Util: Add format string parsing

src/util/format/builder.h

@@ -15,6 +15,20 @@ namespace Util::Format {
 
 class Builder {
 public:
+	enum class Align : uint8_t {
+		None,
+		Left = 60,   // '<'
+		Right = 62,  // '>'
+		Center = 94, // '^'
+	};
+
+	enum class Sign : uint8_t {
+		None,
+		Negative = 45, // '-'
+		Both = 43,     // '+'
+		Space = 32,    // ' '
+	};
+
 	explicit Builder(std::stringstream& builder)
 		: m_builder(builder)
 	{

src/util/format/format.h

@@ -21,13 +21,14 @@ class Parser;
 
 struct Parameter {
 	const void* value;
-	void (*format)(Builder& builder, const void* value);
+	void (*format)(Builder& builder, Parser& parser, const void* value);
 };
 
 template<typename T>
-void formatParameterValue(Builder& builder, const void* value)
+void formatParameterValue(Builder& builder, Parser& parser, const void* value)
 {
 	Formatter<T> formatter;
+	formatter.parse(parser);
 	formatter.format(builder, *static_cast<const T*>(value));
 }
 

src/util/format/formatter.h

@@ -7,9 +7,12 @@
 #pragma once
 
 #include <cstddef> // size_t
-#include <cstdint> // int32_t, uint8_t, uint32_t, int64_t,
+#include <cstdint> // int8_t, int32_t, int64_t, uint8_t, uint32_t
+#include <map>
 #include <string>
 #include <string_view>
+#include <type_traits> // is_integral_v, is_same
+#include <unordered_map>
 #include <vector>
 
 #include "util/format/builder.h"
@@ -17,9 +20,65 @@
 
 namespace Util::Format {
 
+enum class PresentationType : uint8_t {
+	None, // Defaults are any of: 'dgcsp', depending on the type
+	// Interger
+	Binary = 98,          // 'b'
+	BinaryUppercase = 66, // 'B'
+	Decimal = 100,        // 'd'
+	Octal = 111,          // 'o'
+	Hex = 120,            // 'x'
+	HexUppercase = 88,    // 'X'
+	// Floating-point
+	Hexfloat = 97,            // 'a'
+	HexfloatUppercase = 65,   // 'A'
+	Exponent = 101,           // 'e'
+	ExponentUppercase = 69,   // 'E'
+	FixedPoint = 102,         // 'f'
+	FixedPointUppercase = 70, // 'F'
+	General = 103,            // 'g'
+	GeneralUppercase = 71,    // 'G'
+	// Character
+	Character = 99, // 'c'
+	// String
+	String = 115, // 's'
+	// Pointer
+	Pointer = 112, // 'p'
+};
+
+struct Specifier {
+	char fill = ' ';
+	Builder::Align align = Builder::Align::None;
+
+	Builder::Sign sign = Builder::Sign::None;
+
+	bool alternativeForm = false;
+	bool zeroPadding = false;
+	int width = 0;
+	int8_t precision = -1;
+
+	PresentationType type = PresentationType::None;
+};
 
 template<typename T>
 struct Formatter {
+	Specifier specifier;
+
+	constexpr void parse(Parser& parser)
+	{
+		if (std::is_integral_v<T>) {
+			parser.parseSpecifier(specifier, Parser::SpecifierType::Integral);
+		}
+		else if (std::is_floating_point_v<T>) {
+			parser.parseSpecifier(specifier, Parser::SpecifierType::FloatingPoint);
+		}
+		else if (std::is_same_v<T, char>) {
+			parser.parseSpecifier(specifier, Parser::SpecifierType::Char);
+		}
+		else if (std::is_same_v<T, std::string_view>) {
+			parser.parseSpecifier(specifier, Parser::SpecifierType::String);
+		}
+	}
 
 	void format(Builder& builder, T value) const { (void)builder, (void)value; }
 };
@@ -80,6 +139,13 @@ struct Formatter<char[N]> : Formatter<const char*> {
 
 template<typename T>
 struct Formatter<T*> {
+	Specifier specifier;
+
+	constexpr void parse(Parser& parser)
+	{
+		parser.parseSpecifier(specifier, Parser::SpecifierType::Pointer);
+	}
+
 	void format(Builder& builder, T* value) const
 	{
 		value == nullptr

src/util/format/parser.cpp

@@ -6,11 +6,13 @@
 
 #include <algorithm> // replace
 #include <cstddef>   // size_t
+#include <cstdint>   // int8_t
 #include <limits>    // numeric_limits
 #include <string>
 #include <string_view>
 
 #include "util/format/builder.h"
+#include "util/format/formatter.h"
 #include "util/format/parser.h"
 #include "util/meta/assert.h"
 
@@ -75,6 +77,19 @@ void Parser::checkFormatParameterConsistency()
 	// VERIFY(!(braceOpen > m_parameterCount), "format string references nonexistent parameter");
 }
 
+size_t Parser::stringToNumber(std::string_view value)
+{
+	size_t result = 0;
+
+	for (size_t i = 0; i < value.length(); ++i) {
+		VERIFY(value[i] >= '0' && value[i] <= '9', "unexpected '%c'", value[i]);
+		result *= 10;
+		result += value[i] - '0'; // Subtract ASCII 48 to get the number
+	}
+
+	return result;
+}
+
 std::string_view Parser::consumeLiteral()
 {
 	const auto begin = tell();
@@ -143,22 +158,131 @@ std::optional<size_t> Parser::consumeIndex()
 	VERIFY_NOT_REACHED();
 }
 
-size_t Parser::stringToNumber(std::string_view value)
+void Parser::parseSpecifier(Specifier& specifier, SpecifierType type)
 {
-	size_t result = 0;
+	if (consumeSpecific('}') || isEOF()) {
+		return;
+	}
 
-	for (size_t i = 0; i < value.length(); ++i) {
-		VERIFY(value[i] >= '0' && value[i] <= '9', "unexpected '%c'", value[i]);
-		result *= 10;
-		result += value[i] - '0'; // Subtract ASCII 48 to get the number
+	// Alignment
+	char peek0 = peek();
+	char peek1 = peek(1);
+	if (peek1 == '<' || peek1 == '>' || peek1 == '^') {
+		specifier.fill = peek0;
+		specifier.align = static_cast<Builder::Align>(peek1);
+		ignore(2);
 	}
 
-	return result;
-}
+	enum State {
+		AfterAlign,
+		AfterSign,
+		AfterAlternativeForm,
+		AfterZeroPadding,
+		AfterWidth,
+		AfterDot,
+		AfterPrecision,
+		AfterType,
+	} state { State::AfterAlign };
+
+	size_t widthBegin = std::numeric_limits<size_t>::max();
+	size_t precisionBegin = std::numeric_limits<size_t>::max();
+	size_t widthEnd = 0;
+	size_t precisionEnd = 0;
+	std::string_view acceptedTypes = "bdoxaefgscpBXAFG";
+
+	while (true) {
+		char peek0 = peek();
+
+		if (peek0 == '}') {
+			ignore();
+			break;
+		}
+		if (isEOF()) {
+			break;
+		}
+
+		// Sign is only valid for numeric types
+		if (peek0 == '+' || peek0 == '-' || peek0 == ' ') {
+			VERIFY(state < State::AfterSign, "unexpected '%c' at this position", peek0);
+			VERIFY(type == SpecifierType::Integral || type == SpecifierType::FloatingPoint,
+			       "sign option is only valid for numeric types");
+			state = State::AfterSign;
+			specifier.sign = static_cast<Builder::Sign>(peek0);
+		}
+
+		// Alternative form is only valid for numeric types
+		if (peek0 == '#') {
+			VERIFY(state < State::AfterAlternativeForm, "unexpected '#' at this position");
+			VERIFY(type == SpecifierType::Integral || type == SpecifierType::FloatingPoint,
+			       "'#' option is only valid for numeric types");
+			state = State::AfterAlternativeForm;
+			specifier.alternativeForm = true;
+		}
+
+		// Sign aware zero padding is only valid for numeric types
+		if (peek0 == '0') {
+			if (state < State::AfterWidth) {
+				VERIFY(state < State::AfterZeroPadding, "unexpected '0' at this position");
+				VERIFY(type == SpecifierType::Integral || type == SpecifierType::FloatingPoint,
+				       "zero padding option is only valid for numeric types");
+				state = State::AfterZeroPadding;
+				specifier.zeroPadding = true;
+			}
+		}
+
+		if (peek0 >= '0' && peek0 <= '9') {
+			if (widthBegin == std::numeric_limits<size_t>::max() && state < State::AfterDot) {
+				VERIFY(state < State::AfterWidth, "unexpected '%c' at this position", peek0);
+				state = State::AfterWidth;
+				widthBegin = tell();
+			}
+			if (precisionBegin == std::numeric_limits<size_t>::max() && state == State::AfterDot) {
+				state = State::AfterPrecision;
+				precisionBegin = tell();
+			}
+		}
 
+		if (peek0 == '.') {
+			if (state == State::AfterWidth) {
+				widthEnd = tell();
 			}
+
+			VERIFY(state < State::AfterDot, "unexpected '.' at this position");
+			state = State::AfterDot;
 		}
 
+		if ((peek0 >= 'a' && peek0 <= 'z')
+		    || (peek0 >= 'A' && peek0 <= 'Z')) {
+			if (state == State::AfterWidth) {
+				widthEnd = tell();
+			}
+			if (state == State::AfterPrecision) {
+				precisionEnd = tell();
+			}
+
+			VERIFY(state < State::AfterType, "unexpected '%c' at this position", peek0);
+			state = State::AfterType;
+			VERIFY(acceptedTypes.find(peek0) != std::string_view::npos, "unexpected '%c' at this position", peek0);
+			specifier.type = static_cast<PresentationType>(peek0);
+		}
+
+		ignore();
+	}
+
+	if (widthBegin != std::numeric_limits<size_t>::max()) {
+		if (widthEnd == 0) {
+			// We parse until after the closing '}', so take this into account
+			widthEnd = tell() - 1;
+		}
+		specifier.width = static_cast<int>(stringToNumber(m_input.substr(widthBegin, widthEnd - widthBegin)));
+	}
+
+	if (precisionBegin != std::numeric_limits<size_t>::max()) {
+		if (precisionEnd == 0) {
+			// We parse until after the closing '}', so take this into account
+			precisionEnd = tell() - 1;
+		}
+		specifier.precision = static_cast<int8_t>(stringToNumber(m_input.substr(precisionBegin, precisionEnd - precisionBegin)));
 	}
 }
 

src/util/format/parser.h

@@ -15,6 +15,7 @@
 namespace Util::Format {
 
 class Builder;
+struct Specifier;
 
 class Parser final : public GenericLexer {
 public:
@@ -23,16 +24,24 @@ public:
 		Manual,    // {0},{1}
 	};
 
+	enum class SpecifierType {
+		Integral,
+		FloatingPoint,
+		Char,
+		String,
+		Pointer,
+	};
+
 	Parser(std::string_view format, size_t parameterCount);
 	virtual ~Parser();
 
 	void checkFormatParameterConsistency();
+	size_t stringToNumber(std::string_view value);
 
 	std::string_view consumeLiteral();
 	std::optional<size_t> consumeIndex();
 
-	size_t stringToNumber(std::string_view value);
-
+	void parseSpecifier(Specifier& specifier, SpecifierType type);
 
 private:
 	ArgumentIndexingMode m_mode { ArgumentIndexingMode::Automatic };