refactor operator>>(istream&, value&)

Author: grisumbras

include/boost/json/impl/value.ipp

@@ -352,22 +352,30 @@ operator>>(
 {
     using Traits = std::istream::traits_type;
 
+    // sentry prepares the stream for reading and finalizes it in destructor
     std::istream::sentry sentry(is);
     if( !sentry )
         return is;
 
-    unsigned char parser_buf[BOOST_JSON_STACK_BUFFER_SIZE];
+    unsigned char parser_buf[BOOST_JSON_STACK_BUFFER_SIZE / 2];
     stream_parser p({}, {}, parser_buf);
     p.reset( jv.storage() );
 
+    char read_buf[BOOST_JSON_STACK_BUFFER_SIZE / 2];
+    std::streambuf& buf = *is.rdbuf();
     std::ios::iostate err = std::ios::goodbit;
     try
     {
-        std::istream::int_type c = is.rdbuf()->sgetc();
         while( true )
         {
             error_code ec;
 
+            // we peek the buffer; this either makes sure that there's no
+            // more input, or makes sure there's something in the internal
+            // buffer (so in_avail will return a positive number)
+            std::istream::int_type c = is.rdbuf()->sgetc();
+            // if we indeed reached EOF, we check if we parsed a full JSON
+            // document; if not, we error out
             if( Traits::eq_int_type(c, Traits::eof()) )
             {
                 err |= std::ios::eofbit;
@@ -376,29 +384,56 @@ operator>>(
                     break;
             }
 
+            // regardless of reaching EOF, we might have parsed a full JSON
+            // document; if so, we successfully finish
             if( p.done() )
             {
                 jv = p.release();
                 return is;
             }
 
-            char read_buf[1];
-            read_buf[0] = Traits::to_char_type(c);
-            c = is.rdbuf()->snextc();
+            // at this point we definitely have more input, specifically in
+            // buf's internal buffer; we also definitely haven't parsed a whole
+            // document
+            std::streamsize available = buf.in_avail();
+            // if this assert fails, the streambuf is buggy
+            BOOST_ASSERT( available > 0 );
+
+            available = std::min(
+                static_cast<std::size_t>(available), sizeof(read_buf) );
+            // we read from the internal buffer of buf into our buffer
+            available = buf.sgetn( read_buf, available );
+
+            std::size_t consumed = p.write_some( read_buf, available, ec );
+            // if the parser hasn't consumed the entire input we've took from
+            // buf, we put the remaining data back; this should succeed,
+            // because we only read data from buf's internal buffer
+            while( consumed++ < static_cast<std::size_t>(available) )
+            {
+                std::istream::int_type const status = buf.sungetc();
+                BOOST_ASSERT( status != Traits::eof() );
+                (void)status;
+            }
 
-            p.write_some(read_buf, 1, ec);
             if( ec.failed() )
                 break;
         }
     }
     catch(...)
     {
-        is.setstate(std::ios::failbit);
-        throw;
+        try
+        {
+            is.setstate(std::ios::badbit);
+        }
+        // we ignore the exception, because we need to throw the original
+        // exception instead
+        catch( std::ios::failure const& ) { }
+
+        if( is.exceptions() & std::ios::badbit )
+            throw;
     }
 
-    err |= std::ios::failbit;
-    is.setstate(err);
+    is.setstate(err | std::ios::failbit);
     return is;
 }
 

include/boost/json/value.hpp

@@ -3343,12 +3343,16 @@ class value
 
         This function parses JSON from an input stream into a `value`. If
         parsing fails, `std::ios_base::failbit` will be set for `is` and
-        `jv` will be left unchanged.<br>
+        `jv` will be left unchanged. Regardless of whether `skipws` flag is set
+        on `is`, consumes whitespace before and after JSON, because whitespace
+        is considered a part of JSON. Behaves as [_FormattedInputFunction_]
+        (https://en.cppreference.com/w/cpp/named_req/FormattedInputFunction).<br>
 
         Note: this operator cannot assume that the stream only contains a
-        single JSON document, which results in **very underwhelming
-        performance**. If you know that your input consists of a single
-        JSON document, consider using @ref parse function instead.
+        single JSON document, which may result in **very underwhelming
+        performance**, if the stream isn't cooperative. If you know that your
+        input consists of a single JSON document, consider using @ref parse
+        function instead.
 
         @return Reference to `is`.
 
@@ -3358,7 +3362,7 @@ class value
         @par Exception Safety
         Basic guarantee.
         Calls to `memory_resource::allocate` may throw.
-        The stream may throw as described by
+        The stream may throw as configured by
         [`std::ios::exceptions`](https://en.cppreference.com/w/cpp/io/basic_ios/exceptions).
 
         @param is The input stream to parse from.

test/value.cpp

@@ -2261,7 +2261,7 @@ class value_test
             R"({ "x": 1
                , "y": 2
                , "z": [77, null, true, "qwerty uiop"]
-               } 12)");
+               }_12)");
         value jv;
 
         ss >> jv;
@@ -2275,7 +2275,7 @@ class value_test
         // check we didn't consume any extra characters
         std::string s;
         std::getline(ss, s);
-        BOOST_TEST( s == " 12" );
+        BOOST_TEST( s == "_12" );
 
         ss.clear();
         ss.str("23");
@@ -2285,22 +2285,35 @@ class value_test
         ss.clear();
         ss.str("");
         ss >> jv;
+        BOOST_TEST( jv == 23 );
         BOOST_TEST( ss.rdstate() == (std::ios::failbit | std::ios::eofbit) );
 
         ss.clear();
         ss.str("nu");
         ss >> jv;
+        BOOST_TEST( jv == 23 );
         BOOST_TEST( ss.rdstate() == (std::ios::failbit | std::ios::eofbit) );
 
         ss.clear();
         ss.str("[1,2,3,4,]");
         ss >> jv;
+        BOOST_TEST( jv == 23 );
         BOOST_TEST( ss.rdstate() == std::ios::failbit );
 
         {
             throwing_buffer buf;
             std::istream is(&buf);
-            BOOST_TEST_THROWS( is >> jv, std::exception );
+            is >> jv;
+            BOOST_TEST( jv == 23 );
+            BOOST_TEST( is.rdstate() & std::ios::badbit );
+        }
+        {
+            throwing_buffer buf;
+            std::istream is(&buf);
+            is.exceptions(std::ios::badbit);
+            BOOST_TEST_THROWS( is >> jv, std::invalid_argument );
+            BOOST_TEST( jv == 23 );
+            BOOST_TEST( is.rdstate() & std::ios::badbit );
         }
     }