Additional comments and a little fix for rum_extract_tsquery_internal()

Zakirov Artur · Zakirov Artur · commit a59ed37b1ee3 · 2016-11-13T16:56:38.000+03:00
diff --git a/src/rum_ts_utils.c b/src/rum_ts_utils.c
@@ -490,6 +490,14 @@ SortAndUniqItems(TSQuery q, int *size)
 	return res;
 }
 
+/*
+ * Extracting tsvector lexems.
+ */
+
+/*
+ * Extracts tsvector lexemes from **vector**. Uses **build_tsvector_entry**
+ * callback to extract entry.
+ */
 static Datum *
 rum_extract_tsvector_internal(TSVector	vector,
 							  int32	   *nentries,
@@ -515,6 +523,7 @@ rum_extract_tsvector_internal(TSVector	vector,
 			bytea	   *posData;
 			int			posDataSize;
 
+			/* Extract entry using specified method */
 			entries[i] = build_tsvector_entry(vector, we);
 
 			if (we->haspos)
@@ -540,6 +549,10 @@ rum_extract_tsvector_internal(TSVector	vector,
 	return entries;
 }
 
+/*
+ * Used as callback for rum_extract_tsvector_internal().
+ * Just extract entry from tsvector.
+ */
 static Datum
 build_tsvector_entry(TSVector vector, WordEntry *we)
 {
@@ -549,6 +562,10 @@ build_tsvector_entry(TSVector vector, WordEntry *we)
 	return PointerGetDatum(txt);
 }
 
+/*
+ * Used as callback for rum_extract_tsvector_internal.
+ * Returns hashed entry from tsvector.
+ */
 static Datum
 build_tsvector_hash_entry(TSVector vector, WordEntry *we)
 {
@@ -559,6 +576,9 @@ build_tsvector_hash_entry(TSVector vector, WordEntry *we)
 	return Int32GetDatum(hash_value);
 }
 
+/*
+ * Extracts lexemes from tsvector with additional information.
+ */
 Datum
 rum_extract_tsvector(PG_FUNCTION_ARGS)
 {
@@ -575,6 +595,9 @@ rum_extract_tsvector(PG_FUNCTION_ARGS)
 	PG_RETURN_POINTER(entries);
 }
 
+/*
+ * Extracts hashed lexemes from tsvector with additional information.
+ */
 Datum
 rum_extract_tsvector_hash(PG_FUNCTION_ARGS)
 {
@@ -592,6 +615,14 @@ rum_extract_tsvector_hash(PG_FUNCTION_ARGS)
 	PG_RETURN_POINTER(entries);
 }
 
+/*
+ * Extracting tsquery lexemes.
+ */
+
+/*
+ * Extracts tsquery lexemes from **query**. Uses **build_tsquery_entry**
+ * callback to extract lexeme.
+ */
 static Datum *
 rum_extract_tsquery_internal(TSQuery query,
 							 int32 *nentries,
@@ -640,11 +671,6 @@ rum_extract_tsquery_internal(TSQuery query,
 
 		for (i = 0; i < (*nentries); i++)
 		{
-			text       *txt;
-
-			txt = cstring_to_text_with_len(GETOPERAND(query) + operands[i]->distance,
-										   operands[i]->length);
-			entries[i] = PointerGetDatum(txt);
 			entries[i] = build_tsquery_entry(query, operands[i]);
 			partialmatch[i] = operands[i]->prefix;
 			(*extra_data)[i] = (Pointer) map_item_operand;
@@ -679,6 +705,9 @@ rum_extract_tsquery_internal(TSQuery query,
 	return entries;
 }
 
+/*
+ * Extract lexeme from tsquery.
+ */
 static Datum
 build_tsquery_entry(TSQuery query, QueryOperand *operand)
 {
@@ -689,6 +718,9 @@ build_tsquery_entry(TSQuery query, QueryOperand *operand)
 	return PointerGetDatum(txt);
 }
 
+/*
+ * Extract hashed lexeme from tsquery.
+ */
 static Datum
 build_tsquery_hash_entry(TSQuery query, QueryOperand *operand)
 {
@@ -700,6 +732,9 @@ build_tsquery_hash_entry(TSQuery query, QueryOperand *operand)
 	return hash_value;
 }
 
+/*
+ * Extracts lexemes from tsquery with information about prefix search syntax.
+ */
 Datum
 rum_extract_tsquery(PG_FUNCTION_ARGS)
 {
@@ -723,6 +758,10 @@ rum_extract_tsquery(PG_FUNCTION_ARGS)
 	PG_RETURN_POINTER(entries);
 }
 
+/*
+ * Extracts hashed lexemes from tsquery with information about prefix search
+ * syntax.
+ */
 Datum
 rum_extract_tsquery_hash(PG_FUNCTION_ARGS)
 {
@@ -746,6 +785,10 @@ rum_extract_tsquery_hash(PG_FUNCTION_ARGS)
 	PG_RETURN_POINTER(entries);
 }
 
+/*
+ * Functions used for ranking.
+ */
+
 static int
 compareDocR(const void *va, const void *vb)
 {
@@ -1297,6 +1340,10 @@ calc_score(float4 *arrdata, TSVector txt, TSQuery query, int method)
 	return (float4) Wdoc;
 }
 
+/*
+ * Calculates distance inside index. Uses additional information with lexemes
+ * positions.
+ */
 Datum
 rum_tsquery_distance(PG_FUNCTION_ARGS)
 {
@@ -1320,6 +1367,9 @@ rum_tsquery_distance(PG_FUNCTION_ARGS)
 		PG_RETURN_FLOAT8(1.0 / res);
 }
 
+/*
+ * Implementation of <=> operator. Uses default normalization method.
+ */
 Datum
 rum_ts_distance_tt(PG_FUNCTION_ARGS)
 {
@@ -1337,6 +1387,9 @@ rum_ts_distance_tt(PG_FUNCTION_ARGS)
 		PG_RETURN_FLOAT4(1.0 / res);
 }
 
+/*
+ * Implementation of <=> operator. Uses specified normalization method.
+ */
 Datum
 rum_ts_distance_ttf(PG_FUNCTION_ARGS)
 {
@@ -1355,6 +1408,9 @@ rum_ts_distance_ttf(PG_FUNCTION_ARGS)
 		PG_RETURN_FLOAT4(1.0 / res);
 }
 
+/*
+ * Implementation of <=> operator. Uses specified normalization method.
+ */
 Datum
 rum_ts_distance_td(PG_FUNCTION_ARGS)
 {
@@ -1401,6 +1457,9 @@ rum_ts_distance_td(PG_FUNCTION_ARGS)
 		PG_RETURN_FLOAT4(1.0 / res);
 }
 
+/*
+ * Casts tsquery to rum_distance_query type.
+ */
 Datum
 tsquery_to_distance_query(PG_FUNCTION_ARGS)
 {
@@ -1426,6 +1485,9 @@ tsquery_to_distance_query(PG_FUNCTION_ARGS)
 	PG_RETURN_DATUM(HeapTupleGetDatum(htup));
 }
 
+/*
+ * Specifies additional information type for operator class.
+ */
 Datum
 rum_tsvector_config(PG_FUNCTION_ARGS)
 {

Original file line number	Diff line number	Diff line change
`@@ -490,6 +490,14 @@ SortAndUniqItems(TSQuery q, int *size)`
`490`	`490`	`return res;`
`491`	`491`	`}`
`492`	`492`
	`493`	`+/*`
	`494`	`+ * Extracting tsvector lexems.`
	`495`	`+ */`
	`496`	`+`
	`497`	`+/*`
	`498`	`+ * Extracts tsvector lexemes from vector. Uses build_tsvector_entry`
	`499`	`+ * callback to extract entry.`
	`500`	`+ */`
`493`	`501`	`static Datum *`
`494`	`502`	`rum_extract_tsvector_internal(TSVector vector,`
`495`	`503`	`int32 *nentries,`
`@@ -515,6 +523,7 @@ rum_extract_tsvector_internal(TSVector vector,`
`515`	`523`	`bytea *posData;`
`516`	`524`	`int posDataSize;`
`517`	`525`
	`526`	`+ /* Extract entry using specified method */`
`518`	`527`	`entries[i] = build_tsvector_entry(vector, we);`
`519`	`528`
`520`	`529`	`if (we->haspos)`
`@@ -540,6 +549,10 @@ rum_extract_tsvector_internal(TSVector vector,`
`540`	`549`	`return entries;`
`541`	`550`	`}`
`542`	`551`
	`552`	`+/*`
	`553`	`+ * Used as callback for rum_extract_tsvector_internal().`
	`554`	`+ * Just extract entry from tsvector.`
	`555`	`+ */`
`543`	`556`	`static Datum`
`544`	`557`	`build_tsvector_entry(TSVector vector, WordEntry *we)`
`545`	`558`	`{`
`@@ -549,6 +562,10 @@ build_tsvector_entry(TSVector vector, WordEntry *we)`
`549`	`562`	`return PointerGetDatum(txt);`
`550`	`563`	`}`
`551`	`564`
	`565`	`+/*`
	`566`	`+ * Used as callback for rum_extract_tsvector_internal.`
	`567`	`+ * Returns hashed entry from tsvector.`
	`568`	`+ */`
`552`	`569`	`static Datum`
`553`	`570`	`build_tsvector_hash_entry(TSVector vector, WordEntry *we)`
`554`	`571`	`{`
`@@ -559,6 +576,9 @@ build_tsvector_hash_entry(TSVector vector, WordEntry *we)`
`559`	`576`	`return Int32GetDatum(hash_value);`
`560`	`577`	`}`
`561`	`578`
	`579`	`+/*`
	`580`	`+ * Extracts lexemes from tsvector with additional information.`
	`581`	`+ */`
`562`	`582`	`Datum`
`563`	`583`	`rum_extract_tsvector(PG_FUNCTION_ARGS)`
`564`	`584`	`{`
`@@ -575,6 +595,9 @@ rum_extract_tsvector(PG_FUNCTION_ARGS)`
`575`	`595`	`PG_RETURN_POINTER(entries);`
`576`	`596`	`}`
`577`	`597`
	`598`	`+/*`
	`599`	`+ * Extracts hashed lexemes from tsvector with additional information.`
	`600`	`+ */`
`578`	`601`	`Datum`
`579`	`602`	`rum_extract_tsvector_hash(PG_FUNCTION_ARGS)`
`580`	`603`	`{`
`@@ -592,6 +615,14 @@ rum_extract_tsvector_hash(PG_FUNCTION_ARGS)`
`592`	`615`	`PG_RETURN_POINTER(entries);`
`593`	`616`	`}`
`594`	`617`
	`618`	`+/*`
	`619`	`+ * Extracting tsquery lexemes.`
	`620`	`+ */`
	`621`	`+`
	`622`	`+/*`
	`623`	`+ * Extracts tsquery lexemes from query. Uses build_tsquery_entry`
	`624`	`+ * callback to extract lexeme.`
	`625`	`+ */`
`595`	`626`	`static Datum *`
`596`	`627`	`rum_extract_tsquery_internal(TSQuery query,`
`597`	`628`	`int32 *nentries,`
`@@ -640,11 +671,6 @@ rum_extract_tsquery_internal(TSQuery query,`
`640`	`671`
`641`	`672`	`for (i = 0; i < (*nentries); i++)`
`642`	`673`	`{`
`643`		`- text *txt;`
`644`		`-`
`645`		`- txt = cstring_to_text_with_len(GETOPERAND(query) + operands[i]->distance,`
`646`		`- operands[i]->length);`
`647`		`- entries[i] = PointerGetDatum(txt);`
`648`	`674`	`entries[i] = build_tsquery_entry(query, operands[i]);`
`649`	`675`	`partialmatch[i] = operands[i]->prefix;`
`650`	`676`	`(*extra_data)[i] = (Pointer) map_item_operand;`
`@@ -679,6 +705,9 @@ rum_extract_tsquery_internal(TSQuery query,`
`679`	`705`	`return entries;`
`680`	`706`	`}`
`681`	`707`
	`708`	`+/*`
	`709`	`+ * Extract lexeme from tsquery.`
	`710`	`+ */`
`682`	`711`	`static Datum`
`683`	`712`	`build_tsquery_entry(TSQuery query, QueryOperand *operand)`
`684`	`713`	`{`
`@@ -689,6 +718,9 @@ build_tsquery_entry(TSQuery query, QueryOperand *operand)`
`689`	`718`	`return PointerGetDatum(txt);`
`690`	`719`	`}`
`691`	`720`
	`721`	`+/*`
	`722`	`+ * Extract hashed lexeme from tsquery.`
	`723`	`+ */`
`692`	`724`	`static Datum`
`693`	`725`	`build_tsquery_hash_entry(TSQuery query, QueryOperand *operand)`
`694`	`726`	`{`
`@@ -700,6 +732,9 @@ build_tsquery_hash_entry(TSQuery query, QueryOperand *operand)`
`700`	`732`	`return hash_value;`
`701`	`733`	`}`
`702`	`734`
	`735`	`+/*`
	`736`	`+ * Extracts lexemes from tsquery with information about prefix search syntax.`
	`737`	`+ */`
`703`	`738`	`Datum`
`704`	`739`	`rum_extract_tsquery(PG_FUNCTION_ARGS)`
`705`	`740`	`{`
`@@ -723,6 +758,10 @@ rum_extract_tsquery(PG_FUNCTION_ARGS)`
`723`	`758`	`PG_RETURN_POINTER(entries);`
`724`	`759`	`}`
`725`	`760`
	`761`	`+/*`
	`762`	`+ * Extracts hashed lexemes from tsquery with information about prefix search`
	`763`	`+ * syntax.`
	`764`	`+ */`
`726`	`765`	`Datum`
`727`	`766`	`rum_extract_tsquery_hash(PG_FUNCTION_ARGS)`
`728`	`767`	`{`
`@@ -746,6 +785,10 @@ rum_extract_tsquery_hash(PG_FUNCTION_ARGS)`
`746`	`785`	`PG_RETURN_POINTER(entries);`
`747`	`786`	`}`
`748`	`787`
	`788`	`+/*`
	`789`	`+ * Functions used for ranking.`
	`790`	`+ */`
	`791`	`+`
`749`	`792`	`static int`
`750`	`793`	`compareDocR(const void va, const void vb)`
`751`	`794`	`{`
`@@ -1297,6 +1340,10 @@ calc_score(float4 *arrdata, TSVector txt, TSQuery query, int method)`
`1297`	`1340`	`return (float4) Wdoc;`
`1298`	`1341`	`}`
`1299`	`1342`
	`1343`	`+/*`
	`1344`	`+ * Calculates distance inside index. Uses additional information with lexemes`
	`1345`	`+ * positions.`
	`1346`	`+ */`
`1300`	`1347`	`Datum`
`1301`	`1348`	`rum_tsquery_distance(PG_FUNCTION_ARGS)`
`1302`	`1349`	`{`
`@@ -1320,6 +1367,9 @@ rum_tsquery_distance(PG_FUNCTION_ARGS)`
`1320`	`1367`	`PG_RETURN_FLOAT8(1.0 / res);`
`1321`	`1368`	`}`
`1322`	`1369`
	`1370`	`+/*`
	`1371`	`+ * Implementation of <=> operator. Uses default normalization method.`
	`1372`	`+ */`
`1323`	`1373`	`Datum`
`1324`	`1374`	`rum_ts_distance_tt(PG_FUNCTION_ARGS)`
`1325`	`1375`	`{`
`@@ -1337,6 +1387,9 @@ rum_ts_distance_tt(PG_FUNCTION_ARGS)`
`1337`	`1387`	`PG_RETURN_FLOAT4(1.0 / res);`
`1338`	`1388`	`}`
`1339`	`1389`
	`1390`	`+/*`
	`1391`	`+ * Implementation of <=> operator. Uses specified normalization method.`
	`1392`	`+ */`
`1340`	`1393`	`Datum`
`1341`	`1394`	`rum_ts_distance_ttf(PG_FUNCTION_ARGS)`
`1342`	`1395`	`{`
`@@ -1355,6 +1408,9 @@ rum_ts_distance_ttf(PG_FUNCTION_ARGS)`
`1355`	`1408`	`PG_RETURN_FLOAT4(1.0 / res);`
`1356`	`1409`	`}`
`1357`	`1410`
	`1411`	`+/*`
	`1412`	`+ * Implementation of <=> operator. Uses specified normalization method.`
	`1413`	`+ */`
`1358`	`1414`	`Datum`
`1359`	`1415`	`rum_ts_distance_td(PG_FUNCTION_ARGS)`
`1360`	`1416`	`{`
`@@ -1401,6 +1457,9 @@ rum_ts_distance_td(PG_FUNCTION_ARGS)`
`1401`	`1457`	`PG_RETURN_FLOAT4(1.0 / res);`
`1402`	`1458`	`}`
`1403`	`1459`
	`1460`	`+/*`
	`1461`	`+ * Casts tsquery to rum_distance_query type.`
	`1462`	`+ */`
`1404`	`1463`	`Datum`
`1405`	`1464`	`tsquery_to_distance_query(PG_FUNCTION_ARGS)`
`1406`	`1465`	`{`
`@@ -1426,6 +1485,9 @@ tsquery_to_distance_query(PG_FUNCTION_ARGS)`
`1426`	`1485`	`PG_RETURN_DATUM(HeapTupleGetDatum(htup));`
`1427`	`1486`	`}`
`1428`	`1487`
	`1488`	`+/*`
	`1489`	`+ * Specifies additional information type for operator class.`
	`1490`	`+ */`
`1429`	`1491`	`Datum`
`1430`	`1492`	`rum_tsvector_config(PG_FUNCTION_ARGS)`
`1431`	`1493`	`{`