duckdb 0.7.1-dev407.0 → 0.7.1-dev415.0

package/package.json

@@ -2,7 +2,7 @@
   "name": "duckdb",
   "main": "./lib/duckdb.js",
   "types": "./lib/duckdb.d.ts",
-  "version": "0.7.1-dev407.0",
+  "version": "0.7.1-dev415.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/duckdb/src/catalog/catalog_entry/schema_catalog_entry.cpp

@@ -24,7 +24,7 @@ SimilarCatalogEntry SchemaCatalogEntry::GetSimilarEntry(CatalogTransaction trans
                                                         const string &name) {
 	SimilarCatalogEntry result;
 	Scan(transaction.GetContext(), type, [&](CatalogEntry *entry) {
-		auto ldist = StringUtil::LevenshteinDistance(entry->name, name);
+		auto ldist = StringUtil::SimilarityScore(entry->name, name);
 		if (ldist < result.distance) {
 			result.distance = ldist;
 			result.name = entry->name;

package/src/duckdb/src/catalog/catalog_set.cpp

@@ -460,7 +460,7 @@ SimilarCatalogEntry CatalogSet::SimilarEntry(CatalogTransaction transaction, con
 	for (auto &kv : mapping) {
 		auto mapping_value = GetMapping(transaction, kv.first);
 		if (mapping_value && !mapping_value->deleted) {
-			auto ldist = StringUtil::LevenshteinDistance(kv.first, name);
+			auto ldist = StringUtil::SimilarityScore(kv.first, name);
 			if (ldist < result.distance) {
 				result.distance = ldist;
 				result.name = kv.first;

package/src/duckdb/src/common/string_util.cpp

@@ -249,7 +249,7 @@ private:
 };
 
 // adapted from https://en.wikibooks.org/wiki/Algorithm_Implementation/Strings/Levenshtein_distance#C++
-idx_t StringUtil::LevenshteinDistance(const string &s1_p, const string &s2_p) {
+idx_t StringUtil::LevenshteinDistance(const string &s1_p, const string &s2_p, idx_t not_equal_penalty) {
 	auto s1 = StringUtil::Lower(s1_p);
 	auto s2 = StringUtil::Lower(s2_p);
 	idx_t len1 = s1.size();
@@ -273,7 +273,7 @@ idx_t StringUtil::LevenshteinDistance(const string &s1_p, const string &s2_p) {
 			// d[i][j] = std::min({ d[i - 1][j] + 1,
 			//                      d[i][j - 1] + 1,
 			//                      d[i - 1][j - 1] + (s1[i - 1] == s2[j - 1] ? 0 : 1) });
-			int equal = s1[i - 1] == s2[j - 1] ? 0 : 1;
+			int equal = s1[i - 1] == s2[j - 1] ? 0 : not_equal_penalty;
 			idx_t adjacent_score1 = array.Score(i - 1, j) + 1;
 			idx_t adjacent_score2 = array.Score(i, j - 1) + 1;
 			idx_t adjacent_score3 = array.Score(i - 1, j - 1) + equal;
@@ -285,15 +285,19 @@ idx_t StringUtil::LevenshteinDistance(const string &s1_p, const string &s2_p) {
 	return array.Score(len1, len2);
 }
 
+idx_t StringUtil::SimilarityScore(const string &s1, const string &s2) {
+	return LevenshteinDistance(s1, s2, 3);
+}
+
 vector<string> StringUtil::TopNLevenshtein(const vector<string> &strings, const string &target, idx_t n,
                                            idx_t threshold) {
 	vector<pair<string, idx_t>> scores;
 	scores.reserve(strings.size());
 	for (auto &str : strings) {
 		if (target.size() < str.size()) {
-			scores.emplace_back(str, LevenshteinDistance(str.substr(0, target.size()), target));
+			scores.emplace_back(str, SimilarityScore(str.substr(0, target.size()), target));
 		} else {
-			scores.emplace_back(str, LevenshteinDistance(str, target));
+			scores.emplace_back(str, SimilarityScore(str, target));
 		}
 	}
 	return TopNStrings(scores, n, threshold);

package/src/duckdb/src/function/table/version/pragma_version.cpp

@@ -1,8 +1,8 @@
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "0.7.1-dev407"
+#define DUCKDB_VERSION "0.7.1-dev415"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "66ba97b5f9"
+#define DUCKDB_SOURCE_ID "ae3510f069"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"

package/src/duckdb/src/include/duckdb/common/string_util.hpp

@@ -145,8 +145,15 @@ public:
 	DUCKDB_API static string Replace(string source, const string &from, const string &to);
 
 	//! Get the levenshtein distance from two strings
-	DUCKDB_API static idx_t LevenshteinDistance(const string &s1, const string &s2);
-
+	//! The not_equal_penalty is the penalty given when two characters in a string are not equal
+	//! The regular levenshtein distance has a not equal penalty of 1, which means changing a character is as expensive
+	//! as adding or removing one For similarity searches we often want to give extra weight to changing a character For
+	//! example: with an equal penalty of 1, "pg_am" is closer to "depdelay" than "depdelay_minutes"
+	//! with an equal penalty of 3, "depdelay_minutes" is closer to "depdelay" than to "pg_am"
+	DUCKDB_API static idx_t LevenshteinDistance(const string &s1, const string &s2, idx_t not_equal_penalty = 1);
+
+	//! Returns the similarity score between two strings
+	DUCKDB_API static idx_t SimilarityScore(const string &s1, const string &s2);
 	//! Get the top-n strings (sorted by the given score distance) from a set of scores.
 	//! At least one entry is returned (if there is one).
 	//! Strings are only returned if they have a score less than the threshold.

package/src/duckdb/src/planner/bind_context.cpp

@@ -44,7 +44,7 @@ vector<string> BindContext::GetSimilarBindings(const string &column_name) {
 	for (auto &kv : bindings) {
 		auto binding = kv.second.get();
 		for (auto &name : binding->names) {
-			idx_t distance = StringUtil::LevenshteinDistance(name, column_name);
+			idx_t distance = StringUtil::SimilarityScore(name, column_name);
 			scores.emplace_back(binding->alias + "." + name, distance);
 		}
 	}

package/src/duckdb/src/planner/logical_operator.cpp

@@ -167,7 +167,8 @@ idx_t LogicalOperator::EstimateCardinality(ClientContext &context) {
 		max_cardinality = MaxValue(child->EstimateCardinality(context), max_cardinality);
 	}
 	has_estimated_cardinality = true;
-	return max_cardinality;
+	estimated_cardinality = max_cardinality;
+	return estimated_cardinality;
 }
 
 void LogicalOperator::Print() {