/*
	src.c	WJ100
*/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#define COL_END			"\x1b[33;1m"
#define COL_TOKEN		"\x1b[0;1m"
#define COL_KEYWORD		"\x1b[0;1m"
#define COL_FUNC		"\x1b[0;1m"
#define COL_QUOTE		"\x1b[0;1m"
#define COL_STRING		"\x1b[32;1m"
#define COL_COMMENT		"\x1b[35;1m"
#define COL_DIRECTIVE	"\x1b[32;1m"
#define COL_WEIRDCHAR	"\x1b[31;1m"
#define COL_HEX			"\x1b[31;1m"
#define COL_OCTAL		"\x1b[35;1m"
#define COL_NUMBER		"\x1b[36;1m"

char *c_keywords[] = {
	"typedef",
	"struct",
	"union",
	"void",
	"int",
	"char",
	"long",
	"signed",
	"unsigned",
	"const",
	"volatile",
	"static",
	"extern",
	"if",
	"else",
	"while",
	"for",
	"switch",
	"case",
	"default",
	"break",
	"return",

	NULL
};

char *c_funcs[] = {
	"main",

	"printf",
	"fprintf",
	"sprintf",
	"vprintf",
	"vfprintf",
	"vsprintf",

	"scanf",
	"sscanf",
	"fscanf",
	"vscanf",
	"vfscanf",
	"vsscanf",

	"malloc",
	"free",

	"strcpy",
	"strdup",
	"strcat",
	"strncpy",
	"strncat",
	"strcmp",
	"stricmp",
	"strcasecmp",
	"strncmp",
	"strnicmp",
	"strncasecmp",
	"strchr",
	"strrchr",
	"index",
	"rindex",
	"strstr",
	"strspn",
	"strtok",
	"memcpy",
	"memset",
	"bzero",

	"atoi",
	"atol",
	"strtoul",
	"min",
	"max",

	"fopen",
	"fclose",
	"fread",
	"fwrite",
	"fgets",
	"fgetc",
	"fputc",

	"putc",
	"getc",
	"getch",
	"puts",
	"gets",

	"open",
	"close",
	"read",
	"write",

	"dup",
	"dup2",

	"ioctl",
	"fcntl",

	"time",
	"localtime",
	"gmtime",
	"ctime",
	"strftime",

	"stat",
	"fstat",
	"lstat",

	"socket",
	"bind",
	"connect",
	"listen",

	"perror",
	"strerror",

/* variables */
	"errno",
	"NULL",

/* types */
	"size_t",
	"mode_t",
	"time_t",
	"socklen_t",

	NULL
};


int in_comment = 0, in_string = 0, in_quote = 0, in_directive = 0, in_function = 0;

void print_color(char *color) {
	if (in_comment || in_string || in_directive)
		return;

	printf("%s", color);
}


int in_array(char *buf, char **arr) {
int i;

	for(i = 0; arr[i] != NULL; i++) {
		if (!strcmp(buf, arr[i]))
			return 1;
	}
	return 0;
}


int is_keyword(char *buf) {
	return in_array(buf, c_keywords);
}

int is_func(char *buf) {
	return in_array(buf, c_funcs);
}


int is_hexnumber(char *buf) {
	if (*buf == '0')
		buf++;
	else
		return 0;

	if (*buf == 'x')
		buf++;
	else
		return 0;

	if (!*buf)
		return 0;

	while(*buf) {
		if (strchr("0123456789abcdefABCDEF", *buf) == NULL) {
			if (*buf == 'u' || *buf == 'U')
				buf++;
			if (*buf == 'l' || *buf == 'L')
				buf++;
			if (!*buf)
				return 1;
			return 0;
		}
		buf++;
	}
	return 1;
}

int is_octalnumber(char *buf) {
	if (*buf != '0')
		return 0;

	buf++;
	if (!*buf)
		return 0;

	while(*buf) {
		if (strchr("01234567", *buf) == NULL) {
			if (*buf == 'u' || *buf == 'U')
				buf++;
			if (*buf == 'l' || *buf == 'L')
				buf++;
			if (!*buf)
				return 1;
			return 0;
		}
		buf++;
	}
	return 1;
}

int is_number(char *buf) {
	if (!*buf || strchr("0123456789", *buf) == NULL)
		return 0;

/*
	the extensive check for '0' or '0[UL]' is needed for octal numbers...
	if the number is '0123', it is an octal number
*/
	if (*buf == '0') {
		buf++;
		if (!*buf)
			return 1;
		else {
			if (*buf == 'u' || *buf == 'U') {
				buf++;
				if (!*buf)
					return 1;
			}
			if (*buf == 'l' || *buf == 'L') {
				buf++;
				if (!*buf)
					return 1;
			}
			return 0;
		}
	}
	while(*buf) {
		if (strchr("0123456789", *buf) == NULL) {
			if (*buf == 'u' || *buf == 'U')
				buf++;
			if (*buf == 'l' || *buf == 'L')
				buf++;
			if (!*buf)
				return 1;
			return 0;
		}
		buf++;
	}
	return 1;
}


char *my_strtok(char *str, char *delim) {
	while(*str) {
		if (strchr(delim, *str) != NULL)
			return str;
		str++;
	}
	return NULL;
}


void color_src(char *filename) {
char buf[256], *p, last_token = 0, *endp, kar;
int end_color = 0, was_comment = 0;
FILE *f;

	if ((f = fopen(filename, "r")) == NULL) {
		fprintf(stderr, "failed to open file '%s'\n", filename);
		return;
	}
	printf("\n"
		"/*** --8<-- %s --8<-- ***/\n"
		"\n", filename);

	while(fgets(buf, 256, f) != NULL) {
		p = buf;
		while(*p) {
			if (!in_comment) {
				if ((p == buf || strchr(" \t-=~!%^&*()+|{}[]:;<>,.?/", last_token) != NULL)
					&& (endp = my_strtok(p, " \t-=~!%^&*()+|{}[]:;<>,.?/")) != NULL) {
					kar = *endp;
					*endp = 0;

					if (*p) {
						if (is_keyword(p)) {
							print_color(COL_KEYWORD);
							printf("%s", p);
							print_color(COL_END);
							*endp = kar;
							p = endp;
							continue;
						}
						if (is_func(p)) {
							print_color(COL_FUNC);
							printf("%s", p);
							print_color(COL_END);
							*endp = kar;
							p = endp;
							continue;
						}
						if (is_number(p)) {
							print_color(COL_NUMBER);
							printf("%s", p);
							print_color(COL_END);
							*endp = kar;
							p = endp;
							continue;
						}
						if (is_hexnumber(p)) {
							print_color(COL_HEX);
							printf("%s", p);
							print_color(COL_END);
							*endp = kar;
							p = endp;
							continue;
						}
						if (is_octalnumber(p)) {
							print_color(COL_OCTAL);
							printf("%s", p);
							print_color(COL_END);
							*endp = kar;
							p = endp;
							continue;
						}
					}
					*endp = kar;
				}
			}
			switch(*p) {
				case '\\':
					printf("%c", *p);
					p++;
					break;

				case '\'':
					if (in_string || in_comment)
						break;

					if (in_quote) {
						in_quote--;
						end_color = 1;
					} else {
						print_color(COL_QUOTE);
						in_quote++;
					}
					break;

				case '"':
					if (in_quote)
						break;

					if (in_string) {
						if (last_token != '\\') {
							in_string--;
							end_color = 1;
						}
					} else {
						print_color(COL_STRING);
						in_string++;
					}
					break;

				case '/':
					if (!in_string && !in_quote && p[1] == '*') {
						print_color(COL_COMMENT);
						in_comment++;
					}
					if (!in_comment && was_comment) {
						was_comment = 0;
						end_color = 1;
					}
					break;

				case '*':
					if (!in_string && !in_quote) {
						if (in_comment) {
							if (p[1] == '/') {
								in_comment--;
								if (!in_comment)
									was_comment = 1;
							}
						} else {
							print_color(COL_TOKEN);
							end_color = 1;
						}
					}
					break;

				case '#':
					if (!in_string && !in_quote) {
						print_color(COL_DIRECTIVE);
						in_directive++;
					}
					break;

/*
				case '{':
					if (!in_string && !in_quote)
						in_function++;
					break;

				case '}':
					if (!in_string && !in_quote)
						in_function--;
					break;
*/

				case '\t':
					*p = ' ';
					printf("   ");
					break;

				default:
					if (*p < ' ' && *p != '\r' && *p != '\n' && *p != '\t') {
						print_color(COL_WEIRDCHAR);
						printf("[%02x", *p);
						*p = ']';
						end_color = 1;
						break;
					}
					if (in_string || in_quote)
						break;

					if (strchr("-=~!%^&*()+|{}[]:;<>,.?/", *p) != NULL) {
						print_color(COL_TOKEN);
						end_color = 1;
					}
			}
			last_token = *p;
			if (*p) {
				printf("%c", *p);
				p++;
			}
			if (end_color) {
				print_color(COL_END);
				end_color = 0;
			}
		}
		if (in_directive && last_token != '\\') {
			in_directive = 0;
			print_color(COL_END);
		}
		if (in_string && last_token != '\\') {
			in_string = 0;
			print_color(COL_END);
		}
		if (!in_comment && !in_string && !in_directive)
			print_color(COL_END);

/* --More-- prompt
		line_num++;
		if (line_num > 22) {
			printf("--More--"); fflush(stdout);
			if (fgets(buf, 4096, stdin) == NULL)
				break;

			if (!strcmp(buf, "q") || !strcmp(buf, "Q"))
				break;

			line_num = 0;
		}
*/
	}
	fclose(f);
}


int main(int argc, char **argv) {
int i;

	if (argc <= 1) {
		fprintf(stderr, "usage: src <file.c>\n"
			"Example: src hello.c | less -r\n");
		return 1;
	}
	for(i = 1; i < argc; i++)
		color_src(argv[i]);
	return 0;
}

/* EOB */