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

		   (strstr(LYchar_set_names[current_char_set],
			   "ISO-8859") ||
		    strstr(LYchar_set_names[current_char_set],
			   "windows-")));

	    if (given_and_display_8859like) {
		*cp1 = '\0';
		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(anchor->charset, cp4);
		HTPassEightBitRaw = TRUE;
	    }
	    HTAlert(*cp4 ? cp4 : anchor->charset);
	}
	FREE(cp3);
    } else if (cp1 != NULL) {
	/*
	**  No charset parameter is present.
	**  Ignore all other parameters, as
	**  we do when charset is present. - FM
	*/
	*cp1 = '\0';
	format = HTAtom_for(cp);
    }
    FREE(cp);

    /*
    **	Set up defaults, if needed. - FM
    */
    if (!chartrans_ok && !anchor->charset && default_LYhndl >= 0) {
	HTAnchor_setUCInfoStage(anchor, default_LYhndl,
				UCT_STAGE_MIME,
				UCT_SETBY_DEFAULT);
    }
    HTAnchor_copyUCInfoStage(anchor,
			    UCT_STAGE_PARSER,
			    UCT_STAGE_MIME,
			    -1);

    return format;
}



/*	Get various pieces of meta info from file name.
**	-----------------------------------------------
**
**  LYGetFileInfo fills in information that can be determined without
**  an actual (new) access to the filesystem, based on current suffix
**  and character set configuration.  If the file has been loaded and
**  parsed before  (with the same URL generated here!) and the anchor
**  is still around, some results may be influenced by that (in
**  particular, charset info from a META tag - this is not actually
**  tested!).
**  The caller should not keep pointers to the returned objects around
**  for too long, the valid lifetimes vary. In particular, the returned
**  charset string should be copied if necessary.  If return of the
**  file_anchor is requested, that one can be used to retrieve
**  additional bits of info that are stored in the anchor object and
**  are not covered here; as usual, don't keep pointers to the
**  file_anchor longer than necessary since the object may disappear
**  through HTuncache_current_document or at the next document load.
**  - kw
*/
PUBLIC void LYGetFileInfo ARGS7(
	CONST char *,		filename,
	HTParentAnchor **,	pfile_anchor,
	HTFormat *,		pformat,
	HTAtom **,		pencoding,
	CONST char**,		pdesc,
	CONST char**,		pcharset,
	int *,			pfile_cs)
{
	char *Afn;
	char *Aname = NULL;
	HTFormat format;
	HTAtom * myEnc = NULL;
	HTParentAnchor *file_anchor;
	CONST char *file_csname;
	int file_cs;

	/*
	 *  Convert filename to URL.  Note that it is always supposed to
	 *  be a filename, not maybe-filename-maybe-URL, so we don't
	 *  use LYFillLocalFileURL and LYEnsureAbsoluteURL. - kw
	 */
	Afn = HTEscape(filename, URL_PATH);
	LYLocalFileToURL(&Aname, Afn);
	file_anchor = HTAnchor_findSimpleAddress(Aname);

	file_csname = file_anchor->charset;
	format = HTFileFormat(filename, &myEnc, pdesc);
	format = HTCharsetFormat(format, file_anchor, UCLYhndl_HTFile_for_unspec);
	file_cs = HTAnchor_getUCLYhndl(file_anchor, UCT_STAGE_MIME);
	if (!file_csname) {
	    if (file_cs >= 0)
		file_csname = LYCharSet_UC[file_cs].MIMEname;
	    else file_csname = "display character set";
	}
	CTRACE((tfp, "GetFileInfo: '%s' is a%s %s %s file, charset=%s (%d).\n",
	       filename,
	       ((myEnc && *HTAtom_name(myEnc) == '8') ? "n" : myEnc ? "" :
		*HTAtom_name(format) == 'a' ? "n" : ""),
	       myEnc ? HTAtom_name(myEnc) : "",
	       HTAtom_name(format),
	       file_csname,
	       file_cs));
	FREE(Afn);
	FREE(Aname);
	if (pfile_anchor)
	    *pfile_anchor = file_anchor;
	if (pformat)
	    *pformat = format;
	if (pencoding)
	    *pencoding = myEnc;
	if (pcharset)
	    *pcharset = file_csname;
	if (pfile_cs)
	    *pfile_cs = file_cs;
    }

/*	Determine value from file name.
**	-------------------------------
**
*/
PUBLIC float HTFileValue ARGS1(
	CONST char *,	filename)
{
    HTSuffix * suff;
    int n;
    int i;
    int lf = strlen(filename);

#ifndef NO_INIT
    if (!HTSuffixes)
	HTFileInit();
#endif /* !NO_INIT */
    n = HTList_count(HTSuffixes);
    for (i = 0; i < n; i++) {
	int ls;
	suff = (HTSuffix *)HTList_objectAt(HTSuffixes, i);
	ls = strlen(suff->suffix);
	if ((ls <= lf) && 0==strcmp(suff->suffix, filename + lf - ls)) {
	    CTRACE((tfp, "File: Value of %s is %.3f\n",
			filename, suff->quality));
	    return suff->quality;		/* OK -- found */
	}
    }
    return (float)0.3;		/* Dunno! */
}

/*
**  Determine compression type from file name, by looking at its suffix.
**  Sets as side-effect a pointer to the "dot" that begins the suffix.
*/
PUBLIC CompressFileType HTCompressFileType ARGS3(
	char *,		filename,
	char *,		dots,
	char **,	suffix)
{
    CompressFileType result = cftNone;
    size_t len = strlen(filename);
    char *ftype = filename + len;

    if ((len > 4)
     && !strcasecomp((ftype - 3), "bz2")
     && strchr(dots, ftype[-4]) != 0) {
	result = cftBzip2;
	ftype -= 4;
    } else if ((len > 3)
     && !strcasecomp((ftype - 2), "gz")
     && strchr(dots, ftype[-3]) != 0) {
	result = cftGzip;
	ftype -= 3;
    } else if ((len > 2)
     && !strcmp((ftype - 1), "Z")
     && strchr(dots, ftype[-2]) != 0) {
	result = cftCompress;
	ftype -= 2;
    }

    *suffix = ftype;
    CTRACE((tfp, "HTCompressFileType(%s) returns %d:%s\n",
		 filename, result, *suffix));
    return result;
}

/*	Determine write access to a file.
**	---------------------------------
**
**  On exit:
**	Returns YES if file can be accessed and can be written to.
**
**  Bugs:
**	1.	No code for non-unix systems.
**	2.	Isn't there a quicker way?
*/
PUBLIC BOOL HTEditable ARGS1(
	CONST char *,	filename)
{
#ifndef NO_GROUPS
    GETGROUPS_T groups[NGROUPS];
    uid_t	myUid;
    int		ngroups;			/* The number of groups	 */
    struct stat fileStatus;
    int		i;

    if (stat(filename, &fileStatus))		/* Get details of filename */
	return NO;				/* Can't even access file! */

    ngroups = getgroups(NGROUPS, groups);	/* Groups to which I belong  */
    myUid = geteuid();				/* Get my user identifier */

    if (TRACE) {
	int i2;
	fprintf(tfp,
	    "File mode is 0%o, uid=%d, gid=%d. My uid=%d, %d groups (",
	    (unsigned int) fileStatus.st_mode,
	    (int) fileStatus.st_uid,
	    (int) fileStatus.st_gid,
	    (int) myUid,
	    (int) ngroups);
	for (i2 = 0; i2 < ngroups; i2++)
	    fprintf(tfp, " %d", (int) groups[i2]);
	fprintf(tfp, ")\n");
    }

    if (fileStatus.st_mode & 0002)		/* I can write anyway? */
	return YES;

    if ((fileStatus.st_mode & 0200)		/* I can write my own file? */
     && (fileStatus.st_uid == myUid))
	return YES;

    if (fileStatus.st_mode & 0020)		/* Group I am in can write? */
    {
	for (i = 0; i < ngroups; i++) {
	    if (groups[i] == fileStatus.st_gid)
		return YES;
	}
    }
    CTRACE((tfp, "\tFile is not editable.\n"));
#endif /* NO_GROUPS */
    return NO;					/* If no excuse, can't do */
}

/*	Make a save stream.
**	-------------------
**
**	The stream must be used for writing back the file.
**	@@@ no backup done
*/
PUBLIC HTStream * HTFileSaveStream ARGS1(
	HTParentAnchor *,	anchor)
{
    CONST char * addr = anchor->address;
    char * localname = HTLocalName(addr);
    FILE * fp = fopen(localname, BIN_W);

    FREE(localname);
    if (!fp)
	return NULL;

    return HTFWriter_new(fp);
}

/*	Output one directory entry.
**	---------------------------
*/
PUBLIC void HTDirEntry ARGS3(
	HTStructured *, target,
	CONST char *,	tail,
	CONST char *,	entry)
{
    char * relative = NULL;
    char * stripped = NULL;
    char * escaped = NULL;
    int len;

    StrAllocCopy(escaped, entry);
    LYTrimPathSep(escaped);
    if (strcmp(escaped, "..") != 0) {
	stripped = escaped;
	escaped = HTEscape(stripped, URL_XPALPHAS);
	if (((len = strlen(escaped)) > 2) &&
	    escaped[(len - 3)] == '%' &&
	    escaped[(len - 2)] == '2' &&
	    TOUPPER(escaped[(len - 1)]) == 'F') {
	    escaped[(len - 3)] = '\0';
	}
    }

    if (tail == NULL || *tail == '\0') {
	/*
	**  Handle extra slash at end of path.
	*/
	HTStartAnchor(target, NULL, (escaped[0] != '\0' ? escaped : "/"));
    } else {
	/*
	**  If empty tail, gives absolute ref below.
	*/
	relative = 0;
	HTSprintf0(&relative, "%s%s%s",
			   tail,
			   (*escaped != '\0' ? "/" : ""),
			   escaped);
	HTStartAnchor(target, NULL, relative);
	FREE(relative);
    }
    FREE(stripped);
    FREE(escaped);
}

/*	Output parent directory entry.
**	------------------------------
**
**    This gives the TITLE and H1 header, and also a link
**    to the parent directory if appropriate.
**
**  On exit:
**	Returns TRUE if an "Up to <parent>" link was not created
**	for a readable local directory because LONG_LIST is defined
**	and NO_PARENT_DIR_REFERENCE is not defined, so that the
**	calling function should use LYListFmtParse() to create a link
**	to the parent directory.  Otherwise, it returns FALSE. - FM
*/
PUBLIC BOOL HTDirTitles ARGS3(
	HTStructured *, target,
	HTParentAnchor *, anchor,
	BOOL,		tildeIsTop)
{
    CONST char * logical = anchor->address;
    char * path = HTParse(logical, "", PARSE_PATH + PARSE_PUNCTUATION);
    char * current;
    char * cp = NULL;
    BOOL need_parent_link = FALSE;
    int i;
#if defined(USE_DOS_DRIVES)
    BOOL local_link = (strlen(logical) > 18
		     && !strncasecomp(logical, "file://localhost/", 17)
		     && LYIsDosDrive(logical + 17));
    BOOL is_remote = !local_link;
#else
#define is_remote TRUE
#endif

    /*
    **	Check tildeIsTop for treating home directory as Welcome
    **	(assume the tilde is not followed by a username). - FM
    */
    if (tildeIsTop && !strncmp(path, "/~", 2)) {
	if (path[2] == '\0') {
	    path[1] = '\0';
	} else {
	    for (i = 0; path[(i + 2)]; i++) {
		path[i] = path[(i + 2)];
	    }
	    path[i] = '\0';
	}
    }

    /*
    **	Trim out the ;type= parameter, if present. - FM
    */
    if ((cp = strrchr(path, ';')) != NULL) {
	if (!strncasecomp((cp+1), "type=", 5)) {
	    if (TOUPPER(*(cp+6)) == 'D' ||
		TOUPPER(*(cp+6)) == 'A' ||
		TOUPPER(*(cp+6)) == 'I')
		*cp = '\0';
	}
	cp = NULL;
    }
    current = LYPathLeaf (path);	/* last part or "" */

    {
      char * printable = NULL;

#ifdef DIRED_SUPPORT
      printable = HTURLPath_toFile(
	    (0 == strncasecomp(path, "/%2F", 4))	/* "//" ? */
	    ? (path+1)
	    : path,
	    TRUE,
	    is_remote);
      if (0 == strncasecomp(printable, "/vmsysu:", 8) ||
	  0 == strncasecomp(printable, "/anonymou.", 10)) {
	  StrAllocCopy(cp, (printable+1));
	  StrAllocCopy(printable, cp);
	  FREE(cp);
      }
#else
      StrAllocCopy(printable, current);
      HTUnEscape(printable);
#endif /* DIRED_SUPPORT */

      START(HTML_HEAD);
      PUTC('\n');
      START(HTML_TITLE);
      PUTS(*printable ? printable : WELCOME_MSG);
      PUTS(SEGMENT_DIRECTORY);
      END(HTML_TITLE);
      PUTC('\n');
      END(HTML_HEAD);
      PUTC('\n');

#ifdef DIRED_SUPPORT
      START(HTML_H2);
      PUTS(*printable ? SEGMENT_CURRENT_DIR : "");
      PUTS(*printable ? printable : WELCOME_MSG);
      END(HTML_H2);
      PUTC('\n');
#else
      START(HTML_H1);
      PUTS(*printable ? printable : WELCOME_MSG);
      END(HTML_H1);
      PUTC('\n');
#endif /* DIRED_SUPPORT */
      if (((0 == strncasecomp(printable, "vmsysu:", 7)) &&
	   (cp = strchr(printable, '.')) != NULL &&
	   strchr(cp, '/') == NULL) ||
	  (0 == strncasecomp(printable, "anonymou.", 9) &&
	   strchr(printable, '/') == NULL)) {
	  FREE(printable);
	  FREE(path);
	  return(need_parent_link);
      }
      FREE(printable);
    }

#ifndef NO_PARENT_DIR_REFERENCE
    /*
    **	Make link back to parent directory.
    */
    if (current - path > 0
      && LYIsPathSep(current[-1])
      && current[0] != '\0') {	/* was a slash AND something else too */
	char * parent = NULL;
	char * relative = NULL;

	current[-1] = '\0';
	parent = strrchr(path, '/');  /* penultimate slash */

	if ((parent &&
	     (!strcmp(parent, "/..") ||
	      !strncasecomp(parent, "/%2F", 4))) ||
	    !strncasecomp(current, "%2F", 3)) {
	    FREE(path);
	    return(need_parent_link);
	}

	relative = 0;
	HTSprintf0(&relative, "%s/..", current);

#if defined(DOSPATH) || defined(__EMX__)
	if (local_link) {
	    if (parent != 0 && strlen(parent) == 3 ) {
		StrAllocCat(relative, "/.");
	    }
	}
	else
#endif

#if !defined (VMS)
	{
	    /*
	    **	On Unix, if it's not ftp and the directory cannot
	    **	be read, don't put out a link.
	    **
	    **	On VMS, this problem is dealt with internally by
	    **	HTVMSBrowseDir().
	    */
	    DIR  * dp = NULL;

	    if (LYisLocalFile(logical)) {
		/*
		**  We need an absolute file path for the opendir.
		**  We also need to unescape for this test.
		**  Don't worry about %2F now, they presumably have been
		**  dealt with above, and shouldn't appear for local
		**  files anyway...  Assume OS / filesystem will just
		**  ignore superfluous slashes. - KW
		*/
		char * fullparentpath = NULL;

		/*
		**  Path has been shortened above.
		*/
		StrAllocCopy(fullparentpath, *path ? path : "/");

		/*
		**  Guard against weirdness.
		*/
		if (0 == strcmp(current,"..")) {
		    StrAllocCat(fullparentpath,"/../..");
		} else if (0 == strcmp(current,".")) {
		    StrAllocCat(fullparentpath,"/..");
		}

		HTUnEscape(fullparentpath);
		if ((dp = opendir(fullparentpath)) == NULL) {