{ "extracted_sections": [ [ "section_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "backend", "VARCHAR" ], [ "n_mentions", "INTEGER" ], [ "extracted_at", "TIMESTAMP" ] ], "project_mentions": [ [ "mention_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "cik", "VARCHAR" ], [ "section_id", "VARCHAR" ], [ "section_name", "VARCHAR" ], [ "project_name", "VARCHAR" ], [ "project_type", "VARCHAR" ], [ "description", "VARCHAR" ], [ "objective", "VARCHAR" ], [ "amount_usd", "DOUBLE" ], [ "amount_raw", "VARCHAR" ], [ "locations", "VARCHAR[]" ], [ "technologies", "VARCHAR[]" ], [ "partners", "VARCHAR[]" ], [ "suppliers", "VARCHAR[]" ], [ "benefits", "VARCHAR[]" ], [ "risks", "VARCHAR[]" ], [ "years", "INTEGER[]" ], [ "status", "VARCHAR" ], [ "confidence", "DOUBLE" ], [ "backend", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "form_type", "VARCHAR" ], [ "filing_date", "DATE" ], [ "fiscal_year", "INTEGER" ], [ "sector", "VARCHAR" ] ], "projects": [ [ "project_id", "VARCHAR" ], [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "sector", "VARCHAR" ], [ "project_type", "VARCHAR" ], [ "project_name", "VARCHAR" ], [ "canonical_location", "VARCHAR" ], [ "technologies", "VARCHAR[]" ], [ "status", "VARCHAR" ], [ "total_amount_usd", "DOUBLE" ], [ "n_mentions", "INTEGER" ], [ "n_filings", "INTEGER" ], [ "first_seen", "DATE" ], [ "last_seen", "DATE" ], [ "first_year", "INTEGER" ], [ "last_year", "INTEGER" ], [ "avg_confidence", "DOUBLE" ], [ "description", "VARCHAR" ] ], "project_timeline": [ [ "project_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "filing_date", "DATE" ], [ "form_type", "VARCHAR" ], [ "status", "VARCHAR" ], [ "amount_usd", "DOUBLE" ], [ "confidence", "DOUBLE" ], [ "snippet", "VARCHAR" ] ], "spid_sections": [ [ "section_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "section_name", "VARCHAR" ], [ "item_number", "VARCHAR" ], [ "text", "VARCHAR" ], [ "word_count", "INTEGER" ] ], "v_sections": [ [ "section_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "section_name", "VARCHAR" ], [ "item_number", "VARCHAR" ], [ "text", "VARCHAR" ], [ "word_count", "INTEGER" ], [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "filing_company", "VARCHAR" ], [ "form_type", "VARCHAR" ], [ "filing_date", "DATE" ], [ "fiscal_year", "INTEGER" ], [ "company_name", "VARCHAR" ], [ "sector", "VARCHAR" ], [ "sub_industry", "VARCHAR" ], [ "sic_code", "VARCHAR" ], [ "state", "VARCHAR" ] ], "src_filings": [ [ "accession_number", "VARCHAR" ], [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "form_type", "VARCHAR" ], [ "filing_date", "DATE" ], [ "report_date", "DATE" ], [ "fiscal_year", "INTEGER" ], [ "fiscal_quarter", "INTEGER" ], [ "sec_url", "VARCHAR" ], [ "local_path", "VARCHAR" ], [ "primary_document", "VARCHAR" ], [ "file_size_bytes", "BIGINT" ], [ "download_status", "VARCHAR" ], [ "parse_status", "VARCHAR" ], [ "created_at", "TIMESTAMP" ], [ "updated_at", "TIMESTAMP" ] ], "src_sections": [ [ "section_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "section_name", "VARCHAR" ], [ "item_number", "VARCHAR" ], [ "section_order", "INTEGER" ], [ "text", "VARCHAR" ], [ "word_count", "INTEGER" ], [ "char_count", "INTEGER" ], [ "created_at", "TIMESTAMP" ] ], "src_companies": [ [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "sector", "VARCHAR" ], [ "sub_industry", "VARCHAR" ], [ "sic_code", "VARCHAR" ], [ "state", "VARCHAR" ], [ "date_added", "DATE" ], [ "created_at", "TIMESTAMP" ], [ "updated_at", "TIMESTAMP" ], [ "in_sp500", "BOOLEAN" ], [ "in_russell3000", "BOOLEAN" ], [ "sic_description", "VARCHAR" ], [ "state_of_incorporation", "VARCHAR" ], [ "exchange", "VARCHAR" ], [ "fiscal_year_end", "VARCHAR" ], [ "russell_weight_pct", "DOUBLE" ], [ "shard_node", "VARCHAR" ] ], "views": [ [ "src_companies", "CREATE VIEW src_companies AS SELECT * FROM read_parquet('/Users/simon-pierreboucher/UQO_RESEARCH_DB/data/parquet/companies/companies.parquet');" ], [ "src_filings", "CREATE VIEW src_filings AS SELECT * FROM read_parquet('/Users/simon-pierreboucher/UQO_RESEARCH_DB/data/parquet/filings/filings.parquet');" ], [ "src_sections", "CREATE VIEW src_sections AS SELECT * FROM read_parquet('/Users/simon-pierreboucher/UQO_RESEARCH_DB/data/parquet/filing_sections/filing_sections.parquet');" ], [ "v_sections", "CREATE VIEW v_sections AS WITH parsed AS (SELECT s.section_id, s.accession_number, s.section_name, s.item_number, s.\"text\", s.word_count, f.form_type FROM src_sections AS s INNER JOIN src_filings AS f USING (accession_number) WHERE (NOT ((f.form_type = '10-K') AND (s.accession_number = ANY(SELECT DISTINCT accession_number FROM spid_sections))))), all_sec AS ((SELECT section_id, accession_number, section_name, item_number, \"text\", word_count FROM parsed WHERE (((form_type = '10-K') AND (section_name IN ('business', 'risk_factors', 'properties', 'md_and_a', 'quantitative_disclosures'))) OR (NOT (form_type IN ('10-K'))))) UNION ALL (SELECT section_id, accession_number, section_name, item_number, \"text\", word_count FROM spid_sections)) SELECT s.section_id, s.accession_number, s.section_name, s.item_number, s.\"text\", s.word_count, f.cik, f.ticker, f.company_name AS filing_company, f.form_type, CAST(f.filing_date AS DATE) AS filing_date, f.fiscal_year, co.company_name AS company_name, co.sector, co.sub_industry, co.sic_code, co.state FROM all_sec AS s INNER JOIN src_filings AS f USING (accession_number) LEFT JOIN src_companies AS co ON ((f.cik = co.cik));" ] ], "counts": { "extracted_sections": 1123098, "project_mentions": 308065, "projects": 121099, "project_timeline": 308065, "spid_sections": 27164 }, "ledger_words": [ 1123098, 3605932059 ], "parquet_files": [ 1342283, 29.13085581 ], "edgar_filings": [ [ "accession_number", "VARCHAR" ], [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "form_type", "VARCHAR" ], [ "filing_date", "DATE" ], [ "report_date", "DATE" ], [ "fiscal_year", "INTEGER" ], [ "fiscal_quarter", "INTEGER" ], [ "sec_url", "VARCHAR" ], [ "local_path", "VARCHAR" ], [ "primary_document", "VARCHAR" ], [ "file_size_bytes", "BIGINT" ], [ "download_status", "VARCHAR" ], [ "parse_status", "VARCHAR" ], [ "created_at", "TIMESTAMP" ], [ "updated_at", "TIMESTAMP" ] ], "edgar_filing_sections": [ [ "section_id", "VARCHAR" ], [ "accession_number", "VARCHAR" ], [ "section_name", "VARCHAR" ], [ "item_number", "VARCHAR" ], [ "section_order", "INTEGER" ], [ "text", "VARCHAR" ], [ "word_count", "INTEGER" ], [ "char_count", "INTEGER" ], [ "created_at", "TIMESTAMP" ] ], "edgar_companies": [ [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "company_name", "VARCHAR" ], [ "sector", "VARCHAR" ], [ "sub_industry", "VARCHAR" ], [ "sic_code", "VARCHAR" ], [ "state", "VARCHAR" ], [ "date_added", "DATE" ], [ "created_at", "TIMESTAMP" ], [ "updated_at", "TIMESTAMP" ], [ "in_sp500", "BOOLEAN" ], [ "in_russell3000", "BOOLEAN" ], [ "sic_description", "VARCHAR" ], [ "state_of_incorporation", "VARCHAR" ], [ "exchange", "VARCHAR" ], [ "fiscal_year_end", "VARCHAR" ], [ "russell_weight_pct", "DOUBLE" ], [ "shard_node", "VARCHAR" ] ], "edgar_annual_panel": [ [ "cik", "VARCHAR" ], [ "ticker", "VARCHAR" ], [ "fiscal_year", "INTEGER" ], [ "filing_count", "BIGINT" ], [ "total_chunks", "BIGINT" ], [ "ai_adoption_mean", "DOUBLE" ], [ "ai_adoption_max", "DOUBLE" ], [ "ai_adoption_std", "DOUBLE" ], [ "ai_adoption_top5pct", "DOUBLE" ], [ "ai_adoption_wmean", "DOUBLE" ], [ "ai_disruption_risk_mean", "DOUBLE" ], [ "ai_disruption_risk_max", "DOUBLE" ], [ "ai_disruption_risk_std", "DOUBLE" ], [ "ai_disruption_risk_top5pct", "DOUBLE" ], [ "ai_disruption_risk_wmean", "DOUBLE" ], [ "ai_infrastructure_mean", "DOUBLE" ], [ "ai_infrastructure_max", "DOUBLE" ], [ "ai_infrastructure_std", "DOUBLE" ], [ "ai_infrastructure_top5pct", "DOUBLE" ], [ "ai_infrastructure_wmean", "DOUBLE" ], [ "esg_commitment_mean", "DOUBLE" ], [ "esg_commitment_max", "DOUBLE" ], [ "esg_commitment_std", "DOUBLE" ], [ "esg_commitment_top5pct", "DOUBLE" ], [ "esg_commitment_wmean", "DOUBLE" ], [ "climate_risk_mean", "DOUBLE" ], [ "climate_risk_max", "DOUBLE" ], [ "climate_risk_std", "DOUBLE" ], [ "climate_risk_top5pct", "DOUBLE" ], [ "climate_risk_wmean", "DOUBLE" ], [ "cybersecurity_mean", "DOUBLE" ], [ "cybersecurity_max", "DOUBLE" ], [ "cybersecurity_std", "DOUBLE" ], [ "cybersecurity_top5pct", "DOUBLE" ], [ "cybersecurity_wmean", "DOUBLE" ], [ "automation_labor_mean", "DOUBLE" ], [ "automation_labor_max", "DOUBLE" ], [ "automation_labor_std", "DOUBLE" ], [ "automation_labor_top5pct", "DOUBLE" ], [ "automation_labor_wmean", "DOUBLE" ], [ "macro_risk_mean", "DOUBLE" ], [ "macro_risk_max", "DOUBLE" ], [ "macro_risk_std", "DOUBLE" ], [ "macro_risk_top5pct", "DOUBLE" ], [ "macro_risk_wmean", "DOUBLE" ] ], "edgar_tables": [ "anchor_embeddings", "annual_final", "annual_panel", "chunk_scores", "companies", "embeddings_metadata", "filings", "filing_chunks", "filing_scores", "filing_sections", "processing_log", "quarterly_final", "quarterly_panel" ], "temp_ok": true }