urlcp settings control miscellaneous page
If on (the default), try to allow certain bad chunked
Transfer-Encoding information if encountered in a response
(e.g. missing chunk size), by just passing through remaining data
as-is (because chunked coding is mostly clear text anyway). If
off, fail the transaction. Note that regardless of the setting's
value, bad chunked information may indicate corrupt data in the
response. May help recover a fetch when the server erroneously
reports chunked coding in the response. Added in version
6.00.1315620000 20110909; previous versions behaved as if setting
alarmclose(boolean) Whether to use an
alarm()to terminate a blocking connection
close(). The default is off. Added in version 4.04.1050700000 20030418.
badheadermidok(boolean) Whether to accept malformed headers in the middle (i.e. not last) of the response headers. If on, malformed headers will be discarded if followed by at least one valid header. If off, or no valid header follows, the malformed header will be considered the start of the body (which it might very well be). Added (and defaults to on) in version 5.01.1102538511 20041208. Returns previous setting.
checkidleconneof(boolean) If on (the default), idle connections in the Keep-Alive cache are checked for EOF (i.e. server closure) before being reused. The server may have timed out the connection while the socket was idle, which would otherwise cause the next fetch to fail. Added in version 5.01.1115130972 20050503 (default off in previous versions). Returns previous setting.
Clears the cache of "bad" (non-responsive) proxies, which are
set to lower priority in PAC responses when other proxies are
here). Also clears the last PAC
fetch timestamp (see
here). Added in version 7.05.
closeidleconn(no arguments) Closes any currently idle connections in the Keep-Alive cache. Returns 1 if successful, 0 on error. Added in version 5.00.1093895662 20040830.
defaults(no arguments) Resets all
urlcpsettings to their default values.
delaysave(boolean) Sets whether to delay the saving of output when using
<submit TOFILE=$file>until the connection starts returning data. The default is off, e.g. open the file immediately (always deleting previous copy). Turning this setting on is useful when repeatedly downloading to the same file, e.g. obtaining a periodic update of a large data file. The original file will then be preserved if the new fetch fails immediately (e.g. remote server down), yet saves the disk space of a separate backup copy. Added in version 4.0.997840000 20010814.
domvalue $dom $valueSets the value of the DOM item indicated by
<urlinfo domvalue>. Added in version 5.
emptyhttp09ok(boolean) Whether to accept empty HTTP/0.9 responses, i.e. a 0-byte response with no headers. Such responses are technically legal (an empty HTTP/0.9 document), but since few pre-HTTP/1.0 servers exist, are more likely indicative of a server error. If off, an error message is issued and an error is set. Added (and defaults to off) in version 5.01.1097502096 20041011. Returns previous setting.
linger(boolean) Sets whether to set
SO_LINGERtime of 4 seconds on sockets. Added (and defaults to off) in version 5.01.1105153893 20050107. Returns previous setting.
reparent(string) If given a full path (e.g. "
rerootreparent mode and uses that path as the local tree root. If given a full URL (e.g. "
absreparent mode and uses that URL as the page's URL (this is not recommended; links may become incorrect).
reparentimg(boolean) If true (default), image links will be reparented; if false, they will not. Only significant if
reparentmode(string) The returned HTML from a page will be reparented: all the links will be changed in the raw document returned. How the links are modified depends on the mode:
absor 1 Make all links absolute. With this mode, a page can be fetched from a remote site and the returned document placed directly in a local source tree, and even relative links will correctly point to the original locations. If a URL is set with
reparent, the page is reparented as if it were fetched from there, instead of its actual location (the default). (Setting a URL is not recommended, as the links may become incorrect.)
rerootor 2 Re-path same-site links as if the entire remote site were being copied locally to a subtree rooted at the URL path given with
reparent. For example, with a
/local/tree, if the URL
http://somesite.com/dir/page.htmlis fetched, it is assumed it will be saved to
/local/tree/dir/page.html. Thus a link such as
/local/tree/top/list.html. The link
/local/tree/upone.html. If no path is set with
reparent, links become relative, as if the root were
mirroror 3 Make all links absolute, URL-encode them, and prefix the
For an email message, change all internal links (to other
parts of the message) to their safe filenames, as same-dir
relative links. All other (external) links will be made
absolute. This mode is used internally by
when reparenting. Note: since it requires additional
parsed email message information, it cannot currently explicitly
be used by Vortex scripts.
Change all external links - those referring to outside the
page's current directory or below - to have a prefix of
thismessage:", to prevent their access. This can be
used to hide/disable external references in HTML email message
bodies during web display, after the HTML has been
message-reparented by the
offor 0 Turn off all reparenting. Added in version 3.01.968705387 20000911. Default.
reparentmodesettings do not affect the links returned by <urlinfo links>.
sendemptycontent(boolean) Whether to set
Content-Length: 0for empty requests. Some servers will time out, expecting an EOF from the client, if an empty request is sent with no
Content-Length. Added (and defaults to on) in version 5.01.1097006042 20041005. Returns previous setting.
shutdownwr(boolean) Turns on or off the use of
shutdown(SHUT_WR)on HTTP or Gopher sockets when all data has been sent and the connection is not to be re-used (e.g. Keep-Alive has expired or is not in use). This sends an EOF to the server to indicate that the client has finished sending data. Some broken servers may expect such an EOF even if
Content-Lengthis set properly in the request, and may thus time out the request waiting for one. (Note that for
shutdown()to actually be used, it may be necessary to disable Keep-Alive via
<urlcp maxconnrequests 1>.) Added (and defaults to on) in version 5.01.1105300267 20050109. Returns previous setting.
urlcanonslash(boolean) Whether to canonicalize backslashes ("
\") to forward slashes ("
/") in URLs. On by default. Turning off may impair URL parsing.
Whether to collapse multiple forward slashes (e.g. "
to a single forward slash in the path part of URLs. Note: does
not affect the double-slash that immediately follows the protocol
plus colon in some URL protocols. Off by default. Added in
version 7.03.1434400000 20150615.
urlcp returns 1 on success, or 0 or nothing on error,
except as noted under specific options.
<urlcp "maxpgsize" "1MB"> <urlcp "timeout" 300>
urlcp function was added Mar. 26 1997. Various settings
were added later.