1 |
/** |
2 |
* HyperEstraierWrapper.cpp - C++ wrapper for Hyper Estraier |
3 |
*/ |
4 |
#include <estraier.h> |
5 |
#include <estmtdb.h> |
6 |
#include <cabin.h> |
7 |
#include <cstdlib> |
8 |
#include <string> |
9 |
#include <vector> |
10 |
#include <map> |
11 |
#include <cassert> |
12 |
#include <stdexcept> |
13 |
|
14 |
/* backward compatibility for 0.5.4 */ |
15 |
#ifndef ESTCONDAGITO |
16 |
#define ESTCONDAGITO ESTCONDAGIT |
17 |
#endif |
18 |
|
19 |
namespace estraier { |
20 |
|
21 |
class IOError : public std::runtime_error { |
22 |
public: |
23 |
explicit IOError (const std::string& w) : std::runtime_error(w) {} |
24 |
}; |
25 |
|
26 |
class Condition { |
27 |
public: |
28 |
enum { // enumeration for options |
29 |
SURE = ESTCONDSURE, // check every N-gram key |
30 |
USUAL = ESTCONDUSUAL, // check N-gram keys skipping by one |
31 |
FAST = ESTCONDFAST, // check N-gram keys skipping by two |
32 |
AGITO = ESTCONDAGITO, // check N-gram keys skipping by three |
33 |
NOIDF = ESTCONDNOIDF, // without TF-IDF tuning |
34 |
SIMPLE = ESTCONDSIMPLE, // with the simplefied phrase |
35 |
}; |
36 |
ESTCOND * cond; |
37 |
Condition() { |
38 |
/** |
39 |
* constructor |
40 |
*/ |
41 |
cond = est_cond_new(); |
42 |
} |
43 |
~Condition() { |
44 |
/** |
45 |
* destructor |
46 |
*/ |
47 |
est_cond_delete(cond); |
48 |
} |
49 |
void set_phrase(const char *phrase) { |
50 |
/** |
51 |
* set the search phrase |
52 |
*/ |
53 |
est_cond_set_phrase(cond, phrase); |
54 |
} |
55 |
void add_attr(const char *expr) { |
56 |
/** |
57 |
* set the attribute expression |
58 |
*/ |
59 |
est_cond_add_attr(cond, expr); |
60 |
} |
61 |
void set_order(const char *expr) { |
62 |
/** |
63 |
* set the order of a condition object |
64 |
*/ |
65 |
est_cond_set_order(cond, expr); |
66 |
} |
67 |
void set_max(int _max) { |
68 |
/** |
69 |
* set the maximum number of retrieval of a condition object |
70 |
*/ |
71 |
est_cond_set_max(cond, _max); |
72 |
} |
73 |
void set_options(int options) { |
74 |
/** |
75 |
* set options of retrieval of a condition object |
76 |
*/ |
77 |
est_cond_set_options(cond, options); |
78 |
} |
79 |
}; |
80 |
|
81 |
class Document { |
82 |
private: |
83 |
std::string text_buf; |
84 |
public: |
85 |
ESTDOC *doc; |
86 |
Document() { |
87 |
/** |
88 |
* constructor |
89 |
*/ |
90 |
doc = est_doc_new(); |
91 |
} |
92 |
Document(const char* draft) { |
93 |
/** |
94 |
* constructor |
95 |
*/ |
96 |
doc = est_doc_new_from_draft(draft); |
97 |
} |
98 |
Document(ESTDOC *_doc) { |
99 |
/** |
100 |
* constructor |
101 |
*/ |
102 |
doc = _doc; |
103 |
} |
104 |
~Document() { |
105 |
/** |
106 |
* destructor |
107 |
*/ |
108 |
est_doc_delete(doc); |
109 |
} |
110 |
void add_attr(const char * name, const char*value) { |
111 |
/** |
112 |
* add an attribute to a document object |
113 |
*/ |
114 |
est_doc_add_attr(doc, name, value); |
115 |
} |
116 |
void add_text(const char *text) { |
117 |
/** |
118 |
* add a sentence of text to a document object |
119 |
*/ |
120 |
est_doc_add_text(doc, text); |
121 |
} |
122 |
void add_hidden_text(const char * text) { |
123 |
/** |
124 |
* add a hidden sentence to a document object |
125 |
*/ |
126 |
est_doc_add_hidden_text(doc, text); |
127 |
} |
128 |
int id() { |
129 |
/** |
130 |
* get the ID number of a document object |
131 |
*/ |
132 |
return est_doc_id(doc); |
133 |
} |
134 |
std::vector<std::string> * attr_names() { |
135 |
/** |
136 |
* get a list of attribute names of a document object |
137 |
*/ |
138 |
std::vector<std::string> * vs = new std::vector<std::string>; |
139 |
CBLIST * attr_names = est_doc_attr_names(doc); |
140 |
for (int i=0; i < cblistnum(attr_names); i++) { |
141 |
vs->push_back(cblistval(attr_names, i, NULL)); |
142 |
} |
143 |
cblistclose(attr_names); |
144 |
return vs; |
145 |
} |
146 |
const char * attr(const char *name) { |
147 |
/** |
148 |
* get the value of an attribute of a document object |
149 |
*/ |
150 |
return est_doc_attr(doc, name); |
151 |
} |
152 |
const char * cat_texts() { |
153 |
/** |
154 |
* get a list of sentences of the text of a document object |
155 |
*/ |
156 |
return est_doc_cat_texts(doc); |
157 |
} |
158 |
std::vector<std::string>* texts() { |
159 |
/** |
160 |
* get a list of sentences of the text of a document object |
161 |
*/ |
162 |
std::vector<std::string> * vs = new std::vector<std::string>; |
163 |
const CBLIST *texts; |
164 |
texts = est_doc_texts(doc); |
165 |
for(int i = 0; i < cblistnum(texts); i++) { |
166 |
vs->push_back(cblistval(texts, i, NULL)); |
167 |
} |
168 |
return vs; |
169 |
} |
170 |
const char * dump_draft() { |
171 |
/** |
172 |
* dump draft data of a document object |
173 |
*/ |
174 |
return est_doc_dump_draft(doc); |
175 |
} |
176 |
const char * make_snippet(std::vector<std::string> _words, int wwidth, int hwidth, int awidth) { |
177 |
/** |
178 |
* make a snippet of the body text of a document object |
179 |
*/ |
180 |
CBLIST * words; |
181 |
std::vector<std::string>::iterator iter; |
182 |
words = cblistopen(); |
183 |
for (iter = _words.begin(); _words.end() != iter; iter++) { |
184 |
cblistpush(words, iter->c_str(), -1); |
185 |
} |
186 |
const char *result = est_doc_make_snippet(doc, words, wwidth, hwidth, awidth); |
187 |
cblistclose(words); |
188 |
return result; |
189 |
} |
190 |
const char * hidden_texts() { |
191 |
/** |
192 |
* get the hidden texts of a document object. |
193 |
*/ |
194 |
return est_doc_hidden_texts(doc); |
195 |
} |
196 |
}; |
197 |
|
198 |
class Database { |
199 |
private: |
200 |
ESTMTDB *db; |
201 |
int ecode; |
202 |
public: |
203 |
enum { // enumeration for error codes |
204 |
ERRNOERR = ESTENOERR, // no error |
205 |
ERRINVAL = ESTEINVAL, // invalid argument |
206 |
ERRACCES = ESTEACCES, // access forbidden |
207 |
ERRLOCK = ESTELOCK, // lock failure |
208 |
ERRDB = ESTEDB, // database problem |
209 |
ERRIO = ESTEIO, // I/O problem |
210 |
ERRNOITEM = ESTENOITEM, // no item |
211 |
ERRMISC = ESTEMISC // miscellaneous |
212 |
}; |
213 |
enum { // enumeration for open modes |
214 |
DBREADER = ESTDBREADER, // open as a reader |
215 |
DBWRITER = ESTDBWRITER, // open as a writer |
216 |
DBCREAT = ESTDBCREAT, // a writer creating |
217 |
DBTRUNC = ESTDBTRUNC, // a writer truncating |
218 |
DBNOLCK = ESTDBNOLCK, // open without locking |
219 |
DBLCKNB = ESTDBLCKNB, // lock without blocking |
220 |
DBPERFNG = ESTDBPERFNG // use perfect N-gram analyzer |
221 |
}; |
222 |
enum { // enumeration for options of document registration |
223 |
PDCLEAN = ESTPDCLEAN // clean up dispensable regions |
224 |
}; |
225 |
enum { // enumeration for options of document deletion |
226 |
ODCLEAN = ESTODCLEAN // clean up dispensable regions |
227 |
}; |
228 |
enum { // enumeration for options of optimization |
229 |
OPTNOPURGE = ESTOPTNOPURGE, // omit purging dispensable region of deleted |
230 |
OPTNODBOPT = ESTOPTNODBOPT // omit optimizization of the database files |
231 |
}; |
232 |
enum { // enumeration for options of document retrieval |
233 |
GDNOATTR = ESTGDNOATTR, // no attributes |
234 |
GDNOTEXT = ESTGDNOTEXT // no text |
235 |
}; |
236 |
Database() { |
237 |
/** |
238 |
* constructor(dummy) |
239 |
*/ |
240 |
db = NULL; |
241 |
ecode = ERRNOERR; |
242 |
} |
243 |
~Database() { |
244 |
if (db) close(); |
245 |
} |
246 |
bool open(const char * dbname, int mode) { |
247 |
/** |
248 |
* open the database |
249 |
*/ |
250 |
if (db) close(); |
251 |
int ec; |
252 |
db = est_mtdb_open(dbname, mode, &ec); |
253 |
if (!db) ecode = ec; |
254 |
return db; |
255 |
} |
256 |
bool close() { |
257 |
/** |
258 |
* close the database |
259 |
*/ |
260 |
if (!db) throw IOError("closed database"); |
261 |
int ec; |
262 |
bool result = est_mtdb_close(db, &ec); |
263 |
if (!result) ecode = ec; |
264 |
db = NULL; |
265 |
return result; |
266 |
} |
267 |
bool put_doc(Document *doc, int options) { |
268 |
/** |
269 |
* add a document to a database |
270 |
*/ |
271 |
if (!db) throw IOError("closed database"); |
272 |
bool result = est_mtdb_put_doc(db, doc->doc, options); |
273 |
if (!result) ecode = est_mtdb_error(db); |
274 |
return result; |
275 |
} |
276 |
std::vector<int> * search(Condition * cond, int options) { |
277 |
/** |
278 |
* search documents corresponding a condition for a database |
279 |
*/ |
280 |
if (!db) throw IOError("closed database"); |
281 |
int resnum; |
282 |
int * result = est_mtdb_search(db, cond->cond, &resnum, NULL); |
283 |
std::vector<int> *numbers = new std::vector<int>; |
284 |
for (int i=0; i<resnum; i++) { |
285 |
numbers->push_back(result[i]); |
286 |
} |
287 |
return numbers; |
288 |
} |
289 |
static const char * err_msg(int ecode) { |
290 |
/** |
291 |
* get the string of an error |
292 |
*/ |
293 |
return est_err_msg(ecode); |
294 |
} |
295 |
int error() { |
296 |
/** |
297 |
* get the last happended error code of a database |
298 |
*/ |
299 |
return ecode; |
300 |
} |
301 |
bool fatal() { |
302 |
/** |
303 |
* check whether a database has a fatal error |
304 |
*/ |
305 |
if (!db) throw IOError("closed database"); |
306 |
return est_mtdb_fatal(db); |
307 |
} |
308 |
bool flush(int _max) { |
309 |
/** |
310 |
* flush index words in the cache of a database |
311 |
*/ |
312 |
if (!db) throw IOError("closed database"); |
313 |
bool result = est_mtdb_flush(db, _max); |
314 |
if (!result) ecode = est_mtdb_error(db); |
315 |
return result; |
316 |
} |
317 |
bool sync() { |
318 |
/** |
319 |
* synchronize updating contents of a database |
320 |
*/ |
321 |
if (!db) throw IOError("closed database"); |
322 |
bool result = est_mtdb_sync(db); |
323 |
if (!result) ecode = est_mtdb_error(db); |
324 |
return result; |
325 |
} |
326 |
bool optimize(int options) { |
327 |
/** |
328 |
* optimize a database |
329 |
*/ |
330 |
if (!db) throw IOError("closed database"); |
331 |
bool result = est_mtdb_optimize(db, options); |
332 |
if (!result) ecode = est_mtdb_error(db); |
333 |
return result; |
334 |
} |
335 |
bool out_doc(int id, int options) { |
336 |
/** |
337 |
* remove a document from a database |
338 |
*/ |
339 |
if (!db) throw IOError("closed database"); |
340 |
bool result = est_mtdb_out_doc(db, id, options); |
341 |
if (!result) ecode = est_mtdb_error(db); |
342 |
return result; |
343 |
} |
344 |
bool edit_doc(Document *doc) { |
345 |
/** |
346 |
* edit an attribute of a document in a database |
347 |
*/ |
348 |
if (!db) throw IOError("closed database"); |
349 |
bool result = est_mtdb_edit_doc(db, doc->doc); |
350 |
if (!result) ecode = est_mtdb_error(db); |
351 |
return result; |
352 |
} |
353 |
Document * get_doc(int id, int options) { |
354 |
/** |
355 |
* retrieve a document in a database |
356 |
*/ |
357 |
if (!db) throw IOError("closed database"); |
358 |
ESTDOC *doc = est_mtdb_get_doc(db, id, options); |
359 |
if (!doc) { |
360 |
ecode = est_mtdb_error(db); |
361 |
throw est_err_msg(est_mtdb_error(db)); |
362 |
} else { |
363 |
return new Document(doc); |
364 |
} |
365 |
} |
366 |
int uri_to_id(const char *uri) { |
367 |
/** |
368 |
* get the ID of a document spacified by URI |
369 |
*/ |
370 |
if (!db) throw IOError("closed database"); |
371 |
int result = est_mtdb_uri_to_id(db, uri); |
372 |
if(result == -1) ecode = est_mtdb_error(db); |
373 |
return result; |
374 |
} |
375 |
std::map<std::string, std::string> * etch_doc(Document * doc, int max) { |
376 |
/** |
377 |
* extract keywords of a document object |
378 |
*/ |
379 |
if (!db) throw IOError("closed database"); |
380 |
std::map<std::string, std::string> * mss = new std::map<std::string, std::string>; |
381 |
CBMAP * keys = est_mtdb_etch_doc(db, doc->doc, max); |
382 |
cbmapiterinit(keys); |
383 |
int ksiz; |
384 |
while (const char *key = cbmapiternext(keys, &ksiz)) { |
385 |
mss->insert(std::make_pair(key, cbmapget(keys, key, ksiz, NULL))); |
386 |
} |
387 |
return mss; |
388 |
} |
389 |
const char * name() { |
390 |
/** |
391 |
* get the name of a database |
392 |
*/ |
393 |
if (!db) throw IOError("closed database"); |
394 |
return est_mtdb_name(db); |
395 |
} |
396 |
int doc_num() { |
397 |
/** |
398 |
* get the number of documents in a database |
399 |
*/ |
400 |
if (!db) throw IOError("closed database"); |
401 |
return est_mtdb_doc_num(db); |
402 |
} |
403 |
int word_num() { |
404 |
/** |
405 |
* get the number of unique words in a database |
406 |
*/ |
407 |
if (!db) throw IOError("closed database"); |
408 |
return est_mtdb_word_num(db); |
409 |
} |
410 |
double size() { |
411 |
/** |
412 |
* get the size of a database |
413 |
*/ |
414 |
if (!db) throw IOError("closed database"); |
415 |
return est_mtdb_size(db); |
416 |
} |
417 |
void set_cache_size(size_t size, int anum, int tnum) { |
418 |
/** |
419 |
* set the maximum size of the cache memory of a database |
420 |
*/ |
421 |
if (!db) throw IOError("closed database"); |
422 |
est_mtdb_set_cache_size(db, size, anum, tnum); |
423 |
} |
424 |
void set_special_cache(const char *name, int num) { |
425 |
/** |
426 |
* Set the special cache for narrowing and sorting |
427 |
* with document attributes |
428 |
*/ |
429 |
est_mtdb_set_special_cache(db, name, num); |
430 |
} |
431 |
}; |
432 |
|
433 |
static std::vector<std::string> * break_text(const char *text, bool norm, bool tail) { |
434 |
std::vector<std::string> * vs = new std::vector<std::string>; |
435 |
CBLIST *list; |
436 |
list = cblistopen(); |
437 |
est_break_text(text, list, norm, tail); |
438 |
for (int i=0; i < cblistnum(list); i++) { |
439 |
vs->push_back(cblistval(list, i, NULL)); |
440 |
} |
441 |
cblistclose(list); |
442 |
return vs; |
443 |
} |
444 |
|
445 |
static std::vector<std::string> * break_text_perfng(const char *text, bool norm, bool tail) { |
446 |
std::vector<std::string> * vs = new std::vector<std::string>; |
447 |
CBLIST *list; |
448 |
list = cblistopen(); |
449 |
est_break_text_perfng(text, list, norm, tail); |
450 |
for (int i=0; i < cblistnum(list); i++) { |
451 |
vs->push_back(cblistval(list, i, NULL)); |
452 |
} |
453 |
cblistclose(list); |
454 |
return vs; |
455 |
} |
456 |
|
457 |
}; |