1 | /*------------------------------------------------------------------------- |
2 | * |
3 | * dict_synonym.c |
4 | * Synonym dictionary: replace word by its synonym |
5 | * |
6 | * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group |
7 | * |
8 | * |
9 | * IDENTIFICATION |
10 | * src/backend/tsearch/dict_synonym.c |
11 | * |
12 | *------------------------------------------------------------------------- |
13 | */ |
14 | #include "postgres.h" |
15 | |
16 | #include "commands/defrem.h" |
17 | #include "tsearch/ts_locale.h" |
18 | #include "tsearch/ts_utils.h" |
19 | #include "utils/builtins.h" |
20 | |
21 | typedef struct |
22 | { |
23 | char *in; |
24 | char *out; |
25 | int outlen; |
26 | uint16 flags; |
27 | } Syn; |
28 | |
29 | typedef struct |
30 | { |
31 | int len; /* length of syn array */ |
32 | Syn *syn; |
33 | bool case_sensitive; |
34 | } DictSyn; |
35 | |
36 | /* |
37 | * Finds the next whitespace-delimited word within the 'in' string. |
38 | * Returns a pointer to the first character of the word, and a pointer |
39 | * to the next byte after the last character in the word (in *end). |
40 | * Character '*' at the end of word will not be threated as word |
41 | * character if flags is not null. |
42 | */ |
43 | static char * |
44 | findwrd(char *in, char **end, uint16 *flags) |
45 | { |
46 | char *start; |
47 | char *lastchar; |
48 | |
49 | /* Skip leading spaces */ |
50 | while (*in && t_isspace(in)) |
51 | in += pg_mblen(in); |
52 | |
53 | /* Return NULL on empty lines */ |
54 | if (*in == '\0') |
55 | { |
56 | *end = NULL; |
57 | return NULL; |
58 | } |
59 | |
60 | lastchar = start = in; |
61 | |
62 | /* Find end of word */ |
63 | while (*in && !t_isspace(in)) |
64 | { |
65 | lastchar = in; |
66 | in += pg_mblen(in); |
67 | } |
68 | |
69 | if (in - lastchar == 1 && t_iseq(lastchar, '*') && flags) |
70 | { |
71 | *flags = TSL_PREFIX; |
72 | *end = lastchar; |
73 | } |
74 | else |
75 | { |
76 | if (flags) |
77 | *flags = 0; |
78 | *end = in; |
79 | } |
80 | |
81 | return start; |
82 | } |
83 | |
84 | static int |
85 | compareSyn(const void *a, const void *b) |
86 | { |
87 | return strcmp(((const Syn *) a)->in, ((const Syn *) b)->in); |
88 | } |
89 | |
90 | |
91 | Datum |
92 | dsynonym_init(PG_FUNCTION_ARGS) |
93 | { |
94 | List *dictoptions = (List *) PG_GETARG_POINTER(0); |
95 | DictSyn *d; |
96 | ListCell *l; |
97 | char *filename = NULL; |
98 | bool case_sensitive = false; |
99 | tsearch_readline_state trst; |
100 | char *starti, |
101 | *starto, |
102 | *end = NULL; |
103 | int cur = 0; |
104 | char *line = NULL; |
105 | uint16 flags = 0; |
106 | |
107 | foreach(l, dictoptions) |
108 | { |
109 | DefElem *defel = (DefElem *) lfirst(l); |
110 | |
111 | if (strcmp(defel->defname, "synonyms" ) == 0) |
112 | filename = defGetString(defel); |
113 | else if (strcmp(defel->defname, "casesensitive" ) == 0) |
114 | case_sensitive = defGetBoolean(defel); |
115 | else |
116 | ereport(ERROR, |
117 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
118 | errmsg("unrecognized synonym parameter: \"%s\"" , |
119 | defel->defname))); |
120 | } |
121 | |
122 | if (!filename) |
123 | ereport(ERROR, |
124 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
125 | errmsg("missing Synonyms parameter" ))); |
126 | |
127 | filename = get_tsearch_config_filename(filename, "syn" ); |
128 | |
129 | if (!tsearch_readline_begin(&trst, filename)) |
130 | ereport(ERROR, |
131 | (errcode(ERRCODE_CONFIG_FILE_ERROR), |
132 | errmsg("could not open synonym file \"%s\": %m" , |
133 | filename))); |
134 | |
135 | d = (DictSyn *) palloc0(sizeof(DictSyn)); |
136 | |
137 | while ((line = tsearch_readline(&trst)) != NULL) |
138 | { |
139 | starti = findwrd(line, &end, NULL); |
140 | if (!starti) |
141 | { |
142 | /* Empty line */ |
143 | goto skipline; |
144 | } |
145 | if (*end == '\0') |
146 | { |
147 | /* A line with only one word. Ignore silently. */ |
148 | goto skipline; |
149 | } |
150 | *end = '\0'; |
151 | |
152 | starto = findwrd(end + 1, &end, &flags); |
153 | if (!starto) |
154 | { |
155 | /* A line with only one word (+whitespace). Ignore silently. */ |
156 | goto skipline; |
157 | } |
158 | *end = '\0'; |
159 | |
160 | /* |
161 | * starti now points to the first word, and starto to the second word |
162 | * on the line, with a \0 terminator at the end of both words. |
163 | */ |
164 | |
165 | if (cur >= d->len) |
166 | { |
167 | if (d->len == 0) |
168 | { |
169 | d->len = 64; |
170 | d->syn = (Syn *) palloc(sizeof(Syn) * d->len); |
171 | } |
172 | else |
173 | { |
174 | d->len *= 2; |
175 | d->syn = (Syn *) repalloc(d->syn, sizeof(Syn) * d->len); |
176 | } |
177 | } |
178 | |
179 | if (case_sensitive) |
180 | { |
181 | d->syn[cur].in = pstrdup(starti); |
182 | d->syn[cur].out = pstrdup(starto); |
183 | } |
184 | else |
185 | { |
186 | d->syn[cur].in = lowerstr(starti); |
187 | d->syn[cur].out = lowerstr(starto); |
188 | } |
189 | |
190 | d->syn[cur].outlen = strlen(starto); |
191 | d->syn[cur].flags = flags; |
192 | |
193 | cur++; |
194 | |
195 | skipline: |
196 | pfree(line); |
197 | } |
198 | |
199 | tsearch_readline_end(&trst); |
200 | |
201 | d->len = cur; |
202 | qsort(d->syn, d->len, sizeof(Syn), compareSyn); |
203 | |
204 | d->case_sensitive = case_sensitive; |
205 | |
206 | PG_RETURN_POINTER(d); |
207 | } |
208 | |
209 | Datum |
210 | dsynonym_lexize(PG_FUNCTION_ARGS) |
211 | { |
212 | DictSyn *d = (DictSyn *) PG_GETARG_POINTER(0); |
213 | char *in = (char *) PG_GETARG_POINTER(1); |
214 | int32 len = PG_GETARG_INT32(2); |
215 | Syn key, |
216 | *found; |
217 | TSLexeme *res; |
218 | |
219 | /* note: d->len test protects against Solaris bsearch-of-no-items bug */ |
220 | if (len <= 0 || d->len <= 0) |
221 | PG_RETURN_POINTER(NULL); |
222 | |
223 | if (d->case_sensitive) |
224 | key.in = pnstrdup(in, len); |
225 | else |
226 | key.in = lowerstr_with_len(in, len); |
227 | |
228 | key.out = NULL; |
229 | |
230 | found = (Syn *) bsearch(&key, d->syn, d->len, sizeof(Syn), compareSyn); |
231 | pfree(key.in); |
232 | |
233 | if (!found) |
234 | PG_RETURN_POINTER(NULL); |
235 | |
236 | res = palloc0(sizeof(TSLexeme) * 2); |
237 | res[0].lexeme = pnstrdup(found->out, found->outlen); |
238 | res[0].flags = found->flags; |
239 | |
240 | PG_RETURN_POINTER(res); |
241 | } |
242 | |