ast stuff

CMakeLists.txt

@@ -25,10 +25,24 @@ add_executable(
     src/Arguments.h
     src/Normalizer.cpp
     src/Normalizer.h
-    src/ASTNode.cpp
-    src/ASTNode.h
+    src/AST/ASTNode.cpp
+    src/AST/ASTNode.h
+    src/AST/BinaryOpNode.cpp
+    src/AST/BinaryOpNode.h
+    src/AST/ExprNode.cpp
+    src/AST/ExprNode.h
+    src/AST/MulNode.cpp
+    src/AST/MulNode.h
+    src/AST/NumberNode.cpp
+    src/AST/NumberNode.h
+    src/AST/StatementNode.cpp
+    src/AST/StatementNode.h
+    src/AST/SumNode.cpp
+    src/AST/SumNode.h
     src/replace.cpp
     src/replace.h
+    src/Parser.cpp
+    src/Parser.h
 )
 
 target_include_directories(sapphirec PUBLIC src)

core/__internal/allocate/linux.sp

@@ -0,0 +1,11 @@
+namespace allocate {
+
+    @memalloc (void*) (u64 size) {
+
+    }
+
+    @memfree (void* ptr) {
+
+    }
+
+}

core/__internal/linux/x64.sp

@@ -19,5 +19,4 @@ namespace linux {
     @sys_exit (i32 code) {
         syscall1(60,code);
     }
-
 }

core/allocate.sp

@@ -0,0 +1 @@
+import core/__internal/allocate/linux;

core/string.sp

@@ -1,3 +1,5 @@
+import core/allocate;
+
 namespace string {
 
     @len (u64) (str string) {
@@ -14,7 +16,7 @@ namespace string {
         u64 len_a = len(a);
         u64 len_b = len(b);
         u64 final_size = len_a + len_b;
-        i8* chars = i8*(final_size + 1); // TODO: work on allocation
+        i8* chars = (i8*)allocate.memalloc(final_size + 1);
         clone(a,chars,len_a);
         clone(b,chars + len_a,len_b);
         chars[final_size] = 0;
@@ -24,7 +26,7 @@ namespace string {
     @clone (str string, i8* buffer, u64 max_copy_size) {
         u64 chars_cloned = 0;
         i8* ptr = (i8*)string;
-        while(chars_cloned <= max_copy_size && ptr[chars_cloned] != 0)
+        while(chars_cloned <= max_copy_size and ptr[chars_cloned] != 0)
         {
             buffer[chars_cloned] = ptr[chars_cloned];
             chars_cloned += 1;

examples/float128.sp

@@ -0,0 +1,8 @@
+import core/io;
+import core/string;
+
+@main {
+    f128 float1 = 234.6;
+    f128 float2 = 2934748348291930404.5;
+    io.outln(string.concat('Result is: ',(str)(float1 + float2)));
+}

src/AST/ASTNode.cpp

@@ -0,0 +1,9 @@
+#include "ASTNode.h"
+
+ASTNode::ASTNode()
+{
+}
+
+ASTNode::~ASTNode()
+{
+}

src/AST/ASTNode.h

@@ -0,0 +1,9 @@
+#pragma once
+#include <memory>
+
+class ASTNode
+{
+public:
+    ASTNode();
+    ~ASTNode();
+};

src/AST/BinaryOpNode.cpp

@@ -0,0 +1,10 @@
+#include "BinaryOpNode.h"
+
+BinaryOpNode::BinaryOpNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right)
+    : left(left), right(right), ExprNode()
+{
+}
+
+BinaryOpNode::~BinaryOpNode()
+{
+}

src/AST/BinaryOpNode.h

@@ -0,0 +1,12 @@
+#pragma once
+#include "ExprNode.h"
+
+class BinaryOpNode : public ExprNode
+{
+protected:
+    std::shared_ptr<ExprNode> left;
+    std::shared_ptr<ExprNode> right;
+public:
+    BinaryOpNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right);
+    ~BinaryOpNode();
+};

src/AST/ExprNode.cpp

@@ -0,0 +1,16 @@
+#include "ExprNode.h"
+
+ExprNode::ExprNode(std::shared_ptr<ExprNode> child): child(child), ASTNode() 
+{
+
+}
+
+ExprNode::ExprNode(): ASTNode()
+{
+    
+}
+
+ExprNode::~ExprNode() 
+{
+
+}

src/AST/ExprNode.h

@@ -0,0 +1,11 @@
+#pragma once
+#include "ASTNode.h"
+
+class ExprNode : public ASTNode
+{
+    std::shared_ptr<ExprNode> child;
+public:
+    ExprNode(std::shared_ptr<ExprNode> child);
+    ExprNode();
+    ~ExprNode();
+};

src/AST/MulNode.cpp

@@ -0,0 +1,10 @@
+#include "MulNode.h"
+
+MulNode::MulNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right, char op)
+    : BinaryOpNode(left,right), op(op)
+{
+}
+
+MulNode::~MulNode()
+{
+}

src/AST/MulNode.h

@@ -0,0 +1,10 @@
+#pragma once
+#include "BinaryOpNode.h"
+
+class MulNode final : public BinaryOpNode 
+{
+    char op;
+public:
+    MulNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right, char op);
+    ~MulNode();
+};

src/AST/NumberNode.cpp

@@ -0,0 +1,2 @@
+#include "NumberNode.h"
+

src/AST/NumberNode.h

@@ -0,0 +1,10 @@
+#pragma once
+#include "ExprNode.h"
+#include <string>
+
+class NumberNode : public ExprNode 
+{
+public:
+    NumberNode();
+    ~NumberNode();
+};

src/AST/StatementNode.cpp

@@ -0,0 +1,11 @@
+#include "StatementNode.h"
+
+StatementNode::StatementNode(std::shared_ptr<ExprNode> child): child(child), ASTNode() 
+{
+
+}
+
+StatementNode::~StatementNode() 
+{
+
+}

src/AST/StatementNode.h

@@ -0,0 +1,10 @@
+#pragma once
+#include "ExprNode.h"
+
+class StatementNode : public ASTNode
+{
+    std::shared_ptr<ExprNode> child;
+public:
+    StatementNode(std::shared_ptr<ExprNode> child);
+    ~StatementNode();
+};

src/AST/SumNode.cpp

@@ -0,0 +1,10 @@
+#include "SumNode.h"
+
+SumNode::SumNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right, char op)
+    : BinaryOpNode(left,right), op(op)
+{
+}
+
+SumNode::~SumNode()
+{
+}

src/AST/SumNode.h

@@ -0,0 +1,10 @@
+#pragma once
+#include "BinaryOpNode.h"
+
+class SumNode final : public BinaryOpNode 
+{
+    char op;
+public:
+    SumNode(std::shared_ptr<ExprNode> left,std::shared_ptr<ExprNode> right, char op);
+    ~SumNode();
+};

src/ASTNode.cpp

@@ -1,27 +0,0 @@
-#include "ASTNode.h"
-
-ASTNode::ASTNode()
-{
-}
-
-ASTNode::~ASTNode()
-{
-}
-
-BinaryOpNode::BinaryOpNode(std::shared_ptr<ASTNode> left,std::shared_ptr<ASTNode> right)
-    : left(left), right(right)
-{
-}
-
-BinaryOpNode::~BinaryOpNode()
-{
-}
-
-PlusNode::PlusNode(std::shared_ptr<ASTNode> left,std::shared_ptr<ASTNode> right)
-    : BinaryOpNode(left,right)
-{
-}
-
-PlusNode::~PlusNode()
-{
-}

src/ASTNode.h

@@ -1,26 +0,0 @@
-#pragma once
-#include <memory>
-
-class ASTNode
-{
-public:
-    ASTNode();
-    ~ASTNode();
-};
-
-class BinaryOpNode : public ASTNode
-{
-protected:
-    std::shared_ptr<ASTNode> left;
-    std::shared_ptr<ASTNode> right;
-public:
-    BinaryOpNode(std::shared_ptr<ASTNode> left,std::shared_ptr<ASTNode> right);
-    ~BinaryOpNode();
-};
-
-class PlusNode final : public BinaryOpNode 
-{
-public:
-    PlusNode(std::shared_ptr<ASTNode> left,std::shared_ptr<ASTNode> right);
-    ~PlusNode();
-};

src/FileIO.cpp

@@ -1,14 +1,29 @@
 #include "FileIO.h"
+#include "Error.h"
 #include <fstream>
 #include <vector>
+#include <errno.h>
+#include <cstring>
+#include <filesystem>
 
 std::string FileIO::read_all(const std::string& filename)
 {
-    std::ifstream file(filename);
-    if(!file.is_open()) return "";
+    std::ifstream file;
+    file.exceptions(std::ios::badbit | std::ios::failbit);
+    try
+    {
+        file.open(filename);
+    }
+    catch(const std::exception& e)
+    {
+        Error::throw_error_without_location("unable to open file "+ filename + ": " + strerror(errno));
+        return "";
+    }
+    if(std::filesystem::is_directory(std::filesystem::status(filename))) Error::throw_error_without_location("unable to open file "+ filename + ": Is a directory");
+    file.exceptions(std::ios::goodbit);
     std::vector<char> file_chars;
     char fchar;
-    while (file) {
+    while (file.good()) {
         fchar = file.get();
         if(fchar != -1 ) file_chars.push_back(fchar);
     }

src/Importer.cpp

@@ -22,7 +22,7 @@ TokenStream Importer::evaluate(const TokenStream& original)
     {
         Token current_token = original[i];
 
-        if(current_token.tk_type == TT_Keyword && current_token.string_value == Lexer::keywords[1])
+        if(current_token.tk_type == TT_Import)
         {
             Token next_token = original[i+1];
             if(next_token.tk_type == TT_EOF)

src/Lexer.cpp

@@ -7,7 +7,7 @@
 #define IDENTIFIERS "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWYZ_0123456789"
 #define DIGITS "0123456789"
 
-const std::array<std::string,KEYWORD_COUNT> Lexer::keywords = {"var","import","out"};
+const std::array<std::string,TYPE_COUNT> Lexer::types = {"void","bool","str","i8","i16","i32","i64","u8","u16","u32","u64","f32","f64","f128"};
 
 Lexer::Lexer(const std::string& fname)
     : loc(1,0,fname), index(-1), prev_loc(1,0,fname)
@@ -163,6 +163,12 @@ TokenStream Lexer::lex(const std::string& text)
             case '!':
                 result.push_back(Token::make_with_line({TT_Exclamation,loc},current_line_text));
                 break;
+            case '[':
+                result.push_back(Token::make_with_line({TT_Exclamation,loc},current_line_text));
+                break;
+            case ']':
+                result.push_back(Token::make_with_line({TT_Exclamation,loc},current_line_text));
+                break;
             default:
                 Error::throw_error(loc,current_line_text,"unknown character");
         }
@@ -215,21 +221,36 @@ Token Lexer::create_identifier()
             this->rewind();
             std::string identifier(characters.begin(), characters.end());
             if(is_path) return Token::make_with_line({TT_Path,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
-            auto location = std::find(keywords.begin(),keywords.end(),identifier);
-            if(location != keywords.end())
+            auto location = std::find(types.begin(),types.end(),identifier);
+            if(location != types.end())
             {
-                return Token::make_with_line({TT_Keyword,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
+                return Token::make_with_line({TT_Type,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
             }
+            if (identifier == "import") return Token::make_with_line({TT_Import,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall0") return Token::make_with_line({TT_Syscall0,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall1") return Token::make_with_line({TT_Syscall1,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall2") return Token::make_with_line({TT_Syscall2,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall3") return Token::make_with_line({TT_Syscall3,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall4") return Token::make_with_line({TT_Syscall4,{prev_line,prev_column,loc.fname}},current_line_text);
+            if (identifier == "syscall5") return Token::make_with_line({TT_Syscall5,{prev_line,prev_column,loc.fname}},current_line_text);
             return Token::make_with_line({TT_Identifier,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
         }
     }
 
     std::string identifier(characters.begin(), characters.end());
-    auto location = std::find(keywords.begin(),keywords.end(),identifier);
-    if(location != keywords.end())
+    if(is_path) return Token::make_with_line({TT_Path,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
+    auto location = std::find(types.begin(),types.end(),identifier);
+    if(location != types.end())
     {
-        return Token::make_with_line({TT_Keyword,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
+        return Token::make_with_line({TT_Type,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
     }
+    if (identifier == "import") return Token::make_with_line({TT_Import,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall0") return Token::make_with_line({TT_Syscall0,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall1") return Token::make_with_line({TT_Syscall1,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall2") return Token::make_with_line({TT_Syscall2,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall3") return Token::make_with_line({TT_Syscall3,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall4") return Token::make_with_line({TT_Syscall4,{prev_line,prev_column,loc.fname}},current_line_text);
+    if (identifier == "syscall5") return Token::make_with_line({TT_Syscall5,{prev_line,prev_column,loc.fname}},current_line_text);
     return Token::make_with_line({TT_Identifier,identifier,{prev_line,prev_column,loc.fname}},current_line_text);
 }
 

src/Lexer.h

@@ -6,7 +6,7 @@
 #include <array>
 
 typedef std::vector<Token> TokenStream;
-#define KEYWORD_COUNT 3
+#define TYPE_COUNT 14
 
 class Lexer
 {
@@ -34,7 +34,7 @@ private:
 
     bool is_in_string(const std::string& string, const char& character);
 public:
-    static const std::array<std::string,KEYWORD_COUNT> keywords;
+    static const std::array<std::string,TYPE_COUNT> types;
     ~Lexer();
 
     TokenStream lex(const std::string& text);

src/Parser.cpp

@@ -0,0 +1,9 @@
+#include "Parser.h"
+
+Parser::Parser(/* args */)
+{
+}
+
+Parser::~Parser()
+{
+}

src/Parser.h

@@ -0,0 +1,10 @@
+#pragma once
+
+class Parser
+{
+private:
+    /* data */
+public:
+    Parser(/* args */);
+    ~Parser();
+};

src/Token.cpp

@@ -32,7 +32,17 @@ const std::string token_strings[] = {
     "TT_COMMA",
     "TT_PATH",
     "TT_EXCLAMATION",
-    "TT_NEQUAL"
+    "TT_NEQUAL",
+    "TT_LSQB",
+    "TT_RSQB",
+    "TT_TYPE",
+    "TT_IMPORT",
+    "TT_SYSCALL0",
+    "TT_SYSCALL1",
+    "TT_SYSCALL2",
+    "TT_SYSCALL3",
+    "TT_SYSCALL4",
+    "TT_SYSCALL5"
 };
 
 Token::Token(const TokenType& type)
@@ -101,6 +111,10 @@ std::string Token::to_string() const
     {
         return format_string("KEYWORD:%s %s",string_value,details);
     }
+    else if (tk_type == TT_Type)
+    {
+        return format_string("TYPE:%s %s",string_value,details);
+    }
     else if (tk_type == TT_String)
     {
         replace(const_cast<std::string&>(string_value),"\n","\\n");
@@ -154,6 +168,24 @@ std::string Token::to_string() const
             return "EXCLAMATION " + details;
         case TT_NEqual:
             return "NEQUAL " + details;
+        case TT_LSQB:
+            return "LEFTSQB " + details;
+        case TT_RSQB:
+            return "RIGHTSQB " + details;
+        case TT_Import:
+            return "IMPORT " + details;
+        case TT_Syscall0:
+            return "SYSCALL0 " + details;
+        case TT_Syscall1:
+            return "SYSCALL1 " + details;
+        case TT_Syscall2:
+            return "SYSCALL2 " + details;
+        case TT_Syscall3:
+            return "SYSCALL3 " + details;
+        case TT_Syscall4:
+            return "SYSCALL4 " + details;
+        case TT_Syscall5:
+            return "SYSCALL5 " + details;
     }
     return "";
 }

src/Token.h

@@ -33,7 +33,17 @@ enum TokenType
     TT_Comma,
     TT_Path,
     TT_Exclamation,
-    TT_NEqual
+    TT_NEqual,
+    TT_LSQB,
+    TT_RSQB,
+    TT_Type,
+    TT_Import,
+    TT_Syscall0,
+    TT_Syscall1,
+    TT_Syscall2,
+    TT_Syscall3,
+    TT_Syscall4,
+    TT_Syscall5
 };
 
 extern const std::string token_strings[];

src/TopLevelParser.cpp

@@ -1,65 +0,0 @@
-#include "TopLevelParser.h"
-#include "Error.h"
-
-TopLevelParseNode::TopLevelParseNode(const ParseContext& ctx, const TokenStream& tk_stream)
-    : context(ctx), tokens(tk_stream)
-{
-}
-
-TopLevelParseNode::~TopLevelParseNode()
-{
-}
-
-TopLevelParser::TopLevelParser()
-{
-}
-
-TopLevelParser::~TopLevelParser()
-{
-}
-
-std::vector<TopLevelParseNode> TopLevelParser::ParseTokenStream(const TokenStream& stream)
-{
-    current = stream;
-    index = -1;
-    std::vector<TopLevelParseNode> result;
-
-    while(index < current.size())
-    {
-        ++index;
-        if(current[index].tk_type == TT_Keyword)
-        {
-            if(current[index].string_value == "var")
-            {
-                result.push_back(ParseGlobalVariable());
-            } else {
-                Error::throw_error(current[index].loc,current[index].line(),"global sections must start with var or @");
-            }
-        }
-        else if (current[index].tk_type == TT_At)
-        {
-            result.push_back(ParseFunctionDefinition());
-        } else {
-            Error::throw_error(current[index].loc,current[index].line(),"global sections must start with var or @");
-        }
-    }
-
-    return result;
-}
-
-TopLevelParseNode TopLevelParser::ParseGlobalVariable()
-{
-    int prev_index = index;
-    if(!Token::match_token_types(current,{
-        {TT_Keyword},
-        {TT_Identifier},
-        {TT_LParen},
-        {TT_Identifier},
-        {TT_RParen}
-    },index))
-    {
-        Error::throw_error(current[index].loc,current[index].line(),"invalid syntax");
-    }
-    index += 5;
-    
-}

src/TopLevelParser.h

@@ -1,28 +0,0 @@
-#pragma once
-#include "Token.h"
-#include "Lexer.h" //for TokenStream
-
-enum ParseContext {
-    ParseContext_GlobalVariable,
-    ParseContext_FunctionDefinition
-};
-
-struct TopLevelParseNode {
-    ParseContext context;
-    TokenStream tokens;
-    TopLevelParseNode(const ParseContext& ctx, const TokenStream& tk_stream);
-    ~TopLevelParseNode();
-};
-
-class TopLevelParser
-{
-private:
-    TokenStream current;
-    int index;
-    TopLevelParseNode ParseGlobalVariable();
-    TopLevelParseNode ParseFunctionDefinition();
-public:
-    std::vector<TopLevelParseNode> ParseTokenStream(const TokenStream& stream);
-    TopLevelParser();
-    ~TopLevelParser();
-};