From 736f4a24b16ca2a0a7198f3372c4ed15acd2cb7e Mon Sep 17 00:00:00 2001 From: Brandon Thomas Date: Sat, 31 Aug 2024 19:16:04 -0400 Subject: [PATCH] useful analytics queries --- .../backfill_job_product_categories.sql | 8 ++++ .../backfills/set_robot_accounts.sql | 17 +++++++ .../media_files/media_files_types_by_user.sql | 10 ++++ .../tts/tts_results_by_user.sql | 47 +++++++++++++++++++ .../users/top_users_by_usage.sql | 29 +++++++----- 5 files changed, 99 insertions(+), 12 deletions(-) create mode 100644 _sql/actions_and_analytics/backfills/set_robot_accounts.sql create mode 100644 _sql/actions_and_analytics/media_files/media_files_types_by_user.sql create mode 100644 _sql/actions_and_analytics/tts/tts_results_by_user.sql diff --git a/_sql/actions_and_analytics/backfills/backfill_job_product_categories.sql b/_sql/actions_and_analytics/backfills/backfill_job_product_categories.sql index a975d3ab02..74121b72f5 100644 --- a/_sql/actions_and_analytics/backfills/backfill_job_product_categories.sql +++ b/_sql/actions_and_analytics/backfills/backfill_job_product_categories.sql @@ -153,6 +153,14 @@ AND created_at <= '2024-08-30' LIMIT 100000; +-- Update GptSoVits Download Jobs +UPDATE generic_inference_jobs +SET product_category = 'download_gpt_so_vits' +WHERE job_type = 'gpt_sovits' +AND maybe_download_url IS NOT NULL +LIMIT 10000; + + --- TODO: Not working --- Update RVCv2 VC jobs UPDATE generic_inference_jobs diff --git a/_sql/actions_and_analytics/backfills/set_robot_accounts.sql b/_sql/actions_and_analytics/backfills/set_robot_accounts.sql new file mode 100644 index 0000000000..af90f4e918 --- /dev/null +++ b/_sql/actions_and_analytics/backfills/set_robot_accounts.sql @@ -0,0 +1,17 @@ +-- Doc: https://docs.google.com/document/d/14AyEV_0PYwa71a5W8xRRZ8vbI0uZHQF4ycqi539qmAc/edit +-- bluemo69: SpongeBob (abandoned) +-- dzth: likely AI streamer - 144,653 media files (all audio, mostly SpongeBob, using old tokens) in August 2024 +-- johnloberger: AI South Park +-- rewritten_code: AI Breaking Bad +UPDATE users +SET is_api_user = TRUE +WHERE username IN ( + 'bluemo69', + 'dzth', + 'johnloberger', + 'rewritten_code' +); + +-- TODO: +-- leon_says: ???? - 209,745 media files in August 2024 +-- robertmctague: ???? - 285,664 media files in August 2024 diff --git a/_sql/actions_and_analytics/media_files/media_files_types_by_user.sql b/_sql/actions_and_analytics/media_files/media_files_types_by_user.sql new file mode 100644 index 0000000000..624920776f --- /dev/null +++ b/_sql/actions_and_analytics/media_files/media_files_types_by_user.sql @@ -0,0 +1,10 @@ +SELECT + m.media_type, + COUNT(*) +FROM media_files as m +JOIN users as u + ON m.maybe_creator_user_token = u.token +WHERE u.username = lower('dzth') +GROUP BY m.media_type + + diff --git a/_sql/actions_and_analytics/tts/tts_results_by_user.sql b/_sql/actions_and_analytics/tts/tts_results_by_user.sql new file mode 100644 index 0000000000..5fc6c3ab0e --- /dev/null +++ b/_sql/actions_and_analytics/tts/tts_results_by_user.sql @@ -0,0 +1,47 @@ + +-- Select most used tts models by user +-- This query has old token support! +SELECT + mw.token, + mw.title, + count(*) as use_count +FROM model_weights as mw +JOIN ( + select + coalesce(w.token, w_migrated.token) as token + from media_files as f + left outer join model_weights as w + on f.maybe_origin_model_token = w.token + left outer join model_weights as w_migrated + on f.maybe_origin_model_token = w_migrated.maybe_migration_old_model_token + join users as u + on f.maybe_creator_user_token = u.token + where u.username = lower('dzth') +) as x +on mw.token = x.token +group by mw.token, mw.title +order by use_count desc + + +-- See a sample of tts results by user +-- This includes old token support +select + f.token, + f.origin_product_category, + f.created_at, + w.token, + w.title, + w_migrated.token, + w_migrated.title, + SUBSTRING(TRIM(f.maybe_text_transcript), 1, 100) as transcript +from media_files as f +left outer join model_weights as w + on f.maybe_origin_model_token = w.token +left outer join model_weights as w_migrated + on f.maybe_origin_model_token = w_migrated.maybe_migration_old_model_token +join users as u + on f.maybe_creator_user_token = u.token +where u.username = lower('dzth') +order by f.id desc +limit 100 + diff --git a/_sql/actions_and_analytics/users/top_users_by_usage.sql b/_sql/actions_and_analytics/users/top_users_by_usage.sql index 319e607b69..e9f69e0f9d 100644 --- a/_sql/actions_and_analytics/users/top_users_by_usage.sql +++ b/_sql/actions_and_analytics/users/top_users_by_usage.sql @@ -2,6 +2,23 @@ -- noinspection SqlNoDataSourceInspectionForFile -- noinspection SqlResolveForFile +SELECT + m.maybe_creator_user_token as user_token, + u.username, + count(*) as usage_count, + u.created_at +FROM media_files as m +JOIN users as u +ON u.token = m.maybe_creator_user_token +WHERE m.maybe_creator_user_token IS NOT NULL +AND m.created_at > NOW() - INTERVAL 31 DAY +GROUP BY m.maybe_creator_user_token +ORDER BY usage_count DESC +LIMIT 50000; + + +----- Older queries ----- + SELECT u.username as username, j.token as token @@ -11,18 +28,6 @@ ON u.token = j.maybe_creator_user_token AND j.created_at > NOW() - INTERVAL 31 DAY -SELECT - t.maybe_creator_user_token as user_token, - u.username, - count(*) as usage_count, - u.created_at -FROM tts_results as t -JOIN users as u -ON u.token = t.maybe_creator_user_token -WHERE t.maybe_creator_user_token IS NOT NULL -GROUP BY t.maybe_creator_user_token -ORDER BY usage_count DESC -LIMIT 50000; SELECT