parse.c

一个很有名的浏览器

/* Parsing of FTP `ls' directory output. */
/* $Id: parse.c,v 1.38.2.1 2005/04/06 09:30:31 jonas Exp $ */

/* Parts of this file was part of GNU Wget
 * Copyright (C) 1995, 1996, 1997, 2000, 2001 Free Software Foundation, Inc. */

#ifdef HAVE_CONFIG_H
#include "config.h"
#endif

#include <errno.h>
#include <stdio.h>
#include <stdlib.h>
#ifdef HAVE_STRING_H
# include <string.h>
#else
# include <strings.h>
#endif
#ifdef HAVE_UNISTD_H
# include <unistd.h>
#endif
#include <sys/stat.h>
#include <sys/types.h>

#include "elinks.h"

#include "osdep/ascii.h"
#include "protocol/date.h"
#include "protocol/ftp/parse.h"
#include "util/conv.h"
#include "util/string.h"
#include "util/ttime.h"


#define skip_space_end(src, end) \
	do { while ((src) < (end) && isspace(*(src))) (src)++; } while (0)

#define skip_nonspace_end(src, end) \
	do { while ((src) < (end) && !isspace(*(src))) (src)++; } while (0)

static long
parse_ftp_number(unsigned char **src, unsigned char *end, long from, long to)
{
	long number = 0;
	unsigned char *pos = *src;

	for (; pos < end && isdigit(*pos); pos++)
		number = (*pos - '0') + 10 * number;

	*src = pos;

	if (number < from || to < number)
		return -1;

	return number;
}


/* Parser for the EPLF format (see http://pobox.com/~djb/proto/eplf.txt).
 *
 * Lines end with \r\n (CR-LF), but that is handled elsewhere.
 *
 * Some example EPLF response, with the filename separator (tab) displayed as a
 * space:
 */
#ifdef DEBUG_FTP_PARSER
static unsigned char ftp_eplf_responses[] =
 "+i8388621.48594,m825718503,r,s280,\tdjb.html\r\n"
 "+i8388621.50690,m824255907,/,\t514\r\n"
 "+i8388621.48598,m824253270,r,s612,\t514.html\r\n";
#endif

enum ftp_eplf {
	FTP_EPLF_FILENAME	= ASCII_TAB,	/* Filename follows */
	FTP_EPLF_PLAINFILE	= 'r',		/* RETR is possible */
	FTP_EPLF_DIRECTORY	= '/',		/* CWD is possible */
	FTP_EPLF_SIZE		= 's',		/* File size follows */
	FTP_EPLF_MTIME		= 'm',		/* Modification time follows */
	FTP_EPLF_ID		= 'i',		/* Unique file id follows */
};

static struct ftp_file_info *
parse_ftp_eplf_response(struct ftp_file_info *info, unsigned char *src, int len)
{
	/* Skip the '+'-char which starts the line. */
	unsigned char *end = src + len;
	unsigned char *pos = src++;

	/* Handle the series of facts about the file. */

	for (; src < end && pos; src = pos + 1) {
		/* Find the end of the current fact. */
		pos = memchr(src, ',', end - src);
		if (!pos) pos = end;

		switch (*src++) {
		case FTP_EPLF_FILENAME:
			if (src >= end) break;
			info->name.source = src;
			info->name.length = end - src;
			return info;

		case FTP_EPLF_DIRECTORY:
			info->type = FTP_FILE_DIRECTORY;
			break;

		case FTP_EPLF_PLAINFILE:
			info->type = FTP_FILE_PLAINFILE;
			break;

		case FTP_EPLF_SIZE:
			if (src >= pos) break;
			info->size = parse_ftp_number(&src, pos, 0, LONG_MAX);
			break;

		case FTP_EPLF_MTIME:
			if (src >= pos) break;
			info->mtime = (time_t) parse_ftp_number(&src, pos, 0, LONG_MAX);
			break;
		case FTP_EPLF_ID:
			/* Not used */
			break;
		}
	}

	return NULL;
}


/* Parser for UNIX-style listing: */
#ifdef DEBUG_FTP_PARSER
static unsigned char ftp_unix_responses[] =
  /* UNIX-style listing, without inum and without blocks: */
 "-rw-r--r--   1 root     other        531 Jan 29 03:26 README\r\n"
 "dr-xr-xr-x   2 root     other        512 Apr  8  1994 etc\r\n"
 "dr-xr-xr-x   2 root     512 Apr  8  1994 et\r\nc"
 "lrwxrwxrwx   1 root     other          7 Jan 25 00:17 bin -> usr/bin\r\n"
 /* Also produced by Microsoft's FTP servers for Windows: */
 "----------   1 owner    group         1803128 Jul 10 10:18 ls-lR.Z\r\n"
 "d---------   1 owner    group               0 May  9 19:45 Softlib\r\n"
 /* Also WFTPD for MSDOS: */
 "-rwxrwxrwx   1 noone    nogroup      322 Aug 19  1996 message.ftp\r\n"
 /* Also NetWare: */
 "d [R----F--] supervisor            512       Jan 16 18:53    login\r\n"
 "- [R----F--] rhesus             214059       Oct 20 15:27    cx.exe\r\n"
 /* Also NetPresenz for the Mac: */
 "-------r--         326  1391972  1392298 Nov 22  1995 MegaPhone.sit\r\n"
 "drwxrwxr-x               folder        2 May 10  1996 network\r\n";
#endif

enum ftp_unix {
	FTP_UNIX_PERMISSIONS,
	FTP_UNIX_SIZE,
	FTP_UNIX_DAY,
	FTP_UNIX_TIME,
	FTP_UNIX_NAME
};

/* Converts Un*x-style symbolic permissions to number-style ones, e.g. string
 * rwxr-xr-x to 755.
 * Borrowed from lftp source code by Alexander V. Lukyanov.
 * On parse error, it returns 0. */
static int
parse_ftp_unix_permissions(const unsigned char *src, int len)
{
	int perms = 0;

	if (len != 9
	    && !(len == 10 && src[9] == '+'))   /* ACL tag */
		return 0;

	/* User permissions */
	switch (src[0]) {
	case('r'): perms |= S_IRUSR; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[1]) {
	case('w'): perms |= S_IWUSR; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[2]) {
	case('S'): perms |= S_ISUID; break;
	case('s'): perms |= S_ISUID; /* fall-through */
	case('x'): perms |= S_IXUSR; break;
	case('-'): break;
	default: return 0;
	}

	/* Group permissions */
	switch (src[3]) {
	case('r'): perms |= S_IRGRP; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[4]) {
	case('w'): perms |= S_IWGRP; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[5]) {
	case('S'): perms |= S_ISGID; break;
	case('s'): perms |= S_ISGID; /* fall-through */
	case('x'): perms |= S_IXGRP; break;
	case('-'): break;
	default: return 0;
	}

	/* Others permissions */
	switch (src[6]) {
	case('r'): perms |= S_IROTH; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[7]) {
	case('w'): perms |= S_IWOTH; break;
	case('-'): break;
	default: return 0;
	}

	switch (src[8]) {
	case('T'): perms |= S_ISVTX; break;
	case('t'): perms |= S_ISVTX; /* fall-through */
	case('x'): perms |= S_IXOTH; break;
	case('l'):
	case('L'): perms |= S_ISGID; perms &= ~S_IXGRP; break;
	case('-'): break;
	default: return 0;
	}

	return perms;
}

static struct ftp_file_info *
parse_ftp_unix_response(struct ftp_file_info *info, unsigned char *src, int len)
{
	unsigned char *end = src + len;
	enum ftp_file_type type = *src++;
	unsigned char *pos = src;
	struct tm mtime;
	enum ftp_unix fact;

	/* Decide the file type. */
	switch (type) {
	case FTP_FILE_PLAINFILE:
	case FTP_FILE_DIRECTORY:
	case FTP_FILE_SYMLINK:
		info->type = type;
		break;

	default:
		info->type = FTP_FILE_UNKNOWN;
	}

	memset(&mtime, 0, sizeof(mtime));
	mtime.tm_isdst = -1;

	skip_space_end(src, end);
	fact = FTP_UNIX_PERMISSIONS;

	for (pos = src; src < end; src = pos) {
		skip_nonspace_end(pos, end);

		switch (fact) {
		case FTP_UNIX_PERMISSIONS:
			/* We wanna know permissions as well! And I decided to
			 * completely ignore the NetWare perms, they are very
			 * rare and of some nonstandart format.  If you want
			 * them, though, I'll accept patch enabling them.
			 * --pasky */
			if (pos - src == 9)	/* 9 is length of "rwxrwxrwx". */
				info->permissions = parse_ftp_unix_permissions(src, 9);
			fact = FTP_UNIX_SIZE;
			break;

		case FTP_UNIX_SIZE:
			/* Search for the size and month name combo: */
			if (info->size != FTP_SIZE_UNKNOWN
			    && pos - src == 3) {
				int month = parse_month((const unsigned char **) &src, pos);

				if (month != -1) {
					fact = FTP_UNIX_DAY;
					mtime.tm_mon = month;
					break;
				}
			}

			if (!isdigit(*src)) {
				info->size = FTP_SIZE_UNKNOWN;
				break;
			}

			info->size = parse_ftp_number(&src, pos, 0, LONG_MAX);
			break;

		case FTP_UNIX_DAY:
			mtime.tm_mday = parse_day((const unsigned char **) &src, pos);
			fact = FTP_UNIX_TIME;
			break;

		case FTP_UNIX_TIME:
			/* This ought to be either the time, or the
			 * year. Let's be flexible! */
			fact = FTP_UNIX_NAME;

			/* We must deal with digits.  */
			if (!isdigit (*src))
				break;

			/* If we have a number x, it's a year. If we have x:y,
			 * it's hours and minutes. */
			if (!memchr(src, ':', pos - src)) {
				mtime.tm_year = parse_year((const unsigned char **) &src, pos);
				break;
			}

			if (!parse_time((const unsigned char **) &src, &mtime, pos)) {
				mtime.tm_hour = mtime.tm_min = mtime.tm_sec = 0;
			}
			break;

		case FTP_UNIX_NAME:
			/* Since the file name may contain spaces use @end as the
			 * token ending and not @pos. */

			info->name.source = src;
			info->name.length = end - src;

			/* Some FTP sites choose to have ls -F as their default
			 * LIST output, which marks the symlinks with a trailing
			 * `@', directory names with a trailing `/' and
			 * executables with a trailing `*'. This is no problem
			 * unless encountering a symbolic link ending with `@',
			 * or an executable ending with `*' on a server without
			 * default -F output. I believe these cases are very