C语言实现的词法分析器

最新推荐文章于 2024-07-02 22:22:32 发布
C咖咖
最新推荐文章于 2024-07-02 22:22:32 发布
阅读量776
点赞数
文章标签：语言 c token stream file character
#ifndef MYLEX_H 
 
 02#define MYLEX_H 
 
 03  
 
 04#include <stdio.h> 
 
 05#include <stdlib.h> 
 
 06#include <ctype.h> 
 
 07#include <string.h> 
 
 08#include <time.h> 
 
 09  
 
 10#define MAX_ID 20 
 
 11  
 
 12typedef  enum SYMBOL { 
 
 13    nul, eof, plus, minus, times, slash, lparen, rparen, comma, semicolon, period, becomes, 
 
 14    eql, neq, lss, gtr, leq, geq, number, ident, 
 
 15    beginsym, callsym, constsym, dosym, endsym, ifsym, oddsym, proceduresym, readsym, thensym, varsym, whilesym, writesym 
 
 16} SYMBOL; 
 
 17  
 
 18int  read_ch(); 
 
 19int  put_back(int); 
 
 20SYMBOL getsym(); 
 
 21int  w2f(int);  
 
 22  
 
 23int  find_tab(char  *); 
 
 24SYMBOL get_data(int); 
 
 25int  error(); 
 
 26  
 
 27#endif
 
[文件] symbol.c ~ 4KB    下载(21)view source
print?
 
    001#include "mylex.h" 
 
    002  
 
    003extern  FILE *source, *out; 
 
    004  
 
    005const  char *symtype[33] = { 
 
    006    "nul", "eof", "plus", "minus", "times", "slash", "lparen", "rparen", "comma", "seicolon",  "period", "becomes", 
 
    007    "eql", "neq", "lss", "gtr", "leq", "geq", "number", "ident", 
 
    008    "beginsym", "callsym",  "consysym", "dosym", "endsym", "ifsym", "oddsym", "proceduresym", "readsym",  "thensym", "varsym", "whilesym",  "writesym"
 
    009}; 
 
    010  
 
    011int  cur_line, cur_col, err_line, err_col; 
 
    012int  num; 
 
    013char  id[21] = "\0";  
 
    014int  id_len; 
 
    015int  token_num; 
 
    016  
 
    017int  read_ch() { 
 
    018    int ch = fgetc(source); 
 
    019    cur_col++; 
 
    020    if (ch == '\n') { 
 
    021        cur_line++; 
 
    022        cur_col = 0; 
 
    023    }  
 
    024    return ch; 
 
    025} 
 
    026  
 
    027int  put_back(int  ch) { 
 
    028    ungets(ch, source); 
 
    029    cur_col--; 
 
    030    if (ch == '\n') 
 
    031        cur_line--; 
 
    032    return 0; 
 
    033} 
 
    034  
 
    035SYMBOL getsym() { 
 
    036    int ch; 
 
    037    int iskeywords; 
 
    038    char invalid[2] = "\0"; 
 
    039    while ((ch = read_ch()) != EOF && ch <= ' ') 
 
    040        ; 
 
    041    err_line = cur_line; 
 
    042    err_col = cur_col; 
 
    043    switch (ch) { 
 
    044        case EOF : 
 
    045            strcpy(id, "EOF"); 
 
    046            return eof; 
 
    047        case '+' : 
 
    048            strcpy(id, "+"); 
 
    049            return plus; 
 
    050        case '-' : 
 
    051            strcpy(id, "-"); 
 
    052            return minus; 
 
    053        case '*' : 
 
    054            strcpy(id, "*"); 
 
    055            return times; 
 
    056        case '/' : 
 
    057            strcpy(id, "/"); 
 
    058            return slash; 
 
    059        case '(' : 
 
    060            strcpy(id, "("); 
 
    061            return lparen; 
 
    062        case ')' : 
 
    063            strcpy(id, ")"); 
 
    064            return rparen; 
 
    065        case ',' : 
 
    066            strcpy(id, ","); 
 
    067            return comma; 
 
    068        case ';' : 
 
    069            strcpy(id, ";"); 
 
    070            return semicolon; 
 
    071        case '.' : 
 
    072            strcpy(id, "."); 
 
    073            return period; 
 
    074        case ':' : 
 
    075            ch = read_ch(); 
 
    076            if (ch == '=') { 
 
    077                strcpy(id, ":="); 
 
    078                return becomes; 
 
    079            } else
 
    080                return nul; 
 
    081        case '=' : 
 
    082            strcpy(id, "="); 
 
    083            return eql; 
 
    084        case '#' : 
 
    085            strcpy(id, "#"); 
 
    086            return neq; 
 
    087        case '<' : 
 
    088            ch = read_ch(); 
 
    089            if (ch == '=') { 
 
    090                strcpy(id, "<="); 
 
    091                return leq; 
 
    092            } 
 
    093            put_back(ch); 
 
    094            strcpy(id, "<"); 
 
    095            return lss; 
 
    096        case '>' : 
 
    097            ch = read_ch(); 
 
    098            if (ch == '=') { 
 
    099                strcpy(id, ">="); 
 
    100                return geq; 
 
    101            } 
 
    102            put_back(ch); 
 
    103            strcpy(id, ">"); 
 
    104            return gtr; 
 
    105        default : 
 
    106            if (isdigit(ch)) { 
 
    107                num = 0; 
 
    108                do { 
 
    109                    num = 10 * num + ch - '0'; 
 
    110                    ch = read_ch(); 
 
    111                } while (ch != EOF && isdigit(ch)); 
 
    112                _itoa(num, id, 10); 
 
    113                if (isslpha(ch)) { 
 
    114                    invalid[0] = (char)ch; 
 
    115                    strcat(id, invalid); 
 
    116                    ch = read_ch(); 
 
    117                    while (isalnum(ch)) { 
 
    118                        invalid[0] = (char) ch; 
 
    119                        strcat(id, invalid); 
 
    120                        ch = read_ch(); 
 
    121                    } 
 
    122                    error(); 
 
    123                    return nul; 
 
    124                } 
 
    125                put_back(ch); 
 
    126                return number; 
 
    127            } 
 
    128            if (isalpha(ch)) { 
 
    129                id_len = 0; 
 
    130                do { 
 
    131                    if (id_len < MAX_ID) { 
 
    132                        id[id_len] = (char)ch; 
 
    133                        id_len++; 
 
    134                    } 
 
    135                    ch = read_ch(); 
 
    136                } while (ch != EOF && isalnum(ch)); 
 
    137                id[id_len] = '\0'; 
 
    138                put_back(ch); 
 
    139                iskeywords = find_tab(id); 
 
    140                return (iskeywords == -1) ? ident : get_data(iskeywords);  
 
    141            } 
 
    142            error(); 
 
    143            return nul; 
 
    144    }  
 
    145} 
 
    146  
 
    147int  w2f(int flag) { 
 
    148    fprintf(out, "|%4d |%19s |%19s |%19s |\n",token_num, id, symtype[flag], ""); 
 
    149    return 0; 
 
    150}
 
[文件] keyword.c ~ 1KB    下载(22)view source
print?
 
    01#include "mylex.h" 
 
    02  
 
    03extern  FILE *out; 
 
    04extern  int cur_line, cur_col, err_line, err_col; 
 
    05extern  int num; 
 
    06extern  char id[21]; 
 
    07extern  int token_num; 
 
    08  
 
    09char  *keywords[13] = { 
 
    10    "begin", "call", "const", "do", "end", "if", "odd", "procedure",  "read", "then", "var", "while", "write"
 
    11}; 
 
    12  
 
    13int  find_tab(char  id[21]) { 
 
    14    int i; 
 
    15    for (i = 0; i !=13; ++i) { 
 
    16        if (!strcmp(keywords[i], id)) 
 
    17            return i; 
 
    18        else
 
    19            continue; 
 
    20    }  
 
    21    return -1; 
 
    22} 
 
    23  
 
    24SYMBOL get_data(int No) { 
 
    25    switch (No) { 
 
    26        case 0 : 
 
    27            return beginsym; 
 
    28        case 1 : 
 
    29            return callsym; 
 
    30        case 2 : 
 
    31            return constsym; 
 
    32        case 3 : 
 
    33            return dosym; 
 
    34        case 4 : 
 
    35            return endsym; 
 
    36        case 5 : 
 
    37            return ifsym; 
 
    38        case 6 : 
 
    39            return oddsym; 
 
    40        case 7 : 
 
    41            return proceduresym; 
 
    42        case 8 : 
 
    43            return readsym; 
 
    44        case 9 : 
 
    45            return thensym; 
 
    46        case 10 : 
 
    47            return varsym; 
 
    48        case 11 : 
 
    49            return whilesym; 
 
    50        case 12 : 
 
    51            return writesym; 
 
    52        default : 
 
    53            error(">>> Error : "); 
 
    54            return nul; 
 
    55    }  
 
    56} 
 
    57  
 
    58int  error() { 
 
    59    fprintf(out, "|%4d |%19s |%19s %19s |\n", token_num, id, "nul", "invalid character"); 
 
    60    return 0; 
 
    61}
 
[文件] main.c ~ 2KB    下载(22)view source
print?
 
    01#include "mylex.h" 
 
    02  
 
    03extern  int cur_line, cur_col, err_line, err_col; 
 
    04extern  char id[21]; 
 
    05extern  int token_num; 
 
    06  
 
    07FILE  *source, *out; 
 
    08  
 
    09int  main() { 
 
    10    clock_t start, finish; 
 
    11    double duration; 
 
    12    char filename[21] = "\0"; 
 
    13    char outname[21] = "\0"; 
 
    14    int flag; 
 
    15    FILE *stream; 
 
    16    char line[100]; 
 
    17    printf("PL/0 Lexical Analyzer\n\n"); 
 
    18    printf("Please enter the source file: "); 
 
    19    gets_s(filename, 20); 
 
    20    if ((source = fopen(filename, "r")) == NULL) {  
 
    21        printf("Error: the file \"%s\" can not be opened, press any key to exit\n", filename); 
 
    22        _getch(); 
 
    23        return 1; 
 
    24    }  else { 
 
    25        int i; 
 
    26        start = clock();  
 
    27        for (i = 0; i < (int)strlen(filename) - 4; ++i) 
 
    28            outname[i] = filename[i]; 
 
    29        strcat(outname, ".out"); 
 
    30        if ((out = fopen(outname, "w+")) == NULL) {  
 
    31            printf("Error: can not create \"%s\" in current path, press any key to exit\n", outname); 
 
    32            _getch(); 
 
    33            return 1; 
 
    34        } 
 
    35        cur_line = 1; 
 
    36        cur_col = 0; 
 
    37        token_num = 0; 
 
    38        fprintf(out, "+-----+--------------------+--------------------+--------------------+\n"); 
 
    39        fprintf(out, "|%4s |%19s |%19s |%19s |\n", "No", "Symbol", "Symbol Type",  "Notes"); 
 
    40        fprintf(out, "+-----+--------------------+--------------------+--------------------+\n"); 
 
    41        while (!feof(source)) { 
 
    42            token_num++; 
 
    43            flag = getsym(); 
 
    44            if (flag == 0) 
 
    45                continue; 
 
    46            w2f(flag); 
 
    47        } 
 
    48        fprintf(out, "+-----+--------------------+--------------------+--------------------+\n"); 
 
    49        fcloseall(); 
 
    50        stream = fopen(outname, "r"); 
 
    51        while (1) { 
 
    52            fgets(line, 100, stream); 
 
    53            if (feof(stream)) 
 
    54                break; 
 
    55            else
 
    56                printf("%s", line); 
 
    57        } 
 
    58        fclose(stream); 
 
    59        finish = clock();  
 
    60        duration = (double)(finish - start) / CLOCKS_PER_SEC; 
 
    61        printf("completed successfully. (%2.3f secs)\n", duration); 
 
    62    }  
 
    63    return 0; 
 
     }