htmime.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

/*			MIME Message Parse			HTMIME.c
**			==================
**
**	This is RFC 1341-specific code.
**	The input stream pushed into this parser is assumed to be
**	stripped on CRs, ie lines end with LF, not CR LF.
**	(It is easy to change this except for the body part where
**	conversion can be slow.)
**
** History:
**	   Feb 92	Written Tim Berners-Lee, CERN
**
*/
#include <HTUtils.h>
#include <HTMIME.h>		/* Implemented here */
#include <HTTP.h>		/* for redirecting_url */
#include <HTAlert.h>
#include <HTCJK.h>
#include <UCMap.h>
#include <UCDefs.h>
#include <UCAux.h>

#include <LYCookie.h>
#include <LYCharSets.h>
#include <LYCharUtils.h>
#include <LYStrings.h>
#include <LYUtils.h>
#include <LYLeaks.h>

/*		MIME Object
**		-----------
*/

typedef enum {
	MIME_TRANSPARENT,	/* put straight through to target ASAP! */
	miBEGINNING_OF_LINE,	/* first character and not a continuation */
	miA,
	miACCEPT_RANGES,
	miAGE,
	miAL,
	miALLOW,
	miALTERNATES,
	miC,
	miCACHE_CONTROL,
	miCO,
	miCOOKIE,
	miCON,
	miCONNECTION,
	miCONTENT_,
	miCONTENT_BASE,
	miCONTENT_DISPOSITION,
	miCONTENT_ENCODING,
	miCONTENT_FEATURES,
	miCONTENT_L,
	miCONTENT_LANGUAGE,
	miCONTENT_LENGTH,
	miCONTENT_LOCATION,
	miCONTENT_MD5,
	miCONTENT_RANGE,
	miCONTENT_T,
	miCONTENT_TRANSFER_ENCODING,
	miCONTENT_TYPE,
	miDATE,
	miE,
	miETAG,
	miEXPIRES,
	miKEEP_ALIVE,
	miL,
	miLAST_MODIFIED,
	miLINK,
	miLOCATION,
	miP,
	miPR,
	miPRAGMA,
	miPROXY_AUTHENTICATE,
	miPUBLIC,
	miR,
	miRE,
	miREFRESH,
	miRETRY_AFTER,
	miS,
	miSAFE,
	miSE,
	miSERVER,
	miSET_COOKIE,
	miSET_COOKIE1,
	miSET_COOKIE2,
	miT,
	miTITLE,
	miTRANSFER_ENCODING,
	miU,
	miUPGRADE,
	miURI,
	miV,
	miVARY,
	miVIA,
	miW,
	miWARNING,
	miWWW_AUTHENTICATE,
	miSKIP_GET_VALUE,	/* Skip space then get value */
	miGET_VALUE,		/* Get value till white space */
	miJUNK_LINE,		/* Ignore the rest of this folded line */
	miNEWLINE,		/* Just found a LF .. maybe continuation */
	miCHECK,		/* check against check_pointer */
	MIME_NET_ASCII,		/* Translate from net ascii */
	MIME_IGNORE		/* Ignore entire file */
	/* TRANSPARENT and IGNORE are defined as stg else in _WINDOWS */
} MIME_state;

#define VALUE_SIZE 5120		/* @@@@@@@ Arbitrary? */
struct _HTStream {
	CONST HTStreamClass *	isa;

	BOOL			net_ascii;	/* Is input net ascii? */
	MIME_state		state;		/* current state */
	MIME_state		if_ok;		/* got this state if match */
	MIME_state		field;		/* remember which field */
	MIME_state		fold_state;	/* state on a fold */
	BOOL			head_only;	/* only parsing header */
	BOOL			pickup_redirection; /* parsing for location */
	BOOL			no_streamstack; /* use sink directly */
	CONST char *		check_pointer;	/* checking input */

	char *			value_pointer;	/* storing values */
	char			value[VALUE_SIZE];

	HTParentAnchor *	anchor;		/* Given on creation */
	HTStream *		sink;		/* Given on creation */

	char *			boundary;	/* For multipart */
	char *			set_cookie;	/* Set-Cookie */
	char *			set_cookie2;	/* Set-Cookie2 */
	char *			location;	/* Location */

	char *			refresh_url;	/* "Refresh:" URL */

	HTFormat		encoding;	/* Content-Transfer-Encoding */
	char *			compression_encoding;
	HTFormat		format;		/* Content-Type */
	HTStream *		target;		/* While writing out */
	HTStreamClass		targetClass;

	HTAtom *		targetRep;	/* Converting into? */
};

/*
**  This function is for trimming off any paired
**  open- and close-double quotes from header values.
**  It does not parse the string for embedded quotes,
**  and will not modify the string unless both the
**  first and last characters are double-quotes. - FM
*/
PUBLIC void HTMIME_TrimDoubleQuotes ARGS1(
	char *,		value)
{
    int i;
    char *cp = value;

    if (!(cp && *cp) || *cp != '\"')
	return;

    i = strlen(cp);
    if (cp[(i - 1)] != '\"')
	return;
    else
	cp[(i - 1)] = '\0';

    for (i = 0; value[i]; i++)
	value[i] = cp[(i +1)];
}

PRIVATE BOOL content_is_compressed ARGS1(HTStream *, me)
{
    char *encoding = me->anchor->content_encoding;

    return encoding != 0
        && strcmp(encoding, "8bit") != 0
	&& strcmp(encoding, "7bit") != 0
	&& strcmp(encoding, "binary") != 0;
}

/*
 * Strip quotes from a refresh-URL.
 */
PRIVATE void dequote ARGS1(char *, url)
{
    int len;

    len = strlen(url);
    if (*url == '\'' && len > 1 && url[len-1] == url[0]) {
	url[len-1] = '\0';
	while ((url[0] = url[1]) != '\0') {
	    ++url;
	}
    }
}

PRIVATE int pumpData ARGS1(HTStream *, me)
{
    if (strchr(HTAtom_name(me->format), ';') != NULL) {
	char *cp = NULL, *cp1, *cp2, *cp3 = NULL, *cp4;

	CTRACE((tfp, "HTMIME: Extended MIME Content-Type is %s\n",
		HTAtom_name(me->format)));
	StrAllocCopy(cp, HTAtom_name(me->format));
	/*
	** Note that the Content-Type value was converted
	** to lower case when we loaded into me->format,
	** but there may have been a mixed or upper-case
	** atom, so we'll force lower-casing again.  We
	** also stripped spaces and double-quotes, but
	** we'll make sure they're still gone from any
	** charset parameter we check.  - FM
	*/
	LYLowerCase(cp);
	if ((cp1 = strchr(cp, ';')) != NULL) {
	    BOOL chartrans_ok = NO;
	    if ((cp2 = strstr(cp1, "charset")) != NULL) {
		int chndl;

		cp2 += 7;
		while (*cp2 == ' ' || *cp2 == '=' || *cp2 == '\"')
		    cp2++;
		StrAllocCopy(cp3, cp2); /* copy to mutilate more */
		for (cp4 = cp3; (*cp4 != '\0' && *cp4 != '\"' &&
				 *cp4 != ';'  && *cp4 != ':' &&
				 !WHITE(*cp4));	cp4++)
		    ; /* do nothing */
		*cp4 = '\0';
		cp4 = cp3;
		chndl = UCGetLYhndl_byMIME(cp3);
		if (UCCanTranslateFromTo(chndl,
					 current_char_set)) {
		    chartrans_ok = YES;
		    *cp1 = '\0';
		    me->format = HTAtom_for(cp);
		    StrAllocCopy(me->anchor->charset, cp4);
		    HTAnchor_setUCInfoStage(me->anchor, chndl,
					    UCT_STAGE_MIME,
					    UCT_SETBY_MIME);
		}
		else if (chndl < 0) {/* got something but we don't
					recognize it */
		    chndl = UCLYhndl_for_unrec;
		    if (chndl < 0)
			/*
			**  UCLYhndl_for_unrec not defined :-(
			**  fallback to UCLYhndl_for_unspec
			**  which always valid.
			*/
			chndl = UCLYhndl_for_unspec;  /* always >= 0 */
		    if (UCCanTranslateFromTo(chndl,
					     current_char_set)) {
			chartrans_ok = YES;
			*cp1 = '\0';
			me->format = HTAtom_for(cp);
			HTAnchor_setUCInfoStage(me->anchor, chndl,
						UCT_STAGE_MIME,
						UCT_SETBY_DEFAULT);
		    }
		}
		if (chartrans_ok) {
		    LYUCcharset * p_in =
			HTAnchor_getUCInfoStage(me->anchor,
						UCT_STAGE_MIME);
		    LYUCcharset * p_out =
			HTAnchor_setUCInfoStage(me->anchor,
						current_char_set,
						UCT_STAGE_HTEXT,
						UCT_SETBY_DEFAULT);
		    if (!p_out)
			/*
			**	Try again.
			*/
			p_out =
			    HTAnchor_getUCInfoStage(me->anchor,
						    UCT_STAGE_HTEXT);

		    if (!strcmp(p_in->MIMEname,
				"x-transparent")) {
			HTPassEightBitRaw = TRUE;
			HTAnchor_setUCInfoStage(me->anchor,
						HTAnchor_getUCLYhndl(me->anchor,
								     UCT_STAGE_HTEXT),
						UCT_STAGE_MIME,
						UCT_SETBY_DEFAULT);
		    }
		    if (!strcmp(p_out->MIMEname,
				"x-transparent")) {
			HTPassEightBitRaw = TRUE;
			HTAnchor_setUCInfoStage(me->anchor,
						HTAnchor_getUCLYhndl(me->anchor,
								     UCT_STAGE_MIME),
						UCT_STAGE_HTEXT,
						UCT_SETBY_DEFAULT);
		    }
		    if (p_in->enc != UCT_ENC_CJK) {
			HTCJK = NOCJK;
			if (!(p_in->codepoints &
			      UCT_CP_SUBSETOF_LAT1) &&
			    chndl == current_char_set) {
			    HTPassEightBitRaw = TRUE;
			}
		    } else if (p_out->enc == UCT_ENC_CJK) {
			Set_HTCJK(p_in->MIMEname, p_out->MIMEname);
		    }
		} else {
		    /*
		    **  Cannot translate.
		    **  If according to some heuristic the given
		    **  charset and the current display character
		    **  both are likely to be like ISO-8859 in
		    **  structure, pretend we have some kind
		    **  of match.
		    */
		    BOOL given_is_8859
			= (BOOL) (!strncmp(cp4, "iso-8859-", 9) &&
				  isdigit(UCH(cp4[9])));
		    BOOL given_is_8859like
			= (BOOL) (given_is_8859 ||
				  !strncmp(cp4, "windows-", 8) ||
				  !strncmp(cp4, "cp12", 4) ||
				  !strncmp(cp4, "cp-12", 5));
		    BOOL given_and_display_8859like
			= (BOOL) (given_is_8859like &&
				  (strstr(LYchar_set_names[current_char_set],
					  "ISO-8859") ||
				   strstr(LYchar_set_names[current_char_set],
					  "windows-")));

		    if (given_and_display_8859like) {
			*cp1 = '\0';
			me->format = HTAtom_for(cp);
		    }
		    if (given_is_8859) {
			cp1 = &cp4[10];
			while (*cp1 &&
			       isdigit(UCH(*cp1)))
			    cp1++;
			*cp1 = '\0';
		    }
		    if (given_and_display_8859like) {
			StrAllocCopy(me->anchor->charset, cp4);
			HTPassEightBitRaw = TRUE;
		    }
		    HTAlert(*cp4 ? cp4 : me->anchor->charset);
		}
		FREE(cp3);
	    } else {
		/*
		**	No charset parameter is present.
		**	Ignore all other parameters, as
		**	we do when charset is present. - FM
		*/
		*cp1 = '\0';
		me->format = HTAtom_for(cp);
	    }
	}
	FREE(cp);
    }
    /*
    **  If we have an Expires header and haven't
    **  already set the no_cache element for the
    **  anchor, check if we should set it based
    **  on that header. - FM
    */
    if (me->anchor->no_cache == FALSE &&
	me->anchor->expires != NULL) {
	if (!strcmp(me->anchor->expires, "0")) {
	    /*
	     *  The value is zero, which we treat as
	     *  an absolute no-cache directive. - FM
	     */
	    me->anchor->no_cache = TRUE;
	} else if (me->anchor->date != NULL) {
	    /*
	    **  We have a Date header, so check if
	    **  the value is less than or equal to
	    **  that. - FM
	    */
	    if (LYmktime(me->anchor->expires, TRUE) <=
		LYmktime(me->anchor->date, TRUE)) {
		me->anchor->no_cache = TRUE;
	    }
	} else if (LYmktime(me->anchor->expires, FALSE) == 0) {
	    /*
	    **  We don't have a Date header, and
	    **  the value is in past for us. - FM
	    */
	    me->anchor->no_cache = TRUE;
	}
    }
    StrAllocCopy(me->anchor->content_type,
		 HTAtom_name(me->format));

    if (me->set_cookie != NULL || me->set_cookie2 != NULL) {
	LYSetCookie(me->set_cookie,
		    me->set_cookie2,
		    me->anchor->address);
	FREE(me->set_cookie);
	FREE(me->set_cookie2);
    }
    if (me->pickup_redirection) {
	if (me->location && *me->location) {
	    redirecting_url = me->location;
	    me->location = NULL;
	    if (me->targetRep != WWW_DEBUG || me->sink)
		me->head_only = YES;

	} else {
	    permanent_redirection = FALSE;
	    if (me->location) {
		CTRACE((tfp, "HTTP: 'Location:' is zero-length!\n"));
		HTAlert(REDIRECTION_WITH_BAD_LOCATION);
	    }
	    CTRACE((tfp, "HTTP: Failed to pick up location.\n"));
	    if (me->location) {
		FREE(me->location);
	    } else {
		HTAlert(REDIRECTION_WITH_NO_LOCATION);
	    }
	}
    }
    if (me->head_only) {
	/* We are done! - kw */
	me->state = MIME_IGNORE;
	return HT_OK;
    }

    if (me->no_streamstack) {
	me->target = me->sink;
    } else {
	if (!me->compression_encoding) {
	    CTRACE((tfp, "HTMIME: MIME Content-Type is '%s', converting to '%s'\n",
		    HTAtom_name(me->format), HTAtom_name(me->targetRep)));
	} else {
	    /*
	    **	Change the format to that for "www/compressed"
	    **	and set up a stream to deal with it. - FM
	    */
	    CTRACE((tfp, "HTMIME: MIME Content-Type is '%s',\n", HTAtom_name(me->format)));
	    me->format = HTAtom_for("www/compressed");
	    CTRACE((tfp, "        Treating as '%s'.  Converting to '%s'\n",
		    HTAtom_name(me->format), HTAtom_name(me->targetRep)));
	    FREE(me->compression_encoding);
	}
	me->target = HTStreamStack(me->format, me->targetRep,
				   me->sink , me->anchor);
	if (!me->target) {
	    CTRACE((tfp, "HTMIME: Can't translate! ** \n"));
	    me->target = me->sink;	/* Cheat */
	}
    }
    if (me->target) {
	me->targetClass = *me->target->isa;
	/*
	**	Check for encoding and select state from there,
	**	someday, but until we have the relevant code,
	**	from now push straight through. - FM
	*/
	me->state = MIME_TRANSPARENT;	/* Pump rest of data right through */
    } else {
	me->state = MIME_IGNORE;	/* What else to do? */
    }
    if (me->refresh_url != NULL && !content_is_compressed(me)) {
	char *url = NULL;
	char *num = NULL;
	char *txt = NULL;
	char *base = "";	/* FIXME: refresh_url may be relative to doc */

	LYParseRefreshURL(me->refresh_url, &num, &url);
	if (url != NULL && me->format == WWW_HTML) {
	    CTRACE((tfp, "Formatting refresh-url as first line of result\n"));
	    HTSprintf0(&txt, gettext("Refresh: "));
	    HTSprintf(&txt, gettext("%s seconds "), num);
	    dequote(url);
	    HTSprintf(&txt, "<a href=\"%s%s\">%s</a><br>", base, url, url);
	    CTRACE((tfp, "URL %s%s\n", base, url));
	    (me->isa->put_string)(me, txt);
	    free(txt);
	}
	FREE(num);
	FREE(url);
    }
    return HT_OK;
}

PRIVATE int dispatchField ARGS1(HTStream *, me)
{
    int i, j;
    char *cp;

    *me->value_pointer = '\0';
    cp = me->value_pointer;
    while ((cp > me->value) && *(--cp) == ' ')  /* S/390 -- gil -- 0146 */
	/*
	**  Trim trailing spaces.
	*/
	*cp = '\0';

    switch (me->field) {
    case miACCEPT_RANGES:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Accept-Ranges: '%s'\n",
		me->value));
	break;
    case miAGE:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Age: '%s'\n",
		me->value));
	break;
    case miALLOW:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Allow: '%s'\n",
		me->value));
	break;
    case miALTERNATES:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Alternates: '%s'\n",
		me->value));
	break;
    case miCACHE_CONTROL:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Cache-Control: '%s'\n",
		me->value));
	if (!(me->value && *me->value))
	    break;
	/*
	**  Convert to lowercase and indicate in anchor. - FM
	*/
	LYLowerCase(me->value);
	StrAllocCopy(me->anchor->cache_control, me->value);
	/*
	**  Check whether to set no_cache for the anchor. - FM
	*/
	{
	    char *cp1, *cp0 = me->value;

	    while ((cp1 = strstr(cp0, "no-cache")) != NULL) {
		cp1 += 8;
		while (*cp1 != '\0' && WHITE(*cp1))
		    cp1++;
		if (*cp1 == '\0' || *cp1 == ';') {
		    me->anchor->no_cache = TRUE;
		    break;
		}
		cp0 = cp1;
	    }
	    if (me->anchor->no_cache == TRUE)
		break;
	    cp0 = me->value;
	    while ((cp1 = strstr(cp0, "max-age")) != NULL) {
		cp1 += 7;
		while (*cp1 != '\0' && WHITE(*cp1))
		    cp1++;
		if (*cp1 == '=') {
		    cp1++;
		    while (*cp1 != '\0' && WHITE(*cp1))
			cp1++;
		    if (isdigit(UCH(*cp1))) {
			cp0 = cp1;
			while (isdigit(UCH(*cp1)))
			    cp1++;
			if (*cp0 == '0' && cp1 == (cp0 + 1)) {
			    me->anchor->no_cache = TRUE;
			    break;
			}
		    }
		}
		cp0 = cp1;
	    }
	}
	break;
    case miCOOKIE:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Cookie: '%s'\n",
		me->value));
	break;
    case miCONNECTION:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Connection: '%s'\n",
		me->value));
	break;
    case miCONTENT_BASE:
	HTMIME_TrimDoubleQuotes(me->value);
	CTRACE((tfp, "HTMIME: PICKED UP Content-Base: '%s'\n",
		me->value));
	if (!(me->value && *me->value))
	    break;
	/*