compiler.cpp

SSD6 练习4的原版正确答案

/* compiler -- */

#include "flincl.h"
#include "stdio.h"
#include "shortstack.h"
#include "semantics.h"
#include "compiler.h"
#include "ctype.h"
#include "assert.h"
#include "string.h"

#define ctrace if (machine->trace_flag) trace

Compiler::Compiler()
{
}


bool Compiler::init(Machine_ptr m)
{
	machine = m;
	return semantics.init(machine, this);
}


void Compiler::expected_close_squote()
{
    report_error("expected close single quote");
}


void Compiler::expected_close_quote()
{
    report_error("expected close quote");
}


void Compiler::expected_char_const()
{
    report_error("expected character constant");
}


void Compiler::expected_close_paren_or_formal()
{
    report_error("expected close paren or formal parameter");
}


void Compiler::expected_close_paren()
{
    report_error("expected close paren");
}


void Compiler::expected_close_bracket()
{
    report_error("expected close bracket");
}


void Compiler::expected_equal()
{
    report_error("expected equal");
}


void Compiler::expected_indent()
{
    report_error("Compiler:: expected indent");
}


void Compiler::expected_newline()
{
    report_error("expected newline");
}


void Compiler::expected_open_paren()
{
    report_error("expected open paren");
}


void Compiler::expected_open_brace()
{
    report_error("expected open brace");
}

void Compiler::expected_to()
{
    report_error("expected to");
}


void Compiler::expected_token()
{
    report_error("expected more");
}


void Compiler::expected_colon()
{
    report_error("expected colon");
}


void Compiler::expected_comma()
{
    report_error("expected comma");
}


void Compiler::expected_statement()
{
    report_error("expected statement");
}


void Compiler::expected_method_or_field_selector()
{
    report_error("expected method or field selector");
}


void Compiler::expected_decl()
{
    report_error("expected declaration");
}


void Compiler::expected_hex_constant()
{
	report_error("expected hexadecimal constant");
}


void Compiler::expected_oct_constant()
{
	report_error("expected octal constant");
}


void Compiler::expected_number_constant()
{
	report_error("expected number constant");
}


void Compiler::no_digit_in_hex()
{
	report_error("no digits found in hexidecimal constant");
}


void Compiler::no_digit_in_exponent()
{
	report_error("no digit found in exponent");
}



void Compiler::expected_id(char *msg)
{
    char s[100];
    strcpy(s, "expected class name");
    strcat(s, msg);
    report_error(s);
}


void Compiler::expected_variable()
{
    report_error("expected variable name");
}


void Compiler::expected_expression()
{
	report_error("expected expression");
}


void Compiler::indentation_error()
{
    report_error("indentation error");
}

void Compiler::expected_equal_or_in()
{
	report_error("expected = or in after for variable");
}

void Compiler::report_error(char *s)
{
	char msg[128];
	if (!token) {
	    sprintf(msg, "Error: %s", s);
	} else {
		if (token->typ == TOKEN_INDENT) strcpy(token->str, "[indentation]");
		else if (token->typ == TOKEN_DEDENT) strcpy(token->str, "[dedentation]");
		else if (token->typ == TOKEN_NEWLINE) strcpy(token->str, "[newline]");
		sprintf(msg, "Error in line %d near \"%s\": %s", 
		         line_number, token->str, s);
	}
	machine->error_msg(msg);
	error_flag = true;
}


// get_token -- returns true if token is found
//
bool Compiler::get_token()
{
    other_token();
    if (unget_count) {
        unget_count--;
        // ctrace("Token: (ungot), %s\n", token->str);
        return true;
    }
    token->typ = TOKEN_OTHER;
    int i = 0;
    int c;
    if (dedent_count) {
        token->typ = TOKEN_DEDENT;
        dedent_count--;
        token->str[0] = 0;
        ctrace("Token: DEDENT\n");
        return true;
    }
    while ((c = getc(inf)) != EOF) { // this loops over white space
		i = 0;
        token->str[i++] = c;
		if (c == '#') {
			while ((c = getc(inf)) != EOF && c != '\n') ;
			if (c == '\n') ungetc(c, inf);
			continue;
		}
        if (c == '\n') {
			line_number++;
			if (newline_flag) { // (user types from terminal)
				if (interactive_mode) {
					// blank line generates DEDENTs in interactive mode
					while (0 < indentation[istack_top]) {
						dedent_count++;
						istack_top--;
					}
					if (0 != indentation[istack_top]) {
						indentation_error();
					}
					if (dedent_count > 0) {
						dedent_count--;
						token->typ = TOKEN_DEDENT;
						ctrace("Token: DEDENT");
					} else {
						continue; // ignore 2nd blank line
					}
					i = 0;
				} else continue; // ignore blank lines
			} else {
				token->typ = TOKEN_NEWLINE;
				newline_flag = true;
				ctrace("Token: NEWLINE");
			}
        } else if (newline_flag) {
            newline_flag = false;
            int indent = 0;
            while (isspace(c)) {
                indent++;
                if ((c = getc(inf)) == EOF) break;
            }
            ungetc(c, inf);
            if (c == '#' || c == '\n') {
                newline_flag = true;
                continue; // ignore blank lines
            }
            // generate indent or dedent?
            if (indent > indentation[istack_top]) {
                token->typ = TOKEN_INDENT;
                indentation[++istack_top] = indent;
                i = 0; // take space out of token
                ctrace("Token: INDENT");
            } else if (indent < indentation[istack_top]) {
                while (indent < indentation[istack_top]) {
                    dedent_count++;
                    istack_top--;
                }
                if (indent != indentation[istack_top]) {
                    indentation_error();
                }
                dedent_count--;
                token->typ = TOKEN_DEDENT;
                i = 0;
                ctrace("Token: DEDENT");
            } else { // no indentation, newline_flag is off
                continue;
            }
        } else if (isspace(c)) {
            continue; // ignore spaces after initial ones handled above
        } else if (__iscsymf(c)) {
            newline_flag = false;
            /* get alphanumeric */
            while ((c = getc(inf)) != EOF) {
                if (__iscsym(c)) {
                    token->str[i++] = c;
                } else {
                    ungetc(c, inf);
                    break;
                }
            }
            token->typ = TOKEN_ID;
            ctrace("Token: ID");
        } else if (isdigit(c) || c == '-') {
			char first_digit = 0;
            newline_flag = 0;  // index of first digit in string
			if (c == '-') {
				first_digit = 1;
			}
            /* get number */
            while ((c = getc(inf)) != EOF) {
                if (isdigit(c)) {
                    token->str[i++] = c;
                } else {
                    break;
                }
            }
            if (c == '.' || tolower(c) == 'e') {
                token->typ = TOKEN_DOUBLE;
                if (c == '.') token->str[i++] = c;
				else ungetc(c, inf); // unget 'e'

                while ((c = getc(inf)) != EOF) {
                    if (isdigit(c)) {
                        token->str[i++] = c;
                    } else {
                        ungetc(c, inf);
                        ctrace("Token: DOUBLE");
                        break;
                    }
                }
				// may be followed by exponent "e+006"
				if ((c = getc(inf)) != EOF) {
					if (tolower(c) == 'e') {
						token->str[i++] = 'e';
						if ((c = getc(inf)) != EOF) {
							if ((c == '+') || (c == '-')) {
								token->str[i++] = c;
							} else {
								ungetc(c, inf);
							}
							int expcount = 0;
							while ((c = getc(inf)) != EOF) {
								if (isdigit(c)) {
									expcount++;
									token->str[i++] = c;
								} else {
									ungetc(c, inf);
									if (expcount == 0) {
										token->str[i] = 0;
										no_digit_in_exponent();
									}
									break;
								}
							}
						}
					}
				}
			} else if ((i == first_digit + 1) &&           // enough for "0x"
					   (token->str[first_digit] == '0') && // got the "0"
					   (tolower(c) == 'x')) {              // got the "x"
                token->typ = TOKEN_LONG;
				token->str[i++] = c;
				int digit_count = 0;
				while ((c = getc(inf)) != EOF) {
					if (isxdigit(c)) {
						digit_count = 0;
						token->str[i++] = c;
					} else {
						if (digit_count == 0) {
							token->str[i] = 0;
							no_digit_in_hex();
						}
                        ungetc(c, inf);
                        ctrace("Token: LONG");
						break;
					}
				}
            } else {
                ungetc(c, inf);
                token->typ = TOKEN_LONG;
                ctrace("Token: LONG");
            }
        } else if (c == '"') {
            newline_flag = false;
			// leave quotes in... i = 0; // skip the quote
            int quote_count = 0;
            token->typ = TOKEN_STRING;
            while ((c = getc(inf)) != EOF) {
                token->str[i++] = c;
                if (c == '"') {
                    if (quote_count == 0) {
                        quote_count = 1;
                    } else /* quote_count == 1 */ {
                        i = i - 1; // erase second quote
                        quote_count = 0;
                    }
                } else if (quote_count == 1) {
                    // we closed the quote
                    ungetc(c, inf);
					i--;
                    ctrace("Token: STRING");
                    break;
                } else if (c == '\\') { // escape character
                    c = getc(inf);
                    if (c == EOF) break;
                    token->str[i - 1] = c;
                }
            }
            if (quote_count != 1) {
                expected_close_quote();
            } else {
				i--; // back over the close quote
			}
        } else if (c == '\'') {
            newline_flag = false;
			// leave quotes in ... i = 0; // skip the quote
            int quote_count = 0;
            token->typ = TOKEN_SYMBOL;
            while ((c = getc(inf)) != EOF) {
                token->str[i++] = c;
                if (c == '\'') {
                    if (quote_count == 0) {
                        quote_count = 1;
                    } else /* quote_count == 1 */ {
                        i = i - 1; // erase second quote
                        quote_count = 0;
                    }
                } else if (quote_count == 1) {
                    // we closed the quote
                    ungetc(c, inf);
					i--;
                    ctrace("Token: SYMBOL");
                    break;
                } else if (c == '\\') { // escape character
                    c = getc(inf);
                    if (c == EOF) break;
                    token->str[i - 1] = c;
                }
            }
            if (quote_count != 1) {
                expected_close_quote();
            } else {
				i--; // back over the close quote
			}

        } else if (c == '>' || c == '<' || c == '!' || c == '=') {
            newline_flag = false;
			if ((c = getc(inf)) != EOF) {
                if ((c == '=') ||
                    (token->str[i - 1] == '>' && c == '>') ||
                    (token->str[i - 1] == '<' && c == '<')) {
				    token->str[i++] = c;
                } else ungetc(c, inf);
            }
			ctrace("Token: OTHER");
        } else if (c == '*') {
            newline_flag = false;
            if ((c = getc(inf)) != EOF && (c == '*')) {
                token->str[i++] = c;
            } else ungetc(c, inf);
            ctrace("Token: OTHER");
        } else {
            newline_flag = false;
            ctrace("Token: OTHER");
        }
		token->str[i] = 0;
		if (token->typ != TOKEN_NEWLINE) ctrace(", '%s'\n", token->str);
		else ctrace(", '\\n'\n");
        return !error_flag;
    }
	if (istack_top > 0) {
		istack_top--;
		token->typ = TOKEN_DEDENT;
        token->str[0] = 0;
        ctrace("Token: DEDENT\n");
        return true;
	}
	ctrace("Token: none\n");
    return false; // false and no error_flag means end of file
}


// unget_token -- back up to previous token
// (currently supports up to two levels of unget)
// This is more complex than I expected. token is a pointer
// to either token1 or token2, and unget_count tells how
// many levels we have ungotten.  The key "trick" is that
// unget always changes token to the other buffer, and get
// always changes token back.
//
bool Compiler::unget_token()
{
    assert(unget_count < 2);
    // ctrace("unget_token\n");
    unget_count++;
    other_token();
    return true;
}


// compile_file -- compile from opened file handle.
//
// interactive indicates that blank lines should terminate
// compilation of a top-level statement or declaration
// (normally blank lines are ignored, but this does not