From 8f5d0098d6dd18b4967372f9aeac676d2dbbf063 Mon Sep 17 00:00:00 2001
From: zjeffer <4633209+zjeffer@users.noreply.github.com>
Date: Sun, 10 Sep 2023 20:36:27 +0200
Subject: [PATCH] Fixed json parsing with hexadecimal characters

* replace \x with \u00 to follow JSON spec
* fixes #2475 and #2495
* added unit tests for json parsing
---
 include/util/json.hpp | 39 +++++++++++++++++++++++--------------
 test/JsonParser.cpp   | 45 +++++++++++++++++++++++++++++++++++++++++++
 test/meson.build      |  1 +
 3 files changed, 71 insertions(+), 14 deletions(-)
 create mode 100644 test/JsonParser.cpp

diff --git a/include/util/json.hpp b/include/util/json.hpp
index 7cd43552..f0736f9b 100644
--- a/include/util/json.hpp
+++ b/include/util/json.hpp
@@ -3,6 +3,12 @@
 #include <fmt/ostream.h>
 #include <json/json.h>
 
+#include <algorithm>
+#include <codecvt>
+#include <iostream>
+#include <locale>
+#include <regex>
+
 #if (FMT_VERSION >= 90000)
 
 template <>
@@ -12,25 +18,30 @@ struct fmt::formatter<Json::Value> : ostream_formatter {};
 
 namespace waybar::util {
 
-struct JsonParser {
-  JsonParser() {}
+class JsonParser {
+ public:
+  JsonParser() = default;
 
-  const Json::Value parse(const std::string& data) const {
-    Json::Value root(Json::objectValue);
-    if (data.empty()) {
-      return root;
+  Json::Value parse(const std::string& jsonStr) {
+    Json::Value root;
+
+    // replace all occurrences of "\x" with "\u00", because JSON doesn't allow "\x" escape sequences
+    std::string modifiedJsonStr = replaceHexadecimalEscape(jsonStr);
+
+    std::istringstream jsonStream(modifiedJsonStr);
+    std::string errs;
+    if (!Json::parseFromStream(m_readerBuilder, jsonStream, &root, &errs)) {
+      throw std::runtime_error("Error parsing JSON: " + errs);
     }
-    std::unique_ptr<Json::CharReader> const reader(builder_.newCharReader());
-    std::string err;
-    bool res = reader->parse(data.c_str(), data.c_str() + data.size(), &root, &err);
-    if (!res) throw std::runtime_error(err);
     return root;
   }
 
-  ~JsonParser() = default;
-
  private:
-  Json::CharReaderBuilder builder_;
-};
+  Json::CharReaderBuilder m_readerBuilder;
 
+  static std::string replaceHexadecimalEscape(const std::string& str) {
+    static std::regex re("\\\\x");
+    return std::regex_replace(str, re, "\\u00");
+  }
+};
 }  // namespace waybar::util
diff --git a/test/JsonParser.cpp b/test/JsonParser.cpp
new file mode 100644
index 00000000..99a8649e
--- /dev/null
+++ b/test/JsonParser.cpp
@@ -0,0 +1,45 @@
+#include "util/json.hpp"
+
+#if __has_include(<catch2/catch_test_macros.hpp>)
+#include <catch2/catch_test_macros.hpp>
+#else
+#include <catch2/catch.hpp>
+#endif
+
+TEST_CASE("Simple json", "[json]") {
+  SECTION("Parse simple json") {
+    std::string stringToTest = R"({"number": 5, "string": "test"})";
+    waybar::util::JsonParser parser;
+    Json::Value jsonValue = parser.parse(stringToTest);
+    REQUIRE(jsonValue["number"].asInt() == 5);
+    REQUIRE(jsonValue["string"].asString() == "test");
+  }
+}
+
+TEST_CASE("Json with unicode", "[json]") {
+  SECTION("Parse json with unicode") {
+    std::string stringToTest = R"({"test": "\xab"})";
+    waybar::util::JsonParser parser;
+    Json::Value jsonValue = parser.parse(stringToTest);
+    // compare with "\u00ab" because "\xab" is replaced with "\u00ab" in the parser
+    REQUIRE(jsonValue["test"].asString() == "\u00ab");
+  }
+}
+
+TEST_CASE("Json with emoji", "[json]") {
+  SECTION("Parse json with emoji") {
+    std::string stringToTest = R"({"test": "😊"})";
+    waybar::util::JsonParser parser;
+    Json::Value jsonValue = parser.parse(stringToTest);
+    REQUIRE(jsonValue["test"].asString() == "😊");
+  }
+}
+
+TEST_CASE("Json with chinese characters", "[json]") {
+  SECTION("Parse json with chinese characters") {
+    std::string stringToTest = R"({"test": "你好"})";
+    waybar::util::JsonParser parser;
+    Json::Value jsonValue = parser.parse(stringToTest);
+    REQUIRE(jsonValue["test"].asString() == "你好");
+  }
+}
\ No newline at end of file
diff --git a/test/meson.build b/test/meson.build
index 02cbb2a4..7a55a473 100644
--- a/test/meson.build
+++ b/test/meson.build
@@ -8,6 +8,7 @@ test_dep = [
 ]
 test_src = files(
     'main.cpp',
+    'JsonParser.cpp',
     'SafeSignal.cpp',
     'config.cpp',
     '../src/config.cpp',