From: Phil Carmody <phil@dovecot.fi>
Date: Wed, 16 May 2018 13:32:35 +0000 (+0300)
Subject: lib-fts: tokenizer-generic - recognise request for explicit prefix searching
X-Git-Tag: 2.3.9~1280
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=336ddc224efa86d9b2f51243a134f25857d53521;p=thirdparty%2Fdovecot%2Fcore.git

lib-fts: tokenizer-generic - recognise request for explicit prefix searching

Just store a flag in the tokenizer when the setting is seen, nothing more.

Signed-off-by: Phil Carmody <phil@dovecot.fi>
---

diff --git a/src/lib-fts/fts-tokenizer-generic-private.h b/src/lib-fts/fts-tokenizer-generic-private.h
index e9b8f597af..2669023bc5 100644
--- a/src/lib-fts/fts-tokenizer-generic-private.h
+++ b/src/lib-fts/fts-tokenizer-generic-private.h
@@ -40,6 +40,7 @@ enum boundary_algorithm {
 struct generic_fts_tokenizer {
 	struct fts_tokenizer tokenizer;
 	unsigned int max_length;
+	bool prefixsplat; /* for search strings, accept a trailing '*' for explicit prefix */
 	bool wb5a; /* TR29 rule for prefix separation
 	              in e.g. French or Italian. */
 	bool seen_wb5a;
diff --git a/src/lib-fts/fts-tokenizer-generic.c b/src/lib-fts/fts-tokenizer-generic.c
index 1bcc8db4eb..797e431616 100644
--- a/src/lib-fts/fts-tokenizer-generic.c
+++ b/src/lib-fts/fts-tokenizer-generic.c
@@ -36,6 +36,8 @@ fts_tokenizer_generic_create(const char *const *settings,
 	unsigned int max_length = FTS_DEFAULT_TOKEN_MAX_LENGTH;
 	enum boundary_algorithm algo = BOUNDARY_ALGORITHM_SIMPLE;
 	bool wb5a = FALSE;
+	bool search = FALSE;
+	bool explicitprefix = FALSE;
 	unsigned int i;
 
 	for (i = 0; settings[i] != NULL; i += 2) {
@@ -61,17 +63,24 @@ fts_tokenizer_generic_create(const char *const *settings,
 		} else if (strcmp(key, "search") == 0) {
 			/* tokenizing a search string -
 			   makes no difference to us */
+			search = TRUE;
 		} else if (strcasecmp(key, "wb5a") == 0) {
 			if (strcasecmp(value, "no") == 0)
 				wb5a = FALSE;
 			else
 				wb5a = TRUE;
+		} else if (strcasecmp(key, "explicitprefix") == 0) {
+			explicitprefix = TRUE;
 		} else {
 			*error_r = t_strdup_printf("Unknown setting: %s", key);
 			return -1;
 		}
 	}
 
+	/* Tokenise normally unless tokenising an explicit prefix query */
+	if (!search)
+		explicitprefix = FALSE;
+
 	if (wb5a && algo != BOUNDARY_ALGORITHM_TR29) {
 		*error_r = "Can not use WB5a for algorithms other than TR29.";
 		return -1;
@@ -85,6 +94,7 @@ fts_tokenizer_generic_create(const char *const *settings,
 	tok->max_length = max_length;
 	tok->algorithm = algo;
 	tok->wb5a = wb5a;
+	tok->prefixsplat = explicitprefix;
 	tok->token = buffer_create_dynamic(default_pool, 64);
 
 	*tokenizer_r = &tok->tokenizer;