Properly escape strings when converting to text

Properly escape character sequences that need
to be escaped in JSON.

Change-Id: Ia2e3c53447df5b82330701232c868049b108eac6
Reviewed-by: Simons Kevin <kevin.simons@nokia.com>

2 files changed, 135 insertions, 5 deletions

diff --git a/src/qbinaryjsonobject.cpp b/src/qbinaryjsonobject.cpp
index 7086517..4243116 100644
--- a/src/qbinaryjsonobject.cpp
+++ b/src/qbinaryjsonobject.cpp
@@ -87,6 +87,134 @@ QVariantMap JsonObject::toVariantMap() const
 static void objectContentToJson(const Object *o, QByteArray &json, int indent);
 static void arrayContentToJson(const Array *a, QByteArray &json, int indent);
 
+// some code from qutfcodec.cpp, inlined here for performance reasons
+// to allow fast escaping of strings
+static inline bool isUnicodeNonCharacter(uint ucs4)
+{
+    // Unicode has a couple of "non-characters" that one can use internally,
+    // but are not allowed to be used for text interchange.
+    //
+    // Those are the last two entries each Unicode Plane (U+FFFE, U+FFFF,
+    // U+1FFFE, U+1FFFF, etc.) as well as the entries between U+FDD0 and
+    // U+FDEF (inclusive)
+
+    return (ucs4 & 0xfffe) == 0xfffe
+            || (ucs4 - 0xfdd0U) < 16;
+}
+
+static inline uchar hexdig(uint u)
+{
+    return (u < 0xa ? '0' + u : 'a' + u - 0xa);
+}
+
+static QByteArray escapedString(const QString &s)
+{
+    const uchar replacement = '?';
+    QByteArray ba(s.length(), Qt::Uninitialized);
+
+    uchar *cursor = (uchar *)ba.data();
+    const uchar *ba_end = cursor + ba.length();
+
+    const QChar *ch = s.constData();
+    const QChar *end = ch + s.length();
+
+    int surrogate_high = -1;
+
+    while (ch < end) {
+        if (cursor >= ba_end - 6) {
+            // ensure we have enough space
+            int pos = cursor - (uchar *)ba.constData();
+            ba.resize(ba.size()*2);
+            cursor = (uchar *)ba.data() + pos;
+            ba_end = (const uchar *)ba.constData() + ba.length();
+        }
+
+        uint u = ch->unicode();
+        if (surrogate_high >= 0) {
+            if (ch->isLowSurrogate()) {
+                u = QChar::surrogateToUcs4(surrogate_high, u);
+                surrogate_high = -1;
+            } else {
+                // high surrogate without low
+                *cursor = replacement;
+                ++ch;
+                surrogate_high = -1;
+                continue;
+            }
+        } else if (ch->isLowSurrogate()) {
+            // low surrogate without high
+            *cursor = replacement;
+            ++ch;
+            continue;
+        } else if (ch->isHighSurrogate()) {
+            surrogate_high = u;
+            ++ch;
+            continue;
+        }
+
+        if (u < 0x80) {
+            if (u < 0x20 || u == 0x22 || u == 0x5c) {
+                *cursor++ = '\\';
+                switch (u) {
+                case 0x22:
+                    *cursor++ = '"';
+                    break;
+                case 0x5c:
+                    *cursor++ = '\\';
+                    break;
+                case 0x8:
+                    *cursor++ = 'b';
+                    break;
+                case 0xc:
+                    *cursor++ = 'f';
+                    break;
+                case 0xa:
+                    *cursor++ = 'n';
+                    break;
+                case 0xd:
+                    *cursor++ = 'r';
+                    break;
+                case 0x9:
+                    *cursor++ = 't';
+                    break;
+                default:
+                    *cursor++ = 'u';
+                    *cursor++ = '0';
+                    *cursor++ = '0';
+                    *cursor++ = hexdig(u>>4);
+                    *cursor++ = hexdig(u & 0xf);
+               }
+            } else {
+                *cursor++ = (uchar)u;
+            }
+        } else {
+            if (u < 0x0800) {
+                *cursor++ = 0xc0 | ((uchar) (u >> 6));
+            } else {
+                // is it one of the Unicode non-characters?
+                if (isUnicodeNonCharacter(u)) {
+                    *cursor++ = replacement;
+                    ++ch;
+                    continue;
+                }
+
+                if (u > 0xffff) {
+                    *cursor++ = 0xf0 | ((uchar) (u >> 18));
+                    *cursor++ = 0x80 | (((uchar) (u >> 12)) & 0x3f);
+                } else {
+                    *cursor++ = 0xe0 | (((uchar) (u >> 12)) & 0x3f);
+                }
+                *cursor++ = 0x80 | (((uchar) (u >> 6)) & 0x3f);
+            }
+            *cursor++ = 0x80 | ((uchar) (u&0x3f));
+        }
+        ++ch;
+    }
+
+    ba.resize(cursor - (const uchar *)ba.constData());
+    return ba;
+}
+
 static void valueToJson(const Value *v, QByteArray &json, int indent)
 {
     ValueType type = v ? (ValueType)v->type : NullValue;
@@ -99,7 +227,7 @@ static void valueToJson(const Value *v, QByteArray &json, int indent)
         break;
     case StringValue:
         json += '"';
-        json += v->shallowString().toUtf8();
+        json += escapedString(v->shallowString());
         json += '"';
         break;
     case ArrayValue:
@@ -154,7 +282,7 @@ static void objectContentToJson(const Object *o, QByteArray &json, int indent)
         Entry *e = o->entryAt(i);
         json += indentString;
         json += '"';
-        json += e->shallowKey().toUtf8();
+        json += escapedString(e->shallowKey());
         json += "\": ";
         const Value *v = e->value();
         valueToJson(v, json, indent);
diff --git a/tests/auto/tst_qtbinaryjson.cpp b/tests/auto/tst_qtbinaryjson.cpp
index 2b3c042..9063025 100644
--- a/tests/auto/tst_qtbinaryjson.cpp
+++ b/tests/auto/tst_qtbinaryjson.cpp
@@ -477,26 +477,28 @@ void TestQtBinaryJson::toVariantMap()
 void TestQtBinaryJson::toJson()
 {
     JsonObject object;
-    object.insert("Key", QString("Value"));
+    object.insert("\\Key\n", QString("Value"));
     object.insert("null", JsonValue());
     JsonArray array;
     array.append(true);
     array.append(999.);
     array.append(QLatin1String("string"));
     array.append(JsonValue());
+    array.append(QLatin1String("\\\a\n\r\b\tabcABC\""));
     object.insert("Array", array);
 
     QByteArray json = object.toJson();
 
     QByteArray expected =
             "{\n"
-            "    \"Key\": \"Value\",\n"
+            "    \"\\\\Key\\n\": \"Value\",\n"
             "    \"null\": null,\n"
             "    \"Array\": [\n"
             "        true,\n"
             "        999,\n"
             "        \"string\",\n"
-            "        null\n"
+            "        null,\n"
+            "        \"\\\\\\u0007\\n\\r\\b\\tabcABC\\\"\"\n"
             "    ]\n"
             "}\n";
     QCOMPARE(json, expected);