htstring.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, lynx比elinks早的多, 目前好像停止开发, 这是lynx源代码

    if (!delims) delims = " ;,=" ;
    if (!bracks) bracks = "<\"" ;

    get_blanks = (BOOL) (!strchr(delims,' ') && !strchr(bracks,' '));
    get_comments = (BOOL) (strchr(bracks,'(') != NULL);
    skip_comments = (BOOL) (!get_comments && !strchr(delims,'(') && !get_blanks);
#define skipWHITE(c) (!get_blanks && WHITE(c))

    while (*p && skipWHITE(*p))
	p++;				/* Strip white space */
    if (!*p) {
	*pstr = p;
	if (found) *found = '\0';
	return NULL;		/* No first field */
    }
    while (1) {
	/* Strip white space and other delimiters */
	while (*p && (skipWHITE(*p) || strchr(delims,*p))) p++;
	if (!*p) {
	    *pstr = p;
	    if (found) *found = *(p-1);
	    return NULL;					 /* No field */
	}

	if (*p == '(' && (skip_comments || get_comments)) {	  /* Comment */
	    int comment_level = 0;
	    if (get_comments && !start) start = p+1;
	    for(;*p && (*p!=')' || --comment_level>0); p++) {
		if (*p == '(') comment_level++;
		else if (*p == '"') {	      /* quoted field within Comment */
		    for(p++; *p && *p!='"'; p++)
			if (*p == '\\' && *(p+1)) p++; /* Skip escaped chars */
		    if (!*p) break; /* (invalid) end of string found, leave */
		}
		if (*p == '\\' && *(p+1)) p++;	       /* Skip escaped chars */
	    }
	    if (get_comments)
		break;
	    if (*p) p++;
	    if (get_closing_char_too) {
		if (!*p || (!strchr(bracks,*p) && strchr(delims,*p))) {
		    break;
		} else
		    get_closing_char_too = (BOOL) (strchr(bracks,*p) != NULL);
	    }
	} else if (strchr(bracks,*p)) {	       /* quoted or bracketed field */
	    switch (*p) {
	       case '<': closer = '>'; break;
	       case '[': closer = ']'; break;
	       case '{': closer = '}'; break;
	       case ':': closer = ';'; break;
	    default:	 closer = *p;
	    }
	    if (!start) start = ++p;
	    for(;*p && *p!=closer; p++)
		if (*p == '\\' && *(p+1)) p++;	       /* Skip escaped chars */
	    if (get_closing_char_too) {
		p++;
		if (!*p || (!strchr(bracks,*p) && strchr(delims,*p))) {
		    break;
		} else
		    get_closing_char_too = (BOOL) (strchr(bracks,*p) != NULL);
	    } else
	    break;			    /* kr95-10-9: needs to stop here */
	} else {					      /* Spool field */
	    if (!start) start = p;
	    while(*p && !skipWHITE(*p) && !strchr(bracks,*p) &&
					  !strchr(delims,*p))
		p++;
	    if (*p && strchr(bracks,*p)) {
		get_closing_char_too = TRUE;
	    } else {
		if (*p=='(' && skip_comments) {
		    *pstr = p;
		    HTNextTok(pstr, NULL, "(", found);	/*	Advance pstr */
		    *p = '\0';
		    if (*pstr && **pstr) (*pstr)++;
		    return start;
		}
		    break;					   /* Got it */
	    }
	}
    }
    if (found) *found = *p;

    if (*p) *p++ = '\0';
    *pstr = p;
    return start;
}

PRIVATE char *HTAlloc ARGS2(char *, ptr, size_t, length)
{
    if (ptr != 0)
	ptr = (char *)realloc(ptr, length);
    else
	ptr = (char *)malloc(length);
    if (ptr == 0)
	outofmem(__FILE__, "HTAlloc");
    return ptr;
}

/*
 * If SAVE_TIME_NOT_SPACE is defined, StrAllocVsprintf will hang on to
 * its temporary string buffers instead of allocating and freeing them
 * in each invocation.  They only grow and never shrink, and won't be
 * cleaned up on exit. - kw
 */
#if defined(_REENTRANT) || defined(_THREAD_SAFE) || defined(LY_FIND_LEAKS)
#undef SAVE_TIME_NOT_SPACE
#endif

/*
 * Replacement for sprintf, allocates buffer on the fly according to what's
 * needed for its arguments.  Unlike sprintf, this always concatenates to the
 * destination buffer, so we do not have to provide both flavors.
 */
typedef enum { Flags, Width, Prec, Type, Format } PRINTF;

#define VA_INTGR(type) ival = va_arg((*ap), type)
#define VA_FLOAT(type) fval = va_arg((*ap), type)
#define VA_POINT(type) pval = (void *)va_arg((*ap), type)

#define NUM_WIDTH 10	/* allow for width substituted for "*" in "%*s" */
		/* also number of chars assumed to be needed in addition
		   to a given precision in floating point formats */

#define GROW_EXPR(n) (((n) * 3) / 2)
#define GROW_SIZE 256

PUBLIC_IF_FIND_LEAKS char * StrAllocVsprintf ARGS4(
	char **,	pstr,
	size_t,		dst_len,
	CONST char *,	fmt,
	va_list *,	ap)
{
#ifdef SAVE_TIME_NOT_SPACE
    static size_t tmp_len = 0;
    static size_t fmt_len = 0;
    static char *tmp_ptr = NULL;
    static char *fmt_ptr = NULL;
#else
    size_t tmp_len = GROW_SIZE;
    char *tmp_ptr = 0;
    char *fmt_ptr;
#endif /* SAVE_TIME_NOT_SPACE */
    size_t have, need;
    char *dst_ptr = *pstr;
    CONST char *format = fmt;

    if (fmt == 0 || *fmt == '\0')
	return 0;

#ifdef USE_VASPRINTF
    if (pstr && !dst_len) {
	if (*pstr)
	    FREE(*pstr);
	if (vasprintf(pstr, fmt, *ap) >= 0) {
	    mark_malloced(*pstr, strlen(*pstr)+1);
	    return(*pstr);
	}
    }
#endif /* USE_VASPRINTF */

    need = strlen(fmt) + 1;
#ifdef SAVE_TIME_NOT_SPACE
    if (!fmt_ptr || fmt_len < need*NUM_WIDTH) {
	fmt_ptr = HTAlloc(fmt_ptr, fmt_len = need*NUM_WIDTH);
    }
    if (!tmp_ptr || tmp_len < GROW_SIZE) {
	tmp_ptr = HTAlloc(tmp_ptr, tmp_len = GROW_SIZE);
    }
#else
    if ((fmt_ptr = malloc(need*NUM_WIDTH)) == 0
     || (tmp_ptr = malloc(tmp_len)) == 0) {
	outofmem(__FILE__, "StrAllocVsprintf");
    }
#endif /* SAVE_TIME_NOT_SPACE */

    if (dst_ptr == 0) {
	dst_ptr = HTAlloc(dst_ptr, have = GROW_SIZE + need);
    } else {
	have = strlen(dst_ptr) + 1;
	need += dst_len;
	if (have < need)
	    dst_ptr = HTAlloc(dst_ptr, have = GROW_SIZE + need);
    }

    while (*fmt != '\0') {
	if (*fmt == '%') {
	    static char dummy[] = "";
	    PRINTF state = Flags;
	    char *pval   = dummy;	/* avoid const-cast */
	    double fval  = 0.0;
	    int done     = FALSE;
	    int ival     = 0;
	    int prec     = -1;
	    int type     = 0;
	    int used     = 0;
	    int width    = -1;
	    size_t f     = 0;

	    fmt_ptr[f++] = *fmt;
	    while (*++fmt != '\0' && !done) {
		fmt_ptr[f++] = *fmt;

		if (isdigit(UCH(*fmt))) {
		    int num = *fmt - '0';
		    if (state == Flags && num != 0)
			state = Width;
		    if (state == Width) {
			if (width < 0)
			    width = 0;
			width = (width * 10) + num;
		    } else if (state == Prec) {
			if (prec < 0)
			    prec = 0;
			prec = (prec * 10) + num;
		    }
		} else if (*fmt == '*') {
		    VA_INTGR(int);
		    if (state == Flags)
			state = Width;
		    if (state == Width) {
			width = ival;
		    } else if (state == Prec) {
			prec = ival;
		    }
		    sprintf(&fmt_ptr[--f], "%d", ival);
		    f = strlen(fmt_ptr);
		} else if (isalpha(UCH(*fmt))) {
		    done = TRUE;
		    switch (*fmt) {
		    case 'Z': /* FALLTHRU */
		    case 'h': /* FALLTHRU */
		    case 'l': /* FALLTHRU */
		    case 'L': /* FALLTHRU */
			done = FALSE;
			type = *fmt;
			break;
		    case 'o': /* FALLTHRU */
		    case 'i': /* FALLTHRU */
		    case 'd': /* FALLTHRU */
		    case 'u': /* FALLTHRU */
		    case 'x': /* FALLTHRU */
		    case 'X': /* FALLTHRU */
			if (type == 'l')
			    VA_INTGR(long);
			else if (type == 'Z')
			    VA_INTGR(size_t);
			else
			    VA_INTGR(int);
			used = 'i';
			break;
		    case 'f': /* FALLTHRU */
		    case 'e': /* FALLTHRU */
		    case 'E': /* FALLTHRU */
		    case 'g': /* FALLTHRU */
		    case 'G': /* FALLTHRU */
			    VA_FLOAT(double);
			used = 'f';
			break;
		    case 'c':
			VA_INTGR(int);
			used = 'c';
			break;
		    case 's':
			VA_POINT(char *);
			if (prec < 0)
			    prec = strlen(pval);
			used = 's';
			break;
		    case 'p':
			VA_POINT(void *);
			used = 'p';
			break;
		    case 'n':
			VA_POINT(int *);
			used = 0;
			break;
		    default:
			CTRACE((tfp, "unknown format character '%c' in %s\n",
				    *fmt, format));
			break;
		    }
		} else if (*fmt == '.') {
		    state = Prec;
		} else if (*fmt == '%') {
		    done = TRUE;
		    used = '%';
		}
	    }
	    fmt_ptr[f] = '\0';

	    if (prec > 0) {
		switch (used) {
		case 'f':
		    if (width < prec + NUM_WIDTH)
			width = prec + NUM_WIDTH;
		    /* FALLTHRU */
		case 'i':
		    /* FALLTHRU */
		case 'p':
		    if (width < prec + 2)
			width = prec + 2; /* leading sign/space/zero, "0x" */
		    break;
		case 'c':
		    break;
		case '%':
		    break;
		default:
		    if (width < prec)
			width = prec;
		    break;
		}
	    }
	    if (width >= (int)tmp_len) {
		tmp_len = GROW_EXPR(tmp_len + width);
		tmp_ptr = HTAlloc(tmp_ptr, tmp_len);
	    }

	    switch (used) {
	    case 'i':
	    case 'c':
		sprintf(tmp_ptr, fmt_ptr, ival);
		break;
	    case 'f':
		sprintf(tmp_ptr, fmt_ptr, fval);
		break;
	    default:
		sprintf(tmp_ptr, fmt_ptr, pval);
		break;
	    }
	    need = dst_len + strlen(tmp_ptr) + 1;
	    if (need >= have) {
		dst_ptr = HTAlloc(dst_ptr, have = GROW_EXPR(need));
	    }
	    strcpy(dst_ptr + dst_len, tmp_ptr);
	    dst_len += strlen(tmp_ptr);
	} else {
	    if ((dst_len + 2) >= have) {
		dst_ptr = HTAlloc(dst_ptr, (have += GROW_SIZE));
	    }
	    dst_ptr[dst_len++] = *fmt++;
	}
    }

#ifndef SAVE_TIME_NOT_SPACE
    FREE(tmp_ptr);
    FREE(fmt_ptr);
#endif
    dst_ptr[dst_len] = '\0';
    if (pstr)
	*pstr = dst_ptr;
    return (dst_ptr);
}
#undef SAVE_TIME_NOT_SPACE

/*
 * Replacement for sprintf, allocates buffer on the fly according to what's
 * needed for its arguments.  Unlike sprintf, this always concatenates to the
 * destination buffer.
 */
/* Note: if making changes, also check the memory tracking version
 * LYLeakHTSprintf in LYLeaks.c. - kw */
#ifdef HTSprintf		/* if hidden by LYLeaks stuff */
#undef HTSprintf
#endif
#if ANSI_VARARGS
PUBLIC char * HTSprintf (char ** pstr, CONST char * fmt, ...)
#else
PUBLIC char * HTSprintf (va_alist)
    va_dcl
#endif
{
    char *result = 0;
    size_t inuse = 0;
    va_list ap;

    LYva_start(ap,fmt);
    {
#if !ANSI_VARARGS
	char **		pstr = va_arg(ap, char **);
	CONST char *	fmt  = va_arg(ap, CONST char *);
#endif
	if (pstr != 0 && *pstr != 0)
	    inuse = strlen(*pstr);
	result = StrAllocVsprintf(pstr, inuse, fmt, &ap);
    }
    va_end(ap);

    return (result);
}

/*
 * Replacement for sprintf, allocates buffer on the fly according to what's
 * needed for its arguments.  Like sprintf, this always resets the destination
 * buffer.
 */
/* Note: if making changes, also check the memory tracking version
 * LYLeakHTSprintf0 in LYLeaks.c. - kw */
#ifdef HTSprintf0		/* if hidden by LYLeaks stuff */
#undef HTSprintf0
#endif
#if ANSI_VARARGS
PUBLIC char * HTSprintf0 (char ** pstr, CONST char * fmt, ...)
#else
PUBLIC char * HTSprintf0 (va_alist)
    va_dcl
#endif
{
    char *result = 0;
    va_list ap;

    LYva_start(ap,fmt);
    {
#if !ANSI_VARARGS
	char **		pstr = va_arg(ap, char **);
	CONST char *	fmt  = va_arg(ap, CONST char *);
#endif
#ifdef USE_VASPRINTF
	if (pstr) {
	    if (*pstr)
		FREE(*pstr);
	    if (vasprintf(pstr, fmt, ap) >= 0) /* else call outofmem?? */
		mark_malloced(*pstr, strlen(*pstr)+1);
	    result = *pstr;
	} else
#endif /* USE_VASPRINTF */
	result = StrAllocVsprintf(pstr, 0, fmt, &ap);
    }
    va_end(ap);

    return (result);