htgopher.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

/*			GOPHER ACCESS				HTGopher.c
**			=============
**
**  History:
**	26 Sep 90	Adapted from other accesses (News, HTTP) TBL
**	29 Nov 91	Downgraded to C, for portable implementation.
**	10 Mar 96	Foteos Macrides ([e-mail]).  Added a
**			  form-based CSO/PH gateway.  Can be invoked via a
**			  "cso://host[:port]/" or "gopher://host:105/2"
**			  URL.	If a gopher URL is used with a query token
**			  ('?'), the old ISINDEX procedure will be used
**			  instead of the form-based gateway.
**	15 Mar 96	Foteos Macrides ([e-mail]).  Pass
**			  port 79, gtype 0 gopher URLs to the finger
**			  gateway.
*/

#include <HTUtils.h>		/* Coding convention macros */

#ifndef DISABLE_GOPHER
#include <HTAlert.h>
#include <HTParse.h>
#include <HTTCP.h>
#include <HTFinger.h>

/*
**  Implements.
*/
#include <HTGopher.h>

#define GOPHER_PORT 70		/* See protocol spec */
#define CSO_PORT 105		/* See protocol spec */
#define BIG 1024		/* Bug */
#define LINE_LENGTH 256		/* Bug */

/*
**  Gopher entity types.
*/
#define GOPHER_TEXT		'0'
#define GOPHER_MENU		'1'
#define GOPHER_CSO		'2'
#define GOPHER_ERROR		'3'
#define GOPHER_MACBINHEX	'4'
#define GOPHER_PCBINARY		'5'
#define GOPHER_UUENCODED	'6'
#define GOPHER_INDEX		'7'
#define GOPHER_TELNET		'8'
#define GOPHER_BINARY		'9'
#define GOPHER_GIF		'g'
#define GOPHER_HTML		'h'		/* HTML */
#define GOPHER_CHTML		'H'		/* HTML */
#define GOPHER_SOUND		's'
#define GOPHER_WWW		'w'		/* W3 address */
#define GOPHER_IMAGE		'I'
#define GOPHER_TN3270		'T'
#define GOPHER_INFO		'i'
#define GOPHER_DUPLICATE	'+'
#define GOPHER_PLUS_IMAGE	':'		/* Addition from Gopher Plus */
#define GOPHER_PLUS_MOVIE	';'
#define GOPHER_PLUS_SOUND	'<'
#define GOPHER_PLUS_PDF		'P'

#include <HTFormat.h>

/*
**  Hypertext object building machinery.
*/
#include <HTML.h>

#include <LYStrings.h>
#include <LYUtils.h>
#include <LYLeaks.h>

#define PUTC(c) (*targetClass.put_character)(target, c)
#define PUTS(s) (*targetClass.put_string)(target, s)
#define START(e) (*targetClass.start_element)(target, e, 0, 0, -1, 0)
#define END(e) (*targetClass.end_element)(target, e, 0)
#define FREE_TARGET (*targetClass._free)(target)

#define GOPHER_PROGRESS(foo) HTAlert(foo)

#define NEXT_CHAR HTGetCharacter()

/*
**  Module-wide variables.
*/
PRIVATE int s;				/* Socket for gopher or CSO host */

struct _HTStructured {
	CONST HTStructuredClass * isa;	/* For gopher streams */
	/* ... */
};

PRIVATE HTStructured *target;		/* the new gopher hypertext */
PRIVATE HTStructuredClass targetClass;	/* Its action routines */

struct _HTStream
{
  HTStreamClass * isa;			/* For form-based CSO  gateway - FM */
};

typedef struct _CSOfield_info {		/* For form-based CSO gateway - FM */
    struct _CSOfield_info *	next;
    char *			name;
    char *			attributes;
    char *			description;
    int				id;
    int				lookup;
    int				indexed;
    int				url;
    int				max_size;
    int				defreturn;
    int				explicit_return;
    int				reserved;
    int				public;
    char			name_buf[16];	/* Avoid malloc if we can */
    char			desc_buf[32];	/* Avoid malloc if we can */
    char			attr_buf[80];	/* Avoid malloc if we can */
} CSOfield_info;

PRIVATE CSOfield_info *CSOfields = NULL; /* For form-based CSO gateway - FM */

typedef struct _CSOformgen_context {	 /* For form-based CSO gateway - FM */
    char *		host;
    char *		seek;
    CSOfield_info *	fld;
    int			port;
    int			cur_line;
    int			cur_off;
    int			rep_line;
    int			rep_off;
    int			public_override;
    int			field_select;
} CSOformgen_context;

/*	Matrix of allowed characters in filenames
**	=========================================
*/
PRIVATE BOOL acceptable[256];
PRIVATE BOOL acceptable_inited = NO;

PRIVATE void init_acceptable NOARGS
{
    unsigned int i;
    char * good =
      "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789./-_$";
    for(i = 0; i < 256; i++)
	acceptable[i] = NO;
    for(; *good; good++)
	acceptable[(unsigned int)*good] = YES;
    acceptable_inited = YES;
}

/*	Decode one hex character
**	========================
*/
PRIVATE CONST char hex[17] = "0123456789abcdef";

PRIVATE char from_hex ARGS1(char, c)
{
    return (char) (       (c>='0')&&(c<='9') ? c-'0'
			: (c>='A')&&(c<='F') ? c-'A'+10
			: (c>='a')&&(c<='f') ? c-'a'+10
			:		       0);
}

/*	Paste in an Anchor
**	==================
**
**	The title of the destination is set, as there is no way
**	of knowing what the title is when we arrive.
**
** On entry,
**	HT	is in append mode.
**	text	points to the text to be put into the file, 0 terminated.
**	addr	points to the hypertext refernce address 0 terminated.
*/
PUBLIC BOOLEAN HT_Is_Gopher_URL=FALSE;

PRIVATE void write_anchor ARGS2(CONST char *,text, CONST char *,addr)
{
    BOOL present[HTML_A_ATTRIBUTES];
    CONST char * value[HTML_A_ATTRIBUTES];

    int i;

    for (i = 0; i < HTML_A_ATTRIBUTES; i++)
	present[i] = 0;
    present[HTML_A_HREF] = YES;
    ((CONST char **)value)[HTML_A_HREF] = addr;
    present[HTML_A_TITLE] = YES;
    ((CONST char **)value)[HTML_A_TITLE] = text;

    CTRACE((tfp,"HTGopher: adding URL: %s\n",addr));

    HT_Is_Gopher_URL = TRUE;  /* tell HTML.c that this is a Gopher URL */
    (*targetClass.start_element)(target, HTML_A, present,
				 (CONST char **)value, -1, 0);

    PUTS(text);
    END(HTML_A);
}

/*	Parse a Gopher Menu document
**	============================
*/
PRIVATE void parse_menu ARGS2(
	CONST char *,		arg GCC_UNUSED,
	HTParentAnchor *,	anAnchor)
{
    char gtype;
    int ich;
    char line[BIG];
    char *name = NULL, *selector = NULL;	/* Gopher menu fields */
    char *host = NULL;
    char *port;
    char *p = line;
    CONST char *title;
    int bytes = 0;
    int BytesReported = 0;
    char buffer[128];

#define TAB		'\t'
#define HEX_ESCAPE	'%'


    START(HTML_HTML);
    PUTC('\n');
    START(HTML_HEAD);
    PUTC('\n');
    START(HTML_TITLE);
    if ((title = HTAnchor_title(anAnchor)))
	PUTS(title);
    else
	PUTS(GOPHER_MENU_TITLE);
    END(HTML_TITLE);
    PUTC('\n');
    END(HTML_HEAD);
    PUTC('\n');

    START(HTML_BODY);
    PUTC('\n');
    START(HTML_H1);
    if ((title = HTAnchor_title(anAnchor)))
	PUTS(title);
    else
	PUTS(GOPHER_MENU_TITLE);
    END(HTML_H1);
    PUTC('\n');
    START(HTML_PRE);
    while ((ich=NEXT_CHAR) != EOF) {

	if (interrupted_in_htgetcharacter) {
	    CTRACE((tfp, "HTGopher: Interrupted in HTGetCharacter, apparently.\n"));
	    goto end_html;
	}

	if ((char)ich != LF) {
	    *p = (char) ich;    /* Put character in line */
	    if (p< &line[BIG-1]) p++;

	} else {
	    *p++ = '\0';	/* Terminate line */
	    bytes += p-line;	/* add size */
	    p = line;		/* Scan it to parse it */
	    port = 0;		/* Flag "not parsed" */
	    CTRACE((tfp, "HTGopher: Menu item: %s\n", line));
	    gtype = *p++;

	    if (bytes > BytesReported + 1024) {
		sprintf(buffer, TRANSFERRED_X_BYTES, bytes);
		HTProgress(buffer);
		BytesReported = bytes;
	    }

	    /* Break on line with a dot by itself */
	    if ((gtype=='.') && ((*p=='\r') || (*p==0)))
		break;

	    if (gtype && *p) {
		name = p;
		selector = strchr(name, TAB);
		if (selector) {
		    *selector++ = '\0'; /* Terminate name */
		    /*
		     * Gopher+ Type=0+ objects can be binary, and will
		     * have 9 or 5 beginning their selector.  Make sure
		     * we don't trash the terminal by treating them as
		     * text. - FM
		     */
		    if (gtype == GOPHER_TEXT && (*selector == GOPHER_BINARY ||
						 *selector == GOPHER_PCBINARY))
			gtype = *selector;
		    host = strchr(selector, TAB);
		    if (host) {
			*host++ = '\0'; /* Terminate selector */
			port = strchr(host, TAB);
			if (port) {
			    char *junk;
			    port[0] = ':';	/* delimit host a la W3 */
			    junk = strchr(port, TAB);
			    if (junk) *junk++ = '\0';	/* Chop port */
			    if ((port[1]=='0') && (!port[2]))
				port[0] = '\0'; /* 0 means none */
			} /* no port */
		    } /* host ok */
		} /* selector ok */
	    } /* gtype and name ok */

	    /* Nameless files are a separator line */
	    if (gtype == GOPHER_TEXT) {
		int i = strlen(name)-1;
		while (name[i] == ' ' && i >= 0)
		    name[i--] = '\0';
		if (i < 0)
		    gtype = GOPHER_INFO;
	    }

	    if (gtype == GOPHER_WWW) {	/* Gopher pointer to W3 */
		PUTS("(HTML) ");
		write_anchor(name, selector);

	    } else if (gtype == GOPHER_INFO) {
	    /* Information or separator line */
		PUTS("       ");
		PUTS(name);

	    } else if (port) {		/* Other types need port */
		char *address = 0;
		char *format = *selector ? "%s//%s@%s/" : "%s//%s/";

		if (gtype == GOPHER_TELNET) {
		    PUTS(" (TEL) ");
		    HTSprintf0(&address, format, STR_TELNET_URL, selector, host);
		}
		else if (gtype == GOPHER_TN3270)
		{
		    PUTS("(3270) ");
		    HTSprintf0(&address, format, STR_TN3270_URL, selector, host);
		}
		else {			/* If parsed ok */
		    char *r;

		    switch(gtype) {
			case GOPHER_TEXT:
			    PUTS("(FILE) ");
			    break;
			case GOPHER_MENU:
			    PUTS(" (DIR) ");
			    break;
			case GOPHER_CSO:
			    PUTS(" (CSO) ");
			    break;
			case GOPHER_PCBINARY:
			    PUTS(" (BIN) ");
			    break;
			case GOPHER_UUENCODED:
			    PUTS(" (UUE) ");
			    break;
			case GOPHER_INDEX:
			    PUTS("  (?)  ");
			    break;
			case GOPHER_BINARY:
			    PUTS(" (BIN) ");
			    break;
			case GOPHER_GIF:
			case GOPHER_IMAGE:
			case GOPHER_PLUS_IMAGE:
			    PUTS(" (IMG) ");
			    break;
			case GOPHER_SOUND:
			case GOPHER_PLUS_SOUND:
			    PUTS(" (SND) ");
			    break;
			case GOPHER_MACBINHEX:
			    PUTS(" (HQX) ");
			    break;
			case GOPHER_HTML:
			case GOPHER_CHTML:
			    PUTS("(HTML) ");
			    break;
			case 'm':
			    PUTS("(MIME) ");
			    break;
			case GOPHER_PLUS_MOVIE:
			    PUTS(" (MOV) ");
			    break;
			case GOPHER_PLUS_PDF:
			    PUTS(" (PDF) ");
			    break;
			default:
			    PUTS("(UNKN) ");
			    break;
		    }

		    HTSprintf0(&address, "//%s/%c", host, gtype);

		    for(r = selector; *r; r++) { /* Encode selector string */
			if (acceptable[UCH(*r)]) {
			    HTSprintf(&address, "%c", *r);
			} else {
			    HTSprintf(&address, "%c%c%c",
				HEX_ESCAPE,	/* Means hex coming */
				hex[(TOASCII(*r)) >> 4],
				hex[(TOASCII(*r)) & 15]);
			}
		    }
		}
		/* Error response from Gopher doesn't deserve to
		   be a hyperlink. */
		if (strcmp (address, "gopher://error.host:1/0"))
		    write_anchor(name, address);
		else
		    PUTS(name);
		FREE(address);
	    } else { /* parse error */
		CTRACE((tfp, "HTGopher: Bad menu item.\n"));
		PUTS(line);

	    } /* parse error */

	    PUTC('\n');
	    p = line;	/* Start again at beginning of line */

	} /* if end of line */

    } /* Loop over characters */

end_html:
    END(HTML_PRE);
    PUTC('\n');
    END(HTML_BODY);
    PUTC('\n');
    END(HTML_HTML);
    PUTC('\n');
    FREE_TARGET;

    return;
}

/*	Parse a Gopher CSO document from an ISINDEX query.
**	==================================================
**
**   Accepts an open socket to a CSO server waiting to send us
**   data and puts it on the screen in a reasonable manner.
**
**   Perhaps this data can be automatically linked to some
**   other source as well???
**
**  Taken from hacking by Lou [e-mail]
**  on XMosaic-1.1, and put on libwww 2.11 by Arthur Secret,
**  [e-mail] .
*/
PRIVATE void parse_cso ARGS2(
	CONST char *,		arg,
	HTParentAnchor *,	anAnchor)
{
    int ich;
    char line[BIG];
    char *p = line;
    char *second_colon, last_char='\0';
    CONST char *title;

    START(HTML_HEAD);
    PUTC('\n');
    START(HTML_TITLE);
    if ((title = HTAnchor_title(anAnchor)))
	PUTS(title);
    else
	PUTS(GOPHER_CSO_SEARCH_RESULTS);
    END(HTML_TITLE);
    PUTC('\n');
    END(HTML_HEAD);
    PUTC('\n');
    START(HTML_H1);
    if ((title = HTAnchor_title(anAnchor)))
	PUTS(title);
    else {
	PUTS(arg);
	PUTS(GOPHER_SEARCH_RESULTS);
    }
    END(HTML_H1);
    PUTC('\n');
    START(HTML_PRE);

    /*
    **	Start grabbing chars from the network.