |
|
1546 | + } |
|
1547 | + |
|
1548 | + $this->_MBPop (); |
|
1549 | + return $res; |
|
1550 | + } |
|
1551 | + |
|
1552 | + |
|
1553 | + ///////////////////////////////////////////////////////////////////////////// |
|
1554 | + // keyword generation |
|
1555 | + ///////////////////////////////////////////////////////////////////////////// |
|
1556 | + |
|
1557 | + /// connect to searchd server, and generate keyword list for a given query |
|
1558 | + /// returns false on failure, |
|
1559 | + /// an array of words on success |
|
1560 | + function BuildKeywords($query, $index, $hits) |
|
1561 | + { |
|
1562 | + assert(is_string($query)); |
|
1563 | + assert(is_string($index)); |
|
1564 | + assert(is_bool($hits)); |
|
1565 | + |
|
1566 | + $this->_MBPush (); |
|
1567 | + |
|
1568 | + if (!( $fp = $this->_Connect())) |
|
1569 | + { |
|
1570 | + $this->_MBPop(); |
|
1571 | + return false; |
|
1572 | + } |
|
1573 | + |
|
1574 | + ///////////////// |
|
1575 | + // build request |
|
1576 | + ///////////////// |
|
1577 | + |
|
1578 | + // v.1.0 req |
|
1579 | + $req = pack("N", strlen($query)) . $query; // req query |
|
1580 | + $req .= pack("N", strlen($index)) . $index; // req index |
|
1581 | + $req .= pack("N", (int)$hits); |
|
1582 | + |
|
1583 | + //////////////////////////// |
|
1584 | + // send query, get response |
|
1585 | + //////////////////////////// |
|
1586 | + |
|
1587 | + $len = strlen($req); |
|
1588 | + $req = pack("nnN", SEARCHD_COMMAND_KEYWORDS, VER_COMMAND_KEYWORDS, $len) . $req; // add header |
|
1589 | + if (!( $this->_Send($fp, $req, $len+8)) || |
|
1590 | + !( $response = $this->_GetResponse($fp, VER_COMMAND_KEYWORDS))) |
|
1591 | + { |
|
1592 | + $this->_MBPop (); |
|
1593 | + return false; |
|
1594 | + } |
|
1595 | + |
|
1596 | + ////////////////// |
|
1597 | + // parse response |
|
1598 | + ////////////////// |
|
1599 | + |
|
1600 | + $pos = 0; |
|
1601 | + $res = array (); |
|
1602 | + $rlen = strlen($response); |
|
1603 | + list(,$nwords) = unpack("N*", substr($response, $pos, 4)); |
|
1604 | + $pos += 4; |
|
1605 | + for($i=0; $i<$nwords; $i++) |
|
1606 | + { |
|
1607 | + list(,$len) = unpack("N*", substr($response, $pos, 4)); $pos += 4; |
|
1608 | + $tokenized = $len ? substr($response, $pos, $len) : ""; |
|
1609 | + $pos += $len; |
|
1610 | + |
|
1611 | + list(,$len) = unpack("N*", substr($response, $pos, 4)); $pos += 4; |
|
1612 | + $normalized = $len ? substr($response, $pos, $len) : ""; |
|
1613 | + $pos += $len; |
|
1614 | + |
|
1615 | + $res[] = array("tokenized"=>$tokenized, "normalized"=>$normalized); |
|
1616 | + |
|
1617 | + if ($hits) |
|
1618 | + { |
|
1619 | + list($ndocs,$nhits) = array_values(unpack("N*N*", substr($response, $pos, 8))); |
|
1620 | + $pos += 8; |
|
1621 | + $res [$i]["docs"] = $ndocs; |
|
1622 | + $res [$i]["hits"] = $nhits; |
|
1623 | + } |
|
1624 | + |
|
1625 | + if ($pos > $rlen) |
|
1626 | + { |
|
1627 | + $this->_error = "incomplete reply"; |
|
1628 | + $this->_MBPop (); |
|
1629 | + return false; |
|
1630 | + } |
|
1631 | + } |
|
1632 | + |
|
1633 | + $this->_MBPop (); |
|
1634 | + return $res; |
|
1635 | + } |
|
1636 | + |
|
1637 | + function EscapeString($string) |
|
1638 | + { |
|
1639 | + $from = array('\\', '(',')','|','-','!','@','~','"','&', '/', '^', '$', '=', '<'); |
|
1640 | + $to = array('\\\\', '\(','\)','\|','\-','\!','\@','\~','\"', '\&', '\/', '\^', '\$', '\=', '\<'); |
|
1641 | + |
|
1642 | + return str_replace($from, $to, $string); |
|
1643 | + } |
|
1644 | + |
|
1645 | + ///////////////////////////////////////////////////////////////////////////// |
|
1646 | + // attribute updates |
|
1647 | + ///////////////////////////////////////////////////////////////////////////// |
|
1648 | + |
|
1649 | + /// batch update given attributes in given rows in given indexes |
|
1650 | + /// returns amount of updated documents (0 or more) on success, or -1 on failure |
|
1651 | + function UpdateAttributes($index, $attrs, $values, $mva=false, $ignorenonexistent=false) |
|
1652 | + { |
|
1653 | + // verify everything |
|
1654 | + assert(is_string($index)); |
|
1655 | + assert(is_bool($mva)); |
|
1656 | + assert(is_bool($ignorenonexistent)); |
|
1657 | + |
|
1658 | + assert(is_array($attrs)); |
|
1659 | + foreach ($attrs as $attr) |
|
1660 | + assert(is_string($attr)); |
|
1661 | + |
|
1662 | + assert(is_array($values)); |
|
1663 | + foreach ($values as $id=>$entry) |
|
1664 | + { |
|
1665 | + assert(is_numeric($id)); |
|
1666 | + assert(is_array($entry)); |
|
1667 | + assert(count($entry)==count($attrs)); |
|
1668 | + foreach ($entry as $v) |
|
1669 | + { |
|
1670 | + if ($mva) |
|
1671 | + { |
|
1672 | + assert(is_array($v)); |
|
1673 | + foreach ($v as $vv) |
|
1674 | + assert(is_int($vv)); |
|
1675 | + } else |
|
1676 | + assert(is_int($v)); |
|
1677 | + } |
|
1678 | + } |
|
1679 | + |
|
1680 | + // build request |
|
1681 | + $this->_MBPush (); |
|
1682 | + $req = pack("N", strlen($index)) . $index; |
|
1683 | + |
|
1684 | + $req .= pack("N", count($attrs)); |
|
1685 | + $req .= pack("N", $ignorenonexistent ? 1 : 0); |
|
1686 | + foreach ($attrs as $attr) |
|
1687 | + { |
|
1688 | + $req .= pack("N", strlen($attr)) . $attr; |
|
1689 | + $req .= pack("N", $mva ? 1 : 0); |
|
1690 | + } |
|
1691 | + |
|
1692 | + $req .= pack("N", count($values)); |
|
1693 | + foreach ($values as $id=>$entry) |
|
1694 | + { |
|
1695 | + $req .= sphPackU64($id); |
|
1696 | + foreach ($entry as $v) |
|
1697 | + { |
|
1698 | + $req .= pack("N", $mva ? count($v) : $v); |
|
1699 | + if ($mva) |
|
1700 | + foreach ($v as $vv) |
|
1701 | + $req .= pack("N", $vv); |
|
1702 | + } |
|
1703 | + } |
|
1704 | + |
|
1705 | + // connect, send query, get response |
|
1706 | + if (!( $fp = $this->_Connect())) |
|
1707 | + { |
|
1708 | + $this->_MBPop (); |
|
1709 | + return -1; |
|
1710 | + } |
|
1711 | + |
|
1712 | + $len = strlen($req); |
|
1713 | + $req = pack("nnN", SEARCHD_COMMAND_UPDATE, VER_COMMAND_UPDATE, $len) . $req; // add header |
|
1714 | + if (!$this->_Send($fp, $req, $len+8)) |
|
1715 | + { |
|
1716 | + $this->_MBPop (); |
|
1717 | + return -1; |
|
1718 | + } |
|
1719 | + |
|
1720 | + if (!( $response = $this->_GetResponse($fp, VER_COMMAND_UPDATE))) |
|
1721 | + { |
|
1722 | + $this->_MBPop (); |
|
1723 | + return -1; |
|
1724 | + } |
|
1725 | + |
|
1726 | + // parse response |
|
1727 | + list(,$updated) = unpack("N*", substr($response, 0, 4)); |
|
1728 | + $this->_MBPop (); |
|
1729 | + return $updated; |
|
1730 | + } |
|
1731 | + |
|
1732 | + ///////////////////////////////////////////////////////////////////////////// |
|
1733 | + // persistent connections |
|
1734 | + ///////////////////////////////////////////////////////////////////////////// |
|
1735 | + |
|
1736 | + function Open() |
|
1737 | + { |
|
1738 | + if ($this->_socket !== false) |
|
1739 | + { |
|
1740 | + $this->_error = 'already connected'; |
|
1741 | + return false; |
|
1742 | + } |
|
1743 | + if (!$fp = $this->_Connect()) |
|
1744 | + return false; |
|
1745 | + |
|
1746 | + // command, command version = 0, body length = 4, body = 1 |
|
1747 | + $req = pack("nnNN", SEARCHD_COMMAND_PERSIST, 0, 4, 1); |
|
1748 | + if (!$this->_Send($fp, $req, 12)) |
|
1749 | + return false; |
|
1750 | + |
|
1751 | + $this->_socket = $fp; |
|
1752 | + return true; |
|
1753 | + } |
|
1754 | + |
|
1755 | + function Close() |
|
1756 | + { |
|
1757 | + if ($this->_socket === false) |
|
1758 | + { |
|
1759 | + $this->_error = 'not connected'; |
|
1760 | + return false; |
|
1761 | + } |
|
1762 | + |
|
1763 | + fclose($this->_socket); |
|
1764 | + $this->_socket = false; |
|
1765 | + |
|
1766 | + return true; |
|
1767 | + } |
|
1768 | + |
|
1769 | + ////////////////////////////////////////////////////////////////////////// |
|
1770 | + // status |
|
1771 | + ////////////////////////////////////////////////////////////////////////// |
|
1772 | + |
|
1773 | + function Status ($session=false) |
|
1774 | + { |
|
1775 | + assert(is_bool($session)); |
|
1776 | + |
|
1777 | + $this->_MBPush (); |
|
1778 | + if (!( $fp = $this->_Connect())) |
|
1779 | + { |
|
1780 | + $this->_MBPop(); |
|
1781 | + return false; |
|
1782 | + } |
|
1783 | + |
|
1784 | + $req = pack("nnNN", SEARCHD_COMMAND_STATUS, VER_COMMAND_STATUS, 4, $session?0:1); // len=4, body=1 |
|
1785 | + if (!( $this->_Send($fp, $req, 12)) || |
|
1786 | + !( $response = $this->_GetResponse($fp, VER_COMMAND_STATUS))) |
|
1787 | + { |
|
1788 | + $this->_MBPop (); |
|
1789 | + return false; |
|
1790 | + } |
|
1791 | + |
|
1792 | + $res = substr($response, 4); // just ignore length, error handling, etc |
|
1793 | + $p = 0; |
|
1794 | + list($rows, $cols) = array_values(unpack("N*N*", substr($response, $p, 8))); $p += 8; |
|
1795 | + |
|
1796 | + $res = array(); |
|
1797 | + for($i=0; $i<$rows; $i++) |
|
1798 | + for($j=0; $j<$cols; $j++) |
|
1799 | + { |
|
1800 | + list(,$len) = unpack("N*", substr($response, $p, 4)); $p += 4; |
|
1801 | + $res[$i][] = substr($response, $p, $len); $p += $len; |
|
1802 | + } |
|
1803 | + |
|
1804 | + $this->_MBPop (); |
|
1805 | + return $res; |
|
1806 | + } |
|
1807 | + |
|
1808 | + ////////////////////////////////////////////////////////////////////////// |
|
1809 | + // flush |
|
1810 | + ////////////////////////////////////////////////////////////////////////// |
|
1811 | + |
|
1812 | + function FlushAttributes () |
|
1813 | + { |
|
1814 | + $this->_MBPush (); |
|
1815 | + if (!( $fp = $this->_Connect())) |
|
1816 | + { |
|
1817 | + $this->_MBPop(); |
|
1818 | + return -1; |
|
1819 | + } |
|
1820 | + |
|
1821 | + $req = pack("nnN", SEARCHD_COMMAND_FLUSHATTRS, VER_COMMAND_FLUSHATTRS, 0); // len=0 |
|
1822 | + if (!( $this->_Send($fp, $req, 8)) || |
|
1823 | + !( $response = $this->_GetResponse($fp, VER_COMMAND_FLUSHATTRS))) |
|
1824 | + { |
|
1825 | + $this->_MBPop (); |
|
1826 | + return -1; |
|
1827 | + } |
|
1828 | + |
|
1829 | + $tag = -1; |
|
1830 | + if (strlen($response)==4) |
|
1831 | + list(,$tag) = unpack("N*", $response); |
|
1832 | + else |
|
1833 | + $this->_error = "unexpected response length"; |
|
1834 | + |
|
1835 | + $this->_MBPop (); |
|
1836 | + return $tag; |
|
1837 | + } |
|
1838 | 1838 | } |
1839 | 1839 | |
1840 | 1840 | // |