- Issue #377. Sends Content-Type header for all request types to address compatibility issue with Microsoft Edge.
- Issue #375. Encodes asterisk (*) in file names to fix signature mismatches.
- Issue #355. Addresses memory retention issue that arose in a recent Chrome update. Chrome was keeping a reference to the Xhr object in a system array because EvaporateJS was passing an XHR reference as a parameter to Promise.resolve().
- PR #357. The progress callback response now includes more stats
- Issue #356. Filenames containing "!" are now correctly encoded for S3 so that signatures match
- the example html now allows the aws_key to be changed
- Improved examples for node.js
- Issues #340. Resolved memory retention issue when using V2 signatures without
computeContentMd5.
- Issue #333. The sample page
example/evaporate_example.htmlnow allows AWS conifguration options to be temporarily stored on the page and provides acustomAuthMethodbacked by JavaScript. This should make it easier to test EvaporateJS with specific AWS settings. - Issue #309, #312, #314. Allows AWS_URL and AWS_KEY to be overridden when a file is added for upload.
- Issue #320. Adds an optional localCache polyfill if localStorage is not
available. Must be explicitly enabled with
mockLocalStorage. - Issue #329. A contributor provided a signing example for PHP.Laraval.
- Issues #316, #325. Restores Microsoft Windows compatibilty with Edge and Google
Chrome that broken when Evaporate started to parse URLs using the URL
object and incorrectly use the unsupported
Object.assignmethod.
- Issue #300. Restores compatibility with Windows 11 by removing refernces to Object.assign
- Issue #300. Signer methods and urls paramaters now enable the AWS signing version 4 canonical request to be passed.
- Issue #305. Adds new callback
uploadInitiatedthat returns the S3 upload ID.
- Issue #290. Improved python example for V4 signatures.
- Issue #292. Corrects an issue where the override options on Evaporate#add were not properly applied.
- Issue #199. Adds support for multipart uploads using Node FileSytem (fs) ReadableStreams. This enables intregration with Electron to upload files outside of a browser framework.
- Issue #284. Memory was being retained during an upload but released on completion. Memory is properly managed now.
Note: tagged branch 1.6.4 is available with this fix for 1.x users. To install,
npm install git://github.com/TTLabs/EvaporateJS.git#1.6.4- Correctly encodes all single quotes in an S3 object name. Issue #264 (revisited)
- Simplifies internal 'casting' of ArrayBuffer to Uint8Array (ArrayBufferView)
- Optimizes size of last part, making sure it's not reporting longer than it actually is
- Improves memory reuse of large parts
- Lazily calls getPayload for V4 signatures, not at instantiation
- Adds a sample Python signing routine for V4 signatures
- Adds recipes for using node.js
cryptowith Evaporate
- Issue #277. It was possible for a queued file to not start if the previous file was canceled before it had started. Also addressed edge cases and race conditions when canceling a list of files that have been submitted. Thanks to @cvn for the test cases.
- Adds upload stats to .progress and .complete callbacks. Stats include transfer rate in bytes/second, friendly formatted rate (Kbs, Mbs, etc.) and expected seconds to finish.
- Adds new callback "nameChanged" called when the requested S3 object key was not used because an interrupted upload was resumed or a previously uploaded object was used instead.
- Adds signing example for Go.
- File processing is now distributed. Previously, Evaporate would upload
one file at a time. If you uploaded 5 files, each with a total size
less than the part size, that unused "slots" would go unused. This
version will distribute unused slots to the next file to upload, meaning
that Evaporate can upload several files simultaneously, up to the value
of
maxConcurrentParts. In other words, ifmaxConcurrentPartsis 6, and you want to upload many files whose size is less then the part size, then evaporate will upload 6 files conconcurrently with each file using one upload "slot". Evaporate will ensure that no more thanmaxConcurrentPartsare in play at any one time. - Enhances Evaporate#cancel to cancel all uploads
- Documents the
beforeSigneroption for a file upload.
- Evaporate now requires support for ES6 Promises. Use a polyfill for browsers that that don't support them.
- The default Signature style is now V4 (V2 signature users must set
awsSignatureVersionexplicitly) - Instantiate Evaporate using its create class method.
- Evaporate#add no longer returns a numeric id referencing the file upload. It now returns a Promise when adding a file to upload. To act on the file upload, for example, to Pause, Resume or Cancel, use a composed key consisting of the bucket and object name. Refer to the README for for more information.
- Evaporate#cancel, Evaporate#pause and #Evaporate#resume now return promises that resolve when the action is complete, or reject with a reason.
- Makes authorization (signing) a pluggable feature and as a result, removes
built-in support for AWS Lambda. Consequently, async methods can be used
to calculate a signature. The custom authorization method must return
a Promise and is specified through the
customAuthMethodoption. The README includes examples of how to define authorization through AWSLambda, as do the examples. - Allows the upload to be aborted if the signature url responds with 401 or 403. Additionally, if the customAuthMethod promise rejects, then the upload is aborted.
signReponseHandlernow must return a Promise. It is now only used to post-process a response fromsignUrl.
- Evaporate was not fetching all uploaded parts from S3 if the file had more then 1,000 parts.
- Corrects license name for compatibility with webjars.org.
- Addresses file.lastModifiedDate deprecation warning in FireFox
This is a release with only changes to the readme.
- Correctly calculates the local time offset
- Issue #264, properly encodes S3 object names containing single quotes (')
- Adds more test coverage.
- Corrects an error message that was using a 0-based index rather than 1
- Correctly aborts an upload if start() cancels the upload.
- Fixes a bug where the reported progress doubled after resuming an upload
- Fixes the ability to fetch the uploaded part count for files with more than 1,000 uploaded parts.
- Adds safety checks for parsing the error XML from AWS responses
- Issue #231. Enahances the custom
signResponseHandlermethod signature to include the stringToSign and dateString that are passed to the remotesignerUrl. - Issue #232. Adds new option
xAmzHeadersCommonfor specifying xAms headers for all AWS S3 requests. - Improves test coverage further.
- Issue #232, applies
xAmzHeadersCommonto all AWS S3 requests that previously were lacking them.
- Issue #242, canceling an upload did not probably clear internal state, causing other uploads to fail silently.
- Issue #241, after being offline for a period of time, and returing back online, evaporate did not correctly keep track of internal state, causing unnecessary traffic and triggering unpredictable XHR responses.
- Issues #227, 223. Addresses all known concurrency issues
- Issue #214. Uploads on IE were failing because IE does not add a leading slash to a file path. This has been fixed.
- Issue #222. Object key names with parentheses were not being correctly encoded, resulting in signature mismatches.
- Issue #197. Allows signerUrl to be optional so that developers can leverage AWSLambda or signResponseHandler functions to sign requests.
- Issue #202. Defaults AWS Region to us-east-1 and correctly sets the default AWS URL for non us-east-1 users.
- Documents that the signParams and signHeaders object can use functions.
- Issue #205. SignatureDoesNotMatch for some parts, when
maxConcurrentParts > 1. - Issue #208. Addressed Internet Explorer compatibility issue with parsing XML error responses.
- Issue #209. Addresses when the part uploaded byte count is reset when a part fails to upload and will be retried.
- Issue #191: Adds initital test framework.
- Improves the readme in response to user feedback.
- Issue #203. Configuration options signParams and signHeaders were not properly propagated the as a result, were ignored.
- Issue #196. S3 object names containing a slash '/' were being encoded, causing the Signature V4 calculation to be wrong.
- Issue #17. Allows file uploads to be paused and resumed. Pausing can be forced. In such cases, the in-progress parts are immediately aborted and the file upload paused. Otherwise, the file is paused after all in-progress parts have completed.
- Adds three new file callbacks:
paused,pausingandresumed. - The file
startedcallback is now passed the upload ID of the file being started. - Some internal variables renamed for clarity
- Improves evaporate_example.html to exercise more features, including MD5 checksums, pause/resume and integration with ProgressBar.
Republished 1.4.6 as NPM had a draft version by mistake
- Issue #183. Fixes issue where V4 signature was incorrectly calculated for URLs without a search part.
- Issue #179. Allows XMLHttpRequest
xhrWithCreditialsto be set.
- Issue #180. Adds 'bucket' to the list of configuration options that can
be overridden in the
.add()method. - Issue #183. Revtored maxFileSize validation.
- Adds a python example.
- Issue #174. Reverts changes to the documentation around the .add() method and fixes bug related to how the file is passed to EvaporateJS.
- Issue #168. Resolves issue where the localStorage cache was not being cleaned up, resulting in a potential browser error when limits were exceeded.
- Issue #166. Resolves incorrect signuature for V2 signing on getListParts. This bug was introduced in v1.4.0.
- Issue #163. Adds a File callback used to indicate when Evaporate starts processing a file for upload.
- Adds a Golang example file
- Various documentation corrections
- Issue #61. Adds ability to change some config options when adding a file to EvaporateJS.
- Issue #71. Adds support for AWS Signature Version 4.
- Issue #144. Adds support for enabling AWS S3 Transfer Acceleration using the new options
s3Acceleration. - Issue #148. Adds support for signature processing with a custom method using
signResponseHandler.
- Issue #135: Does not throw an error on uninitiated uploads if user chooses to Cancel
- Issue #139. If you wish Evaporate to only upload to the AWS key specified, and disregard any previous attempts to
upload the same file to a different key, you can now specify
onlyRetryForSameFileName. - Issues #41, #143. Adds a sample app.yml file for Google apps.
- Issue #136. Adds a configuration option to validate maximum uploaded file size. You can now specify
maxFileSize. - Issues #74, #133. Removes synchronous logic to fetch server time to resolve AWS
RequestTimeTooSkewedresponse.
- Corrects XML parsing method correctly use
textContentrather thannodeValue. - Corrects JavaScript style issues
- Adds support for signing reqeusts using AWS Lambdas.
- Checks if window is defined before trying to access the window object in
supported. EvaporateJS can now sucesfully be imported into a node.js environment. - Adds an example using AWS SDK for Javascript for the
cryptoMd5Methodconfiguration option.
- Issue #124. Only enables the history cache if MD5 checksums are also enabled.
- Issue #117. Adds the signParams configuration to the localStorage cache for reuse when resuming uploads
- Issue #116. Fixes issue where the uploader would fail because localStorage is not available in incognito mode on some browsers
- Issue #115. Fixes an issue where the uploader assumed that a Blob object had a last modified date
##Features##
- Adds MD5 checksum/digest validation, Issue #96.
- Adds ability to Abort a multipart upload (releasing AWS S3 resources), Issue #98. This change requires S3 CORS Settings to allow the DELETE method.
- Adds retry logic to the Initiate and Complete methods, Issue #107
- Adds ability to resume uploads, even after client or user error, Issue #104. Using this feature requires S3 CORS Settings to allow the HEAD method. Refer to the updated README.md for more information on the necessary bucket policy.
- The
completecallback method signature has changes: a new parameter has been addedawsObjectKeywhich will contain the S3 object key found when attempting to resume uploads after client failure. This is required because it is generally required to create a randomingly unique S3 object key to prevent namespace clashes in multi-user environments. If an object can be resued, the proposed AWS S3 object key would not be used; instead, the object key of existing object would be required.
##Fixes##
- Addresses memory leak when retrying by releasing finished, failed or aborted XMLHttpRequest objects, Issue #100.