Update path to MarkdownEditing syntax file for Requester: Show Documentation (Local)
Fixes bug in parsing --data-raw cURL arg.
Changes domain of docs site.
Improves curl_to_request; to gracefully ignore unknown CURL args.
Puts all release notes in this file.
For request parsing errors with non-empty selections, retry parsing with empty selection starting at region.begin().
This improves ergonomics, for example when navigating through requests in a requester files using fuzzy match.
Try highlighting the first 4, 5, 6, 7... characters of this request and then execute it:
get('google.com')
This works for multi-line requests as well. Try highlighting "get(" on the first line, then executing the request.
get( 'google.com' )
If response content can't be displayed because it's binary, Requester includes instructions for downloading file.
If response encoding is 'ISO-8859-1' (Latin-1), sets encoding to response.apparent_encoding. This gets encoding right in many cases where encoding is not actually Latin-1.
This works around some old, sort of wacky default behavior in Requests: https://github.com/kennethreitz/requests/issues/1604
Invokes json.dumps with ensure_ascii = False so that non-ascii chars aren't ascii-encoded in JSON responses.
This improves readability of responses like this one:
requests.get('https://api.partiyaedi.ru/api/v3/goods/actual')
Fixes strange, possibly non-deterministic context manager bug in add_path.
Adds detailed docs on how env vars are parsed, and improves error messages, most notably for when an env string can't be evaluated successfully because of a missing env var.
Check out "How Env Vars are Parsed" in the docs for more info: https://kylebebak.github.io/Requester/#how-env-vars-are-parsed
Adds Requester: Open Request History File and Requester: Move Requester File commands.
The first opens the request history file in a read-only view, and the second moves a requester file to a new path and updates request history for all requests sent from this file.
Also, request history now persists combined env string to history file, which means any request can be re-sent even if original requester file is no longer present.
Internal: further improves and simplifies code for computing env and saving to request history file.
Improved merging of env block and env file. Concatenation order of strings parsed from env block and env file determined by relative order of env block and env_file line in requester file.
Read more in Merging Vars from Env Block and Env File.
Internal: cleans up and simplifies get_env method.
Improvement to request history recency weighting.
Awesome news! I decided to vendor all of Requester's deps, which means you can import requests_oauthlib and requests_toolbelt in your env block without any mucking around with pip.
It also means that Requester does graphql query autocompletion out of the box.
Requests was also upgraded from 2.15.1 to 2.18.3, which is the last version of the package that supports Python 3.3, the version of Python run by Sublime Text.
For those interested in package development, I'm also working some tooling to make it easy to vendor dependencies for Sublime Text packages. Given the limitations of Package Control, vendoring deps is, as Will Bond says here, the "one true approach".
Also fixes a bug in request parsing that was caused by a breaking change in Sublime Text's default Python syntax.
Fixes bug due to upgrading Requests from 2.15.1 to 2.18.3. It turns out there is no way to support higher than 2.15.1 using Sublime Text =/
Fixes bug due to upgrading Requests from 2.15.1 to 2.18.3. It turns out there is no way to support higher than 2.15.1 using Sublime Text =/
Awesome news! I decided to vendor all of Requester's deps, which means you can import requests_oauthlib and requests_toolbelt in your env block without any mucking around with pip.
It also means that Requester does graphql query autocompletion out of the box.
Requests was also upgraded from 2.15.1 to 2.18.3, which is the last version of the package that supports Python 3.3, the version of Python run by Sublime Text.
For those interested in package development, I'm also working some tooling to make it easy to vendor dependencies for Sublime Text packages. Given the limitations of Package Control, vendoring deps is, as Will Bond says here, the "one true approach".
Also fixes a bug in request parsing that was caused by a breaking change in Sublime Text's default Python syntax.
When deleting a request from Requester history, requester pages to the previous request, which gives user immediate visual feedback that request has been deleted.
Also includes [ctrl+alt+delete] binding in history tab, so that users are aware this function exists.
Closes #19.
Requester history search has been updated to weight recent request history matches more heavily. This makes history search more predictable and more intuitive.
Closes #18.
Requester's parser has been updated to parse prepared requests. Sending prepared requests can be useful for, e.g., sending requests with a self-signed certificate. Read more in the Sessions section of the docs, or here:
http://docs.python-requests.org/en/master/user/advanced/#prepared-requests
Single requests that don't start at the first character of a line can be parsed correctly. This allows you, for example, to execute requests that have been commented out, or requests that are in doc strings, or requests in a markdown list.
Place your cursor on the first line of either of these requests and attempt to execute them.
- get('google.com')
- get('httpbin.org/get', params={'a': 'b'} )
Explore URL in Response Tab command works for localhost URLs.
Fixes a bug in Import from cURL; parser can now handle "#" characters.
The tabname argument is now included in the fuzzy searchable request history, which makes is considerably more useful.
Check out the Custom Tab Name section of the docs for more info.
Requester requests support a new keyword argument, tabname, that allows you to specify a custom response tab name for a request. This response tab can only be overwritten by requests with the same tabname.
This is useful for differentiating requests whose method and URL are the same but whose meaning is different. GraphQL requests, whose meaning is encoded in the query string, are one example.
Requester's Test Runner now supports custom validation functions, which allow you to make assertions about any aspect of response object using anything Python has to offer.
This is a feature I've wanted to add for a long time. I hope you all like the implementation. Check out the docs to see how it works.
Also, Export to cURL now supports basic authentication.
File downloads don't always display '?' for content length. First, content-length header is inspected, and if it's empty, length of response is taken.
Multiline requests can now be executed by placing cursor anywhere inside the request, not just on the first line, thanks again to @jcberquist =)
This works as long as the view's syntax recognizes Python function calls.
Requester doesn't save original_request to history file if it's the same as request. This decreases size of Requester file.
Requests are saved to history with original value of request sent from requester file.
This means saving requests back to requester file works for requests loaded from history, even if these requests were edited in the response tab.
Requests in history are now keyed for uniqueness on (request_string, requester_file), instead of just request_string, if the requester file exists.
Also, a request can be deleted from request history by pressing ctrl+alt+backspace in the request's history tab.
Finally, there are several improvements to file downloads:
- downloads don't require a '/' suffix for Requester to automatically assign the file a name
- downloads correctly resolve the '~' char to the user's home directory
- downloads from URLs with a trailing query string get the correct file extension
Much better syntax highlighting for requester files, thanks to @jcberquist!
Also, fixed bug in explore request command, which choked on requests with a dangling, trailing comma.
Response tab name: if parsed URL has no path, display netloc (base path), instead of displaying nothing.
Also, the default length of response tab names is limit to 32 characters. You can change this limit in settings.
Fixed bug in saving requests back to requester file when paging through request history.
It's taken me a while to get "save request back to requester file" exactly right, but it's finally good to go.
This fixes a bug that would replace either too many or too few characters in the original response view.
Not exactly new features or a bug fixes, but saving requests back to the requester file was significantly improved and simplified.
- You can save back requests even if your requester file has been modified in the meantime. Save will only fail if the original request itself has been modified.
- You can now save back requests that you load from your history file.
- Scrolling to the original request now works reliably even if the requester file isn't currently open when you save back your request.
You can save a modified request in a response tab back to the Requester file from which you sent it!
- Can be used repeatedly in response view.
- Guaranteed to old overwrite old request and nothing else.
Also, the default length of response tab names is limit to 32 characters. You can change this limit in settings.
Requester doesn't print error messages to status bar if requester file contains non-ascii chars.
Serious GraphQL support! Requester can autocomplete any GraphQL query from a response tab.
- Autocomplete provides field type, name, and arguments, similar to GraphiQL.
- Enable autocomplete by installing
graphql-pyusingpip, see documentation for extending Requester.
As far as I know, Requester make Sublime Text the first text editor with reliable GraphQL support!
-
Added basic GraphQL support
gql,gqlv, andgqlokwargs to automatically add GraphQL queries to querystring for GET requests, and to JSON encoded request body for POST requests.- GraphQL autosuggest coming soon!
-
Added
browsercookieauth option for grabbing Chrome/Firefox cookies and adding them to requests, similar to Postman's Interceptor.- Added information on importing Python libraries that depend on shared objects, like
pycyrpto.
- Added information on importing Python libraries that depend on shared objects, like
Extending Requester: no longer writes writes same packages_path to sys.path multiple times.
Test coverage was improved significantly. Also, Requester now has a docs site! Check it out at https://kylebebak.github.io/Requester/, or by running Requester: Show Documentation (Online).
Requester's on fire! Request history has been totally revamped for this version, combining fuzzy search with paging for request history nirvana.
If you run Requester: Request History from the command palette (add this to your key bindings, the command is "requester_history"), you get the usual fuzzy searchable list of requests. But now, if you select one of these requests and hit Enter, Requester doesn't execute the request.
Instead, it "stages" the request into a "request history" tab, from which you can look at the request and edit it before you execute it. And if you press ctrl+alt+ ←/→, you can page between requests that were executed before and after this historical request.
You can actually start paging through past requests from any open response tab.
Being able to stage requests before executing them is a big help for requests that aren't idempotent, like POST requests... Before you had to execute them to even see them.
And paging combined with search makes it trivial to find past requests. If it's a recent request, just start paging backwards. If it was a week ago, and you know it was one of many you executed against the Twitter API, fuzzy search for "twitter" to land in the general neighborhood, then page around to find the exact one you're looking for.
- Ergonomics for basic GET requests: parser handles URLs with or without wrapping quotes.
More good news! Like 2.13.0, this release introduces a number of improvements to Requester.
-
File downloads whose path has no
basenameget an automatically generated filename. If the filename is already take, a new unique filename is generated.- Also removed a subtle bug in downloads.
-
Requests defined over multiple lines are much easier to execute. You can place cursor anywhere on the first line of a multi-line request and run it without highlighting anything.
-
Requester has a new option to "explore" URLs returned in the response body of a request. Highlight a URL in a response tab and press
ctrl+e(cmd+eon macOS) to send a GET request to the highlighted URL.- If the URL has the same domain as the URL for the original request, the rest of the request args will are left unchanged, which means that the new request will be trivially authenticated.
- If the URL has a different domain, the remaining args are removed from the new request.
- Requests executed from a requester file pull in the current values for all env vars in this file, even if the file has not yet been saved.
Great news! This release introduces some cool improvements to Requester.
-
Requester can now source any Python 3 package in its env! All you have to do is set the
packages_pathsetting to a directory with Python 3 packages you'd like to include with Requester, e.g. to a virtualenv directory:~/.virtualenvs/general/lib/python3.5/site-packages.- Use any plugin in the Python ecosystem, including any auth plugin
requests-oauthlib, "https://github.com/requests/requests-oauthlib"
- Use
requests-toolbelt, "https://github.com/requests/toolbelt" - the
requestsandjsonschemapackages are obviously still bundled with Requester
- Use any plugin in the Python ecosystem, including any auth plugin
-
Sending basic GET requests just got more convenient. Selections spanning only one line, of the form
http://google.comorgoogle.comare now transformed toget('http://google.com')before being sent by Requester. -
Added request navigation, request groups, and improved syntax highlighting to requester files, as long as they are saved with the
.pyrextension- You can jump between requests using fuzzy search using the Goto Symbol command
- Running Requester: New Requester File from the command palette creates a file with this extension, and sets the Requester syntax on the view
- Read more in the Request Navigation section of the README
Multiple file uploads and downloads can be executed in parallel.
Access to persist_requests function is controlled by a lock, to make sure threads concurrently executing this function don't corrupt request history. Also, fixed small bug in importing requests from cURL.
Any method can be used to start a file download, not just GET.
File uploads and downloads are persisted to request history, and downloads get their own response tabs.
Fixed race condition bug in uploads and downloads, where env was being referenced before it had been evaluated.
Fixed a bug for uploads: Requester now displays an error message if the file being uploaded doesn't exist.
Support for streamed and chunked file uploads enabled via simple streamed and chunked special kwargs.
Improvement to request chaining: users can "name" response to requests and reference those responses by name in subsequently executed requests.
Added fmt argument to requests, which controls JSON response formatting. It can be one of one of ('raw', 'indent', 'indent_sort').
Improved syntax highlighting for response views with no valid content type in response headers.
Reorganized project so that command files, and syntax files, each have their own directory. All command files are loaded by top-level requester.py module.
Users can specify in settings whether or not to allow_redirects by default.
Fixed bug in get_exports helper.
Added Requester: Authentication Options command, which contains a reference for common auth schemes to use in requests. Added some support for more cURL options when importing requests from cURL. Added Requester: Settings command to command palette.
Bug fixes for importing requests from cURL, including a bug that caused Sublime Text to hang indefinitely.
Added function to export selected request/assertion pairs to a runnable test script that can be integrated into a test routine. Also, exporting to HTTPie now supports sessions.
Simplified code in RequestCommandMixin.get_env that handles building environment dictionary.
Fixed regression that broke chaining requests by reference.
Added export to HTTPie command.
Fixed subtle bugs in arg parsing for export commands. Also made error messages more explicit for certain syntax and parsing errors.
Added export to cURL and import from cURL commands.
Fixed bug with requests args that had a trailing comma, re-executing these requests from response tab used to raise exception.
Requester supplies default http scheme if no scheme is specified in URL.
Improved syntax highlighting for test run views.
Fixed performance issues with benchmark runs, and fixed potential race conditions with pending requests.
Added request benchmarking tool, similar to ab and siege. Check the README for details.
New major version. Rewrote core API and parser so that eval is only called once per request, to extract args and kwargs from request string. Among other things, this means that method and URL, along with args and kwargs passed to requests, are known before response returns.
Building new clients of the core API, such as the new benchmarking tool, is much easier.
Also, various small bug fixes.
Display request body in response tab, if body was present.
Fixed comment/backslash bug in parser.
Improved persist_requests. Requester creates a backup of request history, and will not lose requests if an exception is raised or there is a crash while file is being written to.
There is now no possibility of request history being lost, minus the user deleting his request history file.
Improvements to request parser: parser doesn't choke on escaped single quotes or escaped double quotes.
Added pin_tabs_by_default setting, which allows users that don't want any reponses to be overwritten to pin newly opened response tabs by default.
Improved behavior of RequesterCancelRequestsCommand.
Added command to pin/unpin response tabs.
Unpinned response tabs are overwritten by requests with the same method and URL. Pinned response tabs are not, which makes it possible to have multiple views open for the same request.
To pin/unpin a response tab, hit cmd+t on OSX, or ctrl+t on Linux/Windows.