Fix isvalid for 3-byte overlong encoded UTF-8 sequences (#29908)

JuliaLang · Nov 3, 2018 · e9d32a6 · e9d32a6 · nanosoldier · Nov 4, 2018
1 parent 447095d
commit e9d32a6
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 2 deletions.
diff --git a/src/support/utf8.c b/src/support/utf8.c
@@ -570,6 +570,8 @@ int u8_isvalid(const char *str, size_t len)
             return 0;
         // Check for surrogate chars
         if (byt == 0xed && *pnt > 0x9f) return 0;
+	// Check for overlong encoding
+	if (byt == 0xe0 && *pnt < 0xa0) return 0;
         pnt += 2;
     } else {                        // 4-byte sequence
         // Must have 3 valid continuation characters

diff --git a/test/strings/basic.jl b/test/strings/basic.jl
@@ -467,9 +467,17 @@ end
             end
         end
     end
+    # Check for short three-byte sequences
+    @test isvalid(String, UInt8[0xe0]) == false
+    for (rng, flg) in ((0x00:0x9f, false), (0xa0:0xbf, true), (0xc0:0xff, false))
+        for cont in rng
+            @test isvalid(String, UInt8[0xe0, cont]) == false
+            @test isvalid(String, UInt8[0xe0, cont, 0x80]) == flg
+        end
+    end
     # Check three-byte sequences
-    for r1 in (0xe0:0xec, 0xee:0xef)
-        for byt = r1
+    for r1 in (0xe1:0xec, 0xee:0xef)
+        for byt in r1
             # Check for short sequence
             @test isvalid(String, UInt8[byt]) == false
             for (rng,flg) in ((0x00:0x7f, false), (0x80:0xbf, true), (0xc0:0xff, false))