man/cat1/wget.1.txt

WGET(1) GNU Wget WGET(1)
 
 
 
NAME
       Wget - The non-interactive network downloader.
 
SYNOPSIS
       wget [option]... [URL]...
 
DESCRIPTION
       GNU Wget is a free utility for non-interactive download
       of files from the Web. It supports HTTP, HTTPS, and FTP
       protocols, as well as retrieval through HTTP proxies.
 
       Wget is non-interactive, meaning that it can work in the
       background, while the user is not logged on. This
       allows you to start a retrieval and disconnect from the
       system, letting Wget finish the work. By contrast, most
       of the Web browsers require constant user's presence,
       which can be a great hindrance when transferring a lot
       of data.
 
       Wget can follow links in HTML and XHTML pages and create
       local versions of remote web sites, fully recreating the
       directory structure of the original site. This is some-
       times referred to as "recursive downloading." While
       doing that, Wget respects the Robot Exclusion Standard
       (/robots.txt). Wget can be instructed to convert the
       links in downloaded HTML files to the local files for
       offline viewing.
 
       Wget has been designed for robustness over slow or
       unstable network connections; if a download fails due to
       a network problem, it will keep retrying until the whole
       file has been retrieved. If the server supports reget-
       ting, it will instruct the server to continue the down-
       load from where it left off.
 
OPTIONS
       Option Syntax
 
       Since Wget uses GNU getopt to process command-line argu-
       ments, every option has a long form along with the short
       one. Long options are more convenient to remember, but
       take time to type. You may freely mix different option
       styles, or specify options after the command-line argu-
       ments. Thus you may write:
 
               wget -r --tries=10 http://fly.srk.fer.hr/ -o log
 
       The space between the option accepting an argument and
       the argument may be omitted. Instead of -o log you can
       write -olog.
 
       You may put several options that do not require argu-
       ments together, like:
 
               wget -drc <URL>
 
       This is a complete equivalent of:
 
               wget -d -r -c <URL>
 
       Since the options can be specified after the arguments,
       you may terminate them with --. So the following will
       try to download URL -x, reporting failure to log:
 
               wget -o log -- -x
 
       The options that accept comma-separated lists all
       respect the convention that specifying an empty list
       clears its value. This can be useful to clear the
       .wgetrc settings. For instance, if your .wgetrc sets
       "exclude_directories" to /cgi-bin, the following example
       will first reset it, and then set it to exclude /~nobody
       and /~somebody. You can also clear the lists in
       .wgetrc.
 
               wget -X " -X /~nobody,/~somebody
 
       Most options that do not accept arguments are boolean
       options, so named because their state can be captured
       with a yes-or-no ("boolean") variable. For example,
       --follow-ftp tells Wget to follow FTP links from HTML
       files and, on the other hand, --no-glob tells it not to
       perform file globbing on FTP URLs. A boolean option is
       either affirmative or negative (beginning with --no).
       All such options share several properties.
 
       Unless stated otherwise, it is assumed that the default
       behavior is the opposite of what the option accom-
       plishes. For example, the documented existence of
       --follow-ftp assumes that the default is to not follow
       FTP links from HTML pages.
 
       Affirmative options can be negated by prepending the
       --no- to the option name; negative options can be
       negated by omitting the --no- prefix. This might seem
       superfluous---if the default for an affirmative option
       is to not do something, then why provide a way to
       explicitly turn it off? But the startup file may in
       fact change the default. For instance, using "fol-
       low_ftp = off" in .wgetrc makes Wget not follow FTP
       links by default, and using --no-follow-ftp is the only
       way to restore the factory default from the command
       line.
 
       Basic Startup Options
 
 
       -V
       --version
           Display the version of Wget.
 
       -h
       --help
           Print a help message describing all of Wget's com-
           mand-line options.
 
       -b
       --background
           Go to background immediately after startup. If no
           output file is specified via the -o, output is redi-
           rected to wget-log.
 
       -e command
       --execute command
           Execute command as if it were a part of .wgetrc. A
           command thus invoked will be executed after the com-
           mands in .wgetrc, thus taking precedence over them.
           If you need to specify more than one wgetrc command,
           use multiple instances of -e.
 
       Logging and Input File Options
 
 
       -o logfile
       --output-file=logfile
           Log all messages to logfile. The messages are nor-
           mally reported to standard error.
 
       -a logfile
       --append-output=logfile
           Append to logfile. This is the same as -o, only it
           appends to logfile instead of overwriting the old
           log file. If logfile does not exist, a new file is
           created.
 
       -d
       --debug
           Turn on debug output, meaning various information
           important to the developers of Wget if it does not
           work properly. Your system administrator may have
           chosen to compile Wget without debug support, in
           which case -d will not work. Please note that com-
           piling with debug support is always safe---Wget com-
           piled with the debug support will not print any
           debug info unless requested with -d.
 
       -q
       --quiet
           Turn off Wget's output.
 
       -v
       --verbose
           Turn on verbose output, with all the available data.
           The default output is verbose.
 
       -nv
       --no-verbose
           Turn off verbose without being completely quiet (use
           -q for that), which means that error messages and
           basic information still get printed.
 
       -i file
       --input-file=file
           Read URLs from file. If - is specified as file,
           URLs are read from the standard input. (Use ./- to
           read from a file literally named -.)
 
           If this function is used, no URLs need be present on
           the command line. If there are URLs both on the
           command line and in an input file, those on the com-
           mand lines will be the first ones to be retrieved.
           The file need not be an HTML document (but no harm
           if it is)---it is enough if the URLs are just listed
           sequentially.
 
           However, if you specify --force-html, the document
           will be regarded as html. In that case you may have
           problems with relative links, which you can solve
           either by adding "<base href="url">" to the docu-
           ments or by specifying --base=url on the command
           line.
 
       -F
       --force-html
           When input is read from a file, force it to be
           treated as an HTML file. This enables you to
           retrieve relative links from existing HTML files on
           your local disk, by adding "<base href="url">" to
           HTML, or using the --base command-line option.
 
       -B URL
       --base=URL
           Prepends URL to relative links read from the file
           specified with the -i option.
 
       Download Options
 
 
       --bind-address=ADDRESS
           When making client TCP/IP connections, bind to
           ADDRESS on the local machine. ADDRESS may be speci-
           fied as a hostname or IP address. This option can
           be useful if your machine is bound to multiple IPs.
 
       -t number
       --tries=number
           Set number of retries to number. Specify 0 or inf
           for infinite retrying. The default is to retry 20
           times, with the exception of fatal errors like "con-
           nection refused" or "not found" (404), which are not
           retried.
 
       -O file
       --output-document=file
           The documents will not be written to the appropriate
           files, but all will be concatenated together and
           written to file. If - is used as file, documents
           will be printed to standard output, disabling link
           conversion. (Use ./- to print to a file literally
           named -.)
 
           Use of -O is not intended to mean simply "use the
           name file instead of the one in the URL;" rather, it
           is analogous to shell redirection: wget -O file
           http://foo is intended to work like wget -O -
           http://foo > file; file will be truncated immedi-
           ately, and all downloaded content will be written
           there.
 
           For this reason, -N (for timestamp-checking) is not
           supported in combination with -O: since file is
           always newly created, it will always have a very new
           timestamp. A warning will be issued if this combina-
           tion is used.
 
           Similarly, using -r or -p with -O may not work as
           you expect: Wget won't just download the first file
           to file and then download the rest to their normal
           names: all downloaded content will be placed in
           file. This was disabled in version 1.11, but has
           been reinstated (with a warning) in 1.11.2, as there
           are some cases where this behavior can actually have
           some use.
 
           Note that a combination with -k is only permitted
           when downloading a single document, as in that case
           it will just convert all relative URIs to external
           ones; -k makes no sense for multiple URIs when
           they're all being downloaded to a single file.
 
       -nc
       --no-clobber
           If a file is downloaded more than once in the same
           directory, Wget's behavior depends on a few options,
           including -nc. In certain cases, the local file
           will be clobbered, or overwritten, upon repeated
           download. In other cases it will be preserved.
 
           When running Wget without -N, -nc, -r, or p, down-
           loading the same file in the same directory will
           result in the original copy of file being preserved
           and the second copy being named file.1. If that
           file is downloaded yet again, the third copy will be
           named file.2, and so on. When -nc is specified,
           this behavior is suppressed, and Wget will refuse to
           download newer copies of file. Therefore,
           ""no-clobber"" is actually a misnomer in this
           mode---it's not clobbering that's prevented (as the
           numeric suffixes were already preventing clobber-
           ing), but rather the multiple version saving that's
           prevented.
 
           When running Wget with -r or -p, but without -N or
           -nc, re-downloading a file will result in the new
           copy simply overwriting the old. Adding -nc will
           prevent this behavior, instead causing the original
           version to be preserved and any newer copies on the
           server to be ignored.
 
           When running Wget with -N, with or without -r or -p,
           the decision as to whether or not to download a
           newer copy of a file depends on the local and remote
           timestamp and size of the file. -nc may not be
           specified at the same time as -N.
 
           Note that when -nc is specified, files with the suf-
           fixes .html or .htm will be loaded from the local
           disk and parsed as if they had been retrieved from
           the Web.
 
       -c
       --continue
           Continue getting a partially-downloaded file. This
           is useful when you want to finish up a download
           started by a previous instance of Wget, or by
           another program. For instance:
 
                   wget -c ftp://sunsite.doc.ic.ac.uk/ls-lR.Z
 
           If there is a file named ls-lR.Z in the current
           directory, Wget will assume that it is the first
           portion of the remote file, and will ask the server
           to continue the retrieval from an offset equal to
           the length of the local file.
 
           Note that you don't need to specify this option if
           you just want the current invocation of Wget to
           retry downloading a file should the connection be
           lost midway through. This is the default behavior.
           -c only affects resumption of downloads started
           prior to this invocation of Wget, and whose local
           files are still sitting around.
 
           Without -c, the previous example would just download
           the remote file to ls-lR.Z.1, leaving the truncated
           ls-lR.Z file alone.
 
           Beginning with Wget 1.7, if you use -c on a non-
           empty file, and it turns out that the server does
           not support continued downloading, Wget will refuse
           to start the download from scratch, which would
           effectively ruin existing contents. If you really
           want the download to start from scratch, remove the
           file.
 
           Also beginning with Wget 1.7, if you use -c on a
           file which is of equal size as the one on the
           server, Wget will refuse to download the file and
           print an explanatory message. The same happens when
           the file is smaller on the server than locally (pre-
           sumably because it was changed on the server since
           your last download attempt)---because "continuing"
           is not meaningful, no download occurs.
 
           On the other side of the coin, while using -c, any
           file that's bigger on the server than locally will
           be considered an incomplete download and only
           "(length(remote) - length(local))" bytes will be
           downloaded and tacked onto the end of the local
           file. This behavior can be desirable in certain
           cases---for instance, you can use wget -c to down-
           load just the new portion that's been appended to a
           data collection or log file.
 
           However, if the file is bigger on the server because
           it's been changed, as opposed to just appended to,
           you'll end up with a garbled file. Wget has no way
           of verifying that the local file is really a valid
           prefix of the remote file. You need to be espe-
           cially careful of this when using -c in conjunction
           with -r, since every file will be considered as an
           "incomplete download" candidate.
 
           Another instance where you'll get a garbled file if
           you try to use -c is if you have a lame HTTP proxy
           that inserts a "transfer interrupted" string into
           the local file. In the future a "rollback" option
           may be added to deal with this case.
 
           Note that -c only works with FTP servers and with
           HTTP servers that support the "Range" header.
 
       --progress=type
           Select the type of the progress indicator you wish
           to use. Legal indicators are "dot" and "bar".
 
           The "bar" indicator is used by default. It draws an
           ASCII progress bar graphics (a.k.a "thermometer"
           display) indicating the status of retrieval. If the
           output is not a TTY, the "dot" bar will be used by
           default.
 
           Use --progress=dot to switch to the "dot" display.
           It traces the retrieval by printing dots on the
           screen, each dot representing a fixed amount of
           downloaded data.
 
           When using the dotted retrieval, you may also set
           the style by specifying the type as dot:style. Dif-
           ferent styles assign different meaning to one dot.
           With the "default" style each dot represents 1K,
           there are ten dots in a cluster and 50 dots in a
           line. The "binary" style has a more "computer"-like
           orientation---8K dots, 16-dots clusters and 48 dots
           per line (which makes for 384K lines). The "mega"
           style is suitable for downloading very large
           files---each dot represents 64K retrieved, there are
           eight dots in a cluster, and 48 dots on each line
           (so each line contains 3M).
 
           Note that you can set the default style using the
           "progress" command in .wgetrc. That setting may be
           overridden from the command line. The exception is
           that, when the output is not a TTY, the "dot"
           progress will be favored over "bar". To force the
           bar output, use --progress=bar:force.
 
       -N
       --timestamping
           Turn on time-stamping.
 
       -S
       --server-response
           Print the headers sent by HTTP servers and responses
           sent by FTP servers.
 
       --spider
           When invoked with this option, Wget will behave as a
           Web spider, which means that it will not download
           the pages, just check that they are there. For
           example, you can use Wget to check your bookmarks:
 
                   wget --spider --force-html -i bookmarks.html
 
           This feature needs much more work for Wget to get
           close to the functionality of real web spiders.
 
       -T seconds
       --timeout=seconds
           Set the network timeout to seconds seconds. This is
           equivalent to specifying --dns-timeout, --con-
           nect-timeout, and --read-timeout, all at the same
           time.
 
           When interacting with the network, Wget can check
           for timeout and abort the operation if it takes too
           long. This prevents anomalies like hanging reads
           and infinite connects. The only timeout enabled by
           default is a 900-second read timeout. Setting a
           timeout to 0 disables it altogether. Unless you
           know what you are doing, it is best not to change
           the default timeout settings.
 
           All timeout-related options accept decimal values,
           as well as subsecond values. For example, 0.1 sec-
           onds is a legal (though unwise) choice of timeout.
           Subsecond timeouts are useful for checking server
           response times or for testing network latency.
 
       --dns-timeout=seconds
           Set the DNS lookup timeout to seconds seconds. DNS
           lookups that don't complete within the specified
           time will fail. By default, there is no timeout on
           DNS lookups, other than that implemented by system
           libraries.
 
       --connect-timeout=seconds
           Set the connect timeout to seconds seconds. TCP
           connections that take longer to establish will be
           aborted. By default, there is no connect timeout,
           other than that implemented by system libraries.
 
       --read-timeout=seconds
           Set the read (and write) timeout to seconds seconds.
           The "time" of this timeout refers to idle time: if,
           at any point in the download, no data is received
           for more than the specified number of seconds, read-
           ing fails and the download is restarted. This
           option does not directly affect the duration of the
           entire download.
 
           Of course, the remote server may choose to terminate
           the connection sooner than this option requires.
           The default read timeout is 900 seconds.
 
       --limit-rate=amount
           Limit the download speed to amount bytes per second.
           Amount may be expressed in bytes, kilobytes with the
           k suffix, or megabytes with the m suffix. For exam-
           ple, --limit-rate=20k will limit the retrieval rate
           to 20KB/s. This is useful when, for whatever rea-
           son, you don't want Wget to consume the entire
           available bandwidth.
 
           This option allows the use of decimal numbers, usu-
           ally in conjunction with power suffixes; for exam-
           ple, --limit-rate=2.5k is a legal value.
 
           Note that Wget implements the limiting by sleeping
           the appropriate amount of time after a network read
           that took less time than specified by the rate.
           Eventually this strategy causes the TCP transfer to
           slow down to approximately the specified rate. How-
           ever, it may take some time for this balance to be
           achieved, so don't be surprised if limiting the rate
           doesn't work well with very small files.
 
       -w seconds
       --wait=seconds
           Wait the specified number of seconds between the
           retrievals. Use of this option is recommended, as
           it lightens the server load by making the requests
           less frequent. Instead of in seconds, the time can
           be specified in minutes using the "m" suffix, in
           hours using "h" suffix, or in days using "d" suffix.
 
           Specifying a large value for this option is useful
           if the network or the destination host is down, so
           that Wget can wait long enough to reasonably expect
           the network error to be fixed before the retry. The
           waiting interval specified by this function is
           influenced by "--random-wait", which see.
 
       --waitretry=seconds
           If you don't want Wget to wait between every
           retrieval, but only between retries of failed down-
           loads, you can use this option. Wget will use lin-
           ear backoff, waiting 1 second after the first fail-
           ure on a given file, then waiting 2 seconds after
           the second failure on that file, up to the maximum
           number of seconds you specify. Therefore, a value
           of 10 will actually make Wget wait up to (1 + 2 +
           ... + 10) = 55 seconds per file.
 
           Note that this option is turned on by default in the
           global wgetrc file.
 
       --random-wait
           Some web sites may perform log analysis to identify
           retrieval programs such as Wget by looking for sta-
           tistically significant similarities in the time
           between requests. This option causes the time
           between requests to vary between 0.5 and 1.5 * wait
           seconds, where wait was specified using the --wait
           option, in order to mask Wget's presence from such
           analysis.
 
           A 2001 article in a publication devoted to develop-
           ment on a popular consumer platform provided code to
           perform this analysis on the fly. Its author sug-
           gested blocking at the class C address level to
           ensure automated retrieval programs were blocked
           despite changing DHCP-supplied addresses.
 
           The --random-wait option was inspired by this ill-
           advised recommendation to block many unrelated users
           from a web site due to the actions of one.
 
       --no-proxy
           Don't use proxies, even if the appropriate *_proxy
           environment variable is defined.
 
       -Q quota
       --quota=quota
           Specify download quota for automatic retrievals.
           The value can be specified in bytes (default), kilo-
           bytes (with k suffix), or megabytes (with m suffix).
 
           Note that quota will never affect downloading a sin-
           gle file. So if you specify wget -Q10k
           ftp://wuarchive.wustl.edu/ls-lR.gz, all of the
           ls-lR.gz will be downloaded. The same goes even
           when several URLs are specified on the command-line.
           However, quota is respected when retrieving either
           recursively, or from an input file. Thus you may
           safely type wget -Q2m -i sites---download will be
           aborted when the quota is exceeded.
 
           Setting quota to 0 or to inf unlimits the download
           quota.
 
       --no-dns-cache
           Turn off caching of DNS lookups. Normally, Wget
           remembers the IP addresses it looked up from DNS so
           it doesn't have to repeatedly contact the DNS server
           for the same (typically small) set of hosts it
           retrieves from. This cache exists in memory only; a
           new Wget run will contact DNS again.
 
           However, it has been reported that in some situa-
           tions it is not desirable to cache host names, even
           for the duration of a short-running application like
           Wget. With this option Wget issues a new DNS lookup
           (more precisely, a new call to "gethostbyname" or
           "getaddrinfo") each time it makes a new connection.
           Please note that this option will not affect caching
           that might be performed by the resolving library or
           by an external caching layer, such as NSCD.
 
           If you don't understand exactly what this option
           does, you probably won't need it.
 
       --restrict-file-names=mode
           Change which characters found in remote URLs may
           show up in local file names generated from those
           URLs. Characters that are restricted by this option
           are escaped, i.e. replaced with %HH, where HH is the
           hexadecimal number that corresponds to the
           restricted character.
 
           By default, Wget escapes the characters that are not
           valid as part of file names on your operating sys-
           tem, as well as control characters that are typi-
           cally unprintable. This option is useful for chang-
           ing these defaults, either because you are download-
           ing to a non-native partition, or because you want
           to disable escaping of the control characters.
 
           When mode is set to "unix", Wget escapes the charac-
           ter / and the control characters in the ranges 0--31
           and 128--159. This is the default on Unix-like
           OS'es.
 
           When mode is set to "windows", Wget escapes the
           characters \, |, /, :, ?, ", *, <, >, and the con-
           trol characters in the ranges 0--31 and 128--159.
           In addition to this, Wget in Windows mode uses +
           instead of : to separate host and port in local file
           names, and uses @ instead of ? to separate the query
           portion of the file name from the rest. Therefore,
           a URL that would be saved as
           www.xemacs.org:4300/search.pl?input=blah in Unix
           mode would be saved as
           www.xemacs.org+4300/search.pl@input=blah in Windows
           mode. This mode is the default on Windows.
 
           If you append ,nocontrol to the mode, as in
           unix,nocontrol, escaping of the control characters
           is also switched off. You can use
           --restrict-file-names=nocontrol to turn off escaping
           of control characters without affecting the choice
           of the OS to use as file name restriction mode.
 
       -4
       --inet4-only
       -6
       --inet6-only
           Force connecting to IPv4 or IPv6 addresses. With
           --inet4-only or -4, Wget will only connect to IPv4
           hosts, ignoring AAAA records in DNS, and refusing to
           connect to IPv6 addresses specified in URLs. Con-
           versely, with --inet6-only or -6, Wget will only
           connect to IPv6 hosts and ignore A records and IPv4
           addresses.
 
           Neither options should be needed normally. By
           default, an IPv6-aware Wget will use the address
           family specified by the host's DNS record. If the
           DNS responds with both IPv4 and IPv6 addresses, Wget
           will try them in sequence until it finds one it can
           connect to. (Also see "--prefer-family" option
           described below.)
 
           These options can be used to deliberately force the
           use of IPv4 or IPv6 address families on dual family
           systems, usually to aid debugging or to deal with
           broken network configuration. Only one of
           --inet6-only and --inet4-only may be specified at
           the same time. Neither option is available in Wget
           compiled without IPv6 support.
 
       --prefer-family=IPv4/IPv6/none
           When given a choice of several addresses, connect to
           the addresses with specified address family first.
           IPv4 addresses are preferred by default.
 
           This avoids spurious errors and connect attempts
           when accessing hosts that resolve to both IPv6 and
           IPv4 addresses from IPv4 networks. For example,
           www.kame.net resolves to
           2001:200:0:8002:203:47ff:fea5:3085 and to
           203.178.141.194. When the preferred family is
           "IPv4", the IPv4 address is used first; when the
           preferred family is "IPv6", the IPv6 address is used
           first; if the specified value is "none", the address
           order returned by DNS is used without change.
 
           Unlike -4 and -6, this option doesn't inhibit access
           to any address family, it only changes the order in
           which the addresses are accessed. Also note that
           the reordering performed by this option is sta-
           ble---it doesn't affect order of addresses of the
           same family. That is, the relative order of all
           IPv4 addresses and of all IPv6 addresses remains
           intact in all cases.
 
       --retry-connrefused
           Consider "connection refused" a transient error and
           try again. Normally Wget gives up on a URL when it
           is unable to connect to the site because failure to
           connect is taken as a sign that the server is not
           running at all and that retries would not help.
           This option is for mirroring unreliable sites whose
           servers tend to disappear for short periods of time.
 
       --user=user
       --password=password
           Specify the username user and password password for
           both FTP and HTTP file retrieval. These parameters
           can be overridden using the --ftp-user and
           --ftp-password options for FTP connections and the
           --http-user and --http-password options for HTTP
           connections.
 
       Directory Options
 
 
       -nd
       --no-directories
           Do not create a hierarchy of directories when
           retrieving recursively. With this option turned on,
           all files will get saved to the current directory,
           without clobbering (if a name shows up more than
           once, the filenames will get extensions .n).
 
       -x
       --force-directories
           The opposite of -nd---create a hierarchy of directo-
           ries, even if one would not have been created other-
           wise. E.g. wget -x http://fly.srk.fer.hr/robots.txt
           will save the downloaded file to fly.srk.fer.hr/ro-
           bots.txt.
 
       -nH
       --no-host-directories
           Disable generation of host-prefixed directories. By
           default, invoking Wget with -r
           http://fly.srk.fer.hr/ will create a structure of
           directories beginning with fly.srk.fer.hr/. This
           option disables such behavior.
 
       --protocol-directories
           Use the protocol name as a directory component of
           local file names. For example, with this option,
           wget -r http://host will save to http/host/...
           rather than just to host/....
 
       --cut-dirs=number
           Ignore number directory components. This is useful
           for getting a fine-grained control over the direc-
           tory where recursive retrieval will be saved.
 
           Take, for example, the directory at
           ftp://ftp.xemacs.org/pub/xemacs/. If you retrieve
           it with -r, it will be saved locally under
           ftp.xemacs.org/pub/xemacs/. While the -nH option
           can remove the ftp.xemacs.org/ part, you are still
           stuck with pub/xemacs. This is where --cut-dirs
           comes in handy; it makes Wget not "see" number
           remote directory components. Here are several exam-
           ples of how --cut-dirs option works.
 
                   No options -> ftp.xemacs.org/pub/xemacs/
                   -nH -> pub/xemacs/
                   -nH --cut-dirs=1 -> xemacs/
                   -nH --cut-dirs=2 -> .
 
                   --cut-dirs=1 -> ftp.xemacs.org/xemacs/
                   ...
 
           If you just want to get rid of the directory struc-
           ture, this option is similar to a combination of -nd
           and -P. However, unlike -nd, --cut-dirs does not
           lose with subdirectories---for instance, with -nH
           --cut-dirs=1, a beta/ subdirectory will be placed to
           xemacs/beta, as one would expect.
 
       -P prefix
       --directory-prefix=prefix
           Set directory prefix to prefix. The directory pre-
           fix is the directory where all other files and sub-
           directories will be saved to, i.e. the top of the
           retrieval tree. The default is . (the current
           directory).
 
       HTTP Options
 
 
       -E
       --html-extension
           If a file of type application/xhtml+xml or text/html
           is downloaded and the URL does not end with the reg-
           exp \.[Hh][Tt][Mm][Ll]?, this option will cause the
           suffix .html to be appended to the local filename.
           This is useful, for instance, when you're mirroring
           a remote site that uses .asp pages, but you want the
           mirrored pages to be viewable on your stock Apache
           server. Another good use for this is when you're
           downloading CGI-generated materials. A URL like
           http://site.com/article.cgi?25 will be saved as
           article.cgi?25.html.
 
           Note that filenames changed in this way will be re-
           downloaded every time you re-mirror a site, because
           Wget can't tell that the local X.html file corre-
           sponds to remote URL X (since it doesn't yet know
           that the URL produces output of type text/html or
           application/xhtml+xml. To prevent this re-download-
           ing, you must use -k and -K so that the original
           version of the file will be saved as X.orig.
 
       --http-user=user
       --http-password=password
           Specify the username user and password password on
           an HTTP server. According to the type of the chal-
           lenge, Wget will encode them using either the
           "basic" (insecure), the "digest", or the Windows
           "NTLM" authentication scheme.
 
           Another way to specify username and password is in
           the URL itself. Either method reveals your password
           to anyone who bothers to run "ps". To prevent the
           passwords from being seen, store them in .wgetrc or
           .netrc, and make sure to protect those files from
           other users with "chmod". If the passwords are
           really important, do not leave them lying in those
           files either---edit the files and delete them after
           Wget has started the download.
 
       --no-cache
           Disable server-side cache. In this case, Wget will
           send the remote server an appropriate directive
           (Pragma: no-cache) to get the file from the remote
           service, rather than returning the cached version.
           This is especially useful for retrieving and flush-
           ing out-of-date documents on proxy servers.
 
           Caching is allowed by default.
 
       --no-cookies
           Disable the use of cookies. Cookies are a mechanism
           for maintaining server-side state. The server sends
           the client a cookie using the "Set-Cookie" header,
           and the client responds with the same cookie upon
           further requests. Since cookies allow the server
           owners to keep track of visitors and for sites to
           exchange this information, some consider them a
           breach of privacy. The default is to use cookies;
           however, storing cookies is not on by default.
 
       --load-cookies file
           Load cookies from file before the first HTTP
           retrieval. file is a textual file in the format
           originally used by Netscape's cookies.txt file.
 
           You will typically use this option when mirroring
           sites that require that you be logged in to access
           some or all of their content. The login process
           typically works by the web server issuing an HTTP
           cookie upon receiving and verifying your creden-
           tials. The cookie is then resent by the browser
           when accessing that part of the site, and so proves
           your identity.
 
           Mirroring such a site requires Wget to send the same
           cookies your browser sends when communicating with
           the site. This is achieved by --load-cookies---sim-
           ply point Wget to the location of the cookies.txt
           file, and it will send the same cookies your browser
           would send in the same situation. Different
           browsers keep textual cookie files in different
           locations:
 
           @asis<Netscape 4.x.>
               The cookies are in ~/.netscape/cookies.txt.
 
           @asis<Mozilla and Netscape 6.x.>
               Mozilla's cookie file is also named cookies.txt,
               located somewhere under ~/.mozilla, in the
               directory of your profile. The full path usu-
               ally ends up looking somewhat like
               ~/.mozilla/default/some-weird-string/cook-
               ies.txt.
 
           @asis<Internet Explorer.>
               You can produce a cookie file Wget can use by
               using the File menu, Import and Export, Export
               Cookies. This has been tested with Internet
               Explorer 5; it is not guaranteed to work with
               earlier versions.
 
           @asis<Other browsers.>
               If you are using a different browser to create
               your cookies, --load-cookies will only work if
               you can locate or produce a cookie file in the
               Netscape format that Wget expects.
 
           If you cannot use --load-cookies, there might still
           be an alternative. If your browser supports a
           "cookie manager", you can use it to view the cookies
           used when accessing the site you're mirroring.
           Write down the name and value of the cookie, and
           manually instruct Wget to send those cookies,
           bypassing the "official" cookie support:
 
                   wget --no-cookies --header "Cookie: <name>=<value>"
 
       --save-cookies file
           Save cookies to file before exiting. This will not
           save cookies that have expired or that have no
           expiry time (so-called "session cookies"), but also
           see --keep-session-cookies.
 
       --keep-session-cookies
           When specified, causes --save-cookies to also save
           session cookies. Session cookies are normally not
           saved because they are meant to be kept in memory
           and forgotten when you exit the browser. Saving
           them is useful on sites that require you to log in
           or to visit the home page before you can access some
           pages. With this option, multiple Wget runs are
           considered a single browser session as far as the
           site is concerned.
 
           Since the cookie file format does not normally carry
           session cookies, Wget marks them with an expiry
           timestamp of 0. Wget's --load-cookies recognizes
           those as session cookies, but it might confuse other
           browsers. Also note that cookies so loaded will be
           treated as other session cookies, which means that
           if you want --save-cookies to preserve them again,
           you must use --keep-session-cookies again.
 
       --ignore-length
           Unfortunately, some HTTP servers (CGI programs, to
           be more precise) send out bogus "Content-Length"
           headers, which makes Wget go wild, as it thinks not
           all the document was retrieved. You can spot this
           syndrome if Wget retries getting the same document
           again and again, each time claiming that the (other-
           wise normal) connection has closed on the very same
           byte.
 
           With this option, Wget will ignore the "Con-
           tent-Length" header---as if it never existed.
 
       --header=header-line
           Send header-line along with the rest of the headers
           in each HTTP request. The supplied header is sent
           as-is, which means it must contain name and value
           separated by colon, and must not contain newlines.
 
           You may define more than one additional header by
           specifying --header more than once.
 
                   wget --header='Accept-Charset: iso-8859-2' \
                        --header='Accept-Language: hr' \
                          http://fly.srk.fer.hr/
 
           Specification of an empty string as the header value
           will clear all previous user-defined headers.
 
           As of Wget 1.10, this option can be used to override
           headers otherwise generated automatically. This
           example instructs Wget to connect to localhost, but
           to specify foo.bar in the "Host" header:
 
                   wget --header="Host: foo.bar" http://localhost/
 
           In versions of Wget prior to 1.10 such use of
           --header caused sending of duplicate headers.
 
       --max-redirect=number
           Specifies the maximum number of redirections to fol-
           low for a resource. The default is 20, which is
           usually far more than necessary. However, on those
           occasions where you want to allow more (or fewer),
           this is the option to use.
 
       --proxy-user=user
       --proxy-password=password
           Specify the username user and password password for
           authentication on a proxy server. Wget will encode
           them using the "basic" authentication scheme.
 
           Security considerations similar to those with
           --http-password pertain here as well.
 
       --referer=url
           Include `Referer: url' header in HTTP request. Use-
           ful for retrieving documents with server-side pro-
           cessing that assume they are always being retrieved
           by interactive web browsers and only come out prop-
           erly when Referer is set to one of the pages that
           point to them.
 
       --save-headers
           Save the headers sent by the HTTP server to the
           file, preceding the actual contents, with an empty
           line as the separator.
 
       -U agent-string
       --user-agent=agent-string
           Identify as agent-string to the HTTP server.
 
           The HTTP protocol allows the clients to identify
           themselves using a "User-Agent" header field. This
           enables distinguishing the WWW software, usually for
           statistical purposes or for tracing of protocol vio-
           lations. Wget normally identifies as Wget/version,
           version being the current version number of Wget.
 
           However, some sites have been known to impose the
           policy of tailoring the output according to the
           "User-Agent"-supplied information. While this is
           not such a bad idea in theory, it has been abused by
           servers denying information to clients other than
           (historically) Netscape or, more frequently,
           Microsoft Internet Explorer. This option allows you
           to change the "User-Agent" line issued by Wget. Use
           of this option is discouraged, unless you really
           know what you are doing.
 
           Specifying empty user agent with --user-agent=""
           instructs Wget not to send the "User-Agent" header
           in HTTP requests.
 
       --post-data=string
       --post-file=file
           Use POST as the method for all HTTP requests and
           send the specified data in the request body.
           "--post-data" sends string as data, whereas
           "--post-file" sends the contents of file. Other
           than that, they work in exactly the same way.
 
           Please be aware that Wget needs to know the size of
           the POST data in advance. Therefore the argument to
           "--post-file" must be a regular file; specifying a
           FIFO or something like /dev/stdin won't work. It's
           not quite clear how to work around this limitation
           inherent in HTTP/1.0. Although HTTP/1.1 introduces
           chunked transfer that doesn't require knowing the
           request length in advance, a client can't use chun-
           ked unless it knows it's talking to an HTTP/1.1
           server. And it can't know that until it receives a
           response, which in turn requires the request to have
           been completed -- a chicken-and-egg problem.
 
           Note: if Wget is redirected after the POST request
           is completed, it will not send the POST data to the
           redirected URL. This is because URLs that process
           POST often respond with a redirection to a regular
           page, which does not desire or accept POST. It is
           not completely clear that this behavior is optimal;
           if it doesn't work out, it might be changed in the
           future.
 
           This example shows how to log to a server using POST
           and then proceed to download the desired pages, pre-
           sumably only accessible to authorized users:
 
                   # Log in to the server. This can be done only once.
                   wget --save-cookies cookies.txt \
                        --post-data 'user=foo&password=bar' \
                        http://server.com/auth.php
 
                   # Now grab the page or pages we care about.
                   wget --load-cookies cookies.txt \
                        -p http://server.com/interesting/article.php
 
           If the server is using session cookies to track user
           authentication, the above will not work because
           --save-cookies will not save them (and neither will
           browsers) and the cookies.txt file will be empty.
           In that case use --keep-session-cookies along with
           --save-cookies to force saving of session cookies.
 
       --content-disposition
           If this is set to on, experimental (not fully-func-
           tional) support for "Content-Disposition" headers is
           enabled. This can currently result in extra round-
           trips to the server for a "HEAD" request, and is
           known to suffer from a few bugs, which is why it is
           not currently enabled by default.
 
           This option is useful for some file-downloading CGI
           programs that use "Content-Disposition" headers to
           describe what the name of a downloaded file should
           be.
 
       --auth-no-challenge
           If this option is given, Wget will send Basic HTTP
           authentication information (plaintext username and
           password) for all requests, just like Wget 1.10.2
           and prior did by default.
 
           Use of this option is not recommended, and is
           intended only to support some few obscure servers,
           which never send HTTP authentication challenges, but
           accept unsolicited auth info, say, in addition to
           form-based authentication.
 
       HTTPS (SSL/TLS) Options
 
       To support encrypted HTTP (HTTPS) downloads, Wget must
       be compiled with an external SSL library, currently
       OpenSSL. If Wget is compiled without SSL support, none
       of these options are available.
 
       --secure-protocol=protocol
           Choose the secure protocol to be used. Legal values
           are auto, SSLv2, SSLv3, and TLSv1. If auto is used,
           the SSL library is given the liberty of choosing the
           appropriate protocol automatically, which is
           achieved by sending an SSLv2 greeting and announcing
           support for SSLv3 and TLSv1. This is the default.
 
           Specifying SSLv2, SSLv3, or TLSv1 forces the use of
           the corresponding protocol. This is useful when
           talking to old and buggy SSL server implementations
           that make it hard for OpenSSL to choose the correct
           protocol version. Fortunately, such servers are
           quite rare.
 
       --no-check-certificate
           Don't check the server certificate against the
           available certificate authorities. Also don't
           require the URL host name to match the common name
           presented by the certificate.
 
           As of Wget 1.10, the default is to verify the
           server's certificate against the recognized certifi-
           cate authorities, breaking the SSL handshake and
           aborting the download if the verification fails.
           Although this provides more secure downloads, it
           does break interoperability with some sites that
           worked with previous Wget versions, particularly
           those using self-signed, expired, or otherwise
           invalid certificates. This option forces an "inse-
           cure" mode of operation that turns the certificate
           verification errors into warnings and allows you to
           proceed.
 
           If you encounter "certificate verification" errors
           or ones saying that "common name doesn't match
           requested host name", you can use this option to
           bypass the verification and proceed with the down-
           load. Only use this option if you are otherwise
           convinced of the site's authenticity, or if you
           really don't care about the validity of its certifi-
           cate. It is almost always a bad idea not to check
           the certificates when transmitting confidential or
           important data.
 
       --certificate=file
           Use the client certificate stored in file. This is
           needed for servers that are configured to require
           certificates from the clients that connect to them.
           Normally a certificate is not required and this
           switch is optional.
 
       --certificate-type=type
           Specify the type of the client certificate. Legal
           values are PEM (assumed by default) and DER, also
           known as ASN1.
 
       --private-key=file
           Read the private key from file. This allows you to
           provide the private key in a file separate from the
           certificate.
 
       --private-key-type=type
           Specify the type of the private key. Accepted val-
           ues are PEM (the default) and DER.
 
       --ca-certificate=file
           Use file as the file with the bundle of certificate
           authorities ("CA") to verify the peers. The cer-
           tificates must be in PEM format.
 
           Without this option Wget looks for CA certificates
           at the system-specified locations, chosen at OpenSSL
           installation time.
 
       --ca-directory=directory
           Specifies directory containing CA certificates in
           PEM format. Each file contains one CA certificate,
           and the file name is based on a hash value derived
           from the certificate. This is achieved by process-
           ing a certificate directory with the "c_rehash"
           utility supplied with OpenSSL. Using --ca-directory
           is more efficient than --ca-certificate when many
           certificates are installed because it allows Wget to
           fetch certificates on demand.
 
           Without this option Wget looks for CA certificates
           at the system-specified locations, chosen at OpenSSL
           installation time.
 
       --random-file=file
           Use file as the source of random data for seeding
           the pseudo-random number generator on systems with-
           out /dev/random.
 
           On such systems the SSL library needs an external
           source of randomness to initialize. Randomness may
           be provided by EGD (see --egd-file below) or read
           from an external source specified by the user. If
           this option is not specified, Wget looks for random
           data in $RANDFILE or, if that is unset, in
           $HOME/.rnd. If none of those are available, it is
           likely that SSL encryption will not be usable.
 
           If you're getting the "Could not seed OpenSSL PRNG;
           disabling SSL." error, you should provide random
           data using some of the methods described above.
 
       --egd-file=file
           Use file as the EGD socket. EGD stands for Entropy
           Gathering Daemon, a user-space program that collects
           data from various unpredictable system sources and
           makes it available to other programs that might need
           it. Encryption software, such as the SSL library,
           needs sources of non-repeating randomness to seed
           the random number generator used to produce crypto-
           graphically strong keys.
 
           OpenSSL allows the user to specify his own source of
           entropy using the "RAND_FILE" environment variable.
           If this variable is unset, or if the specified file
           does not produce enough randomness, OpenSSL will
           read random data from EGD socket specified using
           this option.
 
           If this option is not specified (and the equivalent
           startup command is not used), EGD is never con-
           tacted. EGD is not needed on modern Unix systems
           that support /dev/random.
 
       FTP Options
 
 
       --ftp-user=user
       --ftp-password=password
           Specify the username user and password password on
           an FTP server. Without this, or the corresponding
           startup option, the password defaults to -wget@,
           normally used for anonymous FTP.
 
           Another way to specify username and password is in
           the URL itself. Either method reveals your password
           to anyone who bothers to run "ps". To prevent the
           passwords from being seen, store them in .wgetrc or
           .netrc, and make sure to protect those files from
           other users with "chmod". If the passwords are
           really important, do not leave them lying in those
           files either---edit the files and delete them after
           Wget has started the download.
 
       --no-remove-listing
           Don't remove the temporary .listing files generated
           by FTP retrievals. Normally, these files contain
           the raw directory listings received from FTP
           servers. Not removing them can be useful for debug-
           ging purposes, or when you want to be able to easily
           check on the contents of remote server directories
           (e.g. to verify that a mirror you're running is com-
           plete).
 
           Note that even though Wget writes to a known file-
           name for this file, this is not a security hole in
           the scenario of a user making .listing a symbolic
           link to /etc/passwd or something and asking "root"
           to run Wget in his or her directory. Depending on
           the options used, either Wget will refuse to write
           to .listing, making the globbing/recur-
           sion/time-stamping operation fail, or the symbolic
           link will be deleted and replaced with the actual
           .listing file, or the listing will be written to a
           .listing.number file.
 
           Even though this situation isn't a problem, though,
           "root" should never run Wget in a non-trusted user's
           directory. A user could do something as simple as
           linking index.html to /etc/passwd and asking "root"
           to run Wget with -N or -r so the file will be over-
           written.
 
       --no-glob
           Turn off FTP globbing. Globbing refers to the use
           of shell-like special characters (wildcards), like
           *, ?, [ and ] to retrieve more than one file from
           the same directory at once, like:
 
                   wget ftp://gnjilux.srk.fer.hr/*.msg
 
           By default, globbing will be turned on if the URL
           contains a globbing character. This option may be
           used to turn globbing on or off permanently.
 
           You may have to quote the URL to protect it from
           being expanded by your shell. Globbing makes Wget
           look for a directory listing, which is system-spe-
           cific. This is why it currently works only with
           Unix FTP servers (and the ones emulating Unix "ls"
           output).
 
       --no-passive-ftp
           Disable the use of the passive FTP transfer mode.
           Passive FTP mandates that the client connect to the
           server to establish the data connection rather than
           the other way around.
 
           If the machine is connected to the Internet
           directly, both passive and active FTP should work
           equally well. Behind most firewall and NAT configu-
           rations passive FTP has a better chance of working.
           However, in some rare firewall configurations,
           active FTP actually works when passive FTP doesn't.
           If you suspect this to be the case, use this option,
           or set "passive_ftp=off" in your init file.
 
       --retr-symlinks
           Usually, when retrieving FTP directories recursively
           and a symbolic link is encountered, the linked-to
           file is not downloaded. Instead, a matching sym-
           bolic link is created on the local filesystem. The
           pointed-to file will not be downloaded unless this
           recursive retrieval would have encountered it sepa-
           rately and downloaded it anyway.
 
           When --retr-symlinks is specified, however, symbolic
           links are traversed and the pointed-to files are
           retrieved. At this time, this option does not cause
           Wget to traverse symlinks to directories and recurse
           through them, but in the future it should be
           enhanced to do this.
 
           Note that when retrieving a file (not a directory)
           because it was specified on the command-line, rather
           than because it was recursed to, this option has no
           effect. Symbolic links are always traversed in this
           case.
 
       --no-http-keep-alive
           Turn off the "keep-alive" feature for HTTP down-
           loads. Normally, Wget asks the server to keep the
           connection open so that, when you download more than
           one document from the same server, they get trans-
           ferred over the same TCP connection. This saves
           time and at the same time reduces the load on the
           server.
 
           This option is useful when, for some reason, persis-
           tent (keep-alive) connections don't work for you,
           for example due to a server bug or due to the
           inability of server-side scripts to cope with the
           connections.
 
       Recursive Retrieval Options
 
 
       -r
       --recursive
           Turn on recursive retrieving.
 
       -l depth
       --level=depth
           Specify recursion maximum depth level depth. The
           default maximum depth is 5.
 
       --delete-after
           This option tells Wget to delete every single file
           it downloads, after having done so. It is useful
           for pre-fetching popular pages through a proxy,
           e.g.:
 
                   wget -r -nd --delete-after http://whatever.com/~popular/page/
 
           The -r option is to retrieve recursively, and -nd to
           not create directories.
 
           Note that --delete-after deletes files on the local
           machine. It does not issue the DELE command to
           remote FTP sites, for instance. Also note that when
           --delete-after is specified, --convert-links is
           ignored, so .orig files are simply not created in
           the first place.
 
       -k
       --convert-links
           After the download is complete, convert the links in
           the document to make them suitable for local view-
           ing. This affects not only the visible hyperlinks,
           but any part of the document that links to external
           content, such as embedded images, links to style
           sheets, hyperlinks to non-HTML content, etc.
 
           Each link will be changed in one of the two ways:
 
           * The links to files that have been downloaded by
               Wget will be changed to refer to the file they
               point to as a relative link.
 
               Example: if the downloaded file /foo/doc.html
               links to /bar/img.gif, also downloaded, then the
               link in doc.html will be modified to point to
               ../bar/img.gif. This kind of transformation
               works reliably for arbitrary combinations of
               directories.
 
           * The links to files that have not been downloaded
               by Wget will be changed to include host name and
               absolute path of the location they point to.
 
               Example: if the downloaded file /foo/doc.html
               links to /bar/img.gif (or to ../bar/img.gif),
               then the link in doc.html will be modified to
               point to http://hostname/bar/img.gif.
 
           Because of this, local browsing works reliably: if a
           linked file was downloaded, the link will refer to
           its local name; if it was not downloaded, the link
           will refer to its full Internet address rather than
           presenting a broken link. The fact that the former
           links are converted to relative links ensures that
           you can move the downloaded hierarchy to another
           directory.
 
           Note that only at the end of the download can Wget
           know which links have been downloaded. Because of
           that, the work done by -k will be performed at the
           end of all the downloads.
 
       -K
       --backup-converted
           When converting a file, back up the original version
           with a .orig suffix. Affects the behavior of -N.
 
       -m
       --mirror
           Turn on options suitable for mirroring. This option
           turns on recursion and time-stamping, sets infinite
           recursion depth and keeps FTP directory listings.
           It is currently equivalent to -r -N -l inf
           --no-remove-listing.
 
       -p
       --page-requisites
           This option causes Wget to download all the files
           that are necessary to properly display a given HTML
           page. This includes such things as inlined images,
           sounds, and referenced stylesheets.
 
           Ordinarily, when downloading a single HTML page, any
           requisite documents that may be needed to display it
           properly are not downloaded. Using -r together with
           -l can help, but since Wget does not ordinarily dis-
           tinguish between external and inlined documents, one
           is generally left with "leaf documents" that are
           missing their requisites.
 
           For instance, say document 1.html contains an
           "<IMG>" tag referencing 1.gif and an "<A>" tag
           pointing to external document 2.html. Say that
           2.html is similar but that its image is 2.gif and it
           links to 3.html. Say this continues up to some
           arbitrarily high number.
 
           If one executes the command:
 
                   wget -r -l 2 http://<site>/1.html
 
           then 1.html, 1.gif, 2.html, 2.gif, and 3.html will
           be downloaded. As you can see, 3.html is without
           its requisite 3.gif because Wget is simply counting
           the number of hops (up to 2) away from 1.html in
           order to determine where to stop the recursion.
           However, with this command:
 
                   wget -r -l 2 -p http://<site>/1.html
 
           all the above files and 3.html's requisite 3.gif
           will be downloaded. Similarly,
 
                   wget -r -l 1 -p http://<site>/1.html
 
           will cause 1.html, 1.gif, 2.html, and 2.gif to be
           downloaded. One might think that:
 
                   wget -r -l 0 -p http://<site>/1.html
 
           would download just 1.html and 1.gif, but unfortu-
           nately this is not the case, because -l 0 is equiva-
           lent to -l inf---that is, infinite recursion. To
           download a single HTML page (or a handful of them,
           all specified on the command-line or in a -i URL
           input file) and its (or their) requisites, simply
           leave off -r and -l:
 
                   wget -p http://<site>/1.html
 
           Note that Wget will behave as if -r had been speci-
           fied, but only that single page and its requisites
           will be downloaded. Links from that page to exter-
           nal documents will not be followed. Actually, to
           download a single page and all its requisites (even
           if they exist on separate websites), and make sure
           the lot displays properly locally, this author likes
           to use a few options in addition to -p:
 
                   wget -E -H -k -K -p http://<site>/<document>
 
           To finish off this topic, it's worth knowing that
           Wget's idea of an external document link is any URL
           specified in an "<A>" tag, an "<AREA>" tag, or a
           "<LINK>" tag other than "<LINK REL="stylesheet">".
 
       --strict-comments
           Turn on strict parsing of HTML comments. The
           default is to terminate comments at the first occur-
           rence of -->.
 
           According to specifications, HTML comments are
           expressed as SGML declarations. Declaration is spe-
           cial markup that begins with <! and ends with >,
           such as <!DOCTYPE ...>, that may contain comments
           between a pair of -- delimiters. HTML comments are
           "empty declarations", SGML declarations without any
           non-comment text. Therefore, <!--foo--> is a valid
           comment, and so is <!--one-- --two-->, but
           <!--1--2--> is not.
 
           On the other hand, most HTML writers don't perceive
           comments as anything other than text delimited with
           <!-- and -->, which is not quite the same. For
           example, something like <!------------> works as a
           valid comment as long as the number of dashes is a
           multiple of four (!). If not, the comment techni-
           cally lasts until the next --, which may be at the
           other end of the document. Because of this, many
           popular browsers completely ignore the specification
           and implement what users have come to expect: com-
           ments delimited with <!-- and -->.
 
           Until version 1.9, Wget interpreted comments
           strictly, which resulted in missing links in many
           web pages that displayed fine in browsers, but had
           the misfortune of containing non-compliant comments.
           Beginning with version 1.9, Wget has joined the
           ranks of clients that implements "naive" comments,
           terminating each comment at the first occurrence of
           -->.
 
           If, for whatever reason, you want strict comment
           parsing, use this option to turn it on.
 
       Recursive Accept/Reject Options
 
 
       -A acclist --accept acclist
       -R rejlist --reject rejlist
           Specify comma-separated lists of file name suffixes
           or patterns to accept or reject. Note that if any of
           the wildcard characters, *, ?, [ or ], appear in an
           element of acclist or rejlist, it will be treated as
           a pattern, rather than a suffix.
 
       -D domain-list
       --domains=domain-list
           Set domains to be followed. domain-list is a comma-
           separated list of domains. Note that it does not
           turn on -H.
 
       --exclude-domains domain-list
           Specify the domains that are not to be followed..
 
       --follow-ftp
           Follow FTP links from HTML documents. Without this
           option, Wget will ignore all the FTP links.
 
       --follow-tags=list
           Wget has an internal table of HTML tag / attribute
           pairs that it considers when looking for linked doc-
           uments during a recursive retrieval. If a user
           wants only a subset of those tags to be considered,
           however, he or she should be specify such tags in a
           comma-separated list with this option.
 
       --ignore-tags=list
           This is the opposite of the --follow-tags option.
           To skip certain HTML tags when recursively looking
           for documents to download, specify them in a comma-
           separated list.
 
           In the past, this option was the best bet for down-
           loading a single page and its requisites, using a
           command-line like:
 
                   wget --ignore-tags=a,area -H -k -K -r http://<site>/<document>
 
           However, the author of this option came across a
           page with tags like "<LINK REL="home" HREF="/">" and
           came to the realization that specifying tags to
           ignore was not enough. One can't just tell Wget to
           ignore "<LINK>", because then stylesheets will not
           be downloaded. Now the best bet for downloading a
           single page and its requisites is the dedicated
           --page-requisites option.
 
       --ignore-case
           Ignore case when matching files and directories.
           This influences the behavior of -R, -A, -I, and -X
           options, as well as globbing implemented when down-
           loading from FTP sites. For example, with this
           option, -A *.txt will match file1.txt, but also
           file2.TXT, file3.TxT, and so on.
 
       -H
       --span-hosts
           Enable spanning across hosts when doing recursive
           retrieving.
 
       -L
       --relative
           Follow relative links only. Useful for retrieving a
           specific home page without any distractions, not
           even those from the same hosts.
 
       -I list
       --include-directories=list
           Specify a comma-separated list of directories you
           wish to follow when downloading. Elements of list
           may contain wildcards.
 
       -X list
       --exclude-directories=list
           Specify a comma-separated list of directories you
           wish to exclude from download. Elements of list may
           contain wildcards.
 
       -np
       --no-parent
           Do not ever ascend to the parent directory when
           retrieving recursively. This is a useful option,
           since it guarantees that only the files below a cer-
           tain hierarchy will be downloaded.
 
FILES
       /usr/local/etc/wgetrc
           Default location of the global startup file.
 
       .wgetrc
           User startup file.
 
BUGS
       You are welcome to submit bug reports via the GNU Wget
       bug tracker (see <http://wget.addictivecode.org/Bug-
       Tracker>).
 
       Before actually submitting a bug report, please try to
       follow a few simple guidelines.
 
       1. Please try to ascertain that the behavior you see
           really is a bug. If Wget crashes, it's a bug. If
           Wget does not behave as documented, it's a bug. If
           things work strange, but you are not sure about the
           way they are supposed to work, it might well be a
           bug, but you might want to double-check the documen-
           tation and the mailing lists.
 
       2. Try to repeat the bug in as simple circumstances as
           possible. E.g. if Wget crashes while downloading
           wget -rl0 -kKE -t5 --no-proxy http://yoyodyne.com -o
           /tmp/log, you should try to see if the crash is
           repeatable, and if will occur with a simpler set of
           options. You might even try to start the download
           at the page where the crash occurred to see if that
           page somehow triggered the crash.
 
           Also, while I will probably be interested to know
           the contents of your .wgetrc file, just dumping it
           into the debug message is probably a bad idea.
           Instead, you should first try to see if the bug
           repeats with .wgetrc moved out of the way. Only if
           it turns out that .wgetrc settings affect the bug,
           mail me the relevant parts of the file.
 
       3. Please start Wget with -d option and send us the
           resulting output (or relevant parts thereof). If
           Wget was compiled without debug support, recompile
           it---it is much easier to trace bugs with debug sup-
           port on.
 
           Note: please make sure to remove any potentially
           sensitive information from the debug log before
           sending it to the bug address. The "-d" won't go
           out of its way to collect sensitive information, but
           the log will contain a fairly complete transcript of
           Wget's communication with the server, which may
           include passwords and pieces of downloaded data.
           Since the bug address is publically archived, you
           may assume that all bug reports are visible to the
           public.
 
       4. If Wget has crashed, try to run it in a debugger,
           e.g. "gdb `which wget` core" and type "where" to get
           the backtrace. This may not work if the system
           administrator has disabled core files, but it is
           safe to try.
 
SEE ALSO
       This is not the complete manual for GNU Wget. For more
       complete information, including more detailed explana-
       tions of some of the options, and a number of commands
       available for use with .wgetrc files and the -e option,
       see the GNU Info entry for wget.
 
AUTHOR
       Originally written by Hrvoje Niksic
       <hniksic@xemacs.org>. Currently maintained by Micah
       Cowan <micah@cowan.name>.
 
COPYRIGHT
       Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002,
       2003, 2004, 2005, 2006, 2007, 2008 Free Software Founda-
       tion, Inc.
 
       Permission is granted to copy, distribute and/or modify
       this document under the terms of the GNU Free Documenta-
       tion License, Version 1.2 or any later version published
       by the Free Software Foundation; with no Invariant Sec-
       tions, no Front-Cover Texts, and no Back-Cover Texts. A
       copy of the license is included in the section entitled
       "GNU Free Documentation License".
 
 
 
GNU Wget 1.11.4 2008-06-29 WGET(1)