{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "3b01d7f9", "metadata": {}, "outputs": [], "source": [ "import sys\n", "sys.path.insert(0, \"/home/georg/code/glockenspiel/suno_analytics/\")" ] }, { "cell_type": "code", "execution_count": 35, "id": "4a711a9c", "metadata": {}, "outputs": [], "source": [ "import time\n", "import json\n", "from urllib.parse import quote\n", "import datetime\n", "import pandas as pd\n", "import boto3\n", "import sqlalchemy\n", "\n", "def get_secret():\n", " session = boto3.session.Session()\n", " client = session.client(service_name=\"secretsmanager\", region_name=\"us-east-2\")\n", " get_secret_value_response = client.get_secret_value(SecretId=\"app-user-main-db-secret\")\n", " secret = get_secret_value_response[\"SecretString\"]\n", " return json.loads(secret)\n", "\n", "my_secrets = get_secret()\n", "\n", "engine = sqlalchemy.create_engine(\n", " \"postgresql://suno:%s@suno-main-postgres-prod-analytics.cnfvffydbwvc.us-east-2.rds.amazonaws.com/suno_main\"\n", " % quote(my_secrets[\"password\"]),\n", ")" ] }, { "cell_type": "code", "execution_count": 43, "id": "a1b3d762", "metadata": {}, "outputs": [], "source": [ "# # cutoff_date = \"2025-07-04 19:31:53\"\n", "# df = pd.read_sql_query(\n", "# \"SELECT table_name FROM information_schema.tables WHERE table_schema = 'public'\",\n", "# engine,\n", "# )\n", "# assert df[\"table_name\"].nunique() >= 61" ] }, { "cell_type": "code", "execution_count": 40, "id": "560da53a", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "=== bots_notificationhistory ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- notification_type\n", "- push_sent_at\n", "- push_period_end\n", "- user_id\n", "\n", "=== bots_cliplineage ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- parent_id\n", "- relationship\n", "- clip_id\n", "\n", "=== bots_genreanalytics ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- genre\n", "- ditto_vector\n", "- is_platform_genre\n", "- image_url\n", "- ditto_v2_vector\n", "\n", "=== growth_adsuser ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- idfa\n", "- idfa_source\n", "- user_id\n", "\n", "=== bots_userprofilepin ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- relative_index\n", "- user_id\n", "\n", "=== bots_customfreecreditlimit ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- free_credits_limit\n", "- duration_since_signup_in_days\n", "- user_id\n", "\n", "=== video_videohook ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- original_clip\n", "- status\n", "- title\n", "- start_clip_timestamp\n", "- end_clip_timestamp\n", "- rendered_video_s3_id\n", "- rendered_video_preview_s3_id\n", "- thumbnail_image_s3_id\n", "- raw_video_upload_id\n", "- user_id\n", "- is_test\n", "- allow_comments\n", "- caption\n", "- show_lyrics\n", "- content_rating\n", "- content_rating_tags\n", "- video_duration\n", "- video_streaming_resolutions\n", "- is_deleted\n", "- rendered_watermarked_video_s3_id\n", "\n", "=== growth_usershare ===\n", "- created_at\n", "- updated_at\n", "- id\n", "- source\n", "- platform\n", "- content_id\n", "- content_type\n", "- sharer_id\n", "- content_owner_id\n", "\n", "=== bots_groupedusernotification ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- type\n", "- content_id\n", "- sender_id_1\n", "- sender_id_2\n", "- count\n", "- is_read\n", "- is_active\n", "- last_operation_at\n", "- sent_at\n", "- read_at\n", "- user_id\n", "- sender_id_3\n", "- secondary_content_id\n", "\n", "=== music_player_playbarstate ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- playbar_state\n", "- song_index\n", "- song_play_time\n", "- repeat_state\n", "- action_time\n", "- song_ids_in_queue\n", "- playlist_context\n", "- device_context\n", "- device_id\n", "- device_type\n", "- user_id\n", "- volume\n", "- context_id\n", "- context_type\n", "\n", "=== moderation_rules ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- entity_type\n", "- entity_id\n", "- capability\n", "- status\n", "- reason\n", "- expires_at\n", "- created_by\n", "- shadow_mode\n", "\n", "=== bots_generatedlyrics ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- interaction_type\n", "- resulted_in_error\n", "- is_complete\n", "- lyrics_request_id\n", "- prompt_text\n", "- model_name\n", "- create_session_token\n", "- model_generated_title\n", "- model_generated_lyrics\n", "- model_generated_tags\n", "- title_used\n", "- lyrics_used\n", "- tags_used\n", "- create_custom_used_model_output\n", "- generation_request_id\n", "- user_id\n", "- selected_for_ab\n", "- context_lyrics_edit\n", "- context_lyrics_prefix\n", "- context_lyrics_suffix\n", "- infill_undone\n", "\n", "=== oauth2_provider_grant ===\n", "- id\n", "- code\n", "- expires\n", "- redirect_uri\n", "- scope\n", "- application_id\n", "- user_id\n", "- created\n", "- updated\n", "- code_challenge\n", "- code_challenge_method\n", "- nonce\n", "- claims\n", "\n", "=== bots_userdeleterequest ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- persist_user_id\n", "- is_deleted\n", "- deleted_at\n", "- user_id\n", "- clerk_id\n", "- username\n", "- discord_uid\n", "- is_operated_by_admin\n", "- email\n", "\n", "=== oauth2_provider_application ===\n", "- id\n", "- client_id\n", "- redirect_uris\n", "- client_type\n", "- authorization_grant_type\n", "- client_secret\n", "- name\n", "- user_id\n", "- skip_authorization\n", "- created\n", "- updated\n", "- algorithm\n", "- post_logout_redirect_uris\n", "- hash_client_secret\n", "- allowed_origins\n", "\n", "=== oauth2_provider_idtoken ===\n", "- id\n", "- jti\n", "- expires\n", "- scope\n", "- created\n", "- updated\n", "- application_id\n", "- user_id\n", "\n", "=== oauth2_provider_accesstoken ===\n", "- id\n", "- token\n", "- expires\n", "- scope\n", "- application_id\n", "- user_id\n", "- created\n", "- updated\n", "- source_refresh_token_id\n", "- id_token_id\n", "- token_checksum\n", "\n", "=== bots_clipparent ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- non_owned_parent_clip_id\n", "- owned_parent_clip_id\n", "\n", "=== oauth2_provider_refreshtoken ===\n", "- id\n", "- token\n", "- access_token_id\n", "- application_id\n", "- user_id\n", "- created\n", "- updated\n", "- revoked\n", "- token_family\n", "\n", "=== bots_radiostationstats ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- num_likes\n", "- num_live_user\n", "- play_count\n", "- radio_id\n", "\n", "=== bots_radiostationreaction ===\n", "- id\n", "- reaction_type\n", "- radio_id\n", "- user_id\n", "\n", "=== recommendations_usertouserengagement ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- user_id\n", "- engage_user_id\n", "- total_score\n", "- relationship\n", "- p_date\n", "\n", "=== bots_userhcaptchahistory ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- token\n", "- passed\n", "- score\n", "- ip_address\n", "- hcaptcha_response\n", "- user_id\n", "\n", "=== bots_school ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- country\n", "- domain\n", "\n", "=== bots_revcatevent ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- event_id\n", "- app_id\n", "- app_user_id\n", "- transaction_id\n", "- type\n", "- payload\n", "- result\n", "- presented_offering_id\n", "- product_id\n", "\n", "=== bots_usageplan ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- level\n", "- name\n", "- monthly_price_usd\n", "- features\n", "- stripe_product_id\n", "- stripe_annual_price_id\n", "- stripe_monthly_price_id\n", "- daily_credits\n", "- monthly_credits\n", "- annual_price_usd\n", "- plan_key\n", "\n", "=== bots_projectclip ===\n", "- id\n", "- relative_index\n", "- pinned\n", "- updated_at\n", "- clip_id\n", "- project_id\n", "- updated_by_id\n", "- user_id\n", "- created_session_id\n", "\n", "=== bots_dummymigrationtestingmodel ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- test_field\n", "\n", "=== bots_playlistclip ===\n", "- id\n", "- relative_index\n", "- updated_at\n", "- clip_id\n", "- playlist_id\n", "- updated_by_id\n", "- created_session_id\n", "\n", "=== bots_onetimefreeusage ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- free_cover_generations_used\n", "- free_cover_generations_limit\n", "- free_persona_generations_used\n", "- free_persona_generations_limit\n", "- user_id\n", "- free_web_remasters_limit\n", "- free_web_remasters_used\n", "- free_mobile_remasters_limit\n", "- free_mobile_remasters_used\n", "- free_mobile_v4_gens_limit\n", "- free_mobile_v4_gens_used\n", "- free_web_v4_gens_limit\n", "- free_web_v4_gens_used\n", "\n", "=== bots_personalizedplaylist ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- playlist_type\n", "- owner_user_id\n", "- playlist_id\n", "- user_id\n", "\n", "=== auth_group ===\n", "- id\n", "- name\n", "\n", "=== bots_collabplaylist ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- image_s3_id\n", "- is_public\n", "- is_collaborative\n", "- version_number\n", "- is_discover_playlist\n", "- upvote_count\n", "- dislike_count\n", "- play_count\n", "- skip_count\n", "- flag_count\n", "- is_public_approved\n", "- playlist_type\n", "- is_trashed\n", "- creation_source\n", "- created_session_id\n", "- user_id\n", "\n", "=== auth_permission ===\n", "- id\n", "- name\n", "- content_type_id\n", "- codename\n", "\n", "=== auth_user_groups ===\n", "- id\n", "- user_id\n", "- group_id\n", "\n", "=== bots_actionlogging ===\n", "- id\n", "- email\n", "- discord_id\n", "- action_type\n", "- action_message\n", "- clips\n", "- timestamp\n", "- user_id\n", "\n", "=== bots_affiliate ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- affiliate_code\n", "- is_active\n", "- first_subscribe_at\n", "- user_id\n", "\n", "=== bots_clipalignment ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- history_clip_id\n", "- history_end_seconds\n", "- clip_start_seconds\n", "- future_clip_id\n", "- clip_end_seconds\n", "- future_start_seconds\n", "\n", "=== bots_appversionupdate ===\n", "- id\n", "- updated_at\n", "- app_name\n", "- platform\n", "- version\n", "- version_type\n", "- action_type\n", "- created_at\n", "\n", "=== bots_clipcommentban ===\n", "- id\n", "- reason\n", "- created_at\n", "- expires_at\n", "- banned_user_id\n", "- clip_id\n", "\n", "=== bots_blockedpayments ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- fingerprint\n", "- type\n", "- email\n", "- username\n", "\n", "=== bots_botjail ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- username\n", "- status\n", "- total_jailed_count\n", "- latest_jailed_reason\n", "- latest_jailed_at\n", "- latest_released_at\n", "- total_copycat_violation_count\n", "- latest_copycat_violation_at\n", "- total_api_violation_count\n", "- api_counter\n", "- latest_api_violation_at\n", "- total_hcaptcha_violation_count\n", "- hcaptcha_score\n", "- latest_hcaptcha_violation_at\n", "- user_id\n", "- copycat_name\n", "- user_ip\n", "- api_attribution\n", "\n", "=== bots_clipdiscordmessage ===\n", "- id\n", "- discord_message_id\n", "- created_at\n", "- clip_id\n", "- discord_channel_id\n", "- discord_guild_id\n", "- discord_is_dm\n", "\n", "=== moderation_rule_enforcement_logs ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- entity_type\n", "- entity_id\n", "- capability\n", "- shadow_mode\n", "- was_applied\n", "- rule_status\n", "- expires_at\n", "\n", "=== bots_clipcaption ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- caption\n", "- raw_mentions\n", "\n", "=== bots_contest ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- type\n", "- description\n", "- start_time\n", "- end_time\n", "- submissions_playlist_id\n", "- can_download_remixes\n", "\n", "=== bots_commentban ===\n", "- reason\n", "- created_at\n", "- expires_at\n", "- banned_user_id\n", "- is_released\n", "- id\n", "- updated_at\n", "\n", "=== bots_clipprompt ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- clip_id\n", "- user_id\n", "- is_public\n", "- string_tags\n", "\n", "=== bots_commentreport ===\n", "- id\n", "- updated_at\n", "- created_at\n", "- reason\n", "- comment_id\n", "- user_id\n", "\n", "=== bots_contestwinnernotification ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- contest_name\n", "- is_read\n", "- is_active\n", "- sent_at\n", "- read_at\n", "- clip_id\n", "- user_id\n", "\n", "=== bots_dailytheme ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- theme\n", "- date\n", "\n", "=== bots_mobileusageplan ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- product_id\n", "- subscription_period_type\n", "- usage_plan_id\n", "- app_store\n", "\n", "=== bots_generatedclip ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- time_used\n", "- metadata\n", "- user_id\n", "- status\n", "- discord_message_id\n", "- prompt_id\n", "- request_id\n", "- is_generated\n", "- s3_id\n", "- upvote_count\n", "- batch_index\n", "- model_name\n", "- prompt_text\n", "- daily_theme_id\n", "- is_deleted\n", "- image_s3_id\n", "- is_public\n", "- dislike_count\n", "- flag_count\n", "- play_count\n", "- skip_count\n", "- title\n", "- slug\n", "- creation_source\n", "- allow_comments\n", "- is_hidden\n", "- display_tags\n", "\n", "=== bots_generationrequest ===\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "- id\n", "- created_at\n", "- updated_at\n", "- status\n", "- time_used\n", "- discord_message_id\n", "- batch_size\n", "- model_name\n", "- metadata\n", "- prompt_id\n", "- user_id\n", "- is_deleted\n", "\n", "=== bots_periodcreditusage ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- period_anchor\n", "- credits_used\n", "- period_credit_limit\n", "- extra_credits_balance\n", "- user_id\n", "- period_end\n", "- stripe_subscription_id\n", "- plan_level\n", "- extra_credits_used\n", "- usage_plan_id\n", "- revcat_subscription_id\n", "- is_active\n", "- period_extra_credit_limit\n", "- subscription_period_type\n", "\n", "=== bots_generatedplaylist ===\n", "- id\n", "- type\n", "- tag\n", "- updated_at\n", "- playlist_id\n", "\n", "=== bots_iosinvitepromocodeusage ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- user_country\n", "- promo_code\n", "- redemption_status\n", "- user_id\n", "\n", "=== bots_iosinvitepromocode ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- promo_code\n", "- redeemed_promo_code\n", "- user_id\n", "- admitted_at\n", "- user_country\n", "- waitlisted_at\n", "\n", "=== bots_iosinvitereward ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- credits_granted\n", "- granted_at\n", "- reward_type\n", "- user_id\n", "\n", "=== bots_persona ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- root_clip_id\n", "- name\n", "- description\n", "- image_s3_id\n", "- user_id\n", "- is_suno_persona\n", "- is_trashed\n", "- clip_id\n", "- is_public\n", "- is_public_approved\n", "- upvote_count\n", "- follower_count\n", "- created_session_id\n", "- clip_count\n", "\n", "=== bots_personafollow ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- followed_persona_id\n", "- following_user_id\n", "\n", "=== bots_personaclip ===\n", "- id\n", "- clip_id\n", "- persona_id\n", "- created_at\n", "- updated_at\n", "- user_id\n", "- created_session_id\n", "\n", "=== bots_playlistanalytics ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- ditto_vector\n", "- playlist_id\n", "- user_id\n", "\n", "=== bots_remixcontestsoundclash ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- duration\n", "- clip_id\n", "- user_id\n", "- is_base_clip\n", "\n", "=== bots_projectinvite ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- invited_email\n", "- status\n", "- invited_by_user_id\n", "- invited_user_id\n", "- project_id\n", "\n", "=== bots_profilefollow ===\n", "- id\n", "- updated_at\n", "- created_at\n", "- followed_profile_id\n", "- following_profile_id\n", "\n", "=== bots_project ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- image_s3_id\n", "- is_trashed\n", "- is_public\n", "- user_id\n", "- created_session_id\n", "\n", "=== bots_radiopromptchat ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- prompt\n", "- role\n", "- radio_id\n", "- user_id\n", "- instrumental\n", "- is_ready\n", "- should_gen\n", "- search_keywords\n", "- created_session_id\n", "\n", "=== bots_purchaseinfo ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- stripe_id\n", "- purchase_type\n", "- amount\n", "- user_id\n", "- status\n", "- client_secret\n", "- revcat_id\n", "\n", "=== bots_radiostation ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- image_s3_id\n", "- prompt\n", "- is_liked\n", "- last_listened_at\n", "- playlist_id\n", "- user_id\n", "- is_live\n", "- is_suno_hosted\n", "- query_term\n", "- created_session_id\n", "\n", "=== bots_sessionhistory ===\n", "- session_id\n", "- session_type\n", "- user_id\n", "- city\n", "- country\n", "- ip\n", "- created_at\n", "- updated_at\n", "- expire_time\n", "- length_in_sec\n", "- properties\n", "- client_creation_time\n", "- client_expire_time\n", "- platform\n", "\n", "=== bots_similarclips ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- vector\n", "- similar_clip_ids\n", "\n", "=== bots_staffpickplaylist ===\n", "- created_at\n", "- updated_at\n", "- id\n", "- name\n", "- tags\n", "- image_url\n", "- playlist_id\n", "- redirect_url\n", "\n", "=== bots_uncategorizedpayments ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- charge_id\n", "- type\n", "- email\n", "- username\n", "\n", "=== bots_collabplaylistclip ===\n", "- id\n", "- relative_index\n", "- updated_at\n", "- created_session_id\n", "- clip_id\n", "- collab_playlist_id\n", "- updated_by_id\n", "\n", "=== bots_usercaptchastate ===\n", "- id\n", "- captcha_required\n", "- failed_last_token_validation\n", "- last_hcaptcha_score\n", "- last_captcha_completed_at\n", "- gens_since_last_captcha\n", "- state_checks_since_last_gen\n", "- user_id\n", "- total_interactive_challenges_completed\n", "- created_at\n", "- updated_at\n", "\n", "=== social_followingfeedactivity ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- entity_id\n", "- entity_type\n", "- activity_type\n", "- feed_timestamp\n", "- followed_user_id\n", "- following_user_id\n", "- activity_id\n", "\n", "=== bots_userprofile ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- creation_vector\n", "- listening_vector\n", "- user_id\n", "\n", "=== recommendations_usersongfeedback ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- feedback_type\n", "- feedback_reasons\n", "- custom_reason\n", "- is_current\n", "- clip_id\n", "- user_id\n", "- is_deleted\n", "- entity_id\n", "\n", "=== bots_userjail ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- status\n", "- reason\n", "- source\n", "- user_id\n", "- counter\n", "- hcaptcha_score\n", "- is_bot_by_reactions\n", "\n", "=== bots_usermobiledevicetoken ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- registration_token\n", "- registration_token_timestamp\n", "- user_id\n", "\n", "=== bots_usernotification ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- type\n", "- source_id\n", "- is_read\n", "- is_active\n", "- group_id\n", "- sender_info_id\n", "- user_id\n", "- read_at\n", "- sent_at\n", "\n", "=== bots_userpersonareaction ===\n", "- id\n", "- loved\n", "- updated_at\n", "- persona_id\n", "- user_id\n", "\n", "=== bots_videoshortextra ===\n", "- video_id_id\n", "- created_at\n", "- updated_at\n", "- flag_count\n", "- share_count\n", "- download_count\n", "\n", "=== bots_videoshort ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- is_trashed\n", "- image_s3_id\n", "- video_s3_id\n", "- title\n", "- description\n", "- duration\n", "- like_count\n", "- dislike_count\n", "- play_count\n", "- visibility\n", "- moderation\n", "- inferred_language\n", "- short_key\n", "- original_audio_id\n", "- user_id\n", "\n", "=== clips_clip_tags ===\n", "- id\n", "- clip_id\n", "- tag_id\n", "\n", "=== bots_videoshortreaction ===\n", "- id\n", "- play_count\n", "- skip_count\n", "- flagged\n", "- flagged_reason\n", "- reaction_type\n", "- updated_at\n", "- user_id\n", "- video_id_id\n", "\n", "=== clips_clip ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- s3_audio_path\n", "- origin_type\n", "- created_by_id\n", "- metadata\n", "- is_processing\n", "- is_shared\n", "- string_tags\n", "- ext\n", "- source_type\n", "\n", "=== clips_aggregation_info ===\n", "- clip_id\n", "- total_play_count\n", "- auto_play_count\n", "- total_play_time\n", "- created_at\n", "- updated_at\n", "\n", "=== clips_playlist ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "\n", "=== clips_tag ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- created_by_id\n", "\n", "=== clips_usertoken ===\n", "- id\n", "- token\n", "- is_active\n", "- user_id\n", "\n", "=== clips_team ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "\n", "=== clips_teammembership ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- role\n", "- team_id\n", "- user_id\n", "\n", "=== bots_commentreaction ===\n", "- reaction_type\n", "- comment_id\n", "- user_id\n", "- id\n", "- created_at\n", "- updated_at\n", "\n", "=== clips_textprompt_tags ===\n", "- id\n", "- textprompt_id\n", "- tag_id\n", "\n", "=== bots_playlist ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "- image_s3_id\n", "- user_id\n", "- is_public\n", "- is_discover_playlist\n", "- is_trashed\n", "- dislike_count\n", "- flag_count\n", "- is_public_approved\n", "- play_count\n", "- skip_count\n", "- upvote_count\n", "- playlist_type\n", "- creation_source\n", "- created_session_id\n", "- song_count\n", "\n", "=== bots_royaluser ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- royal_id\n", "- user_id\n", "\n", "=== growth_pricingexp20250501user ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- exp_treatment_group\n", "- post_exp_treatment_group\n", "- user_id\n", "\n", "=== django_migrations ===\n", "- id\n", "- app\n", "- name\n", "- applied\n", "\n", "=== django_content_type ===\n", "- id\n", "- app_label\n", "- model\n", "\n", "=== django_session ===\n", "- session_key\n", "- session_data\n", "- expire_date\n", "\n", "=== bots_usercommentban ===\n", "- reason\n", "- created_at\n", "- expires_at\n", "- banned_user_id\n", "- target_user_id\n", "- id\n", "- updated_at\n", "\n", "=== label_maker_projects ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- project_name\n", "- question\n", "- instructions\n", "- preference_dimensions\n", "- target_review_count\n", "- statsig_gate\n", "- created_by\n", "- enabled\n", "- priority\n", "- credits_per_annotation\n", "- max_annotations_per_task\n", "- max_annotations_per_user\n", "- min_review_time_seconds\n", "\n", "=== billing_creditsgranthistory ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- amount\n", "- activity_type\n", "- reference_id\n", "- user_id\n", "\n", "=== contests_vote ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- contest_id\n", "- loser_id\n", "- user_id\n", "- winner_id\n", "\n", "=== challenge_userchallenge ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- category\n", "- level\n", "- user_level_required\n", "- name\n", "- description\n", "- reward_credits\n", "- is_completed\n", "- goal\n", "- user_id\n", "\n", "=== label_maker_tasks ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- review_items\n", "- project_id\n", "\n", "=== label_maker_annotations ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- user_id\n", "- preferences\n", "- feedback\n", "- listen_time_seconds\n", "- total_listen_time_seconds\n", "- skipped\n", "- credits_earned\n", "- credits_processed\n", "- task_id\n", "\n", "=== bots_savedprompt ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- title\n", "- gpt_description_prompt\n", "- tags\n", "- negative_tags\n", "- lyrics\n", "- pinned_at\n", "- user_id\n", "- sliders\n", "\n", "=== bots_userconfig ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- shown_creation_tour\n", "- user_id\n", "- trending_song_language\n", "- trending_song_rankby\n", "- has_accepted_timbaland_terms\n", "- preferred_tags\n", "- shown_claim_username\n", "- has_set_remix_perm\n", "- creation_song_language\n", "\n", "=== growth_usershareattribution ===\n", "- created_at\n", "- updated_at\n", "- user_id\n", "- source\n", "- share_id\n", "- content_id\n", "- content_type\n", "\n", "=== video_gen_stockvideo ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- source\n", "- source_video_id\n", "- description\n", "- duration\n", "- width\n", "- height\n", "\n", "=== auth_group_permissions ===\n", "- id\n", "- group_id\n", "- permission_id\n", "\n", "=== bots_usercollabplaylistreaction ===\n", "- id\n", "- play_count\n", "- skip_count\n", "- flagged\n", "- flagged_reason\n", "- reaction_type\n", "- updated_at\n", "- playlist_id\n", "- user_id\n", "\n", "=== auth_user ===\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ "- id\n", "- password\n", "- last_login\n", "- is_superuser\n", "- username\n", "- first_name\n", "- last_name\n", "- email\n", "- is_staff\n", "- is_active\n", "- date_joined\n", "\n", "=== auth_user_user_permissions ===\n", "- id\n", "- user_id\n", "- permission_id\n", "\n", "=== bots_creditpack ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- stripe_id\n", "- amount\n", "- price_usd\n", "\n", "=== bots_generatedplaylistoverride ===\n", "- id\n", "- override_type\n", "- created_at\n", "- clip_id\n", "- generated_playlist_id\n", "\n", "=== bots_invitehistory ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- invitee_reward_amount\n", "- inviter_reward_amount\n", "- invitee_id\n", "- inviter_id\n", "- is_active\n", "- is_recorded\n", "- reward_condition_met_at\n", "- has_sent_credit_limit_depletion_email\n", "- has_sent_email_to_use_credits\n", "- has_sent_invite_success_email\n", "\n", "=== bots_promptinspiration ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- description\n", "\n", "=== bots_userplaylistreaction ===\n", "- id\n", "- play_count\n", "- skip_count\n", "- flagged\n", "- flagged_reason\n", "- reaction_type\n", "- updated_at\n", "- playlist_id\n", "- user_id\n", "\n", "=== bots_userreaction ===\n", "- id\n", "- play_count\n", "- skip_count\n", "- flagged\n", "- flagged_reason\n", "- reaction_type\n", "- updated_at\n", "- clip_id\n", "- user_id\n", "- feedback_reason\n", "\n", "=== clips_clip_playlists ===\n", "- id\n", "- clip_id\n", "- playlist_id\n", "\n", "=== clips_textprompt ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- content\n", "- created_by_id\n", "\n", "=== clips_usercliprating ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- rating\n", "- string_tags\n", "- clip_id\n", "- user_id\n", "\n", "=== django_admin_log ===\n", "- id\n", "- action_time\n", "- object_id\n", "- object_repr\n", "- action_flag\n", "- change_message\n", "- content_type_id\n", "- user_id\n", "\n", "=== pg_stat_statements ===\n", "- userid\n", "- dbid\n", "- toplevel\n", "- queryid\n", "- query\n", "- plans\n", "- total_plan_time\n", "- min_plan_time\n", "- max_plan_time\n", "- mean_plan_time\n", "- stddev_plan_time\n", "- calls\n", "- total_exec_time\n", "- min_exec_time\n", "- max_exec_time\n", "- mean_exec_time\n", "- stddev_exec_time\n", "- rows\n", "- shared_blks_hit\n", "- shared_blks_read\n", "- shared_blks_dirtied\n", "- shared_blks_written\n", "- local_blks_hit\n", "- local_blks_read\n", "- local_blks_dirtied\n", "- local_blks_written\n", "- temp_blks_read\n", "- temp_blks_written\n", "- blk_read_time\n", "- blk_write_time\n", "- temp_blk_read_time\n", "- temp_blk_write_time\n", "- wal_records\n", "- wal_fpi\n", "- wal_bytes\n", "- jit_functions\n", "- jit_generation_time\n", "- jit_inlining_count\n", "- jit_inlining_time\n", "- jit_optimization_count\n", "- jit_optimization_time\n", "- jit_emission_count\n", "- jit_emission_time\n", "\n", "=== pg_stat_statements_info ===\n", "- dealloc\n", "- stats_reset\n", "\n", "=== bots_studentplaneligibleuser ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- user_id\n", "\n", "=== chat_chatmessage ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- channel_id\n", "- channel_type\n", "- chat_id\n", "- chat_text\n", "- user_avatar_url\n", "- user_name\n", "- moderation_status\n", "- moderation_reasons\n", "- user_id\n", "- flag_count\n", "\n", "=== chat_chatflag ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- reason\n", "- chat_message_id\n", "- user_id\n", "\n", "=== bots_contestclip ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- duration\n", "- is_base_clip\n", "- clip_id\n", "- contest_id\n", "- user_id\n", "- is_submitted\n", "- user_country\n", "- rating\n", "\n", "=== bots_usertrackingevent ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- has_created_song\n", "- user_id\n", "\n", "=== bots_discordinfo ===\n", "- id\n", "- discord_id\n", "- user_id\n", "- has_accepted_tos\n", "- discord_username\n", "- clerk_id\n", "- max_concurrent_generations\n", "- stripe_customer_id\n", "- stripe_subscription_id\n", "- subscription_status\n", "- subscription_anchor\n", "- uid\n", "- subscription_cancel_at\n", "- extra_credits_balance\n", "- subscription_period_end\n", "- subscription_plan_id\n", "- extra_credits_used\n", "- subscription_changing_to\n", "- subscription_period_type\n", "- is_on_gen_waitlist\n", "- is_beta_tester\n", "- is_merged\n", "- accepted_custom_mode_tos\n", "- accepted_custom_mode_tos_timestamp\n", "- display_name\n", "- handle\n", "- flagged_at\n", "- handle_updated_at\n", "- profile_description\n", "- avatar_image_url\n", "- active_subscription_platform\n", "- revcat_subscription_id\n", "- phone_number\n", "- secondary_email\n", "- birthday\n", "- registered_platform\n", "- subscription_period_paid_amount_usd\n", "- updated_at\n", "\n", "=== bots_clipembedding ===\n", "- clip_id\n", "- vector\n", "- similar_clip_ids\n", "- is_processing\n", "- lyric_semantic_embedding\n", "- ditto_self_sim_embedding\n", "- created_at\n", "- updated_at\n", "- ditto_genre_embedding\n", "\n", "=== recommendations_trendingclips ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- time_period\n", "- language\n", "- clip_ids\n", "- dagster_run_id\n", "\n", "=== bots_modeltype ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- major_version\n", "- name\n", "- internal_key\n", "- description\n", "- flag_name\n", "- display_priority\n", "- external_key\n", "- model_subtype\n", "- capabilities\n", "\n", "=== bots_clipshareasset ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- clip_start_time\n", "- clip_end_time\n", "- config\n", "- s3_id\n", "- status\n", "\n", "=== bots_alexauserinfo ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- alexa_user_id\n", "- alexa_api_endpoint\n", "- alexa_latest_sync_token\n", "- user_id\n", "\n", "=== bots_studioprojectversion ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- revision\n", "- payload\n", "- project_id\n", "\n", "=== bots_feature ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- name\n", "- description\n", "\n", "=== bots_usageplanfeature ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- feature_id\n", "- usage_plan_id\n", "\n", "=== video_gen_generatedvideo ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- status\n", "- video_url\n", "- clip_id\n", "- user_id\n", "\n", "=== bots_studioproject ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- title\n", "- user_id\n", "- archived\n", "\n", "=== bots_twittertosonglog ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- gen_request_id\n", "- state\n", "- tweet_id\n", "- twitter_content\n", "- twitter_context\n", "- error_message\n", "- posted_tweet_id\n", "- image_url\n", "- twitter_context_json\n", "\n", "=== bots_artistnameclaim ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- artist_name\n", "- is_active\n", "- user_id\n", "\n", "=== bots_studioprojectclip ===\n", "- id\n", "- clip_id\n", "- studio_project_id\n", "\n", "=== bots_generatedclipextra ===\n", "- clip_id\n", "- created_at\n", "- updated_at\n", "- download_audio_count\n", "- download_video_count\n", "- share_count\n", "- is_public_approved\n", "- inferred_language\n", "- download_audio_wav_count\n", "- lyrics_copyright_metadata\n", "- content_rating\n", "- content_rating_tags\n", "- lyrics_copyright_v3_metadata\n", "- first_published_at\n", "- reuse_prompt_count\n", "- audio_snippets\n", "\n", "=== video_clipprimaryhook ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip\n", "- primary_hook_id\n", "\n", "=== video_hookreaction ===\n", "- id\n", "- reaction_type\n", "- hook_id\n", "- user_id\n", "- updated_at\n", "\n", "=== video_videohookstats ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- view_count\n", "- like_count\n", "- dislike_count\n", "- video_hook_id\n", "\n", "=== bots_userstats ===\n", "- id\n", "- display_name\n", "- total_clips\n", "- total_followers\n", "- total_following\n", "- first_onebox\n", "- first_custom\n", "- first_instrumental\n", "- last_login\n", "- last_create\n", "- user_id\n", "- total_clips_public\n", "- last_monthly_update\n", "- last_periodic_update\n", "- total_likes\n", "- total_likes_received\n", "- total_likes_received_till_last_month\n", "- total_plays\n", "- total_plays_received\n", "- total_plays_received_till_last_month\n", "- total_playlists\n", "- consecutive_days_active\n", "- longest_login_streak\n", "- consecutive_creation_streak\n", "- longest_creation_streak\n", "- top_creation_tags\n", "\n", "=== video_videosprite ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- s3_id\n", "- thumbnail_width\n", "- thumbnail_height\n", "- image_width\n", "- image_height\n", "- interval\n", "- total_frames\n", "- video_upload_id\n", "\n", "=== bots_clipvideo ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- clip_id\n", "- video_s3_id\n", "- is_cover\n", "- gif_s3_id\n", "- video_preview_s3_id\n", "- streaming_file_sizes\n", "\n", "=== bots_userblock ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- source\n", "- blocked_user_id\n", "- user_id\n", "\n", "=== video_hookcomment ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- content\n", "- track_time_second\n", "- is_deleted\n", "- deleted_by\n", "- num_likes\n", "- num_replies\n", "- raw_mentions\n", "- hook_id\n", "- parent_id\n", "- user_id\n", "\n", "=== video_hookcommentreaction ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- reaction_type\n", "- comment_id\n", "- user_id\n", "\n", "=== bots_videoupload ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- image_s3_id\n", "- video_s3_id\n", "- error_type\n", "- upload_type\n", "- title\n", "- duration\n", "- file_sha256\n", "- audio_upload_id\n", "- user_id\n", "- generated_description\n", "- height\n", "- width\n", "- gif_preview_s3_id\n", "- video_preview_s3_id\n", "- content_rating\n", "- content_rating_tags\n", "\n", "=== video_hookcommentreport ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- reason\n", "- comment_id\n", "- user_id\n", "\n", "=== moderation_warnings ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- user_id\n", "- reason\n", "- acknowledged\n", "- acknowledged_at\n", "- created_by\n", "- count\n", "\n", "=== moderation_entity_state_change_logs ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- entity_type\n", "- entity_id\n", "- action\n", "- reason\n", "- created_by\n", "- user_id\n", "\n", "=== bots_clipcomment ===\n", "- id\n", "- created_at\n", "- updated_at\n", "- content\n", "- track_time_second\n", "- is_deleted\n", "- deleted_by\n", "- num_likes\n", "- clip_id\n", "- parent_id\n", "- user_id\n", "- num_replies\n", "- raw_mentions\n", "\n", "=== bots_audioupload ===\n", "- id\n", "- updated_at\n", "- audio_s3_id\n", "- error_type\n", "- upload_type\n", "- created_at\n", "- clip_id\n", "- user_id\n", "- duration\n", "- has_vocal\n", "- image_s3_id\n", "- requested_clip_id\n", "- title\n", "- inferred_lyrics\n", "- inferred_description\n", "- display_tags\n" ] } ], "source": [ "from sqlalchemy import text\n", "\n", "# Get table names\n", "df = pd.read_sql_query(\n", " \"SELECT table_name FROM information_schema.tables WHERE table_schema = 'public'\",\n", " engine,\n", ")\n", "\n", "table_names = df[\"table_name\"].tolist()\n", "\n", "# Dict to store samples\n", "samples = {}\n", "\n", "# Loop through each table\n", "for table in table_names:\n", " print(f\"\\n=== {table} ===\")\n", " \n", " # Query 5 rows\n", " sample_df = pd.read_sql_query(\n", " f'SELECT * FROM \"{table}\" LIMIT 5;',\n", " engine,\n", " )\n", " \n", " # Print columns\n", " for col in sample_df.columns:\n", " print(f\"- {col}\")\n", " \n", " # Store in dict\n", " samples[table] = sample_df" ] }, { "cell_type": "code", "execution_count": 44, "id": "36b1a9ee", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
idcreated_atupdated_atplaybar_statesong_indexsong_play_timerepeat_stateaction_timesong_ids_in_queueplaylist_contextdevice_contextdevice_iddevice_typeuser_idvolumecontext_idcontext_type
0bfbdc2cb-2be0-4a04-857e-baba141b1ae92025-03-26 11:09:52.349261+00:002025-03-26 11:09:52.378415+00:00paused00.000000no-repeat2025-03-26 11:09:51.908000+00:00[]NoneNone0f76fa90-97da-41d4-8c92-16f11fdfdaa7Android752750391000850f31d-1f6c-428d-884b-af36c84c8dcasong_recommendations
142cad374-2eb3-4603-95df-ba101761db232025-03-21 17:41:16.048727+00:002025-03-21 18:05:51.830577+00:00paused098.172117no-repeat2025-03-21 18:05:51.199000+00:00[9e940068-6c26-4b7e-b333-c909b4d5e00e, e32fffc...NoneNone612c6e72-e30d-43f2-b781-7ed8f7b424acWindows75005489100defaultworkspace
2f1eeef31-e47e-41ea-85ef-644c5bdb89822025-04-07 01:59:04.342358+00:002025-04-07 01:59:16.052607+00:00paused010.259176no-repeat2025-04-07 01:59:15.931000+00:00[80ec4366-acb8-4427-864a-440ff82b1597, 90c80e3...NoneNoned1ec0a17-0aa2-4f35-9726-a7dfe49d4d2fiOS80068789100defaultworkspace
3646ac3a6-23ac-46bf-bcf7-ff46e68dd4a02025-03-24 09:33:19.190131+00:002025-03-24 09:33:42.187334+00:00paused00.000000no-repeat2025-03-24 09:33:38.672000+00:00[36aacc89-ae7f-44ea-b00a-bada8741178f, d39ff74...NoneNonecf06c687-4d5d-4665-b43a-c86b8c3fc60eAndroid77376330100defaultworkspace
471a2a456-347c-49bb-bbbe-255ef651f4d22025-03-21 04:57:04.985426+00:002025-03-21 04:57:14.644862+00:00paused08.324648no-repeat2025-03-21 04:57:14.881000+00:00[1375f826-1766-4f95-9e22-9a3e4b1f7cb3, 82a66bf...NoneNoneba6943ae-d81b-4ae7-9eef-907c70b679beAndroid76791404100defaultworkspace
\n", "
" ], "text/plain": [ " id created_at \\\n", "0 bfbdc2cb-2be0-4a04-857e-baba141b1ae9 2025-03-26 11:09:52.349261+00:00 \n", "1 42cad374-2eb3-4603-95df-ba101761db23 2025-03-21 17:41:16.048727+00:00 \n", "2 f1eeef31-e47e-41ea-85ef-644c5bdb8982 2025-04-07 01:59:04.342358+00:00 \n", "3 646ac3a6-23ac-46bf-bcf7-ff46e68dd4a0 2025-03-24 09:33:19.190131+00:00 \n", "4 71a2a456-347c-49bb-bbbe-255ef651f4d2 2025-03-21 04:57:04.985426+00:00 \n", "\n", " updated_at playbar_state song_index song_play_time \\\n", "0 2025-03-26 11:09:52.378415+00:00 paused 0 0.000000 \n", "1 2025-03-21 18:05:51.830577+00:00 paused 0 98.172117 \n", "2 2025-04-07 01:59:16.052607+00:00 paused 0 10.259176 \n", "3 2025-03-24 09:33:42.187334+00:00 paused 0 0.000000 \n", "4 2025-03-21 04:57:14.644862+00:00 paused 0 8.324648 \n", "\n", " repeat_state action_time \\\n", "0 no-repeat 2025-03-26 11:09:51.908000+00:00 \n", "1 no-repeat 2025-03-21 18:05:51.199000+00:00 \n", "2 no-repeat 2025-04-07 01:59:15.931000+00:00 \n", "3 no-repeat 2025-03-24 09:33:38.672000+00:00 \n", "4 no-repeat 2025-03-21 04:57:14.881000+00:00 \n", "\n", " song_ids_in_queue playlist_context \\\n", "0 [] None \n", "1 [9e940068-6c26-4b7e-b333-c909b4d5e00e, e32fffc... None \n", "2 [80ec4366-acb8-4427-864a-440ff82b1597, 90c80e3... None \n", "3 [36aacc89-ae7f-44ea-b00a-bada8741178f, d39ff74... None \n", "4 [1375f826-1766-4f95-9e22-9a3e4b1f7cb3, 82a66bf... None \n", "\n", " device_context device_id device_type user_id \\\n", "0 None 0f76fa90-97da-41d4-8c92-16f11fdfdaa7 Android 75275039 \n", "1 None 612c6e72-e30d-43f2-b781-7ed8f7b424ac Windows 75005489 \n", "2 None d1ec0a17-0aa2-4f35-9726-a7dfe49d4d2f iOS 80068789 \n", "3 None cf06c687-4d5d-4665-b43a-c86b8c3fc60e Android 77376330 \n", "4 None ba6943ae-d81b-4ae7-9eef-907c70b679be Android 76791404 \n", "\n", " volume context_id context_type \n", "0 100 0850f31d-1f6c-428d-884b-af36c84c8dca song_recommendations \n", "1 100 default workspace \n", "2 100 default workspace \n", "3 100 default workspace \n", "4 100 default workspace " ] }, "execution_count": 44, "metadata": {}, "output_type": "execute_result" } ], "source": [ "samples[\"music_player_playbarstate\"]" ] }, { "cell_type": "code", "execution_count": 24, "id": "34b25e18", "metadata": {}, "outputs": [], "source": [ "t0 = time.time()\n", "df = pd.read_sql_query(\n", " \"\"\"\n", " SELECT \n", " prompt_text,\n", " COUNT(DISTINCT id) AS num_clips\n", " FROM bots_generatedclip\n", " WHERE\n", " created_at >= '2025-06-01' AND created_at < '2025-06-02'\n", " AND is_generated = TRUE\n", " AND is_deleted = FALSE\n", " GROUP BY prompt_text\n", " ORDER BY num_clips DESC\n", " \"\"\",\n", " engine,\n", ")\n", "td = time.time() - t0" ] }, { "cell_type": "code", "execution_count": 26, "id": "753d51e8", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "981.736864566803" ] }, "execution_count": 26, "metadata": {}, "output_type": "execute_result" } ], "source": [ "td" ] }, { "cell_type": "code", "execution_count": 30, "id": "eaae8bcc", "metadata": {}, "outputs": [], "source": [ "t0 = time.time()\n", "df2 = pd.read_sql_query(\n", " \"\"\"\n", " SELECT \n", " id AS clip_id,\n", " prompt_text\n", " FROM bots_generatedclip\n", " WHERE\n", " created_at >= '2025-06-01' AND created_at < '2025-06-02'\n", " AND is_generated = TRUE\n", " AND is_deleted = FALSE\n", " \"\"\",\n", " engine,\n", ")\n", "td2 = time.time() - t0" ] }, { "cell_type": "code", "execution_count": 31, "id": "253ef63a", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "347.49016857147217" ] }, "execution_count": 31, "metadata": {}, "output_type": "execute_result" } ], "source": [ "td2" ] }, { "cell_type": "code", "execution_count": 32, "id": "c4bf3873", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "5.783333333333333" ] }, "execution_count": 32, "metadata": {}, "output_type": "execute_result" } ], "source": [ "347/60" ] }, { "cell_type": "code", "execution_count": 33, "id": "aee03904", "metadata": {}, "outputs": [], "source": [ "df2 = (\n", " df2.groupby(\"prompt_text\")\n", " .agg(num_clips=(\"clip_id\", \"nunique\"))\n", " .reset_index()\n", " .sort_values(\"num_clips\", ascending=False)\n", ")" ] }, { "cell_type": "code", "execution_count": 9, "id": "3a28aebb", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
prompt_textnum_clips
0912160
548303[Instrumental]225739
129\\n\\n1471
1177797[Verse]\\noh, my love\\nMy friend you know\\nit's...1462
530248[Fade Out]964
900832[Verse]\\nFirst verse starts here..\\n\\n[Verse 2...894
482455Wheere there is desire, there iis gonna be a f...890
122\\n889
1306192[instrumental]\\n[instrumental]\\n[instrumental]...560
145\\n\\n\\n\\n509
454801Verse 1 \\n[Instrumental - Quickly Escalate]\\n\\...504
645439[Intro]\\n[Verse 1]\\n[Pre-Chorus]\\n[Chorus 1]\\n...500
100366488
830183[Verse]\\nAyo\\nWho that man who remove the hat?...410
758539[Verse 1]\\nShe slid through the doorway with h...400
669936[Owl hoot]\\n[Intro – close-up spoken- hearbeat...388
425744This is the truth\\n\\n[Verse ]\\nI made you with...383
614756[Intro]\\n(Welcome, initiate...)\\nYou were born...352
2055091400 feet\\n1400 feet\\n1400 feet\\n1400 feet\\n14...304
561614[Intro – 0:00–1:00]\\n\\n[Slow, panoramic synth ...298
\n", "
" ], "text/plain": [ " prompt_text num_clips\n", "0 912160\n", "548303 [Instrumental] 225739\n", "129 \\n\\n 1471\n", "1177797 [Verse]\\noh, my love\\nMy friend you know\\nit's... 1462\n", "530248 [Fade Out] 964\n", "900832 [Verse]\\nFirst verse starts here..\\n\\n[Verse 2... 894\n", "482455 Wheere there is desire, there iis gonna be a f... 890\n", "122 \\n 889\n", "1306192 [instrumental]\\n[instrumental]\\n[instrumental]... 560\n", "145 \\n\\n\\n\\n 509\n", "454801 Verse 1 \\n[Instrumental - Quickly Escalate]\\n\\... 504\n", "645439 [Intro]\\n[Verse 1]\\n[Pre-Chorus]\\n[Chorus 1]\\n... 500\n", "100366 488\n", "830183 [Verse]\\nAyo\\nWho that man who remove the hat?... 410\n", "758539 [Verse 1]\\nShe slid through the doorway with h... 400\n", "669936 [Owl hoot]\\n[Intro – close-up spoken- hearbeat... 388\n", "425744 This is the truth\\n\\n[Verse ]\\nI made you with... 383\n", "614756 [Intro]\\n(Welcome, initiate...)\\nYou were born... 352\n", "205509 1400 feet\\n1400 feet\\n1400 feet\\n1400 feet\\n14... 304\n", "561614 [Intro – 0:00–1:00]\\n\\n[Slow, panoramic synth ... 298" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" } ], "source": [ "lyrics_df.head(20)" ] }, { "cell_type": "code", "execution_count": 15, "id": "b2810b5d", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "[Intro – 0:00–1:00]\n", "\n", "[Slow, panoramic synth swell. Cold, shimmering textures drift in like distant light through fog. Deep sub-bass rumbles gently under a gated pad. No vocals yet — only emotion unfolding in slow motion.]\n", "\n", "⸻\n", "\n", "[Verse 1 – 1:01–1:45]\n", "\n", "[Kick subtly joins. A warm, pulsing bassline breathes beneath the surface. Vocals enter low, calm, and intimate — surrounded by soft atmospheric delay.]\n", "\n", "Kui mina olin veel väikene mees,\n", "üks helin mul helises rinna sees.\n", "Ja kui mina sirgusin suuremaks,\n", "läks helingi rinna sees kangemaks.\n", "\n", "⸻\n", "\n", "[Drop 1 – 1:46–2:35]\n", "\n", "[Full groove unfolds. Gated, layered synths pulse rhythmically. Bass grows thicker. Fragments of vocals echo through reverb (“helin… mu hing…”) — hypnotic, trance-like.]\n", "\n", "See helin mu elu ja minu hing,\n", "tal kitsaks on jäänud maapäälne ring.\n", "See helin mu elu ja minu hing,\n", "tal kitsaks on jäänud maapäälne ring.\n", "\n", "⸻\n", "\n", "[Bridge – 2:36–3:05]\n", "\n", "[Kick and bass pull back. Strings rise subtly. Space opens. Vocals return in a near-whisper, haunted and close.]\n", "\n", "Nüüd on see helin pea matnud mind,\n", "ta alla rusuks on raugenud rind.\n", "Kui mina olin veel väikene mees,\n", "üks helin mul helises rinna sees.\n", "\n", "⸻\n", "\n", "[Drop 2 – 3:06–3:55]\n", "\n", "[Drop returns — deeper and wider. Synths modulate gently. Vocal lines grow more intense. Harmony layers expand.]\n", "\n", "See helin mu elu ja minu hing,\n", "tal kitsaks on jäänud maapäälne ring.\n", "See helin mu elu ja minu hing,\n", "tal kitsaks on jäänud maapäälne ring.\n", "\n", "⸻\n", "\n", "[Power Chorus – 3:56–4:35]\n", "\n", "[All elements bloom. Gated synths, vocal harmonies, thick sidechained bass. Emotional climax.]\n", "\n", "Ta alla rusuks on raugenud rind.\n", "See helin mu elu ja minu hing,\n", "See helin mu elu ja minu hing,\n", "tal kitsaks on jäänud maapäälne ring!\n", "\n", "⸻\n", "\n", "[Outro – 4:36–5:20]\n", "\n", "[Kick fades. Synths dissolve into reverb. Only piano and ambient textures remain. Vocals return one last time — soft, nostalgic, fading into mist.]\n", "\n", "Kui mina olin veel väikene mees,\n", "üks helin mul helises rinna sees.\n", "Ja kui mina sirgusin suuremaks,\n", "läks helingi rinna sees kangemaks.\n", "\n" ] } ], "source": [ "print(lyrics_df.loc[561614][\"prompt_text\"])" ] }, { "cell_type": "code", "execution_count": null, "id": "15b194ad", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 27, "id": "76b4dfb8", "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
clip_idprompt_text
01869069b-5f6b-4e30-bf5d-fc706cdd3d35Verso 1: \\nMe levanto en la mañana, con el so...
1b89c6a2c-428a-498c-90a8-86e9b83efe0fСьогодні випив небагато,\\nЗ друзяками посидів ...
25c5065b4-2b9b-4fba-a0c7-9c6b0e8fbd63[Verse]\\nHow miserable life would be\\nIf black...
33648441f-4fbf-4d58-a757-e9fbc714e51c[Instrumental]
4b1303ae2-cda2-4caf-a631-53dad56ceb65[Verse 1]\\nChrome rims spin, got the shine so ...
\n", "
" ], "text/plain": [ " clip_id \\\n", "0 1869069b-5f6b-4e30-bf5d-fc706cdd3d35 \n", "1 b89c6a2c-428a-498c-90a8-86e9b83efe0f \n", "2 5c5065b4-2b9b-4fba-a0c7-9c6b0e8fbd63 \n", "3 3648441f-4fbf-4d58-a757-e9fbc714e51c \n", "4 b1303ae2-cda2-4caf-a631-53dad56ceb65 \n", "\n", " prompt_text \n", "0 Verso 1: \\nMe levanto en la mañana, con el so... \n", "1 Сьогодні випив небагато,\\nЗ друзяками посидів ... \n", "2 [Verse]\\nHow miserable life would be\\nIf black... \n", "3 [Instrumental] \n", "4 [Verse 1]\\nChrome rims spin, got the shine so ... " ] }, "execution_count": 27, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df.head()" ] }, { "cell_type": "code", "execution_count": null, "id": "f0ea2ac3", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "57e40d02", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 24, "id": "aaebdc3a", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "id ddf3637a-e302-41a3-87e7-21837b35d48c\n", "created_at 2025-06-10 03:24:27.591883+00:00\n", "updated_at 2025-07-03 12:45:53.177334+00:00\n", "time_used 116.427652\n", "metadata {'tags': 'rap, rhythmic, male vocals, electron...\n", "user_id 68830935\n", "status complete\n", "discord_message_id None\n", "prompt_id None\n", "request_id a14cd2a3-c828-43b0-a378-17d3feafe6a3\n", "is_generated True\n", "s3_id ddf3637a-e302-41a3-87e7-21837b35d48c\n", "upvote_count 20\n", "batch_index 0\n", "model_name chirp-v3p5-engine-s-8\n", "prompt_text [Verse]\\nЖи да ши, пишу на битах,\\nСлово остро...\n", "daily_theme_id None\n", "is_deleted False\n", "image_s3_id image_video_upload_3829aacb-be43-4796-82eb-284...\n", "is_public True\n", "dislike_count 0\n", "flag_count 0\n", "play_count 135\n", "skip_count 0\n", "title Жи-ши пиши\n", "slug None\n", "creation_source None\n", "allow_comments True\n", "is_hidden False\n", "display_tags None\n", "Name: 0, dtype: object" ] }, "execution_count": 24, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df.iloc[0]" ] }, { "cell_type": "code", "execution_count": 20, "id": "7355ca06", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "'\\n [Verse1]\\n 绿叶轻摇，岁月静好\\n 风穿过窗，拂过陶盆的沧桑\\n 生命交织，线与枝的舞蹈\\n 坚韧与枯荣，一刹那就是永恒\\n\\n [Verse2]\\n 鲜绿渐隐，棕黄交织\\n 叶尖轻卷，干枯中藏着秘密\\n 苔藓如茵，湿漉漉的生机\\n 悄然守护，这一角的静谧\\n\\n [Chorus]\\n 晨光微曦，电线交织的乐章\\n 谱写着，平凡与非凡的对话\\n 岁月叶语，静听心灵的呼吸\\n 每片绿意，是对生活的执着与向往\\n \\n\\n\\n[End]'" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df.iloc[0][\"prompt_text\"]" ] }, { "cell_type": "code", "execution_count": null, "id": "2bb9b6ed", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "4a00ca40", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.10.14" } }, "nbformat": 4, "nbformat_minor": 5 }