htfile.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

/*			File Access				HTFile.c
**			===========
**
**	This is unix-specific code in general, with some VMS bits.
**	These are routines for file access used by browsers.
**	Development of this module for Unix DIRED_SUPPORT in Lynx
**	 regrettably has has been conducted in a manner with now
**	 creates a major impediment for hopes of adapting Lynx to
**	 a newer version of the library.
**
**  History:
**	   Feb 91	Written Tim Berners-Lee CERN/CN
**	   Apr 91	vms-vms access included using DECnet syntax
**	26 Jun 92 (JFG) When running over DECnet, suppressed FTP.
**			Fixed access bug for relative names on VMS.
**	   Sep 93 (MD)	Access to VMS files allows sharing.
**	15 Nov 93 (MD)	Moved HTVMSname to HTVMSUTILS.C
**	27 Dec 93 (FM)	FTP now works with VMS hosts.
**			FTP path must be Unix-style and cannot include
**			the device or top directory.
*/

#include <HTUtils.h>

#ifndef VMS
#if defined(DOSPATH)
#undef LONG_LIST
#define LONG_LIST  /* Define this for long style unix listings (ls -l),
		     the actual style is configurable from lynx.cfg */
#endif
/* #define NO_PARENT_DIR_REFERENCE */ /* Define this for no parent links */
#endif /* !VMS */

#if defined(DOSPATH)
#define HAVE_READDIR 1
#define USE_DIRENT
#endif

#if defined(USE_DOS_DRIVES)
#include <HTDOS.h>
#endif

#include <HTFile.h>		/* Implemented here */

#ifdef VMS
#include <stat.h>
#endif /* VMS */

#if defined (USE_ZLIB) || defined (USE_BZLIB)
#include <GridText.h>
#endif

#define MULTI_SUFFIX ".multi"	/* Extension for scanning formats */

#include <HTParse.h>
#include <HTTCP.h>
#ifndef DECNET
#include <HTFTP.h>
#endif /* !DECNET */
#include <HTAnchor.h>
#include <HTAtom.h>
#include <HTAAProt.h>
#include <HTFWriter.h>
#include <HTInit.h>
#include <HTBTree.h>
#include <HTAlert.h>
#include <HTCJK.h>
#include <UCDefs.h>
#include <UCMap.h>
#include <UCAux.h>

#include <LYexit.h>
#include <LYCharSets.h>
#include <LYGlobalDefs.h>
#include <LYStrings.h>
#include <LYUtils.h>
#include <LYLeaks.h>

typedef struct _HTSuffix {
	char *		suffix;
	HTAtom *	rep;
	HTAtom *	encoding;
	char *		desc;
	float		quality;
} HTSuffix;

typedef struct {
    struct stat file_info;
    char sort_tags;
    char file_name[1];	/* on the end of the struct, since its length varies */
} DIRED;

#ifndef NGROUPS
#ifdef NGROUPS_MAX
#define NGROUPS NGROUPS_MAX
#else
#define NGROUPS 32
#endif /* NGROUPS_MAX */
#endif /* NGROUPS */

#ifndef GETGROUPS_T
#define GETGROUPS_T int
#endif

#include <HTML.h>		/* For directory object building */

#define PUTC(c) (*target->isa->put_character)(target, c)
#define PUTS(s) (*target->isa->put_string)(target, s)
#define START(e) (*target->isa->start_element)(target, e, 0, 0, -1, 0)
#define END(e) (*target->isa->end_element)(target, e, 0)
#define MAYBE_END(e) if (HTML_dtd.tags[e].contents != SGML_EMPTY) \
			(*target->isa->end_element)(target, e, 0)
#define FREE_TARGET (*target->isa->_free)(target)
#define ABORT_TARGET (*targetClass._abort)(target, NULL);

struct _HTStructured {
	CONST HTStructuredClass *	isa;
	/* ... */
};

/*
**  Controlling globals.
*/
PUBLIC int HTDirAccess = HT_DIR_OK;

#ifdef DIRED_SUPPORT
PUBLIC int HTDirReadme = HT_DIR_README_NONE;
#else
PUBLIC int HTDirReadme = HT_DIR_README_TOP;
#endif /* DIRED_SUPPORT */

PRIVATE char *HTMountRoot = "/Net/";		/* Where to find mounts */
#ifdef VMS
PRIVATE char *HTCacheRoot = "/WWW$SCRATCH";	/* Where to cache things */
#else
PRIVATE char *HTCacheRoot = "/tmp/W3_Cache_";	/* Where to cache things */
#endif /* VMS */

/*
**  Suffix registration.
*/
PRIVATE HTList * HTSuffixes = 0;
PRIVATE HTSuffix no_suffix = { "*", NULL, NULL, NULL, 1.0 };
PRIVATE HTSuffix unknown_suffix = { "*.*", NULL, NULL, NULL, 1.0};


/*	To free up the suffixes at program exit.
**	----------------------------------------
*/
#ifdef LY_FIND_LEAKS
PRIVATE void free_suffixes NOPARAMS;
#endif

#ifdef LONG_LIST
PRIVATE char *FormatStr ARGS3(
    char **,	bufp,
    char *,	start,
    CONST char *,	entry)
{
    char fmt[512];
    if (*start) {
	sprintf(fmt, "%%%.*ss", (int) sizeof(fmt) - 3, start);
	HTSprintf0(bufp, fmt, entry);
    } else if (*bufp && !(entry && *entry)) {
	**bufp = '\0';
    } else if (entry) {
	StrAllocCopy(*bufp, entry);
    }
    return *bufp;
}

PRIVATE char *FormatNum ARGS3(
    char **,	bufp,
    char *,	start,
    int,	entry)
{
    char fmt[512];
    if (*start) {
	sprintf(fmt, "%%%.*sd", (int) sizeof(fmt) - 3, start);
	HTSprintf0(bufp, fmt, entry);
    } else {
	sprintf(fmt, "%d", entry);
	StrAllocCopy(*bufp, fmt);
    }
    return *bufp;
}

PRIVATE void LYListFmtParse ARGS5(
	char *,		fmtstr,
	DIRED *,	data,
	char *,		file,
	HTStructured *, target,
	char *,		tail)
{
	char c;
	char *s;
	char *end;
	char *start;
	char *str = NULL;
	char *buf = NULL;
	char tmp[LY_MAXPATH];
	char type;
#ifndef NOUSERS
	char *name;
#endif
	time_t now;
	char *datestr;
#ifdef S_IFLNK
	int len;
#endif
#define SEC_PER_YEAR	(60 * 60 * 24 * 365)

#ifdef _WINDOWS	/* 1998/01/06 (Tue) 21:20:53 */
	static char *pbits[] = {
		"---", "--x", "-w-", "-wx",
		"r--", "r-x", "rw-", "rwx",
		0 };
#define PBIT(a, n, s)  pbits[((a) >> (n)) & 0x7]

#else
	static char *pbits[] = { "---", "--x", "-w-", "-wx",
		"r--", "r-x", "rw-", "rwx", 0 };
	static char *psbits[] = { "--S", "--s", "-wS", "-ws",
		"r-S", "r-s", "rwS", "rws", 0 };
#define PBIT(a, n, s)  (s) ? psbits[((a) >> (n)) & 0x7] : \
	pbits[((a) >> (n)) & 0x7]
#endif
#ifdef S_ISVTX
	static char *ptbits[] = { "--T", "--t", "-wT", "-wt",
		"r-T", "r-t", "rwT", "rwt", 0 };
#define PTBIT(a, s)  (s) ? ptbits[(a) & 0x7] : pbits[(a) & 0x7]
#else
#define PTBIT(a, s)  PBIT(a, 0, 0)
#endif

	if (data->file_info.st_mode == 0)
		fmtstr = "    %a";	/* can't stat so just do anchor */

	StrAllocCopy(str, fmtstr);
	s = str;
	end = str + strlen(str);
	START(HTML_PRE);
	while (*s) {
		start = s;
		while (*s) {
			if (*s == '%') {
				if (*(s+1) == '%') /* literal % */
					s++;
				else
					break;
			}
			s++;
		}
		/* s is positioned either at a % or at \0 */
		*s = '\0';
		if (s > start) {	/* some literal chars. */
			PUTS(start);
		}
		if (s == end)
			break;
		start = ++s;
		while (isdigit(UCH(*s)) || *s == '.' || *s == '-' || *s == ' ' ||
		    *s == '#' || *s == '+' || *s == '\'')
			s++;
		c = *s;		/* the format char. or \0 */
		*s = '\0';

		switch (c) {
		case '\0':
			PUTS(start);
			continue;

		case 'A':
		case 'a':	/* anchor */
			HTDirEntry(target, tail, data->file_name);
			FormatStr(&buf, start, data->file_name);
			PUTS(buf);
			END(HTML_A);
			*buf = '\0';
#ifdef S_IFLNK
			if (c != 'A' && S_ISLNK(data->file_info.st_mode) &&
			    (len = readlink(file, tmp, sizeof(tmp) - 1)) >= 0) {
				PUTS(" -> ");
				tmp[len] = '\0';
				PUTS(tmp);
			}
#endif
			break;

		case 'T':	/* MIME type */
		case 't':	/* MIME type description */
		    if (S_ISDIR(data->file_info.st_mode)) {
			if (c != 'T') {
			    FormatStr(&buf, start, ENTRY_IS_DIRECTORY);
			} else {
			    FormatStr(&buf, start, "");
			}
		    } else {
			CONST char *cp2;
			HTFormat format;
			format = HTFileFormat(file, NULL, &cp2);

			if (c != 'T') {
			    if (cp2 == NULL) {
				if (!strncmp(HTAtom_name(format),
					     "application",11)) {
				    cp2 = HTAtom_name(format) + 12;
				    if (!strncmp(cp2,"x-",2))
					cp2 += 2;
				} else {
				    cp2 = HTAtom_name(format);
				}
			    }
			    FormatStr(&buf, start, cp2);
			} else {
			    FormatStr(&buf, start, HTAtom_name(format));
			}
		    }
		    break;

		case 'd':	/* date */
			now = time(0);
			datestr = ctime(&data->file_info.st_mtime);
			if ((now - data->file_info.st_mtime) < SEC_PER_YEAR/2)
				/*
				**  MMM DD HH:MM
				*/
				sprintf(tmp, "%.12s", datestr + 4);
			else
				/*
				**  MMM DD  YYYY
				*/
				sprintf(tmp, "%.7s %.4s ", datestr + 4,
					datestr + 20);
			FormatStr(&buf, start, tmp);
			break;

		case 's':	/* size in bytes */
			FormatNum(&buf, start, (int) data->file_info.st_size);
			break;

		case 'K':	/* size in Kilobytes but not for directories */
			if (S_ISDIR(data->file_info.st_mode)) {
				FormatStr(&buf, start, "");
				StrAllocCat(buf, " ");
				break;
			}
			/* FALL THROUGH */
		case 'k':	/* size in Kilobytes */
			FormatNum(&buf, start, (int)((data->file_info.st_size+1023)/1024));
			StrAllocCat(buf, "K");
			break;

		case 'p':	/* unix-style permission bits */
			switch(data->file_info.st_mode & S_IFMT) {
#if defined(_MSC_VER) && defined(_S_IFIFO)
			case _S_IFIFO: type = 'p'; break;
#else
			case S_IFIFO: type = 'p'; break;
#endif
			case S_IFCHR: type = 'c'; break;
			case S_IFDIR: type = 'd'; break;
			case S_IFREG: type = '-'; break;
#ifdef S_IFBLK
			case S_IFBLK: type = 'b'; break;
#endif
#ifdef S_IFLNK
			case S_IFLNK: type = 'l'; break;
#endif
#ifdef S_IFSOCK
# ifdef S_IFIFO		/* some older machines (e.g., apollo) have a conflict */
#  if S_IFIFO != S_IFSOCK
			case S_IFSOCK: type = 's'; break;
#  endif
# else
			case S_IFSOCK: type = 's'; break;
# endif
#endif /* S_IFSOCK */
			default: type = '?'; break;
			}
#ifdef _WINDOWS
			sprintf(tmp, "%c%s", type,
				PBIT(data->file_info.st_mode, 6, data->file_info.st_mode & S_IRWXU));
#else
			sprintf(tmp, "%c%s%s%s", type,
				PBIT(data->file_info.st_mode, 6, data->file_info.st_mode & S_ISUID),
				PBIT(data->file_info.st_mode, 3, data->file_info.st_mode & S_ISGID),
				PTBIT(data->file_info.st_mode,   data->file_info.st_mode & S_ISVTX));
#endif
			FormatStr(&buf, start, tmp);
			break;

		case 'o':	/* owner */
#ifndef NOUSERS
			name = HTAA_UidToName (data->file_info.st_uid);
			if (*name) {
				FormatStr(&buf, start, name);
			} else {
				FormatNum(&buf, start, (int) data->file_info.st_uid);
			}
#endif
			break;

		case 'g':	/* group */
#ifndef NOUSERS
			name = HTAA_GidToName(data->file_info.st_gid);
			if (*name) {
				FormatStr(&buf, start, name);
			} else {
				FormatNum(&buf, start, (int) data->file_info.st_gid);
			}
#endif
			break;

		case 'l':	/* link count */
			FormatNum(&buf, start, (int) data->file_info.st_nlink);
			break;

		case '%':	/* literal % with flags/width */
			FormatStr(&buf, start, "%");
			break;

		default:
			fprintf(stderr,
			"Unknown format character `%c' in list format\n", c);
			break;
		}
		if (buf)
		    PUTS(buf);

		s++;
	}
	FREE(buf);
	END(HTML_PRE);
	PUTC('\n');
	FREE(str);
}
#endif /* LONG_LIST */

/*	Define the representation associated with a file suffix.
**	--------------------------------------------------------
**
**	Calling this with suffix set to "*" will set the default
**	representation.
**	Calling this with suffix set to "*.*" will set the default
**	representation for unknown suffix files which contain a ".".
**
**	The encoding parameter can give a trivial (8bit, 7bit, binary)
**	or real (gzip, compress) encoding.
**
**	If filename suffix is already defined with the same encoding
**	its previous definition is overridden.
*/
PUBLIC void HTSetSuffix5 ARGS5(
	CONST char *,	suffix,
	CONST char *,	representation,
	CONST char *,	encoding,
	CONST char *,	desc,
	double,		value)
{
    HTSuffix * suff;
    BOOL trivial_enc = (BOOL) IsUnityEncStr(encoding);

    if (strcmp(suffix, "*") == 0)
	suff = &no_suffix;
    else if (strcmp(suffix, "*.*") == 0)
	suff = &unknown_suffix;
    else {
	HTList *cur = HTSuffixes;

	while (NULL != (suff = (HTSuffix*)HTList_nextObject(cur))) {
	    if (suff->suffix && 0 == strcmp(suff->suffix, suffix) &&
		((trivial_enc && IsUnityEnc(suff->encoding)) ||
		 (!trivial_enc && !IsUnityEnc(suff->encoding) &&
		     strcmp(encoding, HTAtom_name(suff->encoding)) == 0)))
		break;
	}
	if (!suff) { /* Not found -- create a new node */
	    suff = typecalloc(HTSuffix);
	    if (suff == NULL)
		outofmem(__FILE__, "HTSetSuffix");

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

	    HTList_addObject(HTSuffixes, suff);

	    StrAllocCopy(suff->suffix, suffix);
	}
    }

    if (representation)
	suff->rep = HTAtom_for(representation);

    /*
    **	Memory leak fixed.
    **	05-28-94 Lynx 2-3-1 Garrett Arch Blythe
    **	Invariant code removed.
    */
    suff->encoding = HTAtom_for(encoding);