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

#ifdef _WINDOWS
	/*
	 * Someone claims that stat() doesn't give the proper result for a
	 * directory on Windows.
	 */
	if (result == -1
	 && access(filename, 0) == 0) {
	    data->st_mode = S_IFDIR;
	    result = 0;
	}
#endif
    }
    return result;
}
#endif

/*	Load a document.
**	----------------
**
**  On entry:
**	addr		must point to the fully qualified hypertext reference.
**			This is the physical address of the file
**
**  On exit:
**	returns		<0		Error has occurred.
**			HTLOADED	OK
**
*/
PUBLIC int HTLoadFile ARGS4(
	CONST char *,		addr,
	HTParentAnchor *,	anchor,
	HTFormat,		format_out,
	HTStream *,		sink)
{
    char * filename = NULL;
    char * acc_method = NULL;
    char * ftp_newhost;
    HTFormat format;
    char * nodename = NULL;
    char * newname = NULL;	/* Simplified name of file */
    HTAtom * encoding;		/* @@ not used yet */
    HTAtom * myEncoding = NULL; /* enc of this file, may be gzip etc. */
    int status = -1;
    char *dot;
#ifdef VMS
    struct stat stat_info;
#endif /* VMS */
#ifdef USE_ZLIB
    gzFile gzfp = 0;
#endif /* USE_ZLIB */
#ifdef USE_BZLIB
    BZFILE *bzfp = 0;
#endif /* USE_ZLIB */
#if defined(USE_ZLIB) || defined(USE_BZLIB)
    CompressFileType internal_decompress = cftNone;
    BOOL failed_decompress = NO;
#endif

    /*
    **	Reduce the filename to a basic form (hopefully unique!).
    */
    StrAllocCopy(newname, addr);
    filename=HTParse(newname, "", PARSE_PATH|PARSE_PUNCTUATION);
    nodename=HTParse(newname, "", PARSE_HOST);

    /*
    **	If access is ftp, or file is on another host, invoke ftp now.
    */
    acc_method = HTParse(newname, "", PARSE_ACCESS);
    if (strcmp("ftp", acc_method) == 0 ||
       (!LYSameHostname("localhost", nodename) &&
	!LYSameHostname(nodename, HTHostName()))) {
	status = -1;
	FREE(newname);
	FREE(filename);
	FREE(nodename);
	FREE(acc_method);
#ifndef DISABLE_FTP
	ftp_newhost = HTParse(addr, "", PARSE_HOST);
	if (strcmp(ftp_lasthost, ftp_newhost))
	    ftp_local_passive = ftp_passive;

	status = HTFTPLoad(addr, anchor, format_out, sink);

	if ( ftp_passive == ftp_local_passive ) {
	    if (( status >= 400 ) || ( status < 0 )) {
		ftp_local_passive = !ftp_passive;
		status = HTFTPLoad(addr, anchor, format_out, sink);
	    }
	}

	free(ftp_lasthost);
	ftp_lasthost = ftp_newhost;
#endif /* DISABLE_FTP */
	return status;
    } else {
	FREE(newname);
	FREE(acc_method);
    }
#if defined(VMS) || defined(USE_DOS_DRIVES)
    HTUnEscape(filename);
#endif /* VMS */

    /*
    **	Determine the format and encoding mapped to any suffix.
    */
    if (anchor->content_type && anchor->content_encoding) {
	/*
	 *  If content_type and content_encoding are BOTH already set
	 *  in the anchor object, we believe it and don't try to
	 *  derive format and encoding from the filename. - kw
	 */
	format = HTAtom_for(anchor->content_type);
	myEncoding = HTAtom_for(anchor->content_encoding);
    } else {
	int default_UCLYhndl = UCLYhndl_HTFile_for_unspec;

	if (force_old_UCLYhndl_on_reload) {
	    force_old_UCLYhndl_on_reload = FALSE;
	    default_UCLYhndl = forced_UCLYhdnl;
	}

	format = HTFileFormat(filename, &myEncoding, NULL);

    /*
    **	Check the format for an extended MIME charset value, and
    **	act on it if present.  Otherwise, assume what is indicated
    **	by the last parameter (fallback will effectively be
    **	UCLYhndl_for_unspec, by default ISO-8859-1). - kw
    */
	format = HTCharsetFormat(format, anchor, default_UCLYhndl );
    }

#ifdef VMS
    /*
    **	Check to see if the 'filename' is in fact a directory.	If it is
    **	create a new hypertext object containing a list of files and
    **	subdirectories contained in the directory.  All of these are links
    **	to the directories or files listed.
    */
    if (HTStat(filename, &stat_info) == -1) {
	CTRACE((tfp, "HTLoadFile: Can't stat %s\n", filename));
    } else {
	if (S_ISDIR(stat_info.st_mode)) {
	    if (HTDirAccess == HT_DIR_FORBID) {
		FREE(filename);
		FREE(nodename);
		return HTLoadError(sink, 403, DISALLOWED_DIR_SCAN);
	    }

	    if (HTDirAccess == HT_DIR_SELECTIVE) {
		char * enable_file_name = NULL;

		HTSprintf0(&enable_file_name, "%s/%s", filename, HT_DIR_ENABLE_FILE);
		if (HTStat(enable_file_name, &stat_info) == -1) {
		    FREE(filename);
		    FREE(nodename);
		    FREE(enable_file_name);
		    return HTLoadError(sink, 403, DISALLOWED_SELECTIVE_ACCESS);
		}
	    }

	    FREE(filename);
	    FREE(nodename);
	    return HTVMSBrowseDir(addr, anchor, format_out, sink);
	}
    }

    /*
    **	Assume that the file is in Unix-style syntax if it contains a '/'
    **	after the leading one. @@
    */
    {
	FILE * fp;
	char * vmsname = strchr(filename + 1, '/') ?
		    HTVMS_name(nodename, filename) : filename + 1;
	fp = fopen(vmsname, "r", "shr=put", "shr=upd");

	/*
	**  If the file wasn't VMS syntax, then perhaps it is Ultrix.
	*/
	if (!fp) {
	    char * ultrixname = 0;
	    CTRACE((tfp, "HTLoadFile: Can't open as %s\n", vmsname));
	    HTSprintf0(&ultrixname, "%s::\"%s\"", nodename, filename);
	    fp = fopen(ultrixname, "r", "shr=put", "shr=upd");
	    if (!fp) {
		CTRACE((tfp, "HTLoadFile: Can't open as %s\n",
			    ultrixname));
	    }
	    FREE(ultrixname);
	}
	if (fp) {
	    char *semicolon = NULL;

	    if (HTEditable(vmsname)) {
		HTAtom * put = HTAtom_for("PUT");
		HTList * methods = HTAnchor_methods(anchor);
		if (HTList_indexOf(methods, put) == (-1)) {
		    HTList_addObject(methods, put);
		}
	    }
	    /*
	    **	Trim vmsname at semicolon if a version number was
	    **	included, so it doesn't interfere with the check
	    **	for a compressed file. - FM
	    */
	    if ((semicolon = strchr(vmsname, ';')) != NULL)
		*semicolon = '\0';
	    /*
	    **	Fake a Content-Encoding for compressed files. - FM
	    */
	    if (!IsUnityEnc(myEncoding)) {
		/*
		 *  We already know from the call to HTFileFormat above
		 *  that this is a compressed file, no need to look at
		 *  the filename again. - kw
		 */
#ifdef USE_ZLIB
		if (strcmp(format_out->name, "www/download") != 0 &&
		    (!strcmp(HTAtom_name(myEncoding), "gzip") ||
		     !strcmp(HTAtom_name(myEncoding), "x-gzip"))) {
		    fclose(fp);
		    if (semicolon != NULL)
			*semicolon = ';';
		    gzfp = gzopen(vmsname, BIN_R);

		    CTRACE((tfp, "HTLoadFile: gzopen of `%s' gives %p\n",
				vmsname, (void*)gzfp));
		    internal_decompress = cftGzip;
		} else
#endif	/* USE_ZLIB */
#ifdef USE_BZLIB
		if (strcmp(format_out->name, "www/download") != 0 &&
		    (!strcmp(HTAtom_name(myEncoding), "bzip2") ||
		     !strcmp(HTAtom_name(myEncoding), "x-bzip2"))) {
		    fclose(fp);
		    if (semicolon != NULL)
			*semicolon = ';';
		    bzfp = BZ2_bzopen(vmsname, BIN_R);

		    CTRACE((tfp, "HTLoadFile: bzopen of `%s' gives %p\n",
				vmsname, (void*)bzfp));
		    use_zread = YES;
		} else
#endif	/* USE_BZLIB */
		{
		    StrAllocCopy(anchor->content_type, format->name);
		    StrAllocCopy(anchor->content_encoding, HTAtom_name(myEncoding));
		    format = HTAtom_for("www/compressed");
		}
	    } else {
		/* FIXME: should we check if suffix is after ']' or ':' ? */
		CompressFileType cft = HTCompressFileType(vmsname, "._-", &dot);

		if (cft != cftNone) {
		    char *cp = NULL;

		    StrAllocCopy(cp, vmsname);
		    cp[dot - vmsname] = '\0';
		    format = HTFileFormat(cp, &encoding, NULL);
		    FREE(cp);
		    format = HTCharsetFormat(format, anchor,
					     UCLYhndl_HTFile_for_unspec);
		    StrAllocCopy(anchor->content_type, format->name);
		}

		switch (cft) {
		case cftCompress:
		    StrAllocCopy(anchor->content_encoding, "x-compress");
		    format = HTAtom_for("www/compressed");
		    break;
		case cftGzip:
		    StrAllocCopy(anchor->content_encoding, "x-gzip");
#ifdef USE_ZLIB
		    if (strcmp(format_out->name, "www/download") != 0) {
			fclose(fp);
			if (semicolon != NULL)
			    *semicolon = ';';
			gzfp = gzopen(vmsname, BIN_R);

			CTRACE((tfp, "HTLoadFile: gzopen of `%s' gives %p\n",
				    vmsname, (void*)gzfp));
			internal_decompress = cftGzip;
		    }
#else  /* USE_ZLIB */
		    format = HTAtom_for("www/compressed");
#endif	/* USE_ZLIB */
		    break;
		case cftBzip2:
		    StrAllocCopy(anchor->content_encoding, "x-bzip2");
#ifdef USE_BZLIB
		    if (strcmp(format_out->name, "www/download") != 0) {
			fclose(fp);
			if (semicolon != NULL)
			    *semicolon = ';';
			bzfp = BZ2_bzopen(vmsname, BIN_R);

			CTRACE((tfp, "HTLoadFile: bzopen of `%s' gives %p\n",
				    vmsname, (void*)bzfp));
			internal_decompress = cfgBzip2;
		    }
#else  /* USE_BZLIB */
		    format = HTAtom_for("www/compressed");
#endif	/* USE_BZLIB */
		    break;
		case cftNone:
		    break;
		}
	    }
	    if (semicolon != NULL)
		*semicolon = ';';
	    FREE(filename);
	    FREE(nodename);
#if defined(USE_ZLIB) || defined(USE_BZLIB)
	    if (internal_decompress != cftNone) {
		switch (internal_decompress) {
#ifdef USE_ZLIB
		case cftCompress:
		case cftGzip:
		    failed_decompress = (gzfp == 0);
		    break;
#endif
#ifdef USE_BZLIB
		case cftBzip2:
		    failed_decompress = (bzfp == 0);
		    break;
#endif
		default:
		    failed_decompress = YES;
		    break;
		}
		if (failed_decompress) {
		    status = HTLoadError(NULL,
					 -(HT_ERROR),
					 FAILED_OPEN_COMPRESSED_FILE);
		} else {
		    char * sugfname = NULL;
		    if (anchor->SugFname) {
			StrAllocCopy(sugfname, anchor->SugFname);
		    } else {
			char * anchor_path = HTParse(anchor->address, "",
						     PARSE_PATH + PARSE_PUNCTUATION);
			char * lastslash;
			HTUnEscape(anchor_path);
			lastslash = strrchr(anchor_path, '/');
			if (lastslash)
			    StrAllocCopy(sugfname, lastslash + 1);
			FREE(anchor_path);
		    }
		    FREE(anchor->content_encoding);
		    if (sugfname && *sugfname)
			HTCheckFnameForCompression(&sugfname, anchor,
						   TRUE);
		    if (sugfname && *sugfname)
			StrAllocCopy(anchor->SugFname, sugfname);
		    FREE(sugfname);
#ifdef USE_BZLIB
		    if (bzfp)
			status = HTParseBzFile(format, format_out,
					       anchor,
					       bzfp, sink);
#endif
#ifdef USE_ZLIB
		    if (gzfp)
			status = HTParseGzFile(format, format_out,
					       anchor,
					       gzfp, sink);
#endif
		}
	    } else
#endif /* USE_ZLIB || USE_BZLIB */
	    {
		status = HTParseFile(format, format_out, anchor, fp, sink);
		fclose(fp);
	    }
	    return status;
	}  /* If successful open */
	FREE(filename);
    }

#else /* not VMS: */

    FREE(filename);

    /*
    **	For unix, we try to translate the name into the name of a
    **	transparently mounted file.
    **
    **	Not allowed in secure (HTClientHost) situations. TBL 921019
    */
#ifndef NO_UNIX_IO
    /*	Need protection here for telnet server but not httpd server. */

    if (!HTSecure) {		/* try local file system */
	char * localname = HTLocalName(addr);
	struct stat dir_info;

#ifdef HAVE_READDIR
	/*
	**  Multiformat handling.
	**
	**  If needed, scan directory to find a good file.
	**  Bug:  We don't stat the file to find the length.
	*/
	if ((strlen(localname) > strlen(MULTI_SUFFIX)) &&
	    (0 == strcmp(localname + strlen(localname) - strlen(MULTI_SUFFIX),
			 MULTI_SUFFIX))) {
	    DIR *dp = 0;
	    BOOL forget_multi = NO;

	    STRUCT_DIRENT * dirbuf;
	    float best = (float) NO_VALUE_FOUND; /* So far best is bad */
	    HTFormat best_rep = NULL;	/* Set when rep found */
	    HTAtom * best_enc = NULL;
	    char * best_name = NULL;	/* Best dir entry so far */

	    char *base = strrchr(localname, '/');
	    int baselen = 0;

	    if (!base || base == localname) {
		forget_multi = YES;
	    } else {
		*base++ = '\0';		/* Just got directory name */
		baselen = strlen(base)- strlen(MULTI_SUFFIX);
		base[baselen] = '\0';	/* Chop off suffix */

		dp = opendir(localname);
	    }
	    if (forget_multi || !dp) {
		FREE(localname);
		FREE(nodename);
		return HTLoadError(sink, 500, FAILED_DIR_SCAN);
	    }

	    while ((dirbuf = readdir(dp)) != NULL) {
		/*
		**  While there are directory entries to be read...
		*/
#if !(defined(DOSPATH) || defined(__EMX__))
		if (dirbuf->d_ino == 0)
		    continue;	/* if the entry is not being used, skip it */
#endif
		if ((int)strlen(dirbuf->d_name) > baselen &&	 /* Match? */
		    !strncmp(dirbuf->d_name, base, baselen)) {
		    HTAtom * enc;
		    HTFormat rep = HTFileFormat(dirbuf->d_name, &enc, NULL);
		    float filevalue = HTFileValue(dirbuf->d_name);
		    float value = HTStackValue(rep, format_out,
						filevalue,
						0L  /* @@@@@@ */);
		    if (value <= 0.0) {
			char *atomname = NULL;
			CompressFileType cft = HTCompressFileType(dirbuf->d_name, ".", &dot);
			char * cp = NULL;

			enc = NULL;
			if (cft != cftNone) {
			    StrAllocCopy(cp, dirbuf->d_name);
			    cp[dot - dirbuf->d_name] = '\0';
			    format = HTFileFormat(cp, NULL, NULL);
			    FREE(cp);
			    value = HTStackValue(format, format_out,
						 filevalue, 0);
			    switch (cft) {
			    case cftCompress:
				atomname = "application/x-compressed";
				break;
			    case cftGzip:
				atomname = "application/x-gzip";
				break;
			    case cftBzip2:
				atomname = "application/x-bzip2";
				break;
			    case cftNone:
				break;
			    }
			}

			if (atomname != NULL) {
			    value = HTStackValue(format, format_out,
						 filevalue, 0);
			    if (value <= 0.0) {
				format = HTAtom_for(atomname);
				value = HTStackValue(format, format_out,
						     filevalue, 0);
			    }
			    if (value <= 0.0) {
				format = HTAtom_for("www/compressed");
				value = HTStackValue(format, format_out,
						     filevalue, 0);
			    }
			}
		    }
		    if (value != NO_VALUE_FOUND) {
			CTRACE((tfp, "HTLoadFile: value of presenting %s is %f\n",
				    HTAtom_name(rep), value));
			if  (value > best) {
			    best_rep = rep;
			    best_enc = enc;
			    best = value;
			    StrAllocCopy(best_name, dirbuf->d_name);
			}
		    }	/* if best so far */
		 } /* if match */

	    }