Refactor analytics to not use api.gradio.app #8180

freddyaboulton · 2024-04-30T16:32:03Z

Description

🎯 PRs Should Target Issues

Before your create a PR, please check to see if there is an existing issue for this change. If not, please create an issue before you create this PR, unless the fix is very small.

Not adhering to this guideline will result in the PR being closed.

Tests

PRs will only be merged if tests pass on CI. To run the tests locally, please set up your Gradio environment locally and run the tests: bash scripts/run_all_tests.sh
You may need to run the linters: bash scripts/format_backend.sh and bash scripts/format_frontend.sh

gradio-pr-bot · 2024-04-30T16:32:38Z

🪼 branch checks and previews

•	Name	Status	URL
	Spaces	ready!	Spaces preview
	Website	ready!	Website preview
🦄	Changes	detected!	Details

Install Gradio from this PR

pip install https://gradio-builds.s3.amazonaws.com/522fa68788345948972f96128f3cc6f02a07064a/gradio-4.28.3-py3-none-any.whl

Install Gradio Python Client from this PR

pip install "gradio-client @ git+https://github.com/gradio-app/gradio@522fa68788345948972f96128f3cc6f02a07064a#subdirectory=client/python"

gradio-pr-bot · 2024-04-30T16:32:48Z

🦄 change detected

This Pull Request includes changes to the following packages.

Package	Version
`gradio`	`patch`

Maintainers can select this checkbox to manually select packages to update.

With the following changelog entry.

Refactor analytics to not use api.gradio.app

Maintainers or the PR author can modify the PR title to modify this entry.

Something isn't right?

Maintainers can change the version label to modify the version bump.
If the bot has failed to detect any changes, or if this pull request needs to update multiple packages to different versions or requires a more comprehensive changelog entry, maintainers can update the changelog file directly.

abidlabs · 2024-04-30T19:53:06Z

Instead of using (internal) _send_telemetry_in_thread, why not use send_telemetry? That way we can avoid spinning up our own thread for each analytics request and instead use the single-thread approach that is baked into huggingface_hub (since Lite doesn't support threads, I guess we'd still need to use _send_telemetry_in_thread in that case)

cc @Wauplin

freddyaboulton · 2024-04-30T20:16:32Z

@abidlabs _send_telemetry_in_thread does not actually launch a new thread as far as I can tell, it just sends the request to HF hub. In contrast, send_telemetry will place data in a queue and then start a new thread to pull data off that queue and send to the hub. I think we still need to run the analytics in a separate thread so that the "get ip address" operation does not block, so I think _send_telemetry_in_thread is better to not have an additional background thread running.

abidlabs · 2024-04-30T20:20:20Z

@freddyaboulton yes you're right, the _send_telemetry_in_thread , but create a thread ourselves, e.g. when we do:

threading.Thread(
            target=_do_normal_analytics_request,

it seems like it'd be better to use the documented public method if possible?

freddyaboulton · 2024-04-30T20:20:23Z

gradio/analytics.py

        await asyncio.wait_for(
-            pyodide_pyfetch(url, method="POST", headers=headers, body=body),
+            pyodide_pyfetch(url, method="HEAD", headers=headers),


I originally wrote this to use _send_telemetry_in_thread but the analytics requests were not showing up in the webworker network tab.

If I use pyodide_pyfetch, they show up in the network tab. Still not sure if this works 100% though.

So I think we need this lower-level approach for wasm and can't rely on higher level functions in hub python library.

If _send_telemetry_in_thread uses threading as its name implies, it doesn't work on Lite/Pyodide.

I think using pyodide_pyfetch is a right approach on Lite and it should work.

I reverted the changes for wasm since we can't set the user-agent on chrome!

Ref: https://stackoverflow.com/a/42815264/13103190

abidlabs · 2024-04-30T20:34:10Z

Nvm @freddyaboulton you are right we will need to spin our own thread to get the IP so let's stick with _send_telemetry_in_thread. @Wauplin is that a safe choice or is there a better option you'd suggest?

Wauplin · 2024-05-02T10:55:46Z

we will need to spin our own thread to get the IP so let's stick with _send_telemetry_in_thread. @Wauplin is that a safe choice or is there a better option you'd suggest?

It's a reasonably safe choice yes. Don't see a better option that does not involve re-implementing the request.
Could you add a comment in huggingface_hub's _send_telemetry_in_thread to reference to gradio-lite so that we don't forget? Little chance we change this internal method anytime soon but better safe that sorry :)

freddyaboulton · 2024-05-02T17:36:33Z

Yes will do @Wauplin thanks!

freddyaboulton · 2024-05-02T17:37:02Z

Ok this is ready for review now

abidlabs

Code changes LGTM @freddyaboulton! Jw if you checked to confirm that the analytics went through to kibana?

freddyaboulton · 2024-05-03T00:29:07Z

Yes analytics are going through! Thank you for the review!

* Analytics refactor * add changeset * add changeset * Fix wasm? * Fix python tests' * Revert changes chrome * use util function --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com>

* chatbot components * demoi * add changeset * preprocess fix * add changeset * Make guide for tailwind more verbose (#8152) * Lite wheel optimization (#7855) * Add `pull_request.branches.main` as a trigger of the `publish` workflow * [WIP] Comment out the publish steps * Package and upload the NPM package for debug * Skip the copy_frontend.py hook in the Lite build * add changeset * [WIP] Show gradio files * [WIP] Show gradio files * Comment out installing the gradio and gradio_client libraries * Restore installing gradio_client because it's used by `python js/_website/generate_jsons/generate.py` that follows * Restore installing gradio because it's used by `python js/_website/generate_jsons/generate.py` that follows * Add code * Revert "[WIP] Show gradio files" This reverts commit e15fef2. * Build the gradio wheel with the custom lite target * add changeset * Revert "[WIP] Show gradio files" This reverts commit aef053f. * Revert "Skip the copy_frontend.py hook in the Lite build" This reverts commit ca296d0. * Update .github/actions/install-frontend-deps/action.yml for hatch installation * [WIP] Fix test-functional.yml and .github/actions/install-all-deps/action.yml to call the setup actions in this branch * Revert "[WIP] Fix test-functional.yml and .github/actions/install-all-deps/action.yml to call the setup actions in this branch" This reverts commit 571823b. * Comment-in lines in publish.yml * Move Lite build from publish.yml to deploy-spaces.yml * Use the build_lite option of install-all-deps instead of running the build command * [TMP] Change the branch of action files * Fix the hatch Lite build setting * Return pnpm pack to deploy-space * Revert "[TMP] Change the branch of action files" This reverts commit fe4e1c8. * Remove dependencies for lite build * [TMP] Change the branch of action files * Revert "Remove dependencies for lite build" This reverts commit 856a858. * Install packaging>=23.2 * [TMP] Show packaging version * Fix pip install * Fix * Uninstall packaging once * Use `pip install -U` instead of uninstalling the exiting version * Revert "[TMP] Show packaging version" This reverts commit 910b6bb. * Add `-U` flag * Set packaging version as >=23.2 * Revert the changes on pip install * Set packaging version as >=23.2 in requirements.txt * Revert "Set packaging version as >=23.2" This reverts commit 8aa77c8. * Fix hook name * Revert "Set packaging version as >=23.2 in requirements.txt" This reverts commit fbd605c. * Revert "Revert the changes on pip install" This reverts commit 81ff38a. * Add comments * Revert "[TMP] Change the branch of action files" This reverts commit 0d6aa48. * Revert the trigger of .github/workflows/deploy-spaces.yml * Remove unused `node_auth_token` and `npm_token` inputs from the `install-all-deps` action * [TMP] Trigger CI based on this PR * Remove packging installation * Revert "Remove packging installation" This reverts commit 4a4f18d. * Revert "[TMP] Trigger CI based on this PR" This reverts commit 6cea830. --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: freddyaboulton <alfonsoboulton@gmail.com> * Add ETag to `/custom_component` route to control browser caching (#8170) * Add code * add changeset * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Implement JS Client tests (#8109) * add msw setup and initialisation tests * add changeset * add walk_and_store_blobs improvements and add tests * add changeset * api_info tests * add direct space URL link tests * fix tests * add view_api tests * add post_message test * tweak * add spaces tests * jwt and protocol tests * add post_data tests * test tweaks --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Remove hatch installation in js/app/package.json which is no longer needed (#8174) * Remove hatch installation in js/app/package.json which is no longer needed * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Update test-functional.yml - Fix vulnerability code injection (#8145) * Update test-functional.yml * Update test-functional.yml * tweaks --------- Co-authored-by: pngwn <hello@pngwn.io> * rework upload to be a class method + pass client into each component (#8179) * rework upload to be a class method + pass client into each component * add changeset * Update client/js/src/utils/upload_files.ts * fix storybook * review comments * Apply suggestions from code review Co-authored-by: Hannah <hannahblair@users.noreply.github.com> * format * ts fix --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Hannah <hannahblair@users.noreply.github.com> * chore(deps): update pnpm to v9 (#8123) * chore(deps): update pnpm to v9 * update workflow --------- Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: pngwn <hello@pngwn.io> * Use workspace version for code in _website (#8189) * workspace * add changeset * remove circular import from preview * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Pass Error status in /dev/reload stream (#8106) * get error message * Support multiple clients * add changeset * add changeset * add changeset * Display in UI * console.error the python traceback * lint --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Convert sse calls in client from async to sync (#8182) * convert sse calls in client from async to sync * add changeset * more sync * lint * more sync * fix threadpool * fix timeouts * reuse executor * lint --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * run python reload only if python file changed (#8194) * run python reload only if python file changed * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Freddy Boulton <alfonsoboulton@gmail.com> * fix: handling SIGINT correctly in reload.py, single entrance of block_thread in blocks.py (#8158) * fix: handling SIGINT, single block_thread and fix popen * Use pass --------- Co-authored-by: freddyaboulton <alfonsoboulton@gmail.com> * Add eventsource polyfill for Node.js and browser environments (#8118) * add msw setup and initialisation tests * add changeset * add eventsource polyfill for node and browser envs * add changeset * add changeset * config tweak * types * update eventsource usage * add changeset * add walk_and_store_blobs improvements and add tests * add changeset * api_info tests * add direct space URL link tests * fix tests * add view_api tests * add post_message test * tweak * add spaces tests * jwt and protocol tests * add post_data tests * test tweaks * dynamically import eventsource * revet eventsource imports * add node test * lockfile * add client test in root pkg file * lcokfile * remove eventsource from js/app * add changeset * remove ts ignore * move eventsource polyfill to eventsource factory * add changeset * tweak --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Ensure connectivity to private HF spaces with SSE protocol (#8181) * add msw setup and initialisation tests * add changeset * add eventsource polyfill for node and browser envs * add changeset * add changeset * config tweak * types * update eventsource usage * add changeset * add walk_and_store_blobs improvements and add tests * add changeset * api_info tests * add direct space URL link tests * fix tests * add view_api tests * add post_message test * tweak * add spaces tests * jwt and protocol tests * add post_data tests * test tweaks * dynamically import eventsource * revet eventsource imports * add jwt param to sse requests * add stream test * add changeset * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Support custom components in gr.load (#8200) * Add code * add changeset * Update fuzzy-mirrors-scream.md * Update fuzzy-mirrors-scream.md * Fix tests * Update .changeset/fuzzy-mirrors-scream.md Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * Fix code --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * Refactor analytics to not use api.gradio.app (#8180) * Analytics refactor * add changeset * add changeset * Fix wasm? * Fix python tests' * Revert changes chrome * use util function --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Specify the fastapi version on Lite to avoid ujson installation which is not available on Pyodide yet (#8204) * Specify the fastapi version on Lite to avoid ujson installation which is not available on Pyodide yet * add changeset * Refactoring --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Set the show_api flag on Lite (#8205) * Set the show_api flag on Lite * add changeset * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Extend Interface.from_pipeline() to support Transformers.js.py pipelines on Lite (#8052) * Extend Interface.from_pipeline() to support Transformers.js.py pipelines on Lite (wip: only object-detection in this commit) * add changeset * Add image-classification and image-segmentation * Add zero-shot-image-classification and zero-shot-object-detection * Add document-question-answering * Add feature-extraction and fill-mask * Add question-answering and summarization * Fix an error message * Add text2text-generation, text-classification, and text-generation * Add translation andtranslation_xx_to_yy * Add zero-shot-classification * Add postprocess_takes_inputs to control the args passed to the postprocess function of each pipeline * Add topk option to image-classification * format_backend * Add audio-classification, automatic-speech-recognition, and zero-shot-audio-classification * Add image-to-text * Add token-classification (with JSON component as an output. Is it correct?) * Ignore import type failure of transformers_js_py * Add image-feature-extraction * Add image-to-image * Add text-to-audio * Add depth-estimation * Remove `render=False` * Reorder the if-blocks following the Transformers.js doc * Update gradio/pipelines_utils.py Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * Update gradio/pipelines_utils.py Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * Fix feature-extraction demo * Fix demo title * Add guides/08_gradio-clients-and-lite/gradio-lite-and-transformers-js.md without contents * Rename guides/08_gradio-clients-and-lite/*.md to fix the order * Use pipeline.model.config._name_or_path for the demo title instead of pipeline.model.config.model_type * Fix normal Interface.from_pipeline to use pipeline.model.config.name_or_path as the demo title * Write an article about Gradio-Lite and Transformers.js * Update the doc * tweaks * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * merge * allow the canvas size to be set on the `ImageEditor` (#8127) * add canvas size kwarg to imageeditor * add changeset * fix tests * fix cropsize * changes * notebooks * update docstrings * fix type * fix undefined dimensions * Update image_editor.py Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * fix type * format --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Abubakar Abid <abubakar@huggingface.co> * Rename `eventSource_Factory` and `fetch_implementation` (#8209) * rename eventSource_factory -> stream_factory + rename event_source -> steam * rename fetch_implementation -> fetch * rename fetch to _fetch due to global.fetch conflict * add changeset * format * format * format * format * fix --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * remove redundant event source logic (#8211) * remove redundant event source logic * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * Only connect to heartbeat if needed (#8169) * Add connect_heartbeat field * fix types * add changeset --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> * chore: update versions (#8172) * fixes * type fixes * type fixes * notebook fix * type ignore * data object model * remove component in tuple * more fixes * extend components * remove test var * extend to all components backend * remove loading data models * conflict fix * test and type fixes * playwright test * PR fixes * final changes * Add pltly for 2e2 test * pass loader to Gradio helper class * fix things * add changeset * checks * more fixy * more fixy * more fixy --------- Co-authored-by: gradio-pr-bot <gradio-pr-bot@users.noreply.github.com> Co-authored-by: Simon Duerr <dev@simonduerr.eu> Co-authored-by: Yuichiro Tachibana (Tsuchiya) <t.yic.yt@gmail.com> Co-authored-by: freddyaboulton <alfonsoboulton@gmail.com> Co-authored-by: Hannah <hannahblair@users.noreply.github.com> Co-authored-by: Lê Ngọc Hoa <114990730+h2oa@users.noreply.github.com> Co-authored-by: pngwn <hello@pngwn.io> Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> Co-authored-by: Ali Abdalla <ali.si3luwa@gmail.com> Co-authored-by: Abubakar Abid <abubakar@huggingface.co> Co-authored-by: James Zhou <61927718+jameszhou02@users.noreply.github.com> Co-authored-by: Tiger3018 <tiger3018of02@gmail.com>

freddyaboulton added v: patch A change that requires a patch release t: fix A change that implements a fix labels Apr 30, 2024

freddyaboulton commented Apr 30, 2024

View reviewed changes

freddyaboulton force-pushed the analytics-local branch 2 times, most recently from 0daaf28 to f46379d Compare May 2, 2024 01:27

freddyaboulton marked this pull request as ready for review May 2, 2024 17:36

freddyaboulton requested review from abidlabs, whitphx, aliabid94, aliabd, dawoodkhan82, pngwn and hannahblair May 2, 2024 17:37

freddyaboulton mentioned this pull request May 2, 2024

Support custom components in gr.load #8200

Merged

freddyaboulton and others added 7 commits May 2, 2024 19:12

Analytics refactor

13c969a

add changeset

617a214

add changeset

e9dce94

Fix wasm?

6881079

Fix python tests'

433ee42

Revert changes chrome

428c171

use util function

522fa68

freddyaboulton force-pushed the analytics-local branch from f46379d to 522fa68 Compare May 2, 2024 23:14

abidlabs approved these changes May 2, 2024

View reviewed changes

freddyaboulton merged commit 449d0e6 into main May 3, 2024
7 checks passed

freddyaboulton deleted the analytics-local branch May 3, 2024 00:29

pngwn mentioned this pull request May 3, 2024

chore: update versions #8172

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Refactor analytics to not use api.gradio.app #8180

Refactor analytics to not use api.gradio.app #8180

freddyaboulton commented Apr 30, 2024

gradio-pr-bot commented Apr 30, 2024 •

edited

Loading

gradio-pr-bot commented Apr 30, 2024 •

edited

Loading

Something isn't right?

abidlabs commented Apr 30, 2024 •

edited

Loading

freddyaboulton commented Apr 30, 2024

abidlabs commented Apr 30, 2024

freddyaboulton Apr 30, 2024

whitphx May 1, 2024

freddyaboulton May 2, 2024

whitphx May 2, 2024

abidlabs commented Apr 30, 2024

Wauplin commented May 2, 2024

freddyaboulton commented May 2, 2024

freddyaboulton commented May 2, 2024

abidlabs left a comment

freddyaboulton commented May 3, 2024 •

edited

Loading

Refactor analytics to not use api.gradio.app #8180

Refactor analytics to not use api.gradio.app #8180

Conversation

freddyaboulton commented Apr 30, 2024

Description

🎯 PRs Should Target Issues

Tests

gradio-pr-bot commented Apr 30, 2024 • edited Loading

🪼 branch checks and previews

gradio-pr-bot commented Apr 30, 2024 • edited Loading

🦄 change detected

This Pull Request includes changes to the following packages.

With the following changelog entry.

Something isn't right?

abidlabs commented Apr 30, 2024 • edited Loading

freddyaboulton commented Apr 30, 2024

abidlabs commented Apr 30, 2024

freddyaboulton Apr 30, 2024

Choose a reason for hiding this comment

whitphx May 1, 2024

Choose a reason for hiding this comment

freddyaboulton May 2, 2024

Choose a reason for hiding this comment

whitphx May 2, 2024

Choose a reason for hiding this comment

abidlabs commented Apr 30, 2024

Wauplin commented May 2, 2024

freddyaboulton commented May 2, 2024

freddyaboulton commented May 2, 2024

abidlabs left a comment

Choose a reason for hiding this comment

freddyaboulton commented May 3, 2024 • edited Loading

gradio-pr-bot commented Apr 30, 2024 •

edited

Loading

gradio-pr-bot commented Apr 30, 2024 •

edited

Loading

abidlabs commented Apr 30, 2024 •

edited

Loading

freddyaboulton commented May 3, 2024 •

edited

Loading