src/expat/xmlwf/xmlmime.c

#include <string.h>
#include "xmlmime.h"

static const char *
getTok(const char **pp)
{
  /* inComment means one level of nesting; inComment+1 means two levels etc */
  enum { inAtom, inString, init, inComment };
  int state = init;
  const char *tokStart = 0;
  for (;;) {
    switch (**pp) {
    case '\0':
      if (state == inAtom)
        return tokStart;
      return 0;
    case ' ':
    case '\r':
    case '\t':
    case '\n':
      if (state == inAtom)
        return tokStart;
      break;
    case '(':
      if (state == inAtom)
        return tokStart;
      if (state != inString)
        state++;
      break;
    case ')':
      if (state > init)
        --state;
      else if (state != inString)
        return 0;
      break;
    case ';':
    case '/':
    case '=':
      if (state == inAtom)
        return tokStart;
      if (state == init)
        return (*pp)++;
      break;
    case '\\':
      ++*pp;
      if (**pp == '\0')
        return 0;
      break;
    case '"':
      switch (state) {
      case inString:
        ++*pp;
        return tokStart;
      case inAtom:
        return tokStart;
      case init:
        tokStart = *pp;
        state = inString;
        break;
      }
      break;
    default:
      if (state == init) {
        tokStart = *pp;
        state = inAtom;
      }
      break;
    }
    ++*pp;
  }
  /* not reached */
}

/* key must be lowercase ASCII */

static int
matchkey(const char *start, const char *end, const char *key)
{
  if (!start)
    return 0;
  for (; start != end; start++, key++)
    if (*start != *key && *start != 'A' + (*key - 'a'))
      return 0;
  return *key == '\0';
}

void
getXMLCharset(const char *buf, char *charset)
{
  const char *next, *p;

  charset[0] = '\0';
  next = buf;
  p = getTok(&next);
  if (matchkey(p, next, "text"))
    strcpy(charset, "us-ascii");
  else if (!matchkey(p, next, "application"))
    return;
  p = getTok(&next);
  if (!p || *p != '/')
    return;
  p = getTok(&next);
#if 0
  if (!matchkey(p, next, "xml") && charset[0] == '\0')
    return;
#endif
  p = getTok(&next);
  while (p) {
    if (*p == ';') {
      p = getTok(&next);
      if (matchkey(p, next, "charset")) {
        p = getTok(&next);
        if (p && *p == '=') {
          p = getTok(&next);
          if (p) {
            char *s = charset;
            if (*p == '"') {
              while (++p != next - 1) {
                if (*p == '\\')
                  ++p;
                if (s == charset + CHARSET_MAX - 1) {
                  charset[0] = '\0';
                  break;
                }
                *s++ = *p;
              }
              *s++ = '\0';
            }
            else {
              if (next - p > CHARSET_MAX - 1)
                break;
              while (p != next)
                *s++ = *p++;
              *s = 0;
              break;
            }
          }
        }
        break;
      }
    }
  else
    p = getTok(&next);
  }
}

#ifdef TEST

#include <stdio.h>

int
main(int argc, char *argv[])
{
  char buf[CHARSET_MAX];
  if (argc <= 1)
    return 1;
  printf("%s\n", argv[1]);
  getXMLCharset(argv[1], buf);
  printf("charset=\"%s\"\n", buf);
  return 0;
}

#endif /* TEST */
Commit	Line	Data
	1	#include <string.h>
	2	#include "xmlmime.h"
	3
	4	static const char *
	5	getTok(const char **pp)
	6	{
	7	/* inComment means one level of nesting; inComment+1 means two levels etc */
	8	enum { inAtom, inString, init, inComment };
	9	int state = init;
	10	const char *tokStart = 0;
	11	for (;;) {
	12	switch (**pp) {
	13	case '\0':
	14	if (state == inAtom)
	15	return tokStart;
	16	return 0;
	17	case ' ':
	18	case '\r':
	19	case '\t':
	20	case '\n':
	21	if (state == inAtom)
	22	return tokStart;
	23	break;
	24	case '(':
	25	if (state == inAtom)
	26	return tokStart;
	27	if (state != inString)
	28	state++;
	29	break;
	30	case ')':
	31	if (state > init)
	32	--state;
	33	else if (state != inString)
	34	return 0;
	35	break;
	36	case ';':
	37	case '/':
	38	case '=':
	39	if (state == inAtom)
	40	return tokStart;
	41	if (state == init)
	42	return (*pp)++;
	43	break;
	44	case '\\':
	45	++*pp;
	46	if (**pp == '\0')
	47	return 0;
	48	break;
	49	case '"':
	50	switch (state) {
	51	case inString:
	52	++*pp;
	53	return tokStart;
	54	case inAtom:
	55	return tokStart;
	56	case init:
	57	tokStart = *pp;
	58	state = inString;
	59	break;
	60	}
	61	break;
	62	default:
	63	if (state == init) {
	64	tokStart = *pp;
	65	state = inAtom;
	66	}
	67	break;
	68	}
	69	++*pp;
	70	}
	71	/* not reached */
	72	}
	73
	74	/* key must be lowercase ASCII */
	75
	76	static int
	77	matchkey(const char start, const char end, const char *key)
	78	{
	79	if (!start)
	80	return 0;
	81	for (; start != end; start++, key++)
	82	if (start != key && start != 'A' + (key - 'a'))
	83	return 0;
	84	return *key == '\0';
	85	}
	86
	87	void
	88	getXMLCharset(const char buf, char charset)
	89	{
	90	const char next, p;
	91
	92	charset[0] = '\0';
	93	next = buf;
	94	p = getTok(&next);
	95	if (matchkey(p, next, "text"))
	96	strcpy(charset, "us-ascii");
	97	else if (!matchkey(p, next, "application"))
	98	return;
	99	p = getTok(&next);
	100	if (!p \|\| *p != '/')
	101	return;
	102	p = getTok(&next);
	103	#if 0
	104	if (!matchkey(p, next, "xml") && charset[0] == '\0')
	105	return;
	106	#endif
	107	p = getTok(&next);
	108	while (p) {
	109	if (*p == ';') {
	110	p = getTok(&next);
	111	if (matchkey(p, next, "charset")) {
	112	p = getTok(&next);
	113	if (p && *p == '=') {
	114	p = getTok(&next);
	115	if (p) {
	116	char *s = charset;
	117	if (*p == '"') {
	118	while (++p != next - 1) {
	119	if (*p == '\\')
	120	++p;
	121	if (s == charset + CHARSET_MAX - 1) {
	122	charset[0] = '\0';
	123	break;
	124	}
	125	s++ = p;
	126	}
	127	*s++ = '\0';
	128	}
	129	else {
	130	if (next - p > CHARSET_MAX - 1)
	131	break;
	132	while (p != next)
	133	s++ = p++;
	134	*s = 0;
	135	break;
	136	}
	137	}
	138	}
	139	break;
	140	}
	141	}
	142	else
	143	p = getTok(&next);
	144	}
	145	}
	146
	147	#ifdef TEST
	148
	149	#include <stdio.h>
	150
	151	int
	152	main(int argc, char *argv[])
	153	{
	154	char buf[CHARSET_MAX];
	155	if (argc <= 1)
	156	return 1;
	157	printf("%s\n", argv[1]);
	158	getXMLCharset(argv[1], buf);
	159	printf("charset=\"%s\"\n", buf);
	160	return 0;
	161	}
	162
	163	#endif /* TEST */