Bug 1491137 - Split the JS::Compile that takes const char*/size_t into JS::CompileUtf8 and JS::CompileLatin1 that assert |options.utf8| has consistent value, and a JS::Compile that delegates to one of those as appropriate -- a step toward having entirely separate compilation functions that do direct parsing of UTF-8 without inflating first. r=jandem

--HG--
extra : rebase_source : 4fbc10a7e5e221da4c68c02dc155d925e7a7367d

js/public/CompilationAndEvaluation.h

@@ -86,7 +86,7 @@ JS_ExecuteScript(JSContext* cx, JS::AutoVector<JSObject*>& envChain, JS::Handle<
  * |script| will always be set. On failure, it will be set to nullptr.
  */
 extern JS_PUBLIC_API(bool)
-JS_CompileScript(JSContext* cx, const char* ascii, size_t length,
+JS_CompileScript(JSContext* cx, const char* bytes, size_t length,
                  const JS::CompileOptions& options,
                  JS::MutableHandle<JSScript*> script);
 
@@ -152,9 +152,63 @@ extern JS_PUBLIC_API(bool)
 Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
         SourceBufferHolder& srcBuf, MutableHandle<JSScript*> script);
 
+/**
+ * Compile the provided UTF-8 data into a script.  If the data contains invalid
+ * UTF-8, an error is reported.
+ *
+ * The |options.utf8| flag is asserted to be true.  At a future time this flag
+ * will be removed and UTF-8-ness of source bytes will be entirely determined
+ * by which compilation function is called.
+ *
+ * |script| is always set to the compiled script or to null in case of error.
+ */
 extern JS_PUBLIC_API(bool)
+CompileUtf8(JSContext* cx, const ReadOnlyCompileOptions& options,
+            const char* bytes, size_t length, MutableHandle<JSScript*> script);
+
+/**
+ * Compile the provided Latin-1 data (i.e. each byte directly corresponds to
+ * the same Unicode code point) into a script.
+ *
+ * The |options.utf8| flag is asserted to be false.  At a future time this flag
+ * will be removed and UTF-8-ness of source bytes will be entirely determined
+ * by which compilation function is called.
+ *
+ * This function may eventually be removed, such that *only* bytes containing
+ * UTF-8 source text may be directly compiled.  Avoid using it if you can.
+ *
+ * |script| is always set to the compiled script or to null in case of error.
+ */
+extern JS_PUBLIC_API(bool)
+CompileLatin1(JSContext* cx, const ReadOnlyCompileOptions& options,
+              const char* bytes, size_t length, MutableHandle<JSScript*> script);
+
+/**
+ * DEPRECATED
+ *
+ * Compile the provided bytes into a script.
+ *
+ * If |options.utf8|, the bytes are interpreted as UTF-8 data.  If the data
+ * contains any malformed UTF-8, an error is reported.
+ *
+ * Otherwise they are interpreted as Latin-1, i.e. each byte directly
+ * corresponds to the same Unicode code point.
+ *
+ * |script| is always set to the compiled script or to null in case of error.
+ *
+ * Do not use this API.  The JS::CompileOptions::utf8 flag that indicates how
+ * to interpret |bytes| is currently being replaced by functions indicating an
+ * exact expected encoding.  If you have byte data to compile, you should use
+ * either JS::CompileUtf8 or JS::CompileLatin1, as appropriate.
+ */
+inline bool
 Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
-        const char* bytes, size_t length, MutableHandle<JSScript*> script);
+        const char* bytes, size_t length, MutableHandle<JSScript*> script)
+{
+    return options.utf8
+           ? CompileUtf8(cx, options, bytes, length, script)
+           : CompileLatin1(cx, options, bytes, length, script);
+}
 
 extern JS_PUBLIC_API(bool)
 Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
@@ -168,9 +222,17 @@ extern JS_PUBLIC_API(bool)
 CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& options,
                             SourceBufferHolder& srcBuf, MutableHandle<JSScript*> script);
 
+/**
+ * Compile the given Latin-1 data for non-syntactic scope.
+ *
+ * There is no way to compile UTF-8 data for non-syntactic scope because no
+ * user currently needs it.  Such way could be added in the future if it's ever
+ * needed.
+ */
 extern JS_PUBLIC_API(bool)
-CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& options,
-                            const char* bytes, size_t length, MutableHandle<JSScript*> script);
+CompileLatin1ForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& options,
+                                  const char* bytes, size_t length,
+                                  MutableHandle<JSScript*> script);
 
 extern JS_PUBLIC_API(bool)
 CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& options,

js/src/jsapi-tests/testCompileNonSyntactic.cpp

@@ -85,7 +85,7 @@ testCompile(bool nonSyntactic)
     CHECK(CompileForNonSyntacticScope(cx, options, buf, &script));
     CHECK_EQUAL(script->hasNonSyntacticScope(), true);
 
-    CHECK(CompileForNonSyntacticScope(cx, options, src, length, &script));
+    CHECK(CompileLatin1ForNonSyntacticScope(cx, options, src, length, &script));
     CHECK_EQUAL(script->hasNonSyntacticScope(), true);
 
     {

js/src/jsapi-tests/testExecuteInJSMEnvironment.cpp

@@ -28,7 +28,7 @@ BEGIN_TEST(testExecuteInJSMEnvironment_Basic)
     options.setNoScriptRval(true);
 
     JS::RootedScript script(cx);
-    CHECK(JS::CompileForNonSyntacticScope(cx, options, src, sizeof(src)-1, &script));
+    CHECK(JS::CompileLatin1ForNonSyntacticScope(cx, options, src, sizeof(src) - 1, &script));
 
     JS::RootedObject varEnv(cx, js::NewJSMEnvironment(cx));
     JS::RootedObject lexEnv(cx, JS_ExtensibleLexicalEnvironment(varEnv));
@@ -88,7 +88,7 @@ BEGIN_TEST(testExecuteInJSMEnvironment_Callback)
     options.setNoScriptRval(true);
 
     JS::RootedScript script(cx);
-    CHECK(JS::CompileForNonSyntacticScope(cx, options, src, sizeof(src)-1, &script));
+    CHECK(JS::CompileLatin1ForNonSyntacticScope(cx, options, src, sizeof(src) - 1, &script));
 
     JS::RootedObject nsvo(cx, js::NewJSMEnvironment(cx));
     CHECK(nsvo);

js/src/vm/CompilationAndEvaluation.cpp

@@ -9,6 +9,9 @@
 #include "js/CompilationAndEvaluation.h"
 
 #include "mozilla/Maybe.h" // mozilla::None, mozilla::Some
+#include "mozilla/TextUtils.h" // mozilla::IsAscii
+
+#include <algorithm> // std::all_of
 
 #include "jstypes.h" // JS_PUBLIC_API
 
@@ -50,15 +53,27 @@ Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
 }
 
 static bool
-Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
-        const char* bytes, size_t length, JS::MutableHandleScript script)
+CompileLatin1(JSContext* cx, const ReadOnlyCompileOptions& options,
+              const char* bytes, size_t length, JS::MutableHandleScript script)
 {
-    char16_t* chars;
-    if (options.utf8) {
-        chars = JS::UTF8CharsToNewTwoByteCharsZ(cx, UTF8Chars(bytes, length), &length).get();
-    } else {
-        chars = InflateString(cx, bytes, length);
+    MOZ_ASSERT(!options.utf8);
+
+    char16_t* chars = InflateString(cx, bytes, length);
+    if (!chars) {
+        return false;
     }
+
+    SourceBufferHolder source(chars, length, SourceBufferHolder::GiveOwnership);
+    return ::Compile(cx, options, source, script);
+}
+
+static bool
+CompileUtf8(JSContext* cx, const ReadOnlyCompileOptions& options,
+            const char* bytes, size_t length, JS::MutableHandleScript script)
+{
+    MOZ_ASSERT(options.utf8);
+
+    char16_t* chars = JS::UTF8CharsToNewTwoByteCharsZ(cx, UTF8Chars(bytes, length), &length).get();
     if (!chars) {
         return false;
     }
@@ -76,7 +91,12 @@ Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
         return false;
     }
 
-    return ::Compile(cx, options, reinterpret_cast<const char*>(buffer.begin()), buffer.length(), script);
+    return options.utf8
+           ? ::CompileUtf8(cx, options,
+                           reinterpret_cast<const char*>(buffer.begin()), buffer.length(), script)
+           : ::CompileLatin1(cx, options,
+                             reinterpret_cast<const char*>(buffer.begin()), buffer.length(),
+                             script);
 }
 
 static bool
@@ -100,10 +120,19 @@ JS::Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
 }
 
 bool
-JS::Compile(JSContext* cx, const ReadOnlyCompileOptions& options,
-            const char* bytes, size_t length, JS::MutableHandleScript script)
+JS::CompileLatin1(JSContext* cx, const ReadOnlyCompileOptions& options,
+                  const char* bytes, size_t length, JS::MutableHandleScript script)
 {
-    return ::Compile(cx, options, bytes, length, script);
+    MOZ_ASSERT(!options.utf8);
+    return ::CompileLatin1(cx, options, bytes, length, script);
+}
+
+bool
+JS::CompileUtf8(JSContext* cx, const ReadOnlyCompileOptions& options,
+                const char* bytes, size_t length, JS::MutableHandleScript script)
+{
+    MOZ_ASSERT(options.utf8);
+    return ::CompileUtf8(cx, options, bytes, length, script);
 }
 
 bool
@@ -130,12 +159,16 @@ JS::CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& opt
 }
 
 bool
-JS::CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& optionsArg,
-                                const char* bytes, size_t length, JS::MutableHandleScript script)
+JS::CompileLatin1ForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& optionsArg,
+                                      const char* bytes, size_t length,
+                                      JS::MutableHandleScript script)
 {
+    MOZ_ASSERT(!optionsArg.utf8,
+               "this function only compiles Latin-1 source text");
+
     CompileOptions options(cx, optionsArg);
     options.setNonSyntacticScope(true);
-    return ::Compile(cx, options, bytes, length, script);
+    return ::CompileLatin1(cx, options, bytes, length, script);
 }
 
 bool
@@ -157,10 +190,12 @@ JS::CompileForNonSyntacticScope(JSContext* cx, const ReadOnlyCompileOptions& opt
 }
 
 JS_PUBLIC_API(bool)
-JS_CompileScript(JSContext* cx, const char* ascii, size_t length,
+JS_CompileScript(JSContext* cx, const char* bytes, size_t length,
                  const JS::CompileOptions& options, MutableHandleScript script)
 {
-    return ::Compile(cx, options, ascii, length, script);
+    return options.utf8
+           ? ::CompileUtf8(cx, options, bytes, length, script)
+           : ::CompileLatin1(cx, options, bytes, length, script);
 }
 
 JS_PUBLIC_API(bool)

js/xpconnect/loader/mozJSComponentLoader.cpp

@@ -879,18 +879,18 @@ mozJSComponentLoader::ObjectForLocation(ComponentLoaderInfo& aInfo,
             // don't early return for them here.
             auto buf = map.get<char>();
             if (reuseGlobal) {
-                CompileForNonSyntacticScope(cx, options, buf.get(), map.size(), &script);
+                CompileLatin1ForNonSyntacticScope(cx, options, buf.get(), map.size(), &script);
             } else {
-                Compile(cx, options, buf.get(), map.size(), &script);
+                CompileLatin1(cx, options, buf.get(), map.size(), &script);
             }
         } else {
             nsCString str;
             MOZ_TRY_VAR(str, ReadScript(aInfo));
 
             if (reuseGlobal) {
-                CompileForNonSyntacticScope(cx, options, str.get(), str.Length(), &script);
+                CompileLatin1ForNonSyntacticScope(cx, options, str.get(), str.Length(), &script);
             } else {
-                Compile(cx, options, str.get(), str.Length(), &script);
+                CompileLatin1(cx, options, str.get(), str.Length(), &script);
             }
         }
         // Propagate the exception, if one exists. Also, don't leave the stale

js/xpconnect/loader/mozJSSubScriptLoader.cpp

@@ -178,9 +178,9 @@ PrepareScript(nsIURI* uri,
     // the lazy source loader doesn't know the encoding.
     options.setSourceIsLazy(true);
     if (wantGlobalScript) {
-        return JS::Compile(cx, options, buf, len, script);
+        return JS::CompileLatin1(cx, options, buf, len, script);
     }
-    return JS::CompileForNonSyntacticScope(cx, options, buf, len, script);
+    return JS::CompileLatin1ForNonSyntacticScope(cx, options, buf, len, script);
 }
 
 static bool