🔨 make lexer distinguishes number types

parent 9f5dccba
This diff is collapsed.
...@@ -9444,7 +9444,9 @@ class basic_json ...@@ -9444,7 +9444,9 @@ class basic_json
literal_false, ///< the `false` literal literal_false, ///< the `false` literal
literal_null, ///< the `null` literal literal_null, ///< the `null` literal
value_string, ///< a string -- use get_string() for actual value value_string, ///< a string -- use get_string() for actual value
value_number, ///< a number -- use get_number() for actual value value_unsigned_integer, ///< an unsigned integer -- use get_number() for actual value
value_signed_integer, ///< a signed integer -- use get_number() for actual value
value_float, ///< an floating point number -- use get_number() for actual value
begin_array, ///< the character for array begin `[` begin_array, ///< the character for array begin `[`
begin_object, ///< the character for object begin `{` begin_object, ///< the character for object begin `{`
end_array, ///< the character for array end `]` end_array, ///< the character for array end `]`
...@@ -9596,7 +9598,9 @@ class basic_json ...@@ -9596,7 +9598,9 @@ class basic_json
return "null literal"; return "null literal";
case token_type::value_string: case token_type::value_string:
return "string literal"; return "string literal";
case token_type::value_number: case lexer::token_type::value_unsigned_integer:
case lexer::token_type::value_signed_integer:
case lexer::token_type::value_float:
return "number literal"; return "number literal";
case token_type::begin_array: case token_type::begin_array:
return "'['"; return "'['";
...@@ -9684,18 +9688,22 @@ class basic_json ...@@ -9684,18 +9688,22 @@ class basic_json
"false" { last_token_type = token_type::literal_false; break; } "false" { last_token_type = token_type::literal_false; break; }
// number // number
decimal_point = "."; decimal_point = ".";
digit = [0-9]; digit = [0-9];
digit_1_9 = [1-9]; digit_1_9 = [1-9];
e = "e" | "E"; e = "e" | "E";
minus = "-"; minus = "-";
plus = "+"; plus = "+";
zero = "0"; zero = "0";
exp = e (minus | plus)? digit+; exp = e (minus | plus)? digit+;
frac = decimal_point digit+; frac = decimal_point digit+;
int = (zero | digit_1_9 digit*); int = (zero | digit_1_9 digit*);
number = minus? int frac? exp?; number_unsigned = int;
number { last_token_type = token_type::value_number; break; } number_unsigned { last_token_type = token_type::value_unsigned_integer; break; }
number_signed = minus int;
number_signed { last_token_type = token_type::value_signed_integer; break; }
number_float = minus? int frac? exp?;
number_float { last_token_type = token_type::value_float; break; }
// string // string
quotation_mark = "\""; quotation_mark = "\"";
...@@ -10017,42 +10025,6 @@ class basic_json ...@@ -10017,42 +10025,6 @@ class basic_json
return parse(val, std::is_integral<T>()); return parse(val, std::is_integral<T>());
} }
/*!
This is a helper to determine whether to parse the token into
floating-point or integral type.
@note We wouldn't need it if we had separate token types for
integral and floating-point cases.
@return true iff token matches `^[+-]\d+$`
*/
bool is_integral() const
{
const char* p = m_start;
if (p == nullptr)
{
return false; // LCOV_EXCL_LINE
}
if ((*p == '-') or (*p == '+'))
{
++p;
}
if (p == m_end)
{
return false; // LCOV_EXCL_LINE
}
while ((p < m_end) and (*p >= '0') and (*p <= '9'))
{
++p;
}
return (p == m_end);
}
private: private:
const char* const m_start = nullptr; const char* const m_start = nullptr;
const char* const m_end = nullptr; const char* const m_end = nullptr;
...@@ -10161,7 +10133,7 @@ class basic_json ...@@ -10161,7 +10133,7 @@ class basic_json
return (x == static_cast<decltype(x)>(value)) // x fits into destination T return (x == static_cast<decltype(x)>(value)) // x fits into destination T
and (x < 0) == (value < 0) // preserved sign and (x < 0) == (value < 0) // preserved sign
and ((x != 0) or is_integral()) // strto[u]ll did nto fail //and ((x != 0) or is_integral()) // strto[u]ll did nto fail
and (errno == 0) // strto[u]ll did not overflow and (errno == 0) // strto[u]ll did not overflow
and (m_start < m_end) // token was not empty and (m_start < m_end) // token was not empty
and (endptr == m_end); // parsed entire token exactly and (endptr == m_end); // parsed entire token exactly
...@@ -10185,46 +10157,56 @@ class basic_json ...@@ -10185,46 +10157,56 @@ class basic_json
interpreted as a number interpreted as a number
@param[out] result @ref basic_json object to receive the number. @param[out] result @ref basic_json object to receive the number.
@param[in] token the type of the number token
*/ */
void get_number(basic_json& result) const void get_number(basic_json& result, const token_type token) const
{ {
assert(m_start != nullptr); assert(m_start != nullptr);
assert(m_start < m_cursor); assert(m_start < m_cursor);
assert((token == token_type::value_unsigned_integer) or
(token == token_type::value_signed_integer) or
(token == token_type::value_float));
strtonum num(reinterpret_cast<const char*>(m_start), strtonum num(reinterpret_cast<const char*>(m_start),
reinterpret_cast<const char*>(m_cursor)); reinterpret_cast<const char*>(m_cursor));
const bool is_negative = (*m_start == '-'); switch (token)
result.m_type = value_t::discarded;
if (not num.is_integral())
{ {
// will parse as float below case lexer::token_type::value_unsigned_integer:
}
else if (is_negative)
{
number_integer_t val{0};
if (num.to(val))
{ {
result.m_type = value_t::number_integer; number_unsigned_t val{0};
result.m_value = val; if (num.to(val))
{
result.m_type = value_t::number_unsigned;
result.m_value = val;
return;
}
break;
} }
}
else case lexer::token_type::value_signed_integer:
{
number_unsigned_t val{0};
if (num.to(val))
{ {
result.m_type = value_t::number_unsigned; number_integer_t val{0};
result.m_value = val; if (num.to(val))
{
result.m_type = value_t::number_integer;
result.m_value = val;
return;
}
break;
}
default:
{
break;
} }
} }
number_float_t val{0}; number_float_t val{0};
if (result.m_type != value_t::discarded or (not num.to(val))) if (not num.to(val))
{ {
// already have a value from above or couldn't parse as float_t // couldn't parse as float_t
result.m_type = value_t::discarded;
return; return;
} }
...@@ -10480,9 +10462,11 @@ class basic_json ...@@ -10480,9 +10462,11 @@ class basic_json
break; break;
} }
case lexer::token_type::value_number: case lexer::token_type::value_unsigned_integer:
case lexer::token_type::value_signed_integer:
case lexer::token_type::value_float:
{ {
m_lexer.get_number(result); m_lexer.get_number(result, last_token);
get_token(); get_token();
break; break;
} }
......
...@@ -65,25 +65,37 @@ TEST_CASE("lexer class") ...@@ -65,25 +65,37 @@ TEST_CASE("lexer class")
SECTION("numbers") SECTION("numbers")
{ {
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("0"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("0"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("1"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("1"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("2"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("2"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("3"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("3"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("4"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("4"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("5"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("5"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("6"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("6"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("7"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("7"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("8"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("8"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("9"), CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("9"),
1).scan() == json::lexer::token_type::value_number)); 1).scan() == json::lexer::token_type::value_unsigned_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("-0"),
2).scan() == json::lexer::token_type::value_signed_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("-1"),
2).scan() == json::lexer::token_type::value_signed_integer));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("1.1"),
3).scan() == json::lexer::token_type::value_float));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("-1.1"),
4).scan() == json::lexer::token_type::value_float));
CHECK((json::lexer(reinterpret_cast<const json::lexer::lexer_char_t*>("1E10"),
4).scan() == json::lexer::token_type::value_float));
} }
SECTION("whitespace") SECTION("whitespace")
...@@ -109,7 +121,9 @@ TEST_CASE("lexer class") ...@@ -109,7 +121,9 @@ TEST_CASE("lexer class")
CHECK((json::lexer::token_type_name(json::lexer::token_type::literal_false) == "false literal")); CHECK((json::lexer::token_type_name(json::lexer::token_type::literal_false) == "false literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::literal_null) == "null literal")); CHECK((json::lexer::token_type_name(json::lexer::token_type::literal_null) == "null literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::value_string) == "string literal")); CHECK((json::lexer::token_type_name(json::lexer::token_type::value_string) == "string literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::value_number) == "number literal")); CHECK((json::lexer::token_type_name(json::lexer::token_type::value_unsigned_integer) == "number literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::value_signed_integer) == "number literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::value_float) == "number literal"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::begin_array) == "'['")); CHECK((json::lexer::token_type_name(json::lexer::token_type::begin_array) == "'['"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::begin_object) == "'{'")); CHECK((json::lexer::token_type_name(json::lexer::token_type::begin_object) == "'{'"));
CHECK((json::lexer::token_type_name(json::lexer::token_type::end_array) == "']'")); CHECK((json::lexer::token_type_name(json::lexer::token_type::end_array) == "']'"));
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment