Decode escaped Unicode characters like \u00DE (issue #304, PR #791)

src/ArduinoJson/Configuration.hpp

@@ -120,6 +120,11 @@
 #endif
 #endif
 
+// Convert unicode escape sequence (\u0123) to UTF-8
+#ifndef ARDUINOJSON_DECODE_UNICODE
+#define ARDUINOJSON_DECODE_UNICODE 0
+#endif
+
 // Control the exponentiation threshold for big numbers
 // CAUTION: cannot be more that 1e9 !!!!
 #ifndef ARDUINOJSON_POSITIVE_EXPONENTIATION_THRESHOLD

src/ArduinoJson/Json/JsonDeserializer.hpp

@@ -11,6 +11,7 @@
 #include "../Polyfills/type_traits.hpp"
 #include "../Variant/VariantData.hpp"
 #include "EscapeSequence.hpp"
+#include "Utf8.hpp"
 
 namespace ARDUINOJSON_NAMESPACE {
 
@@ -192,7 +193,18 @@ class JsonDeserializer {
       if (c == '\\') {
         c = current();
         if (c == '\0') return DeserializationError::IncompleteInput;
-        if (c == 'u') return DeserializationError::NotSupported;
+        if (c == 'u') {
+#if ARDUINOJSON_DECODE_UNICODE
+          uint16_t codepoint;
+          move();
+          DeserializationError err = parseCodepoint(codepoint);
+          if (err) return err;
+          Utf8::encodeCodepoint(codepoint, builder);
+          continue;
+#else
+          return DeserializationError::NotSupported;
+#endif
+        }
         // replace char
         c = EscapeSequence::unescapeChar(c);
         if (c == '\0') return DeserializationError::InvalidInput;
@@ -256,6 +268,19 @@ class JsonDeserializer {
     return DeserializationError::Ok;
   }
 
+  DeserializationError parseCodepoint(uint16_t &codepoint) {
+    codepoint = 0;
+    for (uint8_t i = 0; i < 4; ++i) {
+      char digit = current();
+      if (!digit) return DeserializationError::IncompleteInput;
+      uint8_t value = decodeHex(digit);
+      if (value > 0x0F) return DeserializationError::InvalidInput;
+      codepoint = uint16_t((codepoint << 4) | value);
+      move();
+    }
+    return DeserializationError::Ok;
+  }
+
   static inline bool isBetween(char c, char min, char max) {
     return min <= c && c <= max;
   }
@@ -269,6 +294,12 @@ class JsonDeserializer {
     return c == '\'' || c == '\"';
   }
 
+  static inline uint8_t decodeHex(char c) {
+    if (c < 'A') return uint8_t(c - '0');
+    c &= ~0x20;  // uppercase
+    return uint8_t(c - 'A' + 10);
+  }
+
   DeserializationError skipSpacesAndComments() {
     for (;;) {
       switch (current()) {

src/ArduinoJson/Json/Utf8.hpp

@@ -0,0 +1,26 @@
+// ArduinoJson - arduinojson.org
+// Copyright Benoit Blanchon 2014-2018
+// MIT License
+
+#pragma once
+
+namespace ARDUINOJSON_NAMESPACE {
+
+namespace Utf8 {
+template <typename TStringBuilder>
+inline void encodeCodepoint(uint16_t codepoint, TStringBuilder &str) {
+  if (codepoint < 0x80) {
+    str.append(char(codepoint));
+    return;
+  }
+
+  if (codepoint >= 0x00000800) {
+    str.append(char(0xe0 /*0b11100000*/ | (codepoint >> 12)));
+    str.append(char(((codepoint >> 6) & 0x3f /*0b00111111*/) | 0x80));
+  } else {
+    str.append(char(0xc0 /*0b11000000*/ | (codepoint >> 6)));
+  }
+  str.append(char((codepoint & 0x3f /*0b00111111*/) | 0x80));
+}
+}  // namespace Utf8
+}  // namespace ARDUINOJSON_NAMESPACE

src/ArduinoJson/Namespace.hpp

@@ -19,4 +19,5 @@
 #define ARDUINOJSON_NAMESPACE                                                  \
   ARDUINOJSON_CONCAT8(ArduinoJson, ARDUINOJSON_VERSION_MAJOR,                  \
                       ARDUINOJSON_VERSION_MINOR, ARDUINOJSON_VERSION_REVISION, \
-                      _, ARDUINOJSON_USE_LONG_LONG, _, ARDUINOJSON_USE_DOUBLE)
+                      _, ARDUINOJSON_USE_LONG_LONG, ARDUINOJSON_USE_DOUBLE,    \
+                      ARDUINOJSON_DECODE_UNICODE)