LibCMake: Add a lexer for CMakeCache.txt

This is a totally different syntax than for regular CMake files, and
also is undocumented and subject to change, but it's also nice and
simple. :^)

6 files changed, 324 insertions, 5 deletions

diff --git a/Userland/Libraries/LibCMake/CMakeCache/Lexer.cpp b/Userland/Libraries/LibCMake/CMakeCache/Lexer.cpp
new file mode 100644
index 0000000000..b70e57502f
--- /dev/null
+++ b/Userland/Libraries/LibCMake/CMakeCache/Lexer.cpp
@@ -0,0 +1,203 @@
+/*
+ * Copyright (c) 2023, Sam Atkins <atkinssj@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include "Lexer.h"
+#include <AK/CharacterTypes.h>
+#include <AK/Debug.h>
+#include <AK/ScopeLogger.h>
+
+namespace CMake::Cache {
+
+static bool is_identifier_start_character(u32 c)
+{
+    return AK::is_ascii_alpha(c) || c == '_' || c == '-';
+}
+
+static bool is_identifier_character(u32 c)
+{
+    return AK::is_ascii_alphanumeric(c) || c == '_' || c == '-';
+}
+
+Lexer::Lexer(StringView input)
+    : GenericLexer(input)
+{
+}
+
+ErrorOr<Vector<Token>> Lexer::lex(StringView input)
+{
+    Lexer lexer { input };
+    return lexer.lex_file();
+}
+
+ErrorOr<Vector<Token>> Lexer::lex_file()
+{
+    ScopeLogger<CMAKE_DEBUG> logger;
+
+    while (!is_eof()) {
+        skip_whitespace();
+
+        if (is_eof())
+            break;
+
+        if (next_is('#')) {
+            consume_comment();
+            continue;
+        }
+
+        if (next_is("//"sv)) {
+            consume_help_text();
+            continue;
+        }
+
+        if (next_is(is_identifier_start_character)) {
+            consume_variable_definition();
+            continue;
+        }
+
+        consume_garbage();
+    }
+
+    return m_tokens;
+}
+
+void Lexer::skip_whitespace()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    while (!is_eof()) {
+        if (next_is('\n')) {
+            next_line();
+            continue;
+        }
+        auto consumed = consume_while(AK::is_ascii_space);
+        if (consumed.is_empty())
+            break;
+    }
+}
+
+void Lexer::consume_comment()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    VERIFY(consume_specific('#'));
+    auto comment = consume_until('\n');
+    emit_token(Token::Type::Comment, comment, start, position());
+}
+
+void Lexer::consume_help_text()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    VERIFY(consume_specific("//"sv));
+    auto help_text = consume_until('\n');
+    emit_token(Token::Type::HelpText, help_text, start, position());
+}
+
+void Lexer::consume_variable_definition()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    consume_key();
+
+    if (!next_is(':')) {
+        consume_garbage();
+        return;
+    }
+    consume_colon();
+
+    if (!next_is(is_identifier_start_character)) {
+        consume_garbage();
+        return;
+    }
+    consume_type();
+
+    if (!next_is('=')) {
+        consume_garbage();
+        return;
+    }
+    consume_equals();
+
+    consume_value();
+}
+
+void Lexer::consume_key()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    auto key = consume_while(is_identifier_character);
+    emit_token(Token::Type::Key, key, start, position());
+}
+
+void Lexer::consume_colon()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    VERIFY(consume_specific(':'));
+    emit_token(Token::Type::Colon, ":"sv, start, position());
+}
+
+void Lexer::consume_type()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    auto type = consume_while(is_identifier_character);
+    emit_token(Token::Type::Type, type, start, position());
+}
+
+void Lexer::consume_equals()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    VERIFY(consume_specific('='));
+    emit_token(Token::Type::Colon, "="sv, start, position());
+}
+
+void Lexer::consume_value()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    auto value = consume_until('\n');
+    emit_token(Token::Type::Value, value, start, position());
+}
+
+void Lexer::consume_garbage()
+{
+    ScopeLogger<CMAKE_DEBUG> log;
+
+    auto start = position();
+    auto garbage = consume_until('\n');
+    emit_token(Token::Type::Garbage, garbage, start, position());
+}
+
+Position Lexer::position() const
+{
+    return Position {
+        .line = m_line,
+        .column = tell() - m_string_offset_after_previous_newline,
+    };
+}
+
+void Lexer::next_line()
+{
+    VERIFY(consume_specific('\n'));
+    m_string_offset_after_previous_newline = tell();
+    m_line++;
+}
+
+void Lexer::emit_token(Token::Type type, StringView value, Position start, Position end)
+{
+    dbgln_if(CMAKE_DEBUG, "Emitting {} token: `{}` ({}:{} to {}:{})", to_string(type), value, start.line, start.column, end.line, end.column);
+    m_tokens.empend(type, value, start, end);
+}
+
+}
diff --git a/Userland/Libraries/LibCMake/CMakeCache/Lexer.h b/Userland/Libraries/LibCMake/CMakeCache/Lexer.h
new file mode 100644
index 0000000000..f258972fb8
--- /dev/null
+++ b/Userland/Libraries/LibCMake/CMakeCache/Lexer.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2023, Sam Atkins <atkinssj@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/GenericLexer.h>
+#include <AK/Vector.h>
+#include <LibCMake/CMakeCache/Token.h>
+
+namespace CMake::Cache {
+
+class Lexer : private GenericLexer {
+public:
+    static ErrorOr<Vector<Token>> lex(StringView input);
+
+private:
+    Lexer(StringView input);
+
+    ErrorOr<Vector<Token>> lex_file();
+
+    void skip_whitespace();
+
+    void consume_comment();
+    void consume_help_text();
+    void consume_variable_definition();
+    void consume_key();
+    void consume_colon();
+    void consume_type();
+    void consume_equals();
+    void consume_value();
+    void consume_garbage();
+
+    Position position() const;
+    void next_line();
+
+    void emit_token(Token::Type, StringView value, Position start, Position end);
+
+    Vector<Token> m_tokens;
+    size_t m_line { 0 };
+    size_t m_string_offset_after_previous_newline { 0 };
+};
+
+}
diff --git a/Userland/Libraries/LibCMake/CMakeCache/Token.h b/Userland/Libraries/LibCMake/CMakeCache/Token.h
new file mode 100644
index 0000000000..35f84d970f
--- /dev/null
+++ b/Userland/Libraries/LibCMake/CMakeCache/Token.h
@@ -0,0 +1,55 @@
+/*
+ * Copyright (c) 2023, Sam Atkins <atkinssj@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <LibCMake/Position.h>
+
+namespace CMake::Cache {
+
+struct Token {
+    enum class Type {
+        Comment,
+        HelpText,
+        Key,
+        Colon,
+        Type,
+        Equals,
+        Value,
+        Garbage,
+    };
+    Type type;
+    StringView value;
+
+    Position start;
+    Position end;
+};
+
+static constexpr StringView to_string(Token::Type type)
+{
+    switch (type) {
+    case Token::Type::Comment:
+        return "Comment"sv;
+    case Token::Type::HelpText:
+        return "HelpText"sv;
+    case Token::Type::Key:
+        return "Key"sv;
+    case Token::Type::Colon:
+        return "Colon"sv;
+    case Token::Type::Type:
+        return "Type"sv;
+    case Token::Type::Equals:
+        return "Equals"sv;
+    case Token::Type::Value:
+        return "Value"sv;
+    case Token::Type::Garbage:
+        return "Garbage"sv;
+    }
+
+    VERIFY_NOT_REACHED();
+}
+
+}
diff --git a/Userland/Libraries/LibCMake/CMakeLists.txt b/Userland/Libraries/LibCMake/CMakeLists.txt
index 89ac0181af..36af8b8e7e 100644
--- a/Userland/Libraries/LibCMake/CMakeLists.txt
+++ b/Userland/Libraries/LibCMake/CMakeLists.txt
@@ -1,4 +1,5 @@
 set(SOURCES
+    CMakeCache/Lexer.cpp
     Lexer.cpp
     SyntaxHighlighter.cpp
     Token.cpp
diff --git a/Userland/Libraries/LibCMake/Position.h b/Userland/Libraries/LibCMake/Position.h
new file mode 100644
index 0000000000..1870cfffe9
--- /dev/null
+++ b/Userland/Libraries/LibCMake/Position.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (c) 2023, Sam Atkins <atkinssj@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/Types.h>
+
+namespace CMake {
+
+struct Position {
+    size_t line { 0 };
+    size_t column { 0 };
+};
+
+}
diff --git a/Userland/Libraries/LibCMake/Token.h b/Userland/Libraries/LibCMake/Token.h
index b83207b874..ca45a31945 100644
--- a/Userland/Libraries/LibCMake/Token.h
+++ b/Userland/Libraries/LibCMake/Token.h
@@ -8,14 +8,10 @@
 
 #include <AK/StringView.h>
 #include <AK/Vector.h>
+#include <LibCMake/Position.h>
 
 namespace CMake {
 
-struct Position {
-    size_t line { 0 };
-    size_t column { 0 };
-};
-
 struct VariableReference {
     StringView value;
     Position start;