Merge pull request #2186 from schellap/bom

Strip BOM from JSON file streams

src/corehost/cli/deps_format.cpp

@@ -362,6 +362,11 @@ bool deps_json_t::load(bool portable, const pal::string_t& deps_path, const rid_
         return false;
     }
 
+    if (skip_utf8_bom(&file))
+    {
+        trace::verbose(_X("UTF-8 BOM skipped while reading [%s]"), deps_path.c_str());
+    }
+
     try
     {
         const auto json = json_value::parse(file);
@@ -376,8 +381,10 @@ bool deps_json_t::load(bool portable, const pal::string_t& deps_path, const rid_
 
         return (portable) ? load_portable(json, name, rid_fallback_graph) : load_standalone(json, name);
     }
-    catch (...)
+    catch (const web::json::json_exception& je)
     {
+        pal::string_t jes = pal::to_palstring(je.what());
+        trace::error(_X("A JSON parsing exception occurred in [%s]: %s"), deps_path.c_str(), jes.c_str());
         return false;
     }
 }

src/corehost/cli/fxr/fx_muxer.cpp

@@ -83,6 +83,11 @@ pal::string_t fx_muxer_t::resolve_cli_version(const pal::string_t& global_json)
         return retval;
     }
 
+    if (skip_utf8_bom(&file))
+    {
+        trace::verbose(_X("UTF-8 BOM skipped while reading [%s]"), global_json.c_str());
+    }
+
     try
     {
         const auto root = json_value::parse(file);
@@ -106,7 +111,7 @@ pal::string_t fx_muxer_t::resolve_cli_version(const pal::string_t& global_json)
     catch (const web::json::json_exception& je)
     {
         pal::string_t jes = pal::to_palstring(je.what());
-        trace::info(_X("A JSON parsing exception occurred: %s"), jes.c_str());
+        trace::error(_X("A JSON parsing exception occurred in [%s]: %s"), global_json.c_str(), jes.c_str());
     }
     trace::verbose(_X("CLI version is [%s] in global json file [%s]"), retval.c_str(), global_json.c_str());
     return retval;

src/corehost/cli/runtime_config.cpp

@@ -91,6 +91,11 @@ bool runtime_config_t::ensure_parsed()
         return false;
     }
 
+    if (skip_utf8_bom(&file))
+    {
+        trace::verbose(_X("UTF-8 BOM skipped while reading [%s]"), m_path.c_str());
+    }
+
     try
     {
         const auto root = json_value::parse(file);
@@ -104,7 +109,7 @@ bool runtime_config_t::ensure_parsed()
     catch (const web::json::json_exception& je)
     {
         pal::string_t jes = pal::to_palstring(je.what());
-        trace::info(_X("A JSON parsing exception occurred: %s"), jes.c_str());
+        trace::error(_X("A JSON parsing exception occurred in [%s]: %s"), m_path.c_str(), jes.c_str());
         return false;
     }
     return true;

src/corehost/common/pal.windows.cpp

@@ -259,8 +259,12 @@ void pal::readdir(const string_t& path, const string_t& pattern, std::vector<pal
     string_t search_string(path);
     append_path(&search_string, pattern.c_str());
 
-    WIN32_FIND_DATAW data;
-    auto handle = ::FindFirstFileW(search_string.c_str(), &data);
+    WIN32_FIND_DATAW data = { 0 };
+    auto handle = ::FindFirstFileExW(search_string.c_str(), FindExInfoStandard, &data, FindExSearchNameMatch, NULL, 0);
+    if (handle == INVALID_HANDLE_VALUE)
+    {
+        return;
+    }
     do
     {
         string_t filepath(data.cFileName);

src/corehost/common/utils.cpp

@@ -191,3 +191,30 @@ bool parse_known_args(
     return true;
 }
 
+// Try to match 0xEF 0xBB 0xBF byte sequence (no endianness here.)
+bool skip_utf8_bom(pal::ifstream_t* stream)
+{
+    if (stream->eof() || !stream->good())
+    {
+        return false;
+    }
+
+    int peeked = stream->peek();
+    if (peeked == EOF || ((peeked & 0xFF) != 0xEF))
+    {
+        return false;
+    }
+
+    unsigned char bytes[3];
+    stream->read((char*) bytes, 3);
+    if ((stream->gcount() < 3) ||
+            (bytes[1] != 0xBB) || 
+            (bytes[2] != 0xBF))
+    {
+        // Reset to 0 if returning false.
+        stream->seekg(0, stream->beg);
+        return false;
+    }
+
+    return true;
+}

src/corehost/common/utils.h

@@ -26,4 +26,5 @@ bool parse_known_args(
     const std::vector<pal::string_t>& known_opts,
     std::unordered_map<pal::string_t, pal::string_t>* opts,
     int* num_args);
+bool skip_utf8_bom(pal::ifstream_t* stream);
 #endif