util: Replace use of locale dependent atoi(…) with locale-independent…

… std::from_chars(…) (C++17)

src/test/script_tests.cpp

@@ -59,7 +59,7 @@ CScript ParseScript(string s)
  (starts_with(w, "-") && all(string(w.begin()+1, w.end()), ::IsDigit)))
  {
  // Number
- int64_t n = atoi64(w);
+ int64_t n = LocaleIndependentAtoi<int64_t>(w);
  result << n;
  }
  else if (starts_with(w, "0x") && IsHex(string(w.begin()+2, w.end())))

src/test/util_tests.cpp

@@ -544,6 +544,77 @@ BOOST_AUTO_TEST_CASE(test_IsDigit)
  BOOST_CHECK_EQUAL(IsDigit(9), false);
 }
 
+BOOST_AUTO_TEST_CASE(test_LocaleIndependentAtoi)
+{
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1234"), 1'234);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("0"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("01234"), 1'234);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1234"), -1'234);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" 1"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1 "), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1a"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1.1"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("1.9"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+01.9"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1"), -1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" -1"), -1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-1 "), -1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" -1 "), -1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+1"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" +1"), 1);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(" +1 "), 1);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("+-1"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-+1"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("++1"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("--1"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>(""), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("aap"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("0x1"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-32482348723847471234"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("32482348723847471234"), 0);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("-9223372036854775809"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("-9223372036854775808"), -9'223'372'036'854'775'807LL - 1LL);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("9223372036854775807"), 9'223'372'036'854'775'807);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int64_t>("9223372036854775808"), 0);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("-1"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("0"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("18446744073709551615"), 18'446'744'073'709'551'615ULL);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint64_t>("18446744073709551616"), 0U);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-2147483649"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("-2147483648"), -2'147'483'648LL);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("2147483647"), 2'147'483'647);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int32_t>("2147483648"), 0);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("-1"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("0"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("4294967295"), 4'294'967'295U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint32_t>("4294967296"), 0U);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("-32769"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("-32768"), -32'768);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("32767"), 32'767);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int16_t>("32768"), 0);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("-1"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("0"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("65535"), 65'535U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint16_t>("65536"), 0U);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("-129"), 0);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("-128"), -128);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("127"), 127);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<int8_t>("128"), 0);
+
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("-1"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("0"), 0U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("255"), 255U);
+ BOOST_CHECK_EQUAL(LocaleIndependentAtoi<uint8_t>("256"), 0U);
+}
+
 BOOST_AUTO_TEST_CASE(test_ParseInt32)
 {
  int32_t n;

src/util/strencodings.cpp

@@ -440,20 +440,6 @@ std::string FormatParagraph(const std::string& in, size_t width, size_t indent)
  return out.str();
 }
 
-int64_t atoi64(const std::string& str)
-{
-#ifdef _MSC_VER
- return _atoi64(str.c_str());
-#else
- return strtoll(str.c_str(), nullptr, 10);
-#endif
-}
-
-int atoi(const std::string& str)
-{
- return atoi(str.c_str());
-}
-
 /** Upper bound for mantissa.
  * 10^18-1 is the largest arbitrary decimal that will fit in a signed 64-bit integer.
  * Larger integers cannot consist of arbitrary combinations of 0-9:

src/util/strencodings.h

@@ -11,8 +11,10 @@
 
 #include <attributes.h>
 #include <span.h>
+#include <util/string.h>
 
 #include <cassert>
+#include <charconv>
 #include <cstdint>
 #include <iterator>
 #include <string>
@@ -55,8 +57,33 @@ std::string EncodeBase32(const unsigned char* pch, size_t len);
 std::string EncodeBase32(const std::string& str);
 
 void SplitHostPort(std::string in, int &portOut, std::string &hostOut);
-int64_t atoi64(const std::string& str);
-int atoi(const std::string& str);
+
+// LocaleIndependentAtoi is provided for backwards compatibility reasons.
+//
+// New code should use the ParseInt64/ParseUInt64/ParseInt32/ParseUInt32 functions
+// which provide parse error feedback.
+//
+// The goal of LocaleIndependentAtoi is to replicate the exact defined behaviour
+// of atoi and atoi64 as they behave under the "C" locale.
+template <typename T>
+T LocaleIndependentAtoi(const std::string& str)
+{
+ static_assert(std::is_integral<T>::value);
+ T result;
+ // Emulate atoi(...) handling of white space and leading +/-.
+ std::string s = TrimString(str);
+ if (!s.empty() && s[0] == '+') {
+ if (s.length() >= 2 && s[1] == '-') {
+ return 0;
+ }
+ s = s.substr(1);
+ }
+ auto [_, error_condition] = std::from_chars(s.data(), s.data() + s.size(), result);
+ if (error_condition != std::errc{}) {
+ return 0;
+ }
+ return result;
+}
 
 /**
  * Tests if the given character is a decimal digit.

src/util/system.cpp

@@ -54,16 +54,14 @@ ArgsManager gArgs;
 /**
  * Interpret a string argument as a boolean.
  *
- * The definition of atoi() requires that non-numeric string values like "foo",
- * return 0. This means that if a user unintentionally supplies a non-integer
- * argument here, the return value is always false. This means that -foo=false
- * does what the user probably expects, but -foo=true is well defined but does
- * not do what they probably expected.
+ * The definition of LocaleIndependentAtoi<int>() requires that non-numeric string values
+ * like "foo", return 0. This means that if a user unintentionally supplies a
+ * non-integer argument here, the return value is always false. This means that
+ * -foo=false does what the user probably expects, but -foo=true is well defined
+ * but does not do what they probably expected.
  *
- * The return value of atoi() is undefined when given input not representable as
- * an int. On most systems this means string value between "-2147483648" and
- * "2147483647" are well defined (this method will return true). Setting
- * -txindex=2147483648 on most systems, however, is probably undefined.
+ * The return value of LocaleIndependentAtoi<int>(...) is zero when given input not
+ * representable as an int.
  *
  * For a more extensive discussion of this topic (and a wide range of opinions
  * on the Right Way to change this code), see PR12713.

test/lint/lint-locale-dependence.sh

@@ -5,10 +5,8 @@
 
 export LC_ALL=C
 KNOWN_VIOLATIONS=(
- "src/util/strencodings.cpp:.*atoi"
  "src/util/strencodings.cpp:.*strtol"
  "src/util/strencodings.cpp:.*strtoul"
- "src/util/strencodings.h:.*atoi"
  "src/logging.h:.*strftime"
  "src/gridcoin/backup.cpp:.*strftime"
  "src/rpc/protocol.cpp:.*strftime"