scanner.h

PL/0源码

//////////////////////////////////////////////////////////////////////////////
//
//  scanner.[h,cc] -- table-driven lexical scanner
//
//  Scanner tables are created by the program makescan, and are linked
//  into the compiler in the form of scantab.cc.
//
//  assumptions:
//      There is only one 'except' class in the scangen tables, and
//          it constitutes the reserve words.
//      Tokens never contain NULL characters.
//      Case is insignificant.
//

#ifndef SCANNER_H
#define SCANNER_H

#include "inpbuf.h"

#define  MAX_MAJOR_TOKEN_NUM   	100
#define  MAX_MINOR_TOKEN_NUM   	300
#define  MAX_STRING_SPACE      	20000

//  Major and minor token numbers for all tokens can be found in tokens.h,
//  generated automatically by mungegrammar

typedef char major_token_t;             //  0..MAX_MAJOR_TOKEN_NUM 
typedef char minor_token_t;             //  0..MAX_MINOR_TOKEN_NUM
typedef short string_number_t;          //  0..MAX_STRING_SPACE 

class token_attrib_t {
public:
    major_token_t       major;
    minor_token_t       minor;
    string_number_t     text_handle;
    location_t          location;

    major_token_t       get_major(void) const    { return major; }
    minor_token_t       get_minor(void) const    { return minor; }
    location_t          get_location(void) const { return location; }
    const char *        get_text(void) const;
};

int init_scanner(void);

string_number_t define_ident(const char *ident);
//  Allows outside routines to insert pre-defined identifiers into
//  the scanner's internal tables.

token_attrib_t token_fake(major_token_t tok);
//  Used by the corrector to create a hypothetical inserted token.

major_token_t token_get(token_attrib_t& attr);
//  Returns the next token from the source program, and sets a reference
//  parameter to contain its synthetic attributes.  Return value is the
//  same as attr.major.

string token_text(string_number_t where);
//  Return (non-tossed) text of a token.

const char *token_chars(string_number_t where);
//  Return a character pointer to the string's space.

#endif