Merge pull request #9 from nlnwa/fix-slow-state

Fix nonlinear runtime in stateQuery and stateFragment
nlnwa · Jun 14, 2022 · 9a67708 · 9a67708
2 parents 441bfd9 + adac416
commit 9a67708
Show file tree

Hide file tree

Showing 2 changed files with 35 additions and 25 deletions.
diff --git a/url/parser.go b/url/parser.go
@@ -194,7 +194,6 @@ func (p *parser) basicParser(urlOrRef string, base *Url, url *Url, stateOverride
 					input.nextCodePoint()
 				} else {
 					url.cannotBeABaseUrl = true
-					buffer.Reset()
 					state = stateCannotBeABaseUrl
 				}
 			} else if !stateOverridden {
@@ -602,21 +601,15 @@ func (p *parser) basicParser(urlOrRef string, base *Url, url *Url, stateOverride
 			if r == '?' {
 				url.search = new(string)
 				state = stateQuery
-				if len(url.path) == 0 {
-					url.path = append(url.path, "")
-				}
-				url.path[0] += buffer.String()
+				url.path = append(url.path, buffer.String())
 				buffer.Reset()
 			} else if r == '#' {
 				url.hash = new(string)
 				state = stateFragment
-				if len(url.path) == 0 {
-					url.path = append(url.path, "")
-				}
-				url.path[0] += buffer.String()
+				url.path = append(url.path, buffer.String())
 				buffer.Reset()
-			} else {
-				if !input.eof && !isURLCodePoint(r) && r != '%' {
+			} else if !input.eof {
+				if !isURLCodePoint(r) && r != '%' {
 					if err := p.handleError(url, errors.IllegalCodePoint); err != nil {
 						return nil, err
 					}
@@ -626,25 +619,19 @@ func (p *parser) basicParser(urlOrRef string, base *Url, url *Url, stateOverride
 					if err := p.handleError(url, errors.InvalidPercentEncoding); err != nil {
 						return nil, err
 					}
-				}
-				if !input.eof {
-					if invalidPercentEncoding {
-						buffer.WriteString(p.percentEncodeInvalidRune(r, C0PercentEncodeSet))
-					} else {
-						buffer.WriteString(p.percentEncodeRune(r, C0PercentEncodeSet))
-					}
+					buffer.WriteString(p.percentEncodeInvalidRune(r, C0PercentEncodeSet))
 				} else {
-					if len(url.path) == 0 {
-						url.path = append(url.path, "")
-					}
-					url.path[0] += buffer.String()
-					buffer.Reset()
+					buffer.WriteString(p.percentEncodeRune(r, C0PercentEncodeSet))
 				}
+			} else {
+				url.path = append(url.path, buffer.String())
 			}
 		case stateQuery:
 			if !stateOverridden && r == '#' {
 				url.hash = new(string)
 				state = stateFragment
+				*url.search = buffer.String()
+				buffer.Reset()
 			} else if !input.eof {
 				if !isURLCodePoint(r) && r != '%' {
 					if err := p.handleError(url, errors.IllegalCodePoint); err != nil {
@@ -660,7 +647,9 @@ func (p *parser) basicParser(urlOrRef string, base *Url, url *Url, stateOverride
 				if url.isSpecialScheme(url.protocol) {
 					encodeSet = p.opts.specialQueryPercentEncodeSet
 				}
-				*url.search += p.percentEncodeRune(r, encodeSet)
+				buffer.WriteString(p.percentEncodeRune(r, encodeSet))
+			} else {
+				*url.search = buffer.String()
 			}
 		case stateFragment:
 			if !input.eof {
@@ -678,7 +667,9 @@ func (p *parser) basicParser(urlOrRef string, base *Url, url *Url, stateOverride
 				if url.isSpecialScheme(url.protocol) {
 					encodeSet = p.opts.specialFragmentPercentEncodeSet
 				}
-				*url.hash += p.percentEncodeRune(r, encodeSet)
+				buffer.WriteString(p.percentEncodeRune(r, encodeSet))
+			} else {
+				*url.hash = buffer.String()
 			}
 		}
 

diff --git a/url/url_test.go b/url/url_test.go
@@ -629,3 +629,22 @@ func BenchmarkIssue6(b *testing.B) {
 		})
 	}
 }
+
+func BenchmarkIssue8(b *testing.B) {
+	// https://github.com/nlnwa/whatwg-url/issues/8
+	for i := 10; i <= 20; i++ {
+		n := 1 << i
+		b.Run(fmt.Sprint(n), func(b *testing.B) {
+			var buf strings.Builder
+			buf.Grow(n + 32)
+			buf.WriteString("http://example.com/?foo=bar")
+			for j := 0; j <= n; j++ {
+				buf.WriteString("&foo=bar")
+			}
+			b.ResetTimer()
+			for i := 0; i < b.N; i++ {
+				_, _ = Parse(buf.String())
+			}
+		})
+	}
+}