apropos-utils.c revision 1.4.2.4 1 1.4.2.4 yamt /* $NetBSD: apropos-utils.c,v 1.4.2.4 2012/10/30 19:00:37 yamt Exp $ */
2 1.4.2.2 yamt /*-
3 1.4.2.2 yamt * Copyright (c) 2011 Abhinav Upadhyay <er.abhinav.upadhyay (at) gmail.com>
4 1.4.2.2 yamt * All rights reserved.
5 1.4.2.2 yamt *
6 1.4.2.2 yamt * This code was developed as part of Google's Summer of Code 2011 program.
7 1.4.2.2 yamt *
8 1.4.2.2 yamt * Redistribution and use in source and binary forms, with or without
9 1.4.2.2 yamt * modification, are permitted provided that the following conditions
10 1.4.2.2 yamt * are met:
11 1.4.2.2 yamt *
12 1.4.2.2 yamt * 1. Redistributions of source code must retain the above copyright
13 1.4.2.2 yamt * notice, this list of conditions and the following disclaimer.
14 1.4.2.2 yamt * 2. Redistributions in binary form must reproduce the above copyright
15 1.4.2.2 yamt * notice, this list of conditions and the following disclaimer in
16 1.4.2.2 yamt * the documentation and/or other materials provided with the
17 1.4.2.2 yamt * distribution.
18 1.4.2.2 yamt *
19 1.4.2.2 yamt * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 1.4.2.2 yamt * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 1.4.2.2 yamt * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
22 1.4.2.2 yamt * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
23 1.4.2.2 yamt * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
24 1.4.2.2 yamt * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
25 1.4.2.2 yamt * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
26 1.4.2.2 yamt * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
27 1.4.2.2 yamt * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
28 1.4.2.2 yamt * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
29 1.4.2.2 yamt * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 1.4.2.2 yamt * SUCH DAMAGE.
31 1.4.2.2 yamt */
32 1.4.2.2 yamt
33 1.4.2.2 yamt #include <sys/cdefs.h>
34 1.4.2.4 yamt __RCSID("$NetBSD: apropos-utils.c,v 1.4.2.4 2012/10/30 19:00:37 yamt Exp $");
35 1.4.2.2 yamt
36 1.4.2.4 yamt #include <sys/queue.h>
37 1.4.2.2 yamt #include <sys/stat.h>
38 1.4.2.2 yamt
39 1.4.2.2 yamt #include <assert.h>
40 1.4.2.2 yamt #include <ctype.h>
41 1.4.2.2 yamt #include <err.h>
42 1.4.2.2 yamt #include <math.h>
43 1.4.2.2 yamt #include <stdio.h>
44 1.4.2.2 yamt #include <stdlib.h>
45 1.4.2.2 yamt #include <string.h>
46 1.4.2.2 yamt #include <util.h>
47 1.4.2.2 yamt #include <zlib.h>
48 1.4.2.2 yamt
49 1.4.2.2 yamt #include "apropos-utils.h"
50 1.4.2.4 yamt #include "manconf.h"
51 1.4.2.2 yamt #include "mandoc.h"
52 1.4.2.2 yamt #include "sqlite3.h"
53 1.4.2.2 yamt
54 1.4.2.2 yamt typedef struct orig_callback_data {
55 1.4.2.2 yamt void *data;
56 1.4.2.2 yamt int (*callback) (void *, const char *, const char *, const char *,
57 1.4.2.2 yamt const char *, size_t);
58 1.4.2.2 yamt } orig_callback_data;
59 1.4.2.2 yamt
60 1.4.2.2 yamt typedef struct inverse_document_frequency {
61 1.4.2.2 yamt double value;
62 1.4.2.2 yamt int status;
63 1.4.2.2 yamt } inverse_document_frequency;
64 1.4.2.2 yamt
65 1.4.2.2 yamt /* weights for individual columns */
66 1.4.2.2 yamt static const double col_weights[] = {
67 1.4.2.2 yamt 2.0, // NAME
68 1.4.2.2 yamt 2.00, // Name-description
69 1.4.2.2 yamt 0.55, // DESCRIPTION
70 1.4.2.2 yamt 0.10, // LIBRARY
71 1.4.2.2 yamt 0.001, //RETURN VALUES
72 1.4.2.2 yamt 0.20, //ENVIRONMENT
73 1.4.2.2 yamt 0.01, //FILES
74 1.4.2.2 yamt 0.001, //EXIT STATUS
75 1.4.2.2 yamt 2.00, //DIAGNOSTICS
76 1.4.2.2 yamt 0.05, //ERRORS
77 1.4.2.2 yamt 0.00, //md5_hash
78 1.4.2.2 yamt 1.00 //machine
79 1.4.2.2 yamt };
80 1.4.2.2 yamt
81 1.4.2.2 yamt /*
82 1.4.2.2 yamt * lower --
83 1.4.2.2 yamt * Converts the string str to lower case
84 1.4.2.2 yamt */
85 1.4.2.2 yamt char *
86 1.4.2.2 yamt lower(char *str)
87 1.4.2.2 yamt {
88 1.4.2.2 yamt assert(str);
89 1.4.2.2 yamt int i = 0;
90 1.4.2.2 yamt char c;
91 1.4.2.2 yamt while (str[i] != '\0') {
92 1.4.2.2 yamt c = tolower((unsigned char) str[i]);
93 1.4.2.2 yamt str[i++] = c;
94 1.4.2.2 yamt }
95 1.4.2.2 yamt return str;
96 1.4.2.2 yamt }
97 1.4.2.2 yamt
98 1.4.2.2 yamt /*
99 1.4.2.2 yamt * concat--
100 1.4.2.2 yamt * Utility function. Concatenates together: dst, a space character and src.
101 1.4.2.2 yamt * dst + " " + src
102 1.4.2.2 yamt */
103 1.4.2.2 yamt void
104 1.4.2.2 yamt concat(char **dst, const char *src)
105 1.4.2.2 yamt {
106 1.4.2.2 yamt concat2(dst, src, strlen(src));
107 1.4.2.2 yamt }
108 1.4.2.2 yamt
109 1.4.2.2 yamt void
110 1.4.2.2 yamt concat2(char **dst, const char *src, size_t srclen)
111 1.4.2.2 yamt {
112 1.4.2.2 yamt size_t total_len, dst_len;
113 1.4.2.2 yamt assert(src != NULL);
114 1.4.2.2 yamt
115 1.4.2.2 yamt /* If destination buffer dst is NULL, then simply strdup the source buffer */
116 1.4.2.2 yamt if (*dst == NULL) {
117 1.4.2.2 yamt *dst = estrdup(src);
118 1.4.2.2 yamt return;
119 1.4.2.2 yamt }
120 1.4.2.2 yamt
121 1.4.2.2 yamt dst_len = strlen(*dst);
122 1.4.2.2 yamt /*
123 1.4.2.2 yamt * NUL Byte and separator space
124 1.4.2.2 yamt */
125 1.4.2.2 yamt total_len = dst_len + srclen + 2;
126 1.4.2.2 yamt
127 1.4.2.2 yamt *dst = erealloc(*dst, total_len);
128 1.4.2.2 yamt
129 1.4.2.2 yamt /* Append a space at the end of dst */
130 1.4.2.2 yamt (*dst)[dst_len++] = ' ';
131 1.4.2.2 yamt
132 1.4.2.2 yamt /* Now, copy src at the end of dst */
133 1.4.2.2 yamt memcpy(*dst + dst_len, src, srclen + 1);
134 1.4.2.2 yamt }
135 1.4.2.2 yamt
136 1.4.2.2 yamt void
137 1.4.2.2 yamt close_db(sqlite3 *db)
138 1.4.2.2 yamt {
139 1.4.2.2 yamt sqlite3_close(db);
140 1.4.2.2 yamt sqlite3_shutdown();
141 1.4.2.2 yamt }
142 1.4.2.2 yamt
143 1.4.2.2 yamt /*
144 1.4.2.2 yamt * create_db --
145 1.4.2.2 yamt * Creates the database schema.
146 1.4.2.2 yamt */
147 1.4.2.2 yamt static int
148 1.4.2.2 yamt create_db(sqlite3 *db)
149 1.4.2.2 yamt {
150 1.4.2.2 yamt const char *sqlstr = NULL;
151 1.4.2.2 yamt char *schemasql;
152 1.4.2.2 yamt char *errmsg = NULL;
153 1.4.2.2 yamt
154 1.4.2.2 yamt /*------------------------ Create the tables------------------------------*/
155 1.4.2.2 yamt
156 1.4.2.2 yamt #if NOTYET
157 1.4.2.2 yamt sqlite3_exec(db, "PRAGMA journal_mode = WAL", NULL, NULL, NULL);
158 1.4.2.2 yamt #else
159 1.4.2.2 yamt sqlite3_exec(db, "PRAGMA journal_mode = DELETE", NULL, NULL, NULL);
160 1.4.2.2 yamt #endif
161 1.4.2.2 yamt
162 1.4.2.2 yamt schemasql = sqlite3_mprintf("PRAGMA user_version = %d",
163 1.4.2.2 yamt APROPOS_SCHEMA_VERSION);
164 1.4.2.2 yamt sqlite3_exec(db, schemasql, NULL, NULL, &errmsg);
165 1.4.2.2 yamt if (errmsg != NULL)
166 1.4.2.2 yamt goto out;
167 1.4.2.2 yamt sqlite3_free(schemasql);
168 1.4.2.2 yamt
169 1.4.2.2 yamt sqlstr = "CREATE VIRTUAL TABLE mandb USING fts4(section, name, "
170 1.4.2.2 yamt "name_desc, desc, lib, return_vals, env, files, "
171 1.4.2.2 yamt "exit_status, diagnostics, errors, md5_hash UNIQUE, machine, "
172 1.4.2.2 yamt "compress=zip, uncompress=unzip, tokenize=porter); " //mandb
173 1.4.2.2 yamt "CREATE TABLE IF NOT EXISTS mandb_meta(device, inode, mtime, "
174 1.4.2.2 yamt "file UNIQUE, md5_hash UNIQUE, id INTEGER PRIMARY KEY); "
175 1.4.2.2 yamt //mandb_meta
176 1.4.2.2 yamt "CREATE TABLE IF NOT EXISTS mandb_links(link, target, section, "
177 1.4.2.3 yamt "machine, md5_hash); "; //mandb_links
178 1.4.2.2 yamt
179 1.4.2.2 yamt sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg);
180 1.4.2.2 yamt if (errmsg != NULL)
181 1.4.2.2 yamt goto out;
182 1.4.2.2 yamt
183 1.4.2.2 yamt sqlstr = "CREATE INDEX IF NOT EXISTS index_mandb_links ON mandb_links "
184 1.4.2.2 yamt "(link); "
185 1.4.2.2 yamt "CREATE INDEX IF NOT EXISTS index_mandb_meta_dev ON mandb_meta "
186 1.4.2.3 yamt "(device, inode); "
187 1.4.2.3 yamt "CREATE INDEX IF NOT EXISTS index_mandb_links_md5 ON mandb_links "
188 1.4.2.3 yamt "(md5_hash);";
189 1.4.2.2 yamt sqlite3_exec(db, sqlstr, NULL, NULL, &errmsg);
190 1.4.2.2 yamt if (errmsg != NULL)
191 1.4.2.2 yamt goto out;
192 1.4.2.2 yamt return 0;
193 1.4.2.2 yamt
194 1.4.2.2 yamt out:
195 1.4.2.2 yamt warnx("%s", errmsg);
196 1.4.2.2 yamt free(errmsg);
197 1.4.2.2 yamt sqlite3_close(db);
198 1.4.2.2 yamt sqlite3_shutdown();
199 1.4.2.2 yamt return -1;
200 1.4.2.2 yamt }
201 1.4.2.2 yamt
202 1.4.2.2 yamt /*
203 1.4.2.2 yamt * zip --
204 1.4.2.2 yamt * User defined Sqlite function to compress the FTS table
205 1.4.2.2 yamt */
206 1.4.2.2 yamt static void
207 1.4.2.2 yamt zip(sqlite3_context *pctx, int nval, sqlite3_value **apval)
208 1.4.2.2 yamt {
209 1.4.2.2 yamt int nin;
210 1.4.2.2 yamt long int nout;
211 1.4.2.2 yamt const unsigned char * inbuf;
212 1.4.2.2 yamt unsigned char *outbuf;
213 1.4.2.2 yamt
214 1.4.2.2 yamt assert(nval == 1);
215 1.4.2.2 yamt nin = sqlite3_value_bytes(apval[0]);
216 1.4.2.2 yamt inbuf = (const unsigned char *) sqlite3_value_blob(apval[0]);
217 1.4.2.2 yamt nout = nin + 13 + (nin + 999) / 1000;
218 1.4.2.2 yamt outbuf = emalloc(nout);
219 1.4.2.2 yamt compress(outbuf, (unsigned long *) &nout, inbuf, nin);
220 1.4.2.2 yamt sqlite3_result_blob(pctx, outbuf, nout, free);
221 1.4.2.2 yamt }
222 1.4.2.2 yamt
223 1.4.2.2 yamt /*
224 1.4.2.2 yamt * unzip --
225 1.4.2.2 yamt * User defined Sqlite function to uncompress the FTS table.
226 1.4.2.2 yamt */
227 1.4.2.2 yamt static void
228 1.4.2.2 yamt unzip(sqlite3_context *pctx, int nval, sqlite3_value **apval)
229 1.4.2.2 yamt {
230 1.4.2.2 yamt unsigned int rc;
231 1.4.2.2 yamt unsigned char *outbuf;
232 1.4.2.2 yamt z_stream stream;
233 1.4.2.2 yamt
234 1.4.2.2 yamt assert(nval == 1);
235 1.4.2.2 yamt stream.next_in = __UNCONST(sqlite3_value_blob(apval[0]));
236 1.4.2.2 yamt stream.avail_in = sqlite3_value_bytes(apval[0]);
237 1.4.2.2 yamt stream.avail_out = stream.avail_in * 2 + 100;
238 1.4.2.2 yamt stream.next_out = outbuf = emalloc(stream.avail_out);
239 1.4.2.2 yamt stream.zalloc = NULL;
240 1.4.2.2 yamt stream.zfree = NULL;
241 1.4.2.2 yamt
242 1.4.2.2 yamt if (inflateInit(&stream) != Z_OK) {
243 1.4.2.2 yamt free(outbuf);
244 1.4.2.2 yamt return;
245 1.4.2.2 yamt }
246 1.4.2.2 yamt
247 1.4.2.2 yamt while ((rc = inflate(&stream, Z_SYNC_FLUSH)) != Z_STREAM_END) {
248 1.4.2.2 yamt if (rc != Z_OK ||
249 1.4.2.2 yamt (stream.avail_out != 0 && stream.avail_in == 0)) {
250 1.4.2.2 yamt free(outbuf);
251 1.4.2.2 yamt return;
252 1.4.2.2 yamt }
253 1.4.2.2 yamt outbuf = erealloc(outbuf, stream.total_out * 2);
254 1.4.2.2 yamt stream.next_out = outbuf + stream.total_out;
255 1.4.2.2 yamt stream.avail_out = stream.total_out;
256 1.4.2.2 yamt }
257 1.4.2.2 yamt if (inflateEnd(&stream) != Z_OK) {
258 1.4.2.2 yamt free(outbuf);
259 1.4.2.2 yamt return;
260 1.4.2.2 yamt }
261 1.4.2.2 yamt outbuf = erealloc(outbuf, stream.total_out);
262 1.4.2.2 yamt sqlite3_result_text(pctx, (const char *) outbuf, stream.total_out, free);
263 1.4.2.2 yamt }
264 1.4.2.2 yamt
265 1.4.2.4 yamt /*
266 1.4.2.4 yamt * get_dbpath --
267 1.4.2.4 yamt * Read the path of the database from man.conf and return.
268 1.4.2.4 yamt */
269 1.4.2.4 yamt char *
270 1.4.2.4 yamt get_dbpath(const char *manconf)
271 1.4.2.4 yamt {
272 1.4.2.4 yamt TAG *tp;
273 1.4.2.4 yamt char *dbpath;
274 1.4.2.4 yamt
275 1.4.2.4 yamt config(manconf);
276 1.4.2.4 yamt tp = gettag("_mandb", 1);
277 1.4.2.4 yamt if (!tp)
278 1.4.2.4 yamt return NULL;
279 1.4.2.4 yamt
280 1.4.2.4 yamt if (TAILQ_EMPTY(&tp->entrylist))
281 1.4.2.4 yamt return NULL;
282 1.4.2.4 yamt
283 1.4.2.4 yamt dbpath = TAILQ_LAST(&tp->entrylist, tqh)->s;
284 1.4.2.4 yamt return dbpath;
285 1.4.2.4 yamt }
286 1.4.2.4 yamt
287 1.4.2.2 yamt /* init_db --
288 1.4.2.2 yamt * Prepare the database. Register the compress/uncompress functions and the
289 1.4.2.2 yamt * stopword tokenizer.
290 1.4.2.2 yamt * db_flag specifies the mode in which to open the database. 3 options are
291 1.4.2.2 yamt * available:
292 1.4.2.2 yamt * 1. DB_READONLY: Open in READONLY mode. An error if db does not exist.
293 1.4.2.2 yamt * 2. DB_READWRITE: Open in read-write mode. An error if db does not exist.
294 1.4.2.2 yamt * 3. DB_CREATE: Open in read-write mode. It will try to create the db if
295 1.4.2.2 yamt * it does not exist already.
296 1.4.2.2 yamt * RETURN VALUES:
297 1.4.2.2 yamt * The function will return NULL in case the db does not exist and DB_CREATE
298 1.4.2.2 yamt * was not specified. And in case DB_CREATE was specified and yet NULL is
299 1.4.2.2 yamt * returned, then there was some other error.
300 1.4.2.2 yamt * In normal cases the function should return a handle to the db.
301 1.4.2.2 yamt */
302 1.4.2.2 yamt sqlite3 *
303 1.4.2.4 yamt init_db(int db_flag, const char *manconf)
304 1.4.2.2 yamt {
305 1.4.2.2 yamt sqlite3 *db = NULL;
306 1.4.2.2 yamt sqlite3_stmt *stmt;
307 1.4.2.2 yamt struct stat sb;
308 1.4.2.2 yamt int rc;
309 1.4.2.2 yamt int create_db_flag = 0;
310 1.4.2.2 yamt
311 1.4.2.4 yamt char *dbpath = get_dbpath(manconf);
312 1.4.2.4 yamt if (dbpath == NULL)
313 1.4.2.4 yamt errx(EXIT_FAILURE, "_mandb entry not found in man.conf");
314 1.4.2.2 yamt /* Check if the database exists or not */
315 1.4.2.4 yamt if (!(stat(dbpath, &sb) == 0 && S_ISREG(sb.st_mode))) {
316 1.4.2.2 yamt /* Database does not exist, check if DB_CREATE was specified, and set
317 1.4.2.2 yamt * flag to create the database schema
318 1.4.2.2 yamt */
319 1.4.2.2 yamt if (db_flag != (MANDB_CREATE)) {
320 1.4.2.2 yamt warnx("Missing apropos database. "
321 1.4.2.2 yamt "Please run makemandb to create it.");
322 1.4.2.2 yamt return NULL;
323 1.4.2.2 yamt }
324 1.4.2.2 yamt create_db_flag = 1;
325 1.4.2.2 yamt }
326 1.4.2.2 yamt
327 1.4.2.2 yamt /* Now initialize the database connection */
328 1.4.2.2 yamt sqlite3_initialize();
329 1.4.2.4 yamt rc = sqlite3_open_v2(dbpath, &db, db_flag, NULL);
330 1.4.2.2 yamt
331 1.4.2.2 yamt if (rc != SQLITE_OK) {
332 1.4.2.2 yamt warnx("%s", sqlite3_errmsg(db));
333 1.4.2.2 yamt sqlite3_shutdown();
334 1.4.2.2 yamt return NULL;
335 1.4.2.2 yamt }
336 1.4.2.2 yamt
337 1.4.2.2 yamt if (create_db_flag && create_db(db) < 0) {
338 1.4.2.2 yamt warnx("%s", "Unable to create database schema");
339 1.4.2.2 yamt goto error;
340 1.4.2.2 yamt }
341 1.4.2.2 yamt
342 1.4.2.2 yamt rc = sqlite3_prepare_v2(db, "PRAGMA user_version", -1, &stmt, NULL);
343 1.4.2.2 yamt if (rc != SQLITE_OK) {
344 1.4.2.2 yamt warnx("Unable to query schema version: %s",
345 1.4.2.2 yamt sqlite3_errmsg(db));
346 1.4.2.2 yamt goto error;
347 1.4.2.2 yamt }
348 1.4.2.2 yamt if (sqlite3_step(stmt) != SQLITE_ROW) {
349 1.4.2.2 yamt sqlite3_finalize(stmt);
350 1.4.2.2 yamt warnx("Unable to query schema version: %s",
351 1.4.2.2 yamt sqlite3_errmsg(db));
352 1.4.2.2 yamt goto error;
353 1.4.2.2 yamt }
354 1.4.2.2 yamt if (sqlite3_column_int(stmt, 0) != APROPOS_SCHEMA_VERSION) {
355 1.4.2.2 yamt sqlite3_finalize(stmt);
356 1.4.2.2 yamt warnx("Incorrect schema version found. "
357 1.4.2.2 yamt "Please run makemandb -f.");
358 1.4.2.2 yamt goto error;
359 1.4.2.2 yamt }
360 1.4.2.2 yamt sqlite3_finalize(stmt);
361 1.4.2.2 yamt
362 1.4.2.2 yamt sqlite3_extended_result_codes(db, 1);
363 1.4.2.2 yamt
364 1.4.2.2 yamt /* Register the zip and unzip functions for FTS compression */
365 1.4.2.2 yamt rc = sqlite3_create_function(db, "zip", 1, SQLITE_ANY, NULL, zip, NULL, NULL);
366 1.4.2.2 yamt if (rc != SQLITE_OK) {
367 1.4.2.2 yamt warnx("Unable to register function: compress: %s",
368 1.4.2.2 yamt sqlite3_errmsg(db));
369 1.4.2.2 yamt goto error;
370 1.4.2.2 yamt }
371 1.4.2.2 yamt
372 1.4.2.2 yamt rc = sqlite3_create_function(db, "unzip", 1, SQLITE_ANY, NULL,
373 1.4.2.2 yamt unzip, NULL, NULL);
374 1.4.2.2 yamt if (rc != SQLITE_OK) {
375 1.4.2.2 yamt warnx("Unable to register function: uncompress: %s",
376 1.4.2.2 yamt sqlite3_errmsg(db));
377 1.4.2.2 yamt goto error;
378 1.4.2.2 yamt }
379 1.4.2.2 yamt return db;
380 1.4.2.4 yamt
381 1.4.2.2 yamt error:
382 1.4.2.2 yamt sqlite3_close(db);
383 1.4.2.2 yamt sqlite3_shutdown();
384 1.4.2.2 yamt return NULL;
385 1.4.2.2 yamt }
386 1.4.2.2 yamt
387 1.4.2.2 yamt /*
388 1.4.2.2 yamt * rank_func --
389 1.4.2.2 yamt * Sqlite user defined function for ranking the documents.
390 1.4.2.2 yamt * For each phrase of the query, it computes the tf and idf and adds them over.
391 1.4.2.2 yamt * It computes the final rank, by multiplying tf and idf together.
392 1.4.2.2 yamt * Weight of term t for document d = (term frequency of t in d *
393 1.4.2.2 yamt * inverse document frequency of t)
394 1.4.2.2 yamt *
395 1.4.2.2 yamt * Term Frequency of term t in document d = Number of times t occurs in d /
396 1.4.2.2 yamt * Number of times t appears in all
397 1.4.2.2 yamt * documents
398 1.4.2.2 yamt *
399 1.4.2.2 yamt * Inverse document frequency of t = log(Total number of documents /
400 1.4.2.2 yamt * Number of documents in which t occurs)
401 1.4.2.2 yamt */
402 1.4.2.2 yamt static void
403 1.4.2.2 yamt rank_func(sqlite3_context *pctx, int nval, sqlite3_value **apval)
404 1.4.2.2 yamt {
405 1.4.2.2 yamt inverse_document_frequency *idf = sqlite3_user_data(pctx);
406 1.4.2.2 yamt double tf = 0.0;
407 1.4.2.2 yamt const unsigned int *matchinfo;
408 1.4.2.2 yamt int ncol;
409 1.4.2.2 yamt int nphrase;
410 1.4.2.2 yamt int iphrase;
411 1.4.2.2 yamt int ndoc;
412 1.4.2.2 yamt int doclen = 0;
413 1.4.2.2 yamt const double k = 3.75;
414 1.4.2.2 yamt /* Check that the number of arguments passed to this function is correct. */
415 1.4.2.2 yamt assert(nval == 1);
416 1.4.2.2 yamt
417 1.4.2.2 yamt matchinfo = (const unsigned int *) sqlite3_value_blob(apval[0]);
418 1.4.2.2 yamt nphrase = matchinfo[0];
419 1.4.2.2 yamt ncol = matchinfo[1];
420 1.4.2.2 yamt ndoc = matchinfo[2 + 3 * ncol * nphrase + ncol];
421 1.4.2.2 yamt for (iphrase = 0; iphrase < nphrase; iphrase++) {
422 1.4.2.2 yamt int icol;
423 1.4.2.2 yamt const unsigned int *phraseinfo = &matchinfo[2 + ncol+ iphrase * ncol * 3];
424 1.4.2.2 yamt for(icol = 1; icol < ncol; icol++) {
425 1.4.2.2 yamt
426 1.4.2.2 yamt /* nhitcount: number of times the current phrase occurs in the current
427 1.4.2.2 yamt * column in the current document.
428 1.4.2.2 yamt * nglobalhitcount: number of times current phrase occurs in the current
429 1.4.2.2 yamt * column in all documents.
430 1.4.2.2 yamt * ndocshitcount: number of documents in which the current phrase
431 1.4.2.2 yamt * occurs in the current column at least once.
432 1.4.2.2 yamt */
433 1.4.2.2 yamt int nhitcount = phraseinfo[3 * icol];
434 1.4.2.2 yamt int nglobalhitcount = phraseinfo[3 * icol + 1];
435 1.4.2.2 yamt int ndocshitcount = phraseinfo[3 * icol + 2];
436 1.4.2.2 yamt doclen = matchinfo[2 + icol ];
437 1.4.2.2 yamt double weight = col_weights[icol - 1];
438 1.4.2.2 yamt if (idf->status == 0 && ndocshitcount)
439 1.4.2.2 yamt idf->value += log(((double)ndoc / ndocshitcount))* weight;
440 1.4.2.2 yamt
441 1.4.2.2 yamt /* Dividing the tf by document length to normalize the effect of
442 1.4.2.2 yamt * longer documents.
443 1.4.2.2 yamt */
444 1.4.2.2 yamt if (nglobalhitcount > 0 && nhitcount)
445 1.4.2.2 yamt tf += (((double)nhitcount * weight) / (nglobalhitcount * doclen));
446 1.4.2.2 yamt }
447 1.4.2.2 yamt }
448 1.4.2.2 yamt idf->status = 1;
449 1.4.2.2 yamt
450 1.4.2.2 yamt /* Final score = (tf * idf)/ ( k + tf)
451 1.4.2.2 yamt * Dividing by k+ tf further normalizes the weight leading to better
452 1.4.2.2 yamt * results.
453 1.4.2.2 yamt * The value of k is experimental
454 1.4.2.2 yamt */
455 1.4.2.2 yamt double score = (tf * idf->value/ ( k + tf)) ;
456 1.4.2.2 yamt sqlite3_result_double(pctx, score);
457 1.4.2.2 yamt return;
458 1.4.2.2 yamt }
459 1.4.2.2 yamt
460 1.4.2.2 yamt /*
461 1.4.2.2 yamt * run_query --
462 1.4.2.2 yamt * Performs the searches for the keywords entered by the user.
463 1.4.2.2 yamt * The 2nd param: snippet_args is an array of strings providing values for the
464 1.4.2.2 yamt * last three parameters to the snippet function of sqlite. (Look at the docs).
465 1.4.2.2 yamt * The 3rd param: args contains rest of the search parameters. Look at
466 1.4.2.2 yamt * arpopos-utils.h for the description of individual fields.
467 1.4.2.2 yamt *
468 1.4.2.2 yamt */
469 1.4.2.2 yamt int
470 1.4.2.2 yamt run_query(sqlite3 *db, const char *snippet_args[3], query_args *args)
471 1.4.2.2 yamt {
472 1.4.2.2 yamt const char *default_snippet_args[3];
473 1.4.2.2 yamt char *section_clause = NULL;
474 1.4.2.2 yamt char *limit_clause = NULL;
475 1.4.2.2 yamt char *machine_clause = NULL;
476 1.4.2.2 yamt char *query;
477 1.4.2.2 yamt const char *section;
478 1.4.2.2 yamt char *name;
479 1.4.2.2 yamt const char *name_desc;
480 1.4.2.2 yamt const char *machine;
481 1.4.2.2 yamt const char *snippet;
482 1.4.2.2 yamt const char *name_temp;
483 1.4.2.2 yamt char *slash_ptr;
484 1.4.2.2 yamt char *m = NULL;
485 1.4.2.2 yamt int rc;
486 1.4.2.2 yamt inverse_document_frequency idf = {0, 0};
487 1.4.2.2 yamt sqlite3_stmt *stmt;
488 1.4.2.2 yamt
489 1.4.2.2 yamt if (args->machine)
490 1.4.2.2 yamt easprintf(&machine_clause, "AND machine = \'%s\' ", args->machine);
491 1.4.2.2 yamt
492 1.4.2.2 yamt /* Register the rank function */
493 1.4.2.2 yamt rc = sqlite3_create_function(db, "rank_func", 1, SQLITE_ANY, (void *)&idf,
494 1.4.2.2 yamt rank_func, NULL, NULL);
495 1.4.2.2 yamt if (rc != SQLITE_OK) {
496 1.4.2.2 yamt warnx("Unable to register the ranking function: %s",
497 1.4.2.2 yamt sqlite3_errmsg(db));
498 1.4.2.2 yamt sqlite3_close(db);
499 1.4.2.2 yamt sqlite3_shutdown();
500 1.4.2.2 yamt exit(EXIT_FAILURE);
501 1.4.2.2 yamt }
502 1.4.2.2 yamt
503 1.4.2.2 yamt /* We want to build a query of the form: "select x,y,z from mandb where
504 1.4.2.2 yamt * mandb match :query [AND (section LIKE '1' OR section LIKE '2' OR...)]
505 1.4.2.2 yamt * ORDER BY rank DESC..."
506 1.4.2.2 yamt * NOTES: 1. The portion in square brackets is optional, it will be there
507 1.4.2.2 yamt * only if the user has specified an option on the command line to search in
508 1.4.2.2 yamt * one or more specific sections.
509 1.4.2.2 yamt * 2. I am using LIKE operator because '=' or IN operators do not seem to be
510 1.4.2.2 yamt * working with the compression option enabled.
511 1.4.2.2 yamt */
512 1.4.2.2 yamt
513 1.4.2.2 yamt if (args->sec_nums) {
514 1.4.2.2 yamt char *temp;
515 1.4.2.2 yamt int i;
516 1.4.2.2 yamt
517 1.4.2.2 yamt for (i = 0; i < SECMAX; i++) {
518 1.4.2.2 yamt if (args->sec_nums[i] == 0)
519 1.4.2.2 yamt continue;
520 1.4.2.2 yamt easprintf(&temp, " OR section = \'%d\'", i + 1);
521 1.4.2.2 yamt if (section_clause) {
522 1.4.2.2 yamt concat(§ion_clause, temp);
523 1.4.2.2 yamt free(temp);
524 1.4.2.2 yamt } else {
525 1.4.2.2 yamt section_clause = temp;
526 1.4.2.2 yamt }
527 1.4.2.2 yamt }
528 1.4.2.2 yamt if (section_clause) {
529 1.4.2.2 yamt /*
530 1.4.2.2 yamt * At least one section requested, add glue for query.
531 1.4.2.2 yamt */
532 1.4.2.2 yamt temp = section_clause;
533 1.4.2.2 yamt /* Skip " OR " before first term. */
534 1.4.2.2 yamt easprintf(§ion_clause, " AND (%s)", temp + 4);
535 1.4.2.2 yamt free(temp);
536 1.4.2.2 yamt }
537 1.4.2.2 yamt }
538 1.4.2.2 yamt if (args->nrec >= 0) {
539 1.4.2.2 yamt /* Use the provided number of records and offset */
540 1.4.2.2 yamt easprintf(&limit_clause, " LIMIT %d OFFSET %d",
541 1.4.2.2 yamt args->nrec, args->offset);
542 1.4.2.2 yamt }
543 1.4.2.2 yamt
544 1.4.2.2 yamt if (snippet_args == NULL) {
545 1.4.2.2 yamt default_snippet_args[0] = "";
546 1.4.2.2 yamt default_snippet_args[1] = "";
547 1.4.2.2 yamt default_snippet_args[2] = "...";
548 1.4.2.2 yamt snippet_args = default_snippet_args;
549 1.4.2.2 yamt }
550 1.4.2.2 yamt query = sqlite3_mprintf("SELECT section, name, name_desc, machine,"
551 1.4.2.2 yamt " snippet(mandb, %Q, %Q, %Q, -1, 40 ),"
552 1.4.2.2 yamt " rank_func(matchinfo(mandb, \"pclxn\")) AS rank"
553 1.4.2.2 yamt " FROM mandb"
554 1.4.2.2 yamt " WHERE mandb MATCH %Q %s "
555 1.4.2.2 yamt "%s"
556 1.4.2.2 yamt " ORDER BY rank DESC"
557 1.4.2.2 yamt "%s",
558 1.4.2.2 yamt snippet_args[0], snippet_args[1], snippet_args[2], args->search_str,
559 1.4.2.2 yamt machine_clause ? machine_clause : "",
560 1.4.2.2 yamt section_clause ? section_clause : "",
561 1.4.2.2 yamt limit_clause ? limit_clause : "");
562 1.4.2.2 yamt
563 1.4.2.2 yamt free(machine_clause);
564 1.4.2.2 yamt free(section_clause);
565 1.4.2.2 yamt free(limit_clause);
566 1.4.2.2 yamt
567 1.4.2.2 yamt if (query == NULL) {
568 1.4.2.2 yamt *args->errmsg = estrdup("malloc failed");
569 1.4.2.2 yamt return -1;
570 1.4.2.2 yamt }
571 1.4.2.2 yamt rc = sqlite3_prepare_v2(db, query, -1, &stmt, NULL);
572 1.4.2.2 yamt if (rc == SQLITE_IOERR) {
573 1.4.2.2 yamt warnx("Corrupt database. Please rerun makemandb");
574 1.4.2.2 yamt sqlite3_free(query);
575 1.4.2.2 yamt return -1;
576 1.4.2.2 yamt } else if (rc != SQLITE_OK) {
577 1.4.2.2 yamt warnx("%s", sqlite3_errmsg(db));
578 1.4.2.2 yamt sqlite3_free(query);
579 1.4.2.2 yamt return -1;
580 1.4.2.2 yamt }
581 1.4.2.2 yamt
582 1.4.2.2 yamt while (sqlite3_step(stmt) == SQLITE_ROW) {
583 1.4.2.2 yamt section = (const char *) sqlite3_column_text(stmt, 0);
584 1.4.2.2 yamt name_temp = (const char *) sqlite3_column_text(stmt, 1);
585 1.4.2.2 yamt name_desc = (const char *) sqlite3_column_text(stmt, 2);
586 1.4.2.2 yamt machine = (const char *) sqlite3_column_text(stmt, 3);
587 1.4.2.2 yamt snippet = (const char *) sqlite3_column_text(stmt, 4);
588 1.4.2.2 yamt if ((slash_ptr = strrchr(name_temp, '/')) != NULL)
589 1.4.2.2 yamt name_temp = slash_ptr + 1;
590 1.4.2.2 yamt if (machine && machine[0]) {
591 1.4.2.2 yamt m = estrdup(machine);
592 1.4.2.2 yamt easprintf(&name, "%s/%s", lower(m),
593 1.4.2.2 yamt name_temp);
594 1.4.2.2 yamt free(m);
595 1.4.2.2 yamt } else {
596 1.4.2.2 yamt name = estrdup((const char *) sqlite3_column_text(stmt, 1));
597 1.4.2.2 yamt }
598 1.4.2.2 yamt
599 1.4.2.2 yamt (args->callback)(args->callback_data, section, name, name_desc, snippet,
600 1.4.2.2 yamt strlen(snippet));
601 1.4.2.2 yamt
602 1.4.2.2 yamt free(name);
603 1.4.2.2 yamt }
604 1.4.2.2 yamt
605 1.4.2.2 yamt sqlite3_finalize(stmt);
606 1.4.2.2 yamt sqlite3_free(query);
607 1.4.2.2 yamt return *(args->errmsg) == NULL ? 0 : -1;
608 1.4.2.2 yamt }
609 1.4.2.2 yamt
610 1.4.2.2 yamt /*
611 1.4.2.2 yamt * callback_html --
612 1.4.2.2 yamt * Callback function for run_query_html. It builds the html output and then
613 1.4.2.2 yamt * calls the actual user supplied callback function.
614 1.4.2.2 yamt */
615 1.4.2.2 yamt static int
616 1.4.2.2 yamt callback_html(void *data, const char *section, const char *name,
617 1.4.2.2 yamt const char *name_desc, const char *snippet, size_t snippet_length)
618 1.4.2.2 yamt {
619 1.4.2.2 yamt const char *temp = snippet;
620 1.4.2.2 yamt int i = 0;
621 1.4.2.2 yamt size_t sz = 0;
622 1.4.2.2 yamt int count = 0;
623 1.4.2.2 yamt struct orig_callback_data *orig_data = (struct orig_callback_data *) data;
624 1.4.2.2 yamt int (*callback) (void *, const char *, const char *, const char *,
625 1.4.2.2 yamt const char *, size_t) = orig_data->callback;
626 1.4.2.2 yamt
627 1.4.2.2 yamt /* First scan the snippet to find out the number of occurrences of {'>', '<'
628 1.4.2.2 yamt * '"', '&'}.
629 1.4.2.2 yamt * Then allocate a new buffer with sufficient space to be able to store the
630 1.4.2.2 yamt * quoted versions of the special characters {>, <, ", &}.
631 1.4.2.2 yamt * Copy over the characters from the original snippet to this buffer while
632 1.4.2.2 yamt * replacing the special characters with their quoted versions.
633 1.4.2.2 yamt */
634 1.4.2.2 yamt
635 1.4.2.2 yamt while (*temp) {
636 1.4.2.2 yamt sz = strcspn(temp, "<>\"&\002\003");
637 1.4.2.2 yamt temp += sz + 1;
638 1.4.2.2 yamt count++;
639 1.4.2.2 yamt }
640 1.4.2.2 yamt size_t qsnippet_length = snippet_length + count * 5;
641 1.4.2.2 yamt char *qsnippet = emalloc(qsnippet_length + 1);
642 1.4.2.2 yamt sz = 0;
643 1.4.2.2 yamt while (*snippet) {
644 1.4.2.2 yamt sz = strcspn(snippet, "<>\"&\002\003");
645 1.4.2.2 yamt if (sz) {
646 1.4.2.2 yamt memcpy(&qsnippet[i], snippet, sz);
647 1.4.2.2 yamt snippet += sz;
648 1.4.2.2 yamt i += sz;
649 1.4.2.2 yamt }
650 1.4.2.2 yamt
651 1.4.2.2 yamt switch (*snippet++) {
652 1.4.2.2 yamt case '<':
653 1.4.2.2 yamt memcpy(&qsnippet[i], "<", 4);
654 1.4.2.2 yamt i += 4;
655 1.4.2.2 yamt break;
656 1.4.2.2 yamt case '>':
657 1.4.2.2 yamt memcpy(&qsnippet[i], ">", 4);
658 1.4.2.2 yamt i += 4;
659 1.4.2.2 yamt break;
660 1.4.2.2 yamt case '\"':
661 1.4.2.2 yamt memcpy(&qsnippet[i], """, 6);
662 1.4.2.2 yamt i += 6;
663 1.4.2.2 yamt break;
664 1.4.2.2 yamt case '&':
665 1.4.2.2 yamt /* Don't perform the quoting if this & is part of an mdoc escape
666 1.4.2.2 yamt * sequence, e.g. \&
667 1.4.2.2 yamt */
668 1.4.2.2 yamt if (i && *(snippet - 2) != '\\') {
669 1.4.2.2 yamt memcpy(&qsnippet[i], "&", 5);
670 1.4.2.2 yamt i += 5;
671 1.4.2.2 yamt } else {
672 1.4.2.2 yamt qsnippet[i++] = '&';
673 1.4.2.2 yamt }
674 1.4.2.2 yamt break;
675 1.4.2.2 yamt case '\002':
676 1.4.2.2 yamt memcpy(&qsnippet[i], "<b>", 3);
677 1.4.2.2 yamt i += 3;
678 1.4.2.2 yamt break;
679 1.4.2.2 yamt case '\003':
680 1.4.2.2 yamt memcpy(&qsnippet[i], "</b>", 4);
681 1.4.2.2 yamt i += 4;
682 1.4.2.2 yamt break;
683 1.4.2.2 yamt default:
684 1.4.2.2 yamt break;
685 1.4.2.2 yamt }
686 1.4.2.2 yamt }
687 1.4.2.2 yamt qsnippet[++i] = 0;
688 1.4.2.2 yamt (*callback)(orig_data->data, section, name, name_desc,
689 1.4.2.2 yamt (const char *)qsnippet, qsnippet_length);
690 1.4.2.2 yamt free(qsnippet);
691 1.4.2.2 yamt return 0;
692 1.4.2.2 yamt }
693 1.4.2.2 yamt
694 1.4.2.2 yamt /*
695 1.4.2.2 yamt * run_query_html --
696 1.4.2.2 yamt * Utility function to output query result in HTML format.
697 1.4.2.2 yamt * It internally calls run_query only, but it first passes the output to it's
698 1.4.2.2 yamt * own custom callback function, which preprocess the snippet for quoting
699 1.4.2.2 yamt * inline HTML fragments.
700 1.4.2.2 yamt * After that it delegates the call the actual user supplied callback function.
701 1.4.2.2 yamt */
702 1.4.2.2 yamt int
703 1.4.2.2 yamt run_query_html(sqlite3 *db, query_args *args)
704 1.4.2.2 yamt {
705 1.4.2.2 yamt struct orig_callback_data orig_data;
706 1.4.2.2 yamt orig_data.callback = args->callback;
707 1.4.2.2 yamt orig_data.data = args->callback_data;
708 1.4.2.2 yamt const char *snippet_args[] = {"\002", "\003", "..."};
709 1.4.2.2 yamt args->callback = &callback_html;
710 1.4.2.2 yamt args->callback_data = (void *) &orig_data;
711 1.4.2.2 yamt return run_query(db, snippet_args, args);
712 1.4.2.2 yamt }
713 1.4.2.2 yamt
714 1.4.2.2 yamt /*
715 1.4.2.2 yamt * callback_pager --
716 1.4.2.2 yamt * A callback similar to callback_html. It overstrikes the matching text in
717 1.4.2.2 yamt * the snippet so that it appears emboldened when viewed using a pager like
718 1.4.2.2 yamt * more or less.
719 1.4.2.2 yamt */
720 1.4.2.2 yamt static int
721 1.4.2.2 yamt callback_pager(void *data, const char *section, const char *name,
722 1.4.2.2 yamt const char *name_desc, const char *snippet, size_t snippet_length)
723 1.4.2.2 yamt {
724 1.4.2.2 yamt struct orig_callback_data *orig_data = (struct orig_callback_data *) data;
725 1.4.2.2 yamt char *psnippet;
726 1.4.2.2 yamt const char *temp = snippet;
727 1.4.2.2 yamt int count = 0;
728 1.4.2.2 yamt int i = 0;
729 1.4.2.2 yamt size_t sz = 0;
730 1.4.2.2 yamt size_t psnippet_length;
731 1.4.2.2 yamt
732 1.4.2.2 yamt /* Count the number of bytes of matching text. For each of these bytes we
733 1.4.2.2 yamt * will use 2 extra bytes to overstrike it so that it appears bold when
734 1.4.2.2 yamt * viewed using a pager.
735 1.4.2.2 yamt */
736 1.4.2.2 yamt while (*temp) {
737 1.4.2.2 yamt sz = strcspn(temp, "\002\003");
738 1.4.2.2 yamt temp += sz;
739 1.4.2.2 yamt if (*temp == '\003') {
740 1.4.2.2 yamt count += 2 * (sz);
741 1.4.2.2 yamt }
742 1.4.2.2 yamt temp++;
743 1.4.2.2 yamt }
744 1.4.2.2 yamt
745 1.4.2.2 yamt psnippet_length = snippet_length + count;
746 1.4.2.2 yamt psnippet = emalloc(psnippet_length + 1);
747 1.4.2.2 yamt
748 1.4.2.2 yamt /* Copy the bytes from snippet to psnippet:
749 1.4.2.2 yamt * 1. Copy the bytes before \002 as it is.
750 1.4.2.2 yamt * 2. The bytes after \002 need to be overstriked till we encounter \003.
751 1.4.2.2 yamt * 3. To overstrike a byte 'A' we need to write 'A\bA'
752 1.4.2.2 yamt */
753 1.4.2.2 yamt while (*snippet) {
754 1.4.2.2 yamt sz = strcspn(snippet, "\002");
755 1.4.2.2 yamt memcpy(&psnippet[i], snippet, sz);
756 1.4.2.2 yamt snippet += sz;
757 1.4.2.2 yamt i += sz;
758 1.4.2.2 yamt
759 1.4.2.2 yamt /* Don't change this. Advancing the pointer without reading the byte
760 1.4.2.2 yamt * is causing strange behavior.
761 1.4.2.2 yamt */
762 1.4.2.2 yamt if (*snippet == '\002')
763 1.4.2.2 yamt snippet++;
764 1.4.2.2 yamt while (*snippet && *snippet != '\003') {
765 1.4.2.2 yamt psnippet[i++] = *snippet;
766 1.4.2.2 yamt psnippet[i++] = '\b';
767 1.4.2.2 yamt psnippet[i++] = *snippet++;
768 1.4.2.2 yamt }
769 1.4.2.2 yamt if (*snippet)
770 1.4.2.2 yamt snippet++;
771 1.4.2.2 yamt }
772 1.4.2.2 yamt
773 1.4.2.2 yamt psnippet[i] = 0;
774 1.4.2.2 yamt (orig_data->callback)(orig_data->data, section, name, name_desc, psnippet,
775 1.4.2.2 yamt psnippet_length);
776 1.4.2.2 yamt free(psnippet);
777 1.4.2.2 yamt return 0;
778 1.4.2.2 yamt }
779 1.4.2.2 yamt
780 1.4.2.2 yamt /*
781 1.4.2.2 yamt * run_query_pager --
782 1.4.2.2 yamt * Utility function similar to run_query_html. This function tries to
783 1.4.2.2 yamt * pre-process the result assuming it will be piped to a pager.
784 1.4.2.2 yamt * For this purpose it first calls it's own callback function callback_pager
785 1.4.2.2 yamt * which then delegates the call to the user supplied callback.
786 1.4.2.2 yamt */
787 1.4.2.3 yamt int
788 1.4.2.3 yamt run_query_pager(sqlite3 *db, query_args *args)
789 1.4.2.2 yamt {
790 1.4.2.2 yamt struct orig_callback_data orig_data;
791 1.4.2.2 yamt orig_data.callback = args->callback;
792 1.4.2.2 yamt orig_data.data = args->callback_data;
793 1.4.2.2 yamt const char *snippet_args[] = {"\002", "\003", "..."};
794 1.4.2.2 yamt args->callback = &callback_pager;
795 1.4.2.2 yamt args->callback_data = (void *) &orig_data;
796 1.4.2.2 yamt return run_query(db, snippet_args, args);
797 1.4.2.2 yamt }
798