From 81e7a94ca49a1b182841dec79691f840bad9cc66 Mon Sep 17 00:00:00 2001
From: Nuno Cruces <ncruces@users.noreply.github.com>
Date: Mon, 4 Nov 2024 19:30:10 +0000
Subject: [PATCH] Optimize regexp.

---
 ext/regexp/regexp.go      | 30 +++++++++++++++++++++++++-----
 ext/regexp/regexp_test.go | 23 +++++++++++++++++++++++
 2 files changed, 48 insertions(+), 5 deletions(-)

diff --git a/ext/regexp/regexp.go b/ext/regexp/regexp.go
index 1d0aa30..8a1867f 100644
--- a/ext/regexp/regexp.go
+++ b/ext/regexp/regexp.go
@@ -1,11 +1,11 @@
 // Package regexp provides additional regular expression functions.
 //
 // It provides the following Unicode aware functions:
-//   - regexp_like(),
-//   - regexp_count(),
-//   - regexp_instr(),
-//   - regexp_substr(),
-//   - regexp_replace(),
+//   - regexp_like(text, pattern),
+//   - regexp_count(text, pattern [, start]),
+//   - regexp_instr(text, pattern [, start [, N [, endoption [, subexpr ]]]]),
+//   - regexp_substr(text, pattern [, start [, N [, subexpr ]]]),
+//   - regexp_replace(text, pattern, replacement [, start [, N ]]),
 //   - and a REGEXP operator.
 //
 // The implementation uses Go [regexp/syntax] for regular expressions.
@@ -16,6 +16,7 @@ package regexp
 import (
 	"errors"
 	"regexp"
+	"strings"
 
 	"github.com/ncruces/go-sqlite3"
 )
@@ -42,6 +43,25 @@ func Register(db *sqlite3.Conn) error {
 		db.CreateFunction("regexp_replace", 5, flags, regexReplace))
 }
 
+// GlobPrefix returns a GLOB for a regular expression
+// appropriate to take advantage of the [LIKE optimization]
+// in a query such as:
+//
+//	SELECT column WHERE column GLOB :glob_prefix AND column REGEXP :regexp
+//
+// [LIKE optimization]: https://sqlite.org/optoverview.html#the_like_optimization
+func GlobPrefix(re *regexp.Regexp) string {
+	prefix, complete := re.LiteralPrefix()
+	i := strings.IndexAny(prefix, "*?[")
+	if i < 0 {
+		if complete {
+			return prefix
+		}
+		i = len(prefix)
+	}
+	return prefix[:i] + "*"
+}
+
 func load(ctx sqlite3.Context, i int, expr string) (*regexp.Regexp, error) {
 	re, ok := ctx.GetAuxData(i).(*regexp.Regexp)
 	if !ok {
diff --git a/ext/regexp/regexp_test.go b/ext/regexp/regexp_test.go
index d8dc83e..d5852ef 100644
--- a/ext/regexp/regexp_test.go
+++ b/ext/regexp/regexp_test.go
@@ -2,6 +2,7 @@ package regexp
 
 import (
 	"database/sql"
+	"regexp"
 	"testing"
 
 	"github.com/ncruces/go-sqlite3/driver"
@@ -101,3 +102,25 @@ func TestRegister_errors(t *testing.T) {
 		}
 	}
 }
+
+func TestGlobPrefix(t *testing.T) {
+	tests := []struct {
+		re   string
+		want string
+	}{
+		{``, ""},
+		{`a`, "a"},
+		{`a*`, "*"},
+		{`a+`, "a*"},
+		{`ab*`, "a*"},
+		{`ab+`, "ab*"},
+		{`a\?b`, "a*"},
+	}
+	for _, tt := range tests {
+		t.Run(tt.re, func(t *testing.T) {
+			if got := GlobPrefix(regexp.MustCompile(tt.re)); got != tt.want {
+				t.Errorf("GlobPrefix() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}