2011-10-15 20:25:51 +00:00
|
|
|
{- Web url logs.
|
|
|
|
-
|
2023-10-25 17:00:17 +00:00
|
|
|
- Copyright 2011-2023 Joey Hess <id@joeyh.name>
|
2011-10-15 20:25:51 +00:00
|
|
|
-
|
2019-03-13 19:48:14 +00:00
|
|
|
- Licensed under the GNU AGPL version 3 or higher.
|
2011-10-15 20:25:51 +00:00
|
|
|
-}
|
|
|
|
|
2020-07-14 16:44:35 +00:00
|
|
|
{-# LANGUAGE BangPatterns #-}
|
|
|
|
|
2011-10-15 20:25:51 +00:00
|
|
|
module Logs.Web (
|
|
|
|
URLString,
|
2023-10-25 17:00:17 +00:00
|
|
|
URLByteString,
|
2012-11-29 21:01:07 +00:00
|
|
|
getUrls,
|
2023-10-25 17:00:17 +00:00
|
|
|
getUrls',
|
2014-12-08 17:32:27 +00:00
|
|
|
getUrlsWithPrefix,
|
2011-10-15 20:36:56 +00:00
|
|
|
setUrlPresent,
|
2012-11-29 21:01:07 +00:00
|
|
|
setUrlMissing,
|
2013-08-22 22:25:21 +00:00
|
|
|
Downloader(..),
|
|
|
|
getDownloader,
|
|
|
|
setDownloader,
|
2016-01-19 19:55:32 +00:00
|
|
|
setDownloader',
|
2014-12-08 23:14:24 +00:00
|
|
|
setTempUrl,
|
|
|
|
removeTempUrl,
|
2021-04-23 15:44:10 +00:00
|
|
|
parseUrlLog,
|
2011-10-15 20:25:51 +00:00
|
|
|
) where
|
|
|
|
|
2014-12-08 23:14:24 +00:00
|
|
|
import qualified Data.Map as M
|
2023-10-25 17:00:17 +00:00
|
|
|
import qualified Data.ByteString as S
|
2021-04-23 15:44:10 +00:00
|
|
|
import qualified Data.ByteString.Lazy as L
|
2013-07-28 19:27:36 +00:00
|
|
|
|
2016-01-20 20:36:33 +00:00
|
|
|
import Annex.Common
|
2014-12-08 23:14:24 +00:00
|
|
|
import qualified Annex
|
2013-08-29 22:51:22 +00:00
|
|
|
import Logs
|
2011-10-15 20:25:51 +00:00
|
|
|
import Logs.Presence
|
|
|
|
import Logs.Location
|
2014-12-08 17:40:15 +00:00
|
|
|
import Utility.Url
|
2016-01-19 19:55:32 +00:00
|
|
|
import Annex.UUID
|
2021-04-23 17:26:10 +00:00
|
|
|
import qualified Annex.Branch
|
2016-01-19 19:55:32 +00:00
|
|
|
import qualified Types.Remote as Remote
|
2011-10-15 20:25:51 +00:00
|
|
|
|
2023-10-25 17:00:17 +00:00
|
|
|
type URLByteString = S.ByteString
|
|
|
|
|
2011-10-15 20:36:56 +00:00
|
|
|
{- Gets all urls that a key might be available from. -}
|
2011-10-15 20:25:51 +00:00
|
|
|
getUrls :: Key -> Annex [URLString]
|
2014-12-08 23:14:24 +00:00
|
|
|
getUrls key = do
|
2023-10-25 17:00:17 +00:00
|
|
|
l <- map decodeBS <$> getUrls' key
|
2014-12-08 23:14:24 +00:00
|
|
|
tmpl <- Annex.getState (maybeToList . M.lookup key . Annex.tempurls)
|
|
|
|
return (tmpl ++ l)
|
2023-10-25 17:00:17 +00:00
|
|
|
|
|
|
|
{- Note that this does not include temporary urls set with setTempUrl. -}
|
|
|
|
getUrls' :: Key -> Annex [URLByteString]
|
|
|
|
getUrls' key = do
|
|
|
|
config <- Annex.getGitConfig
|
|
|
|
go $ urlLogFile config key : oldurlLogs config key
|
2012-11-11 04:51:07 +00:00
|
|
|
where
|
|
|
|
go [] = return []
|
|
|
|
go (l:ls) = do
|
2023-12-08 17:23:03 +00:00
|
|
|
us <- presentLogInfo l
|
2012-11-11 04:51:07 +00:00
|
|
|
if null us
|
|
|
|
then go ls
|
2023-10-25 17:00:17 +00:00
|
|
|
else return $ map fromLogInfo us
|
2011-10-15 20:25:51 +00:00
|
|
|
|
2014-12-08 17:32:27 +00:00
|
|
|
getUrlsWithPrefix :: Key -> String -> Annex [URLString]
|
2015-03-27 22:49:03 +00:00
|
|
|
getUrlsWithPrefix key prefix = filter (prefix `isPrefixOf`)
|
|
|
|
. map (fst . getDownloader)
|
|
|
|
<$> getUrls key
|
2014-12-08 17:32:27 +00:00
|
|
|
|
2018-10-04 21:33:25 +00:00
|
|
|
setUrlPresent :: Key -> URLString -> Annex ()
|
|
|
|
setUrlPresent key url = do
|
2022-07-18 19:50:36 +00:00
|
|
|
-- Avoid reading the url log when not compacting, for speed.
|
|
|
|
us <- ifM (annexAlwaysCompact <$> Annex.getGitConfig)
|
|
|
|
( getUrls key
|
|
|
|
, pure mempty
|
|
|
|
)
|
2015-01-28 21:17:26 +00:00
|
|
|
unless (url `elem` us) $ do
|
|
|
|
config <- Annex.getGitConfig
|
start implementing hidden git-annex repositories
This adds a separate journal, which does not currently get committed to
an index, but is planned to be committed to .git/annex/index-private.
Changes that are regarding a UUID that is private will get written to
this journal, and so will not be published into the git-annex branch.
All log writing should have been made to indicate the UUID it's
regarding, though I've not verified this yet.
Currently, no UUIDs are treated as private yet, a way to configure that
is needed.
The implementation is careful to not add any additional IO work when
privateUUIDsKnown is False. It will skip looking at the private journal
at all. So this should be free, or nearly so, unless the feature is
used. When it is used, all branch reads will be about twice as expensive.
It is very lucky -- or very prudent design -- that Annex.Branch.change
and maybeChange are the only ways to change a file on the branch,
and Annex.Branch.set is only internal use. That let Annex.Branch.get
always yield any private information that has been recorded, without
the risk that Annex.Branch.set might be called, with a non-private UUID,
and end up leaking the private information into the git-annex branch.
And, this relies on the way git-annex union merges the git-annex branch.
When reading a file, there can be a public and a private version, and
they are just concacenated together. That will be handled the same as if
there were two diverged git-annex branches that got union merged.
2021-04-20 18:32:41 +00:00
|
|
|
addLog (Annex.Branch.RegardingUUID []) (urlLogFile config key)
|
deal better with clock skew situations, using vector clocks
* Deal with clock skew, both forwards and backwards, when logging
information to the git-annex branch.
* GIT_ANNEX_VECTOR_CLOCK can now be set to a fixed value (eg 1)
rather than needing to be advanced each time a new change is made.
* Misuse of GIT_ANNEX_VECTOR_CLOCK will no longer confuse git-annex.
When changing a file in the git-annex branch, the vector clock to use is now
determined by first looking at the current time (or GIT_ANNEX_VECTOR_CLOCK
when set), and comparing it to the newest vector clock already in use in
that file. If a newer time stamp was already in use, advance it forward by
a second instead.
When the clock is set to a time in the past, this avoids logging with
an old timestamp, which would risk that log line later being ignored in favor
of "newer" line that is really not newer.
When a log entry has been made with a clock that was set far ahead in the
future, this avoids newer information being logged with an older timestamp
and so being ignored in favor of that future-timestamped information.
Once all clocks get fixed, this will result in the vector clocks being
incremented, until finally enough time has passed that time gets back ahead
of the vector clock value, and then it will return to usual operation.
(This latter situation is not ideal, but it seems the best that can be done.
The issue with it is, since all writers will be incrementing the last
vector clock they saw, there's no way to tell when one writer made a write
significantly later in time than another, so the earlier write might
arbitrarily be picked when merging. This problem is why git-annex uses
timestamps in the first place, rather than pure vector clocks.)
Advancing forward by 1 second is somewhat arbitrary. setDead
advances a timestamp by just 1 picosecond, and the vector clock could
too. But then it would interfere with setDead, which wants to be
overrulled by any change. So it could use 2 picoseconds or something,
but that seems weird. It could just as well advance it forward by a
minute or whatever, but then it would be harder for real time to catch
up with the vector clock when forward clock slew had happened.
A complication is that many log files contain several different peices of
information, and it may be best to only use vector clocks for the same peice
of information. For example, a key's location log file contains
InfoPresent/InfoMissing for each UUID, and it only looks at the vector
clocks for the UUID that is being changed, and not other UUIDs.
Although exactly where the dividing line is can be hard to determine.
Consider metadata logs, where a field "tag" can have multiple values set
at different times. Should it advance forward past the last tag?
Probably. What about when a different field is set, should it look at
the clocks of other fields? Perhaps not, but currently it does, and
this does not seems like it will cause any problems.
Another one I'm not entirely sure about is the export log, which is
keyed by (fromuuid, touuid). So if multiple repos are exporting to the
same remote, different vector clocks can be used for that remote.
It looks like that's probably ok, because it does not try to determine
what order things occurred when there was an export conflict.
Sponsored-by: Jochen Bartl on Patreon
2021-08-03 20:45:20 +00:00
|
|
|
InfoPresent (LogInfo (encodeBS url))
|
2018-10-04 21:33:25 +00:00
|
|
|
-- If the url does not have an OtherDownloader, it must be present
|
|
|
|
-- in the web.
|
|
|
|
case snd (getDownloader url) of
|
|
|
|
OtherDownloader -> return ()
|
|
|
|
_ -> logChange key webUUID InfoPresent
|
2011-10-15 20:36:56 +00:00
|
|
|
|
2018-10-04 21:33:25 +00:00
|
|
|
setUrlMissing :: Key -> URLString -> Annex ()
|
|
|
|
setUrlMissing key url = do
|
rmurl: remove all forms of an url, no matter what the downloader is set to
* rmurl: When youtube-dl was used for an url, it no longer needs to be
prefixed with "yt:" in order to be removed.
* rmurl: If an url is both used by the web and also claimed by another
special remote, fix a bug that caused the url to to not be removed.
The youtube-dl change is a consequence of how the bug fix is implemented.
But I also think it's the right thing to do. Consider that, before,
git-annex addurl $url followed by git-annex rmurl $url would not remove the
url in the case where youtube-dl was used. That was surprising behavior.
In the unlikely case where a special remote claims an url, and it's been
added using OtherDownloader, but it was also added already as a web url,
it seems better for rmurl to remove both than to arbitrarily remove only one.
And in the case the bug report was filed for, when an url was added as a
web url, but a special remote now claims it, that should not prevent rmurl
removing the web url.
Calling setUrlMissing lets other callers of it behave differently.
Probably the calls to it in eg, Remote.External and Remote.BitTorrent are
fine, since they don't mangle the url and just remove what was provided,
and the OtherDownloader form of a bittorrent url, respectively.
I suspect unregisterurl needs to have a similar change made to rmurl, for
similar reasons.
2021-03-22 16:09:15 +00:00
|
|
|
-- Avoid making any changes if the url was not registered.
|
|
|
|
us <- getUrls key
|
|
|
|
when (url `elem` us) $ do
|
|
|
|
config <- Annex.getGitConfig
|
start implementing hidden git-annex repositories
This adds a separate journal, which does not currently get committed to
an index, but is planned to be committed to .git/annex/index-private.
Changes that are regarding a UUID that is private will get written to
this journal, and so will not be published into the git-annex branch.
All log writing should have been made to indicate the UUID it's
regarding, though I've not verified this yet.
Currently, no UUIDs are treated as private yet, a way to configure that
is needed.
The implementation is careful to not add any additional IO work when
privateUUIDsKnown is False. It will skip looking at the private journal
at all. So this should be free, or nearly so, unless the feature is
used. When it is used, all branch reads will be about twice as expensive.
It is very lucky -- or very prudent design -- that Annex.Branch.change
and maybeChange are the only ways to change a file on the branch,
and Annex.Branch.set is only internal use. That let Annex.Branch.get
always yield any private information that has been recorded, without
the risk that Annex.Branch.set might be called, with a non-private UUID,
and end up leaking the private information into the git-annex branch.
And, this relies on the way git-annex union merges the git-annex branch.
When reading a file, there can be a public and a private version, and
they are just concacenated together. That will be handled the same as if
there were two diverged git-annex branches that got union merged.
2021-04-20 18:32:41 +00:00
|
|
|
addLog (Annex.Branch.RegardingUUID []) (urlLogFile config key)
|
deal better with clock skew situations, using vector clocks
* Deal with clock skew, both forwards and backwards, when logging
information to the git-annex branch.
* GIT_ANNEX_VECTOR_CLOCK can now be set to a fixed value (eg 1)
rather than needing to be advanced each time a new change is made.
* Misuse of GIT_ANNEX_VECTOR_CLOCK will no longer confuse git-annex.
When changing a file in the git-annex branch, the vector clock to use is now
determined by first looking at the current time (or GIT_ANNEX_VECTOR_CLOCK
when set), and comparing it to the newest vector clock already in use in
that file. If a newer time stamp was already in use, advance it forward by
a second instead.
When the clock is set to a time in the past, this avoids logging with
an old timestamp, which would risk that log line later being ignored in favor
of "newer" line that is really not newer.
When a log entry has been made with a clock that was set far ahead in the
future, this avoids newer information being logged with an older timestamp
and so being ignored in favor of that future-timestamped information.
Once all clocks get fixed, this will result in the vector clocks being
incremented, until finally enough time has passed that time gets back ahead
of the vector clock value, and then it will return to usual operation.
(This latter situation is not ideal, but it seems the best that can be done.
The issue with it is, since all writers will be incrementing the last
vector clock they saw, there's no way to tell when one writer made a write
significantly later in time than another, so the earlier write might
arbitrarily be picked when merging. This problem is why git-annex uses
timestamps in the first place, rather than pure vector clocks.)
Advancing forward by 1 second is somewhat arbitrary. setDead
advances a timestamp by just 1 picosecond, and the vector clock could
too. But then it would interfere with setDead, which wants to be
overrulled by any change. So it could use 2 picoseconds or something,
but that seems weird. It could just as well advance it forward by a
minute or whatever, but then it would be harder for real time to catch
up with the vector clock when forward clock slew had happened.
A complication is that many log files contain several different peices of
information, and it may be best to only use vector clocks for the same peice
of information. For example, a key's location log file contains
InfoPresent/InfoMissing for each UUID, and it only looks at the vector
clocks for the UUID that is being changed, and not other UUIDs.
Although exactly where the dividing line is can be hard to determine.
Consider metadata logs, where a field "tag" can have multiple values set
at different times. Should it advance forward past the last tag?
Probably. What about when a different field is set, should it look at
the clocks of other fields? Perhaps not, but currently it does, and
this does not seems like it will cause any problems.
Another one I'm not entirely sure about is the export log, which is
keyed by (fromuuid, touuid). So if multiple repos are exporting to the
same remote, different vector clocks can be used for that remote.
It looks like that's probably ok, because it does not try to determine
what order things occurred when there was an export conflict.
Sponsored-by: Jochen Bartl on Patreon
2021-08-03 20:45:20 +00:00
|
|
|
InfoMissing (LogInfo (encodeBS url))
|
rmurl: remove all forms of an url, no matter what the downloader is set to
* rmurl: When youtube-dl was used for an url, it no longer needs to be
prefixed with "yt:" in order to be removed.
* rmurl: If an url is both used by the web and also claimed by another
special remote, fix a bug that caused the url to to not be removed.
The youtube-dl change is a consequence of how the bug fix is implemented.
But I also think it's the right thing to do. Consider that, before,
git-annex addurl $url followed by git-annex rmurl $url would not remove the
url in the case where youtube-dl was used. That was surprising behavior.
In the unlikely case where a special remote claims an url, and it's been
added using OtherDownloader, but it was also added already as a web url,
it seems better for rmurl to remove both than to arbitrarily remove only one.
And in the case the bug report was filed for, when an url was added as a
web url, but a special remote now claims it, that should not prevent rmurl
removing the web url.
Calling setUrlMissing lets other callers of it behave differently.
Probably the calls to it in eg, Remote.External and Remote.BitTorrent are
fine, since they don't mangle the url and just remove what was provided,
and the OtherDownloader form of a bittorrent url, respectively.
I suspect unregisterurl needs to have a similar change made to rmurl, for
similar reasons.
2021-03-22 16:09:15 +00:00
|
|
|
-- If the url was a web url and none of the remaining urls
|
|
|
|
-- for the key are web urls, the key must not be present
|
|
|
|
-- in the web.
|
|
|
|
when (isweb url && null (filter isweb $ filter (/= url) us)) $
|
2018-10-04 21:33:25 +00:00
|
|
|
logChange key webUUID InfoMissing
|
|
|
|
where
|
|
|
|
isweb u = case snd (getDownloader u) of
|
|
|
|
OtherDownloader -> False
|
|
|
|
_ -> True
|
2013-07-28 19:27:36 +00:00
|
|
|
|
2014-12-08 23:14:24 +00:00
|
|
|
setTempUrl :: Key -> URLString -> Annex ()
|
|
|
|
setTempUrl key url = Annex.changeState $ \s ->
|
|
|
|
s { Annex.tempurls = M.insert key url (Annex.tempurls s) }
|
|
|
|
|
|
|
|
removeTempUrl :: Key -> Annex ()
|
|
|
|
removeTempUrl key = Annex.changeState $ \s ->
|
|
|
|
s { Annex.tempurls = M.delete key (Annex.tempurls s) }
|
|
|
|
|
2017-11-28 21:17:40 +00:00
|
|
|
data Downloader = WebDownloader | YoutubeDownloader | QuviDownloader | OtherDownloader
|
rmurl: remove all forms of an url, no matter what the downloader is set to
* rmurl: When youtube-dl was used for an url, it no longer needs to be
prefixed with "yt:" in order to be removed.
* rmurl: If an url is both used by the web and also claimed by another
special remote, fix a bug that caused the url to to not be removed.
The youtube-dl change is a consequence of how the bug fix is implemented.
But I also think it's the right thing to do. Consider that, before,
git-annex addurl $url followed by git-annex rmurl $url would not remove the
url in the case where youtube-dl was used. That was surprising behavior.
In the unlikely case where a special remote claims an url, and it's been
added using OtherDownloader, but it was also added already as a web url,
it seems better for rmurl to remove both than to arbitrarily remove only one.
And in the case the bug report was filed for, when an url was added as a
web url, but a special remote now claims it, that should not prevent rmurl
removing the web url.
Calling setUrlMissing lets other callers of it behave differently.
Probably the calls to it in eg, Remote.External and Remote.BitTorrent are
fine, since they don't mangle the url and just remove what was provided,
and the OtherDownloader form of a bittorrent url, respectively.
I suspect unregisterurl needs to have a similar change made to rmurl, for
similar reasons.
2021-03-22 16:09:15 +00:00
|
|
|
deriving (Eq, Show, Enum, Bounded)
|
2014-12-08 23:14:24 +00:00
|
|
|
|
|
|
|
{- To keep track of how an url is downloaded, it's mangled slightly in
|
2017-11-28 21:17:40 +00:00
|
|
|
- the log, with a prefix indicating when a Downloader is used. -}
|
2014-12-08 23:14:24 +00:00
|
|
|
setDownloader :: URLString -> Downloader -> String
|
|
|
|
setDownloader u WebDownloader = u
|
|
|
|
setDownloader u QuviDownloader = "quvi:" ++ u
|
2017-11-28 21:17:40 +00:00
|
|
|
setDownloader u YoutubeDownloader = "yt:" ++ u
|
2014-12-08 23:14:24 +00:00
|
|
|
setDownloader u OtherDownloader = ":" ++ u
|
2013-08-22 22:25:21 +00:00
|
|
|
|
2016-01-19 19:55:32 +00:00
|
|
|
setDownloader' :: URLString -> Remote -> String
|
|
|
|
setDownloader' u r
|
|
|
|
| Remote.uuid r == webUUID = setDownloader u WebDownloader
|
|
|
|
| otherwise = setDownloader u OtherDownloader
|
|
|
|
|
2013-08-22 22:25:21 +00:00
|
|
|
getDownloader :: URLString -> (URLString, Downloader)
|
|
|
|
getDownloader u = case separate (== ':') u of
|
2017-11-28 21:17:40 +00:00
|
|
|
("yt", u') -> (u', YoutubeDownloader)
|
|
|
|
-- quvi is not used any longer; youtube-dl should be able to handle
|
|
|
|
-- all urls it did.
|
|
|
|
("quvi", u') -> (u', YoutubeDownloader)
|
2014-12-08 23:14:24 +00:00
|
|
|
("", u') -> (u', OtherDownloader)
|
|
|
|
_ -> (u, WebDownloader)
|
2021-04-23 15:44:10 +00:00
|
|
|
|
|
|
|
{- Parses the content of an url log file, returning the urls that are
|
|
|
|
- currently recorded. -}
|
2023-10-25 17:00:17 +00:00
|
|
|
parseUrlLog :: L.ByteString -> [URLByteString]
|
|
|
|
parseUrlLog = map fromLogInfo . getLog
|