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