* [sbt] set name for assembly jar
* [travis] run assembly prior to release
* [sbt] disable publishing non-cli or thorp-lib modules
* [travis] alternate format for multiple commands in script
* [sbt,travis] Deploy releases and snapshots to sonatype
* [travis] remove oracle JDK builds
* [travis] remove 'compile' job
* [travis] Try adding scala 2.13
* [travis] test using openjdk8 rather than 10
* [travis] force use of scala version, overriding sbt version
* Revert "[travis] Try adding scala 2.13"
This reverts commit 6725199e009ae1f809247bbc0851b62d63121993.
* Revert "[travis] force use of scala version, overriding sbt version"
This reverts commit 21c94992f40cf2ef39c5126e67b66feaabe1b2c0.
* [sbt] Group modules under n.k.thorp organisation
* [storage-aws] ETagGenerator add stub
* [core] MD5HashGenerator add hex and digest helpers
* [domain] MD5Hash can always provide base64 and also digest
Rather that store the base 64 digest some of the time, simply decode
it from the hex hash. The same for the binary digest.
MD5Hash is now cleaner now that it no longer has Option parameters.
* [core] MD5HashGenerator add stubs to allow reading file chunks
* [domain] MD5HashData add sub-objects
* [domain] MD5HashData move back into test where it belongs
* [sbt] add sbt-bloop plugin
* [domain] MD5HashData Add hash of big-file
* [domain] MD5HashData Add hash of big-file
* [core] MD5HashGenerator find end of chunk correctly
* [core] MD5HashGenerator offset is a Long
* [core] MD5HashGenerator don't read past the end of the file
* [storage-aws] ETagGenerator can reproduce ETags
* [storage-aws] ETagGeneratorTest added
* [storate-aws] ETagGenerator refactoring
* [storage-aws] ETageGenerator refactoring
* [core] SyncSuite remove redundant braces
* [storage-api] HashService added
* [storage-aws] S3HashService added
* [core] LocalFileStream refactoring
* [core] integrate HashService and ETagGenerator
* Optimise imports
* [domain] HexEncoder added to replace java 8 only DataTypeConverter
* [core] MD5HashGenerator refactoring
* [core] S3MetaDataEnricher refactoring
* [core] S3MetaDataEnricherSuite refactoring
* [storage-aws] ETagGeneratorTest refactoring
* [storage-aws] StorageServiceSuite refactoring
* [core] S3MetaDataEnricher refactoring
* [core] refactoring
* [storage-aws] refactoring
* [core,storage-{api,aws}] Allow Lister errors to be reported to user
* [cli,core,storage-*] Display simple error message when bucket is invalid
* [core] optimise imports
* [storate-aws] optimise imports
* [core] SyncSuite don't use get on an optional type
* [aws-lib] Uploader provide request with the already calculated md5 hash
* [aws-lib] remove unused accepts method
* [aws-lib] Uploader refactoring
* [domain] Config remove unused threshold and max retries items
* [core] Show upload errors in summary
* [domain] LocalFile add helper to explicitly compare by hash value
Looking to add an optional field to MD5Hash but we want to do our
checks here only on the hash value, not whether a digest is available
or not.
* [core] Sync refactoring
* [core] SyncSuite invoke subject inside it method and after declaring expectations
* [core] SyncSuite use the localfile hash rather than something arbitrary
* [cli] Add `--no-global` and `--no-user` options
* [core] LocalFileStream refactoring
* [core] SyncSuite: ignore user and global configuration files
* [domain] MD5Hash now can optionally store the base64 encoded hash
* [core] MD5HashGenerator pass the digest to MD5Hash
* [aws-lib] Uploader use the base64 encoded hash
* [changelog] updated
* [domain] rewrite and expand ansi codes available
* [aws-api] refactor progress-bar
* [domain] move progress-bar into Terminal
* [aws-api] remove unused value
* [domain] use unicode characters to get a smooth progress bar
* [core] rename the config supplied from CLI as such
This distinguishes it as config supplied from the command line.
* [core] add ConfigOption
* [core] ConfigOption can update a Config
* [core] new validator for config
* [domain] Config doesn't validate source any more
* [cli] PrintLogger default to not print debug messages
* [cli] Use ConfigOptions and new ConfigValidator
* [sbt] Use common settings for project root
* [domain] RemoteKey can handle when prefix is empty
* [cli] remove banner
* [domain] Logger can create version of itself with debug flipped
* [core] Build and validate Config within core module
This means that the `thorp-lib` module can validate its input from a
list of `ConfigOption`s.
* [core] refactor ConfigurationBuilder
* [core] refactor ConfigurationBuilder
* [sbt] starting back from tagless-final by using IO where needed
* [core] Add ParseConfigFile
* [sbt] Make cats-effect available from the domain
* Roll back from tagless-final to just use cat-effect's IO
* [core] extract ParseConfigLines
* [core] ConfigurationBuilder rename apply as buildConfig
* [core] ParseConfig[Files,Lines] rename apply methods
* [core] refactor ParseConfigFile and add tests
* [core] Sync fix call to run
* [core] SyncSuite update tests to use ConfigOptions
* [aws-api] TTFI UploadProgressLogging#logTransfer
* [aws-api] TTFI UploadProgressLogging#logRequestCycle
* [aws-api] TTFI UploadProgressLogging#logByteTransfer
* [aws-api] TTFI UploadProgressListener
* [aws-lib] TTFI UploaderLogging
* [aws-api] UploadProgressListener refactoring
* [aws-api] UploadProgressListener remove IO/Monad
This class is a wrapper for a callback method used by the AWS SDK.
Unfortunately you can't get the IO() created when that listener is
called by the SDK, so unless we manually run unsafeRunSync, as we have
done previously, it would never be executed.
This removes the IO monad entirely and simply runs the code when the
callback is triggered.
* [aws-lib] S3ClientLogging remove unused method
* [aws-lib] TTFI S3ClientLogging
* [aws-lib] TTFI S3ClientCopier
* [aws-lib] TTFI S3ClientObjectLister
* [aws-lib] TTFI Uploader
* [aws-lib] TTFI S3ClientDeleter
* [aws-api] TTFI S3Client
* [aws-lib] TTFI S3ClientBuilder and ThorpS3Client
* [core] TTFI ActionSubmitter
* [cli] TTFI Logger
* [core] TTFI MD5HashGenerator
* [core] TTFI LocalFileStream
* [core] Sync refactoring
* [core] TTFI Sync
* [aws-lib] S3ObjectsByHashSuite truncate lastmodified to match Date
* [aws-lib] ThorpS3ClientSuite truncate lastmodified to match Date
* [core] MD5HashGeneratorTest switch to Id from IO
* [sbt] restrict cats-effect to cli module, cats-core elsewhere
* [core] MD5HashGenerator collapse lines
* [domain] SizeTranslation includes decimals for larger sizes
* [core] MD5HashGenerator rewrite for memory efficiency
No longer attempt to create an Array the size of the file to be
parsed.
Now it creates a single small buffer and reads 8kb chunks in at a
time. Only creating an additional smaller buffer to read the tail of
the file.
Remove methods to parsing only part of a file are they were no longer
used, and remove the relevant tests.
* [aws-lib] Uploader refactoring
* [aws-lib] Uploader remove redundant braces
* [aws-lib] Uploader start upload inside IO
The upload method starts the upload to S3, so should be within the IO.
* [core] log count of errors
* [aws-lib] Uploader handle errors and count them
* [aws-lib] fold S3ClientUploader trait into it's only implementation
This trait was only implemented by S3ClientTransferManager.
* [core] SyncLogging: more robust matching
No longer cares about parameters to case classes, just their types.
* [cli] Logger uses IO for log methods
* [aws-lib] remove 'transfer-manager'prefix and only show tryCount > 1
* [sbt,cli] remove log4j and scala-logging dependencies
* [domain] move QuoteStripper to Domain
Use it directly in MD5Hash to strip quotes from any input.
* [core] SyncLogging call info in proper context
If the IO.unit returned by the info calls isn't part of the chain that
is returned from the function, then the delayed IO action is never
called.
* [aws-lib] Display size in bytes of file being uploaded
* [core] call info in correct context
* [cli] call info in correct context
* [aws-lib] raise summary fetch message to info 1
* [cli] include correct level in info messages
* [aws-lib] S3ClientLogging adjust logging levels
* [aws-lib] display file sizes in english
* [aws-lib] ObjectLister use IO.bracket properly
* [aws-lib] Copier use IO.bracket properly
* [aws-lib] Deleter refactor
* [aws-lib] TransferManagerLogging remove unused methods
* [aws-lib] TransferManager refactor
* [aws-lib] TransferManager refactor
* [aws-lib] TransferManager displays log messages
Use the UploadProgressListener that was being ignored, and use
unsafeRunSync to execute the suspended effect within the IO[Unit].
Using unsafeRunSync is required to render the effects as the listener
returns Unit, meaning the suspended effects would be discarded.
* [domain] Extract SizeTranslation into module
* [aws-api] report bytes transferred in progress
* [core] fix calls to info
info now returns an IO already, so don't need to wrap it in one.
* [aws-lib] remove unused class
* [aws-lib] UploadProgress displays progress bar while uploading
* [aws-api] UploadProgressLogging optimise imports
* [aws-api] UploadProgressLogging rename variables
* [domain] add Terminal object
* [aws-api] UploadProgressLogging use console width and two lines
- Improved clearing of lines after progress bar
- Use console width for progress bar size
* [aws-lib] S3ClientLogging optimise imports
* [aws-lib] TransferManager clear line before logging
* [aws-lib] rename class as TransferManager
* [aws-lib] rename TransferManger as Uploader to not clash
We are using an AWS SDK class with the same name.
* [domain] rename Filter as Include
* [cli]ParseArgs allow exclude and include parameters to be repeated
* [core] don't include include/exclude details in logging
* [domain] combine Include and Exclude into Filter
Config now collect includes and Excludes into a single list and passed
each file to the Filter.isIncluded method, with the list of Filters,
to determine if a file should be included.
* [core] MD5HashGenerator uses IO to return where there is file IO
This required that LocalFile in the domain module no longer be
supplied with a function to convert a File into an MD5Hash. Because
such a function requires reading the file it now must use IO, which we
don't allow in the domain module.
Unfortunate ripple effects out to users of MD5HashGenerator and
LocalFile.
* [aws-lib] Add own copy of test class MD5HashData
* [core] close files after calculating their MD5 hash
FileInputStream was never closed, so eventually ran into
ToManyFilesOpen.
Will come back to look at this again with IO.bracket for better
guarantee that FIS is closed.
Signed-off-by: Paul Campbell <pcampbell@kemitix.net>
* [aws-lib] Fetch all MD5 hashes under prefix
Initial request only returns the first 1000.
