icRdZddlZddlZddlZddlZddlZddlZddlZddlZddl m Z m Z m Z ddl m Z ddlmZmZmZmZmZej(j+ddej,ej.dd ej0d Zej4j7d d Zd Ze dZdZdZ dZ!dZ"dee#de$fdZ%dee#efde&fdZ'd5dee#dee#fdZ(de#deee#effdZ)de#deee#effdZ* d6de#de#de+de,deee#eff d Z-de#d!e#d"ee#effd#Z.d7d$e#d%e,dee#fd&Z/d'ee#deee0fd(Z1d8d$e#d)e,dee0fd*Z2de#d+e,de,fd,Z3dee#effd-Z4 d9dee#d.e&d/e&d0e&dee#eff d1Z5dee#dee#effd2Z6d3Z7e8d4k(re7yy):a YouTube Transcript Extractor ============================= Extracts transcripts for videos identified by the watch history fetcher. Two-tier extraction: 1. PRIMARY: youtube-transcript-api (free, no API key needed) 2. FALLBACK: Supadata.ai API (paid, handles edge cases, AI-generated transcripts) Stores results in: - PostgreSQL: yt_transcripts table (structured storage) - Qdrant: Semantic vector embeddings for similarity search - Local JSON: backup copies for offline access Usage: # Extract transcripts for specific video IDs python youtube_transcript_extractor.py --video-ids dQw4w9WgXcQ abc123 xyz789 # Extract all videos from today's watch history (reads from PostgreSQL) python youtube_transcript_extractor.py --today # Extract from a date range python youtube_transcript_extractor.py --date 2026-02-15 # Dry run (don't store, just print) python youtube_transcript_extractor.py --today --dry-run Author: Genesis System Version: 1.0.0 N)datetime timedeltatimezone)Path)ListDictAnyOptionalTuplez)/mnt/e/genesis-system/data/genesis-memoryz1%(asctime)s [%(levelname)s] %(name)s: %(message)sz%Y-%m-%d %H:%M:%S)levelformatdatefmtyt_transcript_extractorSUPADATA_API_KEY#sd_4b8009caa1fd18698793e2a86117b07czhttps://api.supadata.ai/v1z./mnt/e/genesis-system/data/youtube_transcriptsyt_transcriptsztext-embedding-3-smallcVddl}ddlm}|jdi|j S)z/Get PostgreSQL connection using Elestio config.rN)PostgresConfig)psycopg2elestio_configrconnectget_connection_params)rrs =/mnt/e/genesis-system/scripts/youtube_transcript_extractor.pyget_db_connectionrJs'- 8   EnBBD EEc|j5}|jd|jdddtj dy#1swYxYw)z1Create the transcripts table if it doesn't exist.a CREATE TABLE IF NOT EXISTS yt_transcripts ( id SERIAL PRIMARY KEY, video_id VARCHAR(20) NOT NULL, transcript TEXT, language VARCHAR(10) DEFAULT 'en', word_count INTEGER, extraction_method VARCHAR(20) DEFAULT 'youtube_api', extracted_topics TEXT[], extracted_insights JSONB, created_at TIMESTAMP WITH TIME ZONE DEFAULT NOW(), UNIQUE(video_id) ); CREATE INDEX IF NOT EXISTS idx_yt_transcript_video ON yt_transcripts(video_id); CREATE INDEX IF NOT EXISTS idx_yt_transcript_date ON yt_transcripts(created_at); Nz#Transcript schema verified/created.)cursorexecutecommitloggerinfo)conncurs rensure_transcript_schemar%QsM #   & )* KK56+s "AA video_idsreturnc|s tS|j5}|jd|f|jDchc]}|d c}cdddScc}w#1swYyxYw)z/Check which video IDs already have transcripts.z