Fix parsing of numbers in floating point format to integers

When parsing a string like "123.456" to an integer the ICU backend would
first parse it greedily to a floating point value and then cast/truncate
it to an integer.
Set the flag to only parse integers when parsing to an integral number.
Care must be taken not to set that when parsing e.g. a currency or date
to an integer where the truncation is intended.

Author: Flamefire

doc/changelog.txt

@@ -8,6 +8,8 @@
 /*!
 \page changelog Changelog
 
+- 1.88.0
+    - Fix parsing of numbers in floating point format to integers
 - 1.86.0
     - Make ICU implementation of `to_title` threadsafe
     - Add allocator support to `utf_to_utf`

src/boost/locale/icu/formatter.cpp

@@ -54,7 +54,9 @@ namespace boost { namespace locale { namespace impl_icu {
     public:
         typedef std::basic_string<CharType> string_type;
 
-        number_format(icu::NumberFormat& fmt, std::string codepage) : cvt_(codepage), icu_fmt_(fmt) {}
+        number_format(icu::NumberFormat& fmt, const std::string& codepage, bool isNumberOnly = false) :
+            cvt_(codepage), icu_fmt_(fmt), isNumberOnly_(isNumberOnly)
+        {}
 
         string_type format(double value, size_t& code_points) const override { return do_format(value, code_points); }
         string_type format(int64_t value, size_t& code_points) const override { return do_format(value, code_points); }
@@ -107,6 +109,9 @@ namespace boost { namespace locale { namespace impl_icu {
             icu::ParsePosition pp;
             icu::UnicodeString tmp = cvt_.icu(str.data(), str.data() + str.size());
 
+            // For the plain number parsing (no currency etc) parse "123.456" as 2 ints
+            // not a float later converted to int
+            icu_fmt_.setParseIntegerOnly(std::is_integral<ValueType>::value && isNumberOnly_);
             icu_fmt_.parse(tmp, val, pp);
 
             ValueType tmp_v;
@@ -122,6 +127,7 @@ namespace boost { namespace locale { namespace impl_icu {
 
         icu_std_converter<CharType> cvt_;
         icu::NumberFormat& icu_fmt_;
+        const bool isNumberOnly_;
     };
 
     template<typename CharType>
@@ -355,7 +361,7 @@ namespace boost { namespace locale { namespace impl_icu {
                 icu::NumberFormat& nf =
                   cache.number_format((how == std::ios_base::scientific) ? num_fmt_type::sci : num_fmt_type::number);
                 set_fraction_digits(nf, how, ios.precision());
-                return ptr_type(new number_format<CharType>(nf, encoding));
+                return ptr_type(new number_format<CharType>(nf, encoding, true));
             }
             case currency: {
                 icu::NumberFormat& nf = cache.number_format(

test/formatting_common.hpp

@@ -0,0 +1,82 @@
+//
+// Copyright (c) 2024 Alexander Grund
+//
+// Distributed under the Boost Software License, Version 1.0.
+// https://www.boost.org/LICENSE_1_0.txt
+
+#include <boost/locale/formatting.hpp>
+#include <boost/locale/generator.hpp>
+#include <cstdint>
+#include <limits>
+#include <sstream>
+
+#include "../src/boost/locale/util/foreach_char.hpp"
+#include "boostLocale/test/tools.hpp"
+#include "boostLocale/test/unit_test.hpp"
+
+template<typename CharType, typename IntType>
+void test_parse_multi_number_by_char(const std::locale& locale)
+{
+    std::basic_istringstream<CharType> stream;
+    stream.imbue(locale);
+    stream.str(ascii_to<CharType>("42.12,345"));
+    stream >> boost::locale::as::number;
+
+    IntType value;
+    TEST_REQUIRE(stream >> value);
+    TEST_EQ(value, IntType(42));
+    TEST_EQ(static_cast<char>(stream.get()), '.');
+    TEST_REQUIRE(stream >> value);
+    TEST_EQ(value, IntType(12345));
+    TEST_REQUIRE(!(stream >> value));
+    TEST(stream.eof());
+
+    stream.str(ascii_to<CharType>("42.25,678"));
+    stream.clear();
+    float fValue;
+    TEST_REQUIRE(stream >> fValue);
+    TEST_EQ(fValue, 42.25);
+    TEST_EQ(static_cast<char>(stream.get()), ',');
+    TEST_REQUIRE(stream >> value);
+    TEST_EQ(value, IntType(678));
+    TEST_REQUIRE(!(stream >> value));
+    TEST(stream.eof());
+
+    // Parsing a floating point currency to integer truncates the floating point value but fully parses it
+    stream.str(ascii_to<CharType>("USD1,234.55,67.89"));
+    stream.clear();
+    TEST_REQUIRE(!(stream >> value));
+    stream.clear();
+    stream >> boost::locale::as::currency >> boost::locale::as::currency_iso;
+    if(stream >> value) { // Parsing currencies not fully supported by WinAPI backend
+        TEST_EQ(value, IntType(1234));
+        TEST_EQ(static_cast<char>(stream.get()), ',');
+        TEST_REQUIRE(stream >> boost::locale::as::number >> value);
+        TEST_EQ(value, IntType(67));
+        TEST(!stream.eof());
+    }
+}
+
+/// Test that parsing multiple numbers without any spaces works as expected
+void test_parse_multi_number()
+{
+    const auto locale = boost::locale::generator{}("en_US.UTF-8");
+
+#define BOOST_LOCALE_CALL_I(T, I)      \
+    std::cout << "\t" #I << std::endl; \
+    test_parse_multi_number_by_char<T, I>(locale);
+
+#define BOOST_LOCALE_CALL(T)                                 \
+    std::cout << "test_parse_multi_number " #T << std::endl; \
+    BOOST_LOCALE_CALL_I(T, int16_t);                         \
+    BOOST_LOCALE_CALL_I(T, uint16_t);                        \
+    BOOST_LOCALE_CALL_I(T, int32_t);                         \
+    BOOST_LOCALE_CALL_I(T, uint32_t);                        \
+    BOOST_LOCALE_CALL_I(T, int64_t);                         \
+    BOOST_LOCALE_CALL_I(T, uint64_t);
+
+    BOOST_LOCALE_CALL(char);
+    BOOST_LOCALE_CALL(wchar_t);
+#undef BOOST_LOCALE_CALL
+#undef BOOST_LOCALE_CALL_I
+}

test/test_formatting.cpp

@@ -21,6 +21,7 @@
 
 #include "boostLocale/test/tools.hpp"
 #include "boostLocale/test/unit_test.hpp"
+#include "formatting_common.hpp"
 
 const std::string test_locale_name = "en_US";
 std::string message_path = "./";
@@ -928,6 +929,8 @@ void test_main(int argc, char** argv)
     test_manip<char32_t>();
     test_format_class<char32_t>();
 #endif
+
+    test_parse_multi_number();
 }
 
 // boostinspect:noascii

test/test_posix_formatting.cpp

@@ -19,6 +19,7 @@
 #endif
 #include "boostLocale/test/tools.hpp"
 #include "boostLocale/test/unit_test.hpp"
+#include "formatting_common.hpp"
 
 #ifdef BOOST_LOCALE_NO_POSIX_BACKEND
 // Dummy just to make it compile
@@ -185,6 +186,7 @@ void test_main(int /*argc*/, char** /*argv*/)
             TEST(v == "12345,45" || v == "12 345,45" || v == "12.345,45");
         }
     }
+    test_parse_multi_number();
 }
 
 // boostinspect:noascii

test/test_std_formatting.cpp

@@ -14,6 +14,7 @@
 
 #include "boostLocale/test/tools.hpp"
 #include "boostLocale/test/unit_test.hpp"
+#include "formatting_common.hpp"
 
 template<typename CharType, typename RefCharType>
 void test_by_char(const std::locale& l, const std::locale& lreal)
@@ -230,6 +231,10 @@ void test_main(int /*argc*/, char** /*argv*/)
             }
         }
     }
+    // Std backend silently falls back to the C locale when the locale is not supported
+    // which breaks the test assumptions
+    if(has_std_locale("en_US.UTF-8"))
+        test_parse_multi_number();
 }
 
 // boostinspect:noascii

test/test_winapi_formatting.cpp

@@ -23,6 +23,7 @@
 #include "../src/boost/locale/win32/lcid.hpp"
 #include "boostLocale/test/tools.hpp"
 #include "boostLocale/test/unit_test.hpp"
+#include "formatting_common.hpp"
 
 template<typename CharType>
 void test_by_char(const std::locale& l, std::string name, int lcid)
@@ -176,6 +177,7 @@ void test_main(int /*argc*/, char** /*argv*/)
             test_by_char<wchar_t>(l, name, name_lcid.second);
         }
     }
+    test_parse_multi_number();
     std::cout << "- Testing strftime" << std::endl;
     test_date_time(gen("en_US.UTF-8"));
 }