Remove fstream usage from corehost (#111859)

Contributes to #111665

There's still fstream usage in corerun, so this alone doesn't resolve the Android build errors.

---------

Co-authored-by: Elinor Fung <elfung@microsoft.com>

Author: filipnavara

src/native/corehost/hostmisc/pal.h

@@ -6,7 +6,6 @@
 
 #include <string>
 #include <vector>
-#include <fstream>
 #include <sstream>
 #include <iostream>
 #include <cstring>
@@ -118,14 +117,6 @@ namespace pal
     typedef wchar_t char_t;
     typedef std::wstring string_t;
     typedef std::wstringstream stringstream_t;
-    // TODO: Agree on the correct encoding of the files: The PoR for now is to
-    // temporarily wchar for Windows and char for Unix. Current implementation
-    // implicitly expects the contents on both Windows and Unix as char and
-    // converts them to wchar in code for Windows. This line should become:
-    // typedef std::basic_ifstream<char_t> ifstream_t.
-    typedef std::basic_ifstream<char> ifstream_t;
-    typedef std::istreambuf_iterator<ifstream_t::char_type> istreambuf_iterator_t;
-    typedef std::basic_istream<char> istream_t;
     typedef HRESULT hresult_t;
     typedef HMODULE dll_t;
     typedef FARPROC proc_t;
@@ -207,9 +198,6 @@ namespace pal
     typedef char char_t;
     typedef std::string string_t;
     typedef std::stringstream stringstream_t;
-    typedef std::basic_ifstream<char> ifstream_t;
-    typedef std::istreambuf_iterator<ifstream_t::char_type> istreambuf_iterator_t;
-    typedef std::basic_istream<char> istream_t;
     typedef int hresult_t;
     typedef void* dll_t;
     typedef void* proc_t;

src/native/corehost/hostmisc/pal.unix.cpp

@@ -456,7 +456,7 @@ bool get_install_location_from_file(const pal::string_t& file_path, bool& file_f
 {
     file_found = true;
     bool install_location_found = false;
-    FILE* install_location_file = pal::file_open(file_path, "r");
+    FILE* install_location_file = pal::file_open(file_path, _X("r"));
     if (install_location_file != nullptr)
     {
         if (!get_line_from_file(install_location_file, install_location))
@@ -819,12 +819,10 @@ pal::string_t pal::get_current_os_rid_platform()
     {
         // Read the file to get ID and VERSION_ID data that will be used
         // to construct the RID.
-        std::fstream fsVersionFile;
-
-        fsVersionFile.open(versionFile, std::fstream::in);
+        FILE* fsVersionFile = pal::file_open(versionFile, _X("r"));
 
         // Proceed only if we were able to open the file
-        if (fsVersionFile.good())
+        if (fsVersionFile != nullptr)
         {
             pal::string_t line;
             pal::string_t strID(_X("ID="));
@@ -834,11 +832,8 @@ pal::string_t pal::get_current_os_rid_platform()
 
             bool fFoundID = false, fFoundVersion = false;
 
-            // Read the first line
-            std::getline(fsVersionFile, line);
-
             // Loop until we are at the end of file
-            while (!fsVersionFile.eof())
+            while (get_line_from_file(fsVersionFile, line))
             {
                 // Look for ID if we have not found it already
                 if (!fFoundID)
@@ -872,13 +867,10 @@ pal::string_t pal::get_current_os_rid_platform()
                     // We have everything we need to form the RID - break out of the loop.
                     break;
                 }
-
-                // Read the next line
-                std::getline(fsVersionFile, line);
             }
 
             // Close the file now that we are done with it.
-            fsVersionFile.close();
+            fclose(fsVersionFile);
 
             if (fFoundID)
             {

src/native/corehost/json_parser.cpp

@@ -17,33 +17,9 @@
 
 namespace {
 
-// Try to match 0xEF 0xBB 0xBF byte sequence (no endianness here.)
-std::streampos get_utf8_bom_length(pal::istream_t& stream)
-{
-    if (stream.eof())
-    {
-        return 0;
-    }
-
-    auto peeked = stream.peek();
-    if (peeked == EOF || ((peeked & 0xFF) != 0xEF))
-    {
-        return 0;
-    }
-
-    unsigned char bytes[3];
-    stream.read(reinterpret_cast<char*>(bytes), 3);
-    if ((stream.gcount() < 3) || (bytes[1] != 0xBB) || (bytes[2] != 0xBF))
-    {
-        return 0;
-    }
-
-    return 3;
-}
-
 void get_line_column_from_offset(const char* data, uint64_t size, size_t offset, int *line, int *column)
 {
-    assert(offset < size);
+    assert(offset <= size);
 
     *line = *column = 1;
 
@@ -68,12 +44,6 @@ void get_line_column_from_offset(const char* data, uint64_t size, size_t offset,
 
 } // empty namespace
 
-void json_parser_t::realloc_buffer(size_t size)
-{
-    m_json.resize(size + 1);
-    m_json[size] = '\0';
-}
-
 bool json_parser_t::parse_raw_data(char* data, int64_t size, const pal::string_t& context)
 {
     assert(data != nullptr);
@@ -115,51 +85,64 @@ bool json_parser_t::parse_file(const pal::string_t& path)
 {
     // This code assumes that the caller has checked that the file `path` exists
     // either within the bundle, or as a real file on disk.
-    assert(m_bundle_data == nullptr);
+    assert(m_data == nullptr);
     assert(m_bundle_location == nullptr);
 
     if (bundle::info_t::is_single_file_bundle())
     {
         // Due to in-situ parsing on Linux,
         //  * The json file is mapped as copy-on-write.
         //  * The mapping cannot be immediately released, and will be unmapped by the json_parser destructor.
-        m_bundle_data = bundle::info_t::config_t::map(path, m_bundle_location);
+        m_data = bundle::info_t::config_t::map(path, m_bundle_location);
 
-        if (m_bundle_data != nullptr)
+        if (m_data != nullptr)
         {
-            bool result = parse_raw_data(m_bundle_data, m_bundle_location->size, path);
-            return result;
+            m_size = (size_t)m_bundle_location->size;
         }
     }
 
-    pal::ifstream_t file{ path };
-    if (!file.good())
+    if (m_data == nullptr)
     {
-        trace::error(_X("Cannot use file stream for [%s]: %s"), path.c_str(), pal::strerror(errno).c_str());
-        return false;
-    }
+#ifdef _WIN32
+        // We can't use in-situ parsing on Windows, as JSON data is encoded in
+        // UTF-8 and the host expects wide strings.
+        // We do not need copy-on-write, so read-only mapping will be enough.
+        m_data = (char*)pal::mmap_read(path, &m_size);
+#else // _WIN32
+        m_data = (char*)pal::mmap_copy_on_write(path, &m_size);
+#endif // _WIN32
 
-    auto current_pos = ::get_utf8_bom_length(file);
-    file.seekg(0, file.end);
-    auto stream_size = file.tellg();
-    if (stream_size == -1)
-    {
-        trace::error(_X("Failed to get size of file [%s]"), path.c_str());
-        return false;
+        if (m_data == nullptr)
+        {
+            trace::error(_X("Cannot use file stream for [%s]: %s"), path.c_str(), pal::strerror(errno).c_str());
+            return false;
+        }
     }
 
-    file.seekg(current_pos, file.beg);
+    char *data = m_data;
+    size_t size = m_size;
 
-    realloc_buffer(static_cast<size_t>(stream_size - current_pos));
-    file.read(m_json.data(), stream_size - current_pos);
+    // Skip over UTF-8 BOM, if present
+    if (size >= 3 && data[0] == 0xEF && data[1] == 0xBB && data[1] == 0xBF)
+    {
+        size -= 3;
+        data += 3;
+    }
 
-    return parse_raw_data(m_json.data(), m_json.size(), path);
+    return parse_raw_data(data, size, path);
 }
 
 json_parser_t::~json_parser_t()
 {
-    if (m_bundle_data != nullptr)
+    if (m_data != nullptr)
     {
-        bundle::info_t::config_t::unmap(m_bundle_data, m_bundle_location);
+        if (m_bundle_location != nullptr)
+        {
+            bundle::info_t::config_t::unmap(m_data, m_bundle_location);
+        }
+        else
+        {
+            pal::munmap((void*)m_data, m_size);
+        }
     }
 }

src/native/corehost/json_parser.h

@@ -40,24 +40,22 @@ class json_parser_t {
         bool parse_file(const pal::string_t& path);
 
         json_parser_t()
-            : m_bundle_data(nullptr)
+            : m_data(nullptr)
             , m_bundle_location(nullptr) {}
 
         ~json_parser_t();
 
     private:
-        // This is a vector of char and not pal::char_t because JSON data
-        // parsed by this class is always encoded in UTF-8.  On Windows,
-        // where wide strings are used, m_json is kept in UTF-8, but converted
+        char* m_data; // The memory mapped bytes of the file
+        size_t m_size; // Size of the mapped memory
+
+        // On Windows, where wide strings are used, m_data is kept in UTF-8, but converted
         // to UTF-16 by m_document during load.
-        std::vector<char> m_json;
         document_t m_document;
 
-        // If a json file is parsed from a single-file bundle, the following two fields represent:
-        char* m_bundle_data; // The memory mapped bytes of the application bundle.
-        const bundle::location_t* m_bundle_location; // Location of this json file within the bundle.
-
-        void realloc_buffer(size_t size);
+        // If a json file is parsed from a single-file bundle, the following fields represents
+        // the location of this json file within the bundle.
+        const bundle::location_t* m_bundle_location;
 };
 
 #endif // __JSON_PARSER_H__

src/native/corehost/test/nativehost/host_context_test.cpp

@@ -2,6 +2,7 @@
 // The .NET Foundation licenses this file to you under the MIT license.
 
 #include <iostream>
+#include <fstream>
 #include <pal.h>
 #include <error_codes.h>
 #include <future>