lonely-tiger/tokenizer/Tokenizer.c

#include <ctype.h>
#include <limits.h>
#include <setjmp.h>
#include <stdarg.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "tokenizer/FileTokens.h"
#include "tokenizer/Tokenizer.h"
#include "utils/SnuviUtils.h"

#define TOKEN_BUFFER_LENGTH (1024 * 1024)
#define INCLUDE_STACK_LENGTH 32

static FileTokens fileTokens;
static int fileTokenIndex;
static jmp_buf errorJump;
static Error* error = NULL;

typedef struct {
    const char* path;
    int line;
} IncludeEntry;
static IncludeEntry includeStack[INCLUDE_STACK_LENGTH];
static int includeStackIndex = 0;

static char tokenBuffer[TOKEN_BUFFER_LENGTH];
static int writeIndex = 0;
static int readIndex = 0;
static int16 line = 1;

static void tError(const char* format, ...) {
    va_list args;
    va_start(args, format);
    const char* path = includeStackIndex > 0
                           ? includeStack[includeStackIndex - 1].path
                           : "path not set";
    eInitErrorV(error, path, line, format, args);
    for(int i = includeStackIndex - 2; i >= 0; i--) {
        eAddPath(error, includeStack[i].path);
    }
    va_end(args);
    longjmp(errorJump, 0);
}

static void tAdd(const void* data, int length) {
    if(writeIndex + length > TOKEN_BUFFER_LENGTH) {
        tError("the token buffer is too small");
    }
    memcpy(tokenBuffer + writeIndex, data, length);
    writeIndex += length;
}

static void tAddToken(Token token) {
    unsigned char c = token;
    tAdd(&c, 1);
    tAdd(&line, sizeof(line));
}

static bool tReadTokens(void* dest, int length) {
    if(readIndex + length > writeIndex) {
        return true;
    }
    memcpy(dest, tokenBuffer + readIndex, length);
    readIndex += length;
    return false;
}

static FileToken* tNextToken() {
    if(fileTokenIndex >= fileTokens.length) {
        tError("unexpected end of file");
    }
    return fileTokens.tokens + fileTokenIndex++;
}

static bool tReadSingleIf(char c) {
    if(fileTokenIndex >= fileTokens.length ||
       fileTokens.tokens[fileTokenIndex].type != FT_SINGLE ||
       fileTokens.tokens[fileTokenIndex].single != c) {
        return false;
    }
    fileTokenIndex++;
    return true;
}

static void tParseLiteral(const char* buffer) {
    if(strcmp(buffer, "return") == 0) {
        tAddToken(T_RETURN);
    } else if(strcmp(buffer, "if") == 0) {
        tAddToken(T_IF);
    } else if(strcmp(buffer, "else") == 0) {
        tAddToken(T_ELSE);
    } else if(strcmp(buffer, "while") == 0) {
        tAddToken(T_WHILE);
    } else if(strcmp(buffer, "for") == 0) {
        tAddToken(T_FOR);
    } else if(strcmp(buffer, "break") == 0) {
        tAddToken(T_BREAK);
    } else if(strcmp(buffer, "continue") == 0) {
        tAddToken(T_CONTINUE);
    } else if(strcmp(buffer, "int") == 0) {
        tAddToken(T_INT);
    } else if(strcmp(buffer, "void") == 0) {
        tAddToken(T_VOID);
    } else if(strcmp(buffer, "float") == 0) {
        tAddToken(T_FLOAT);
    } else if(strcmp(buffer, "struct") == 0) {
        tAddToken(T_STRUCT);
    } else if(strcmp(buffer, "new") == 0) {
        tAddToken(T_NEW);
    } else if(strcmp(buffer, "length") == 0) {
        tAddToken(T_LENGTH);
    } else if(strcmp(buffer, "true") == 0) {
        int32 i = 1;
        tAddToken(T_INT_VALUE);
        tAdd(&i, sizeof(int32));
    } else if(strcmp(buffer, "false") == 0) {
        int32 i = 0;
        tAddToken(T_INT_VALUE);
        tAdd(&i, sizeof(int32));
    } else {
        tAddToken(T_LITERAL);
        tAdd(buffer, strlen(buffer) + 1);
    }
}

static long long tParseInt(const char* s) {
    long long l = 0;
    for(int i = 0; s[i] != '\0'; i++) {
        if(l > (LLONG_MAX / 10)) {
            tError("invalid number on line %d", line);
        }
        l *= 10;
        if(!isNumber(s[i])) {
            tError("invalid character in number '%c' on line %d", s[i], line);
        }
        int digit = s[i] - '0';
        if(l > LLONG_MAX - digit) {
            tError("invalid number on line %d", line);
        }
        l += digit;
    }
    return l;
}

static void tParseNumber(const char* buffer) {
    long long l = tParseInt(buffer);
    if(tReadSingleIf('.')) {
        FileToken* t = tNextToken();
        if(t->type != FT_LITERAL) {
            tError("expected literal after comma of number on line %d", line);
        }
        long long comma = tParseInt(t->literal);

        float f = comma;
        while(f > 1.0f) {
            f /= 10.0f;
        }
        f += l;
        tAddToken(T_FLOAT_VALUE);
        tAdd(&f, sizeof(float));
    } else {
        if(l > INT32_MAX) {
            tError("invalid int on line %d", line);
        }
        int32 i = l;
        tAddToken(T_INT_VALUE);
        tAdd(&i, sizeof(int32));
    }
}

static int32 tNextUnicodePart() {
    FileToken* t = tNextToken();
    if(t->type == FT_SINGLE) {
        return t->single;
    } else if(t->type == FT_LITERAL) {
        int length = strlen(t->literal);
        if(length != 1) {
            tError("unicode literal has wrong length %d", length);
        }
        return t->literal[0];
    } else {
        tError("cannot read next unicode character part on line %d", line);
        return 0;
    }
}

static int32 tUnicode(int32 c) {
    if(c == '\\') {
        switch(tNextUnicodePart()) {
            case '"': c = '"'; break;
            case '\\': c = '\\'; break;
            case 'n': c = '\n'; break;
            case 'r': c = '\r'; break;
            case 't': c = '\t'; break;
            default: tError("unknown escaped character at line %d", line);
        }
    }
    if((c & 0xE0) == 0xC0) {
        c = ((c & 0x1F) << 6) | (tNextUnicodePart() & 0x3F);
    } else if((c & 0xF0) == 0xE0) {
        c = ((c & 0xF) << 12) | ((tNextUnicodePart() & 0x3F) << 6);
        c |= tNextUnicodePart(&fileTokens) & 0x3F;
    } else if((c & 0xF8) == 0xF0) {
        c = ((c & 0x7) << 18) | ((tNextUnicodePart() & 0x3F) << 12);
        c |= (tNextUnicodePart() & 0x3F) << 6;
        c |= tNextUnicodePart() & 0x3F;
    }
    return c;
}

static int32 tReadNextPart(const char* s, int* index, int length) {
    if(*index >= length) {
        tError("missing escape character");
    }
    return s[(*index)++];
}

static int32 tStringUnicode(const char* s, int* index, int length) {
    int32 c = s[(*index)++];
    if(c == '\\') {
        switch(tReadNextPart(s, index, length)) {
            case '"': c = '"'; break;
            case '\\': c = '\\'; break;
            case 'n': c = '\n'; break;
            case 'r': c = '\r'; break;
            case 't': c = '\t'; break;
            default: tError("unknown escaped character at line %d", line);
        }
    }
    if((c & 0xE0) == 0xC0) {
        c = ((c & 0x1F) << 6) | (tReadNextPart(s, index, length) & 0x3F);
    } else if((c & 0xF0) == 0xE0) {
        c = ((c & 0xF) << 12) | ((tReadNextPart(s, index, length) & 0x3F) << 6);
        c |= tReadNextPart(s, index, length) & 0x3F;
    } else if((c & 0xF8) == 0xF0) {
        c = ((c & 0x7) << 18) |
            ((tReadNextPart(s, index, length) & 0x3F) << 12);
        c |= (tReadNextPart(s, index, length) & 0x3F) << 6;
        c |= tReadNextPart(s, index, length) & 0x3F;
    }
    return c;
}

static void tAddString() {
    tAddToken(T_TEXT);

    FileToken* t = tNextToken();
    if(t->type == FT_SINGLE) {
        if(t->single != '"') {
            tError("unexpected single '%d'", t->single);
        }
    } else if(t->type == FT_LITERAL) {
        int length = strlen(t->literal);
        const char* s = t->literal;
        int index = 0;
        while(index < length) {
            int32 c = tStringUnicode(s, &index, length);
            tAdd(&c, sizeof(int32));
        }
        if(!tReadSingleIf('"')) {
            tError("unclosed string");
        }
    } else {
        tError("unexpected string file token %d", t);
    }

    int32 c = 0;
    tAdd(&c, sizeof(int32));
}

static void tAddUnicode() {
    FileToken* t = tNextToken();
    if(t->type == FT_LITERAL) {
        int length = strlen(t->literal);
        if(length != 1) {
            tError("invalid character on line %d", line);
        }
        int32 c = t->literal[0];
        tAddToken(T_INT_VALUE);
        tAdd(&c, sizeof(int32));
    } else if(t->type == FT_SINGLE) {
        int32 c = tUnicode(t->single);
        tAddToken(T_INT_VALUE);
        tAdd(&c, sizeof(int32));
    } else {
        tError("invalid character on line %d", line);
    }
    if(!tReadSingleIf('\'')) {
        tError("expecting unicode end");
    }
}

static void tAddToken2(Token te, Token t) {
    if(tReadSingleIf('=')) {
        tAddToken(te);
    } else {
        tAddToken(t);
    }
}

static void tAddToken3(int c, Token tc, Token te, Token t) {
    if(tReadSingleIf(c)) {
        tAddToken(tc);
    } else {
        tAddToken2(te, t);
    }
}

static void tAddToken4(int c, Token tce, Token tc, Token te, Token t) {
    if(tReadSingleIf(c)) {
        tAddToken2(tce, tc);
    } else {
        tAddToken2(te, t);
    }
}

static void tParseToken(FileToken* t) {
    switch(t->type) {
        case FT_PATH:
            if(includeStackIndex >= INCLUDE_STACK_LENGTH) {
                tError("include stack overflow");
            }
            includeStack[includeStackIndex].path = t->literal;
            includeStack[includeStackIndex].line = line;
            includeStackIndex++;
            line = 1;
            tAddToken(T_OPEN_PATH);
            tAdd(t->literal, strlen(t->literal) + 1);
            return;
        case FT_END_PATH:
            if(includeStackIndex <= 0) {
                tError("include stack underflow");
            }
            includeStackIndex--;
            line = includeStack[includeStackIndex].line;
            tAddToken(T_CLOSE_PATH);
            const char* path = includeStackIndex > 0
                                   ? includeStack[includeStackIndex - 1].path
                                   : "path not set";
            tAdd(path, strlen(path) + 1);
            return;
        case FT_NEWLINE: line++; return;
        case FT_LITERAL: {
            const char* buffer = t->literal;
            if(isLetter(buffer[0])) {
                tParseLiteral(buffer);
            } else if(isNumber(buffer[0])) {
                tParseNumber(buffer);
            } else {
                tError("invalid literal string '%s'", buffer);
            }
            return;
        }
        case FT_SPACE: return;
        case FT_SINGLE: {
            char c = t->single;
            switch(c) {
                case ' ': return;
                case '+':
                    tAddToken3('+', T_INCREMENT, T_ADD_SET, T_ADD);
                    return;
                case '-':
                    tAddToken3('-', T_DECREMENT, T_SUB_SET, T_SUB);
                    return;
                case '*': tAddToken2(T_MUL_SET, T_MUL); return;
                case '/': tAddToken2(T_DIV_SET, T_DIV); return;
                case '%': tAddToken2(T_MOD_SET, T_MOD); return;
                case '<':
                    tAddToken4('<', T_LEFT_SHIFT_SET, T_LEFT_SHIFT,
                               T_LESS_EQUAL, T_LESS);
                    return;
                case '>':
                    tAddToken4('>', T_RIGHT_SHIFT_SET, T_RIGHT_SHIFT,
                               T_GREATER_EQUAL, T_GREATER);
                    return;
                case '=': tAddToken2(T_EQUAL, T_SET); return;
                case '!': tAddToken2(T_NOT_EQUAL, T_NOT); return;
                case '&':
                    tAddToken3('&', T_AND, T_BIT_AND_SET, T_BIT_AND);
                    return;
                case '|': tAddToken3('|', T_OR, T_BIT_OR_SET, T_BIT_OR); return;
                case '~': tAddToken(T_BIT_NOT); return;
                case '^': tAddToken2(T_BIT_XOR_SET, T_BIT_XOR); return;
                case ',': tAddToken(T_COMMA); return;
                case ';': tAddToken(T_SEMICOLON); return;
                case '(': tAddToken(T_OPEN_BRACKET); return;
                case ')': tAddToken(T_CLOSE_BRACKET); return;
                case '{': tAddToken(T_OPEN_CURVED_BRACKET); return;
                case '}': tAddToken(T_CLOSE_CURVED_BRACKET); return;
                case '"': tAddString(); return;
                case '\'': tAddUnicode(); return;
                case '.': tAddToken(T_POINT); return;
                case '[': tAddToken(T_OPEN_SQUARE_BRACKET); return;
                case ']': tAddToken(T_CLOSE_SQUARE_BRACKET); return;
            }
            if(isprint(c)) {
                tError("unknown character on line %d: %c", line, c);
            } else {
                tError("unknown character on line %d: %d", line, (int)c);
            }
        }
    }
}

static void tParseFile() {
    readIndex = 0;
    writeIndex = 0;
    line = 1;
    fileTokenIndex = 0;
    while(fileTokenIndex < fileTokens.length) {
        tParseToken(fileTokens.tokens + fileTokenIndex++);
    }
}

void tTokenize(const char* path, Error* e) {
    error = e;
    ftInit(path, &fileTokens, e);
    if(!eHasError(e) && !setjmp(errorJump)) {
        tParseFile();
    }
    if(eHasError(e)) {
        ftPrint(&fileTokens);
    }
    includeStackIndex = 0;
    ftDelete(&fileTokens);
}

Token tPeekToken() {
    if(readIndex >= writeIndex) {
        return T_END;
    }
    return tokenBuffer[readIndex];
}

bool tReadTokenAndLine(Token* t, int16* line) {
    if(readIndex >= writeIndex) {
        return true;
    }
    *t = tokenBuffer[readIndex++];
    return tReadTokens(line, sizeof(int16));
}

bool tReadInt32(int32* i) {
    return tReadTokens(i, sizeof(int32));
}

bool tReadFloat(float* f) {
    return tReadTokens(f, sizeof(float));
}

const char* tReadString() {
    const char* s = tokenBuffer + readIndex;
    while(readIndex <= writeIndex) {
        if(tokenBuffer[readIndex++] == '\0') {
            return s;
        }
    }
    return NULL;
}

int tGetMarker() {
    return readIndex;
}

void tReset(int marker) {
    readIndex = marker;
}