From 9c8f2ccaeee46a3ccca03b3fd5ce1c53b04efc72 Mon Sep 17 00:00:00 2001
From: Gabriel Vasile <gabriel.vasile@email.com>
Date: Sun, 12 Dec 2021 11:38:39 +0200
Subject: [PATCH] Add fuzzing for charset detector functions

---
 internal/charset/charset_test.go | 55 ++++++++++++++++++++++++++++++++
 1 file changed, 55 insertions(+)

diff --git a/internal/charset/charset_test.go b/internal/charset/charset_test.go
index d9b8603..a6e0587 100644
--- a/internal/charset/charset_test.go
+++ b/internal/charset/charset_test.go
@@ -55,3 +55,58 @@ func TestFromPlain(t *testing.T) {
 		}
 	}
 }
+
+func FuzzFromPlain(f *testing.F) {
+	samples := [][]byte{
+		[]byte{0xe6, 0xf8, 0xe5, 0x85, 0x85},
+		[]byte{0xe6, 0xf8, 0xe5},
+		[]byte("æøå"),
+	}
+
+	for _, s := range samples {
+		f.Add(s)
+	}
+
+	f.Fuzz(func(t *testing.T, d []byte) {
+		if charset := FromPlain(d); charset == "" {
+			t.Skip()
+		}
+	})
+}
+func FuzzFromHTML(f *testing.F) {
+	samples := []string{
+		`<meta charset="c">`,
+		`<meta charset="щ">`,
+		`<meta http-equiv="content-type" content="a/b; charset=c">`,
+		`<meta http-equiv="content-type" content="a/b; charset=щ">`,
+		`<f 1=2 /><meta charset="c">`,
+		`<f a=2><meta http-equiv="content-type" content="a/b; charset=c">`,
+		`<f 1=2 /><meta b="b" charset="c">`,
+		`<f a=2><meta b="b" http-equiv="content-type" content="a/b; charset=c">`,
+	}
+
+	for _, s := range samples {
+		f.Add([]byte(s))
+	}
+
+	f.Fuzz(func(t *testing.T, d []byte) {
+		if charset := FromHTML(d); charset == "" {
+			t.Skip()
+		}
+	})
+}
+func FuzzFromXML(f *testing.F) {
+	samples := []string{
+		`<?xml version="1.0" encoding="c"?>`,
+	}
+
+	for _, s := range samples {
+		f.Add([]byte(s))
+	}
+
+	f.Fuzz(func(t *testing.T, d []byte) {
+		if charset := FromXML(d); charset == "" {
+			t.Skip()
+		}
+	})
+}