htformat.c

用于linux和其他unix下面的

/*		Manage different file formats			HTFormat.c
**		=============================
**
** Bugs:
**	Not reentrant.
**
**	Assumes the incoming stream is ASCII, rather than a local file
**	format, and so ALWAYS converts from ASCII on non-ASCII machines.
**	Therefore, non-ASCII machines can't read local files.
**
*/

#include <HTUtils.h>

/* Implements:
*/
#include <HTFormat.h>

PUBLIC float HTMaxSecs = 1e10;		/* No effective limit */
PUBLIC float HTMaxLength = 1e10;	/* No effective limit */
PUBLIC long int HTMaxBytes  = 0;	/* No effective limit */

#ifdef UNIX
#ifdef NeXT
#define PRESENT_POSTSCRIPT "open %s; /bin/rm -f %s\n"
#else
#define PRESENT_POSTSCRIPT "(ghostview %s ; /bin/rm -f %s)&\n"
			   /* Full pathname would be better! */
#endif /* NeXT */
#endif /* UNIX */

#include <HTML.h>
#include <HTMLDTD.h>
#include <HText.h>
#include <HTAlert.h>
#include <HTList.h>
#include <HTInit.h>
#include <HTTCP.h>
#include <HTTP.h>
/*	Streams and structured streams which we use:
*/
#include <HTFWriter.h>
#include <HTPlain.h>
#include <SGML.h>
#include <HTMLGen.h>

#include <LYexit.h>
#include <LYUtils.h>
#include <GridText.h>
#include <LYGlobalDefs.h>
#include <LYLeaks.h>

#ifdef DISP_PARTIAL
#include <LYMainLoop.h>
#endif

PUBLIC	BOOL HTOutputSource = NO;	/* Flag: shortcut parser to stdout */
/* extern  BOOL interactive; LJM */

#ifdef ORIGINAL
struct _HTStream {
      CONST HTStreamClass*	isa;
      /* ... */
};
#endif /* ORIGINAL */

/* this version used by the NetToText stream */
struct _HTStream {
	CONST HTStreamClass *	isa;
	BOOL			had_cr;
	HTStream *		sink;
};

/*	Presentation methods
**	--------------------
*/
PUBLIC	HTList * HTPresentations = NULL;
PUBLIC	HTPresentation * default_presentation = NULL;

/*
 *	To free off the presentation list.
 */
#ifdef LY_FIND_LEAKS
PRIVATE void HTFreePresentations NOPARAMS;
#endif

/*	Define a presentation system command for a content-type
**	-------------------------------------------------------
*/
PUBLIC void HTSetPresentation ARGS6(
	CONST char *,	representation,
	CONST char *,	command,
	double,		quality,
	double,		secs,
	double,		secs_per_byte,
	long int,	maxbytes)
{
    HTPresentation * pres = typecalloc(HTPresentation);
    if (pres == NULL)
	outofmem(__FILE__, "HTSetPresentation");

    pres->rep = HTAtom_for(representation);
    pres->rep_out = WWW_PRESENT;		/* Fixed for now ... :-) */
    pres->converter = HTSaveAndExecute;		/* Fixed for now ...	 */
    pres->quality = (float) quality;
    pres->secs = (float) secs;
    pres->secs_per_byte = (float) secs_per_byte;
    pres->maxbytes = maxbytes;
    pres->command = NULL;
    StrAllocCopy(pres->command, command);

    /*
     *	Memory leak fixed.
     *	05-28-94 Lynx 2-3-1 Garrett Arch Blythe
     */
    if (!HTPresentations)	{
	HTPresentations = HTList_new();
#ifdef LY_FIND_LEAKS
	atexit(HTFreePresentations);
#endif
    }

    if (strcmp(representation, "*")==0) {
	FREE(default_presentation);
	default_presentation = pres;
    } else {
	HTList_addObject(HTPresentations, pres);
    }
}

/*	Define a built-in function for a content-type
**	---------------------------------------------
*/
PUBLIC void HTSetConversion ARGS7(
	CONST char *,	representation_in,
	CONST char *,	representation_out,
	HTConverter*,	converter,
	float,		quality,
	float,		secs,
	float,		secs_per_byte,
	long int,	maxbytes)
{
    HTPresentation * pres = typecalloc(HTPresentation);
    if (pres == NULL)
	outofmem(__FILE__, "HTSetConversion");

    pres->rep = HTAtom_for(representation_in);
    pres->rep_out = HTAtom_for(representation_out);
    pres->converter = converter;
    pres->command = NULL;		/* Fixed */
    pres->quality = quality;
    pres->secs = secs;
    pres->secs_per_byte = secs_per_byte;
    pres->maxbytes = maxbytes;
    pres->command = NULL;

    /*
     *	Memory Leak fixed.
     *	05-28-94 Lynx 2-3-1 Garrett Arch Blythe
     */
    if (!HTPresentations)	{
	HTPresentations = HTList_new();
#ifdef LY_FIND_LEAKS
	atexit(HTFreePresentations);
#endif
    }

    HTList_addObject(HTPresentations, pres);
}

#ifdef LY_FIND_LEAKS
/*
**	Purpose:	Free the presentation list.
**	Arguments:	void
**	Return Value:	void
**	Remarks/Portability/Dependencies/Restrictions:
**		Made to clean up Lynx's bad leakage.
**	Revision History:
**		05-28-94	created Lynx 2-3-1 Garrett Arch Blythe
*/
PRIVATE void HTFreePresentations NOARGS
{
    HTPresentation * pres = NULL;

    /*
     *	Loop through the list.
     */
    while (!HTList_isEmpty(HTPresentations)) {
	/*
	 *  Free off each item.
	 *  May also need to free off it's items, but not sure
	 *  as of yet.
	 */
	pres = (HTPresentation *)HTList_removeLastObject(HTPresentations);
	FREE(pres->command);
	FREE(pres);
    }
    /*
     *	Free the list itself.
     */
    HTList_delete(HTPresentations);
    HTPresentations = NULL;
}
#endif /* LY_FIND_LEAKS */

/*	File buffering
**	--------------
**
**	The input file is read using the macro which can read from
**	a socket or a file.
**	The input buffer size, if large will give greater efficiency and
**	release the server faster, and if small will save space on PCs etc.
*/
#define INPUT_BUFFER_SIZE 4096		/* Tradeoff */
PRIVATE char input_buffer[INPUT_BUFFER_SIZE];
PRIVATE char * input_pointer;
PRIVATE char * input_limit;
PRIVATE int input_file_number;

/*	Set up the buffering
**
**	These routines are public because they are in fact needed by
**	many parsers, and on PCs and Macs we should not duplicate
**	the static buffer area.
*/
PUBLIC void HTInitInput ARGS1 (int,file_number)
{
    input_file_number = file_number;
    input_pointer = input_limit = input_buffer;
}

PUBLIC int interrupted_in_htgetcharacter = 0;
PUBLIC int HTGetCharacter NOARGS
{
    char ch;
    interrupted_in_htgetcharacter = 0;
    do {
	if (input_pointer >= input_limit) {
	    int status = NETREAD(input_file_number,
				 input_buffer, INPUT_BUFFER_SIZE);
	    if (status <= 0) {
		if (status == 0)
		    return EOF;
		if (status == HT_INTERRUPTED) {
		    CTRACE((tfp, "HTFormat: Interrupted in HTGetCharacter\n"));
		    interrupted_in_htgetcharacter = 1;
		    return EOF;
		}
		CTRACE((tfp, "HTFormat: File read error %d\n", status));
		return EOF; /* -1 is returned by UCX
			       at end of HTTP link */
	    }
	    input_pointer = input_buffer;
	    input_limit = input_buffer + status;
	}
	ch = *input_pointer++;
    } while (ch == (char) 13); /* Ignore ASCII carriage return */

    return FROMASCII(UCH(ch));
}

#ifdef USE_SSL
PUBLIC char HTGetSSLCharacter ARGS1(void *, handle)
{
    char ch;
    interrupted_in_htgetcharacter = 0;
    if(!handle)
	return (char)EOF;
    do {
	if (input_pointer >= input_limit) {
	    int status = SSL_read((SSL *)handle,
				 input_buffer, INPUT_BUFFER_SIZE);
	    if (status <= 0) {
		if (status == 0)
		    return (char)EOF;
		if (status == HT_INTERRUPTED) {
		    CTRACE((tfp, "HTFormat: Interrupted in HTGetSSLCharacter\n"));
		    interrupted_in_htgetcharacter = 1;
		    return (char)EOF;
		}
		CTRACE((tfp, "HTFormat: SSL_read error %d\n", status));
		return (char)EOF; /* -1 is returned by UCX
				     at end of HTTP link */
	    }
	    input_pointer = input_buffer;
	    input_limit = input_buffer + status;
	}
	ch = *input_pointer++;
    } while (ch == (char) 13); /* Ignore ASCII carriage return */

    return FROMASCII(ch);
}
#endif /* USE_SSL */

/*  Match maintype to any MIME type starting with maintype,
 *  for example:  image/gif should match image
 */
PRIVATE int half_match ARGS2(char *,trial_type, char *,target)
{
    char *cp = strchr(trial_type, '/');

    /* if no '/' or no '*' */
    if (!cp || *(cp+1) != '*')
	return 0;

    CTRACE((tfp, "HTFormat: comparing %s and %s for half match\n",
		trial_type, target));

	/* main type matches */
    if (!strncmp(trial_type, target, (cp-trial_type)-1))
	return 1;

    return 0;
}

#define WWW_WILDCARD_REP_OUT HTAtom_for("*")

/*		Look up a presentation
**		----------------------
**
**	If fill_in is NULL, only look for an exact match.
**	If a wildcard match is made, *fill_in is used to store
**	a possibly modified presentation, and a pointer to it is
**	returned.  For an exact match, a pointer to the presentation
**	in the HTPresentations list is returned.  Returns NULL if
**	nothing found. - kw
**
*/
PRIVATE HTPresentation * HTFindPresentation ARGS3(
	HTFormat,		rep_in,
	HTFormat,		rep_out,
	HTPresentation*,	fill_in)
{
    HTAtom * wildcard = NULL; /* = HTAtom_for("*"); lookup when needed - kw */

    CTRACE((tfp, "HTFormat: Looking up presentation for %s to %s\n",
		HTAtom_name(rep_in), HTAtom_name(rep_out)));

    /* don't do anymore do it in the Lynx code at startup LJM */
    /* if (!HTPresentations) HTFormatInit(); */ /* set up the list */

    {
	int n = HTList_count(HTPresentations);
	int i;
	HTPresentation * pres, *match,
			*strong_wildcard_match=0,
			*weak_wildcard_match=0,
			*last_default_match=0,
			*strong_subtype_wildcard_match=0;

	for (i = 0; i < n; i++) {
	    pres = (HTPresentation *)HTList_objectAt(HTPresentations, i);
	    if (pres->rep == rep_in) {
		if (pres->rep_out == rep_out) {
		    CTRACE((tfp, "FindPresentation: found exact match: %s\n",
				HTAtom_name(pres->rep)));
		    return pres;

		} else if (!fill_in) {
		    continue;
		} else {
		    if (!wildcard) wildcard = WWW_WILDCARD_REP_OUT;
		    if (pres->rep_out == wildcard) {
			if (!strong_wildcard_match)
			    strong_wildcard_match = pres;
			/* otherwise use the first one */
			CTRACE((tfp, "StreamStack: found strong wildcard match: %s\n",
				    HTAtom_name(pres->rep)));
		    }
		}

	    } else if (!fill_in) {
		continue;

	    } else if (half_match(HTAtom_name(pres->rep),
					      HTAtom_name(rep_in))) {
		if (pres->rep_out == rep_out) {
		    if (!strong_subtype_wildcard_match)
			strong_subtype_wildcard_match = pres;
		    /* otherwise use the first one */
		    CTRACE((tfp, "StreamStack: found strong subtype wildcard match: %s\n",
				HTAtom_name(pres->rep)));
		}
	    }

	    if (pres->rep == WWW_SOURCE) {
		if (pres->rep_out == rep_out) {
		    if (!weak_wildcard_match)
			weak_wildcard_match = pres;
		    /* otherwise use the first one */
		    CTRACE((tfp, "StreamStack: found weak wildcard match: %s\n",
				HTAtom_name(pres->rep_out)));

		} else if (!last_default_match) {
		    if (!wildcard) wildcard = WWW_WILDCARD_REP_OUT;
		    if (pres->rep_out == wildcard)
			 last_default_match = pres;
		    /* otherwise use the first one */
		}
	    }
	}

	match = strong_subtype_wildcard_match ? strong_subtype_wildcard_match :
		strong_wildcard_match ? strong_wildcard_match :
		weak_wildcard_match ? weak_wildcard_match :
		last_default_match;

	if (match) {
	    *fill_in = *match;		/* Specific instance */
	    fill_in->rep = rep_in;		/* yuk */
	    fill_in->rep_out = rep_out; /* yuk */
	    return fill_in;
	}
    }

    return NULL;
}

/*		Create a filter stack
**		---------------------
**
**	If a wildcard match is made, a temporary HTPresentation
**	structure is made to hold the destination format while the
**	new stack is generated. This is just to pass the out format to
**	MIME so far.  Storing the format of a stream in the stream might
**	be a lot neater.
**
*/
PUBLIC HTStream * HTStreamStack ARGS4(
	HTFormat,		rep_in,
	HTFormat,		rep_out,
	HTStream*,		sink,
	HTParentAnchor*,	anchor)
{
    HTPresentation temp;
    HTPresentation *match;
    HTStream *result;

    CTRACE((tfp, "HTFormat: Constructing stream stack for %s to %s\n",
		HTAtom_name(rep_in), HTAtom_name(rep_out)));

    /* don't return on WWW_SOURCE some people might like
     * to make use of the source!!!!  LJM
     */
#if 0
    if (rep_out == WWW_SOURCE || rep_out == rep_in)
	return sink;	/*  LJM */
#endif

    if (rep_out == rep_in) {
	result = sink;

    } else if ((match = HTFindPresentation(rep_in, rep_out, &temp))) {
	if (match == &temp) {
	    CTRACE((tfp, "StreamStack: Using %s\n", HTAtom_name(temp.rep_out)));
	} else {
	    CTRACE((tfp, "StreamStack: found exact match: %s\n",
			HTAtom_name(match->rep)));
	}
	result = (*match->converter)(match, anchor, sink);
    } else {
	result = NULL;
    }
    if (TRACE) {
	if (result && result->isa && result->isa->name) {
	    CTRACE((tfp, "StreamStack: Returning \"%s\"\n", result->isa->name));
	} else if (result) {
	    CTRACE((tfp, "StreamStack: Returning *unknown* stream!\n"));
	} else {
	    CTRACE((tfp, "StreamStack: Returning NULL!\n"));
	    CTRACE_FLUSH(tfp);	/* a crash may be imminent... - kw */
	}
    }
    return result;
}

/*		Put a presentation near start of list
**		-------------------------------------
**
**	Look up a presentation (exact match only) and, if found, reorder
**	it to the start of the HTPresentations list. - kw
*/
PUBLIC void HTReorderPresentation ARGS2(
	HTFormat,		rep_in,
	HTFormat,		rep_out)
{
    HTPresentation *match;
    if ((match = HTFindPresentation(rep_in, rep_out, NULL))) {
	HTList_removeObject(HTPresentations, match);
	HTList_addObject(HTPresentations, match);
    }
}

/*
 * Setup 'get_accept' flag to denote presentations that are not redundant,
 * and will be listed in "Accept:" header.
 */
PUBLIC void HTFilterPresentations NOARGS
{
    int i, j;
    int n = HTList_count(HTPresentations);
    HTPresentation *p, *q;
    BOOL matched;
    char *s, *t, *x, *y;

    for (i = 0; i < n; i++) {