| 1 | CREATE TABLE research_sources ( |
| 2 | source_id VARCHAR(120) PRIMARY KEY, |
| 3 | source_type VARCHAR(60) NOT NULL, |
| 4 | source_name VARCHAR(240) NOT NULL, |
| 5 | reliability_level VARCHAR(20) NOT NULL, |
| 6 | fetch_strategy VARCHAR(40) NOT NULL, |
| 7 | automatic_access VARCHAR(40) NOT NULL, |
| 8 | javascript_required BOOLEAN NOT NULL DEFAULT FALSE, |
| 9 | requests_per_minute INTEGER NOT NULL, |
| 10 | min_delay_seconds DECIMAL(12, 3) NOT NULL |
| 11 | ); |
| 12 | |
| 13 | CREATE TABLE company_research_profiles ( |
| 14 | company_id UUID PRIMARY KEY, |
| 15 | instrument_id UUID NOT NULL UNIQUE, |
| 16 | company_name VARCHAR(240) NOT NULL, |
| 17 | isin VARCHAR(20), |
| 18 | ticker VARCHAR(40) NOT NULL, |
| 19 | exchange VARCHAR(40) NOT NULL, |
| 20 | mic VARCHAR(12), |
| 21 | country VARCHAR(2) NOT NULL, |
| 22 | currency VARCHAR(3) NOT NULL, |
| 23 | official_website VARCHAR(600), |
| 24 | investor_relations_url VARCHAR(600), |
| 25 | press_release_url VARCHAR(600), |
| 26 | annual_reports_url VARCHAR(600), |
| 27 | exchange_announcements_url VARCHAR(600), |
| 28 | regulatory_filings_url VARCHAR(600) |
| 29 | ); |
| 30 | |
| 31 | CREATE TABLE research_documents ( |
| 32 | document_id UUID PRIMARY KEY, |
| 33 | canonical_url VARCHAR(1000) NOT NULL, |
| 34 | original_url VARCHAR(1000) NOT NULL, |
| 35 | title VARCHAR(500), |
| 36 | source_type VARCHAR(60) NOT NULL, |
| 37 | source_name VARCHAR(240) NOT NULL, |
| 38 | publisher VARCHAR(240), |
| 39 | published_at TIMESTAMP, |
| 40 | retrieved_at TIMESTAMP NOT NULL, |
| 41 | language VARCHAR(16), |
| 42 | content_type VARCHAR(120) NOT NULL, |
| 43 | document_type VARCHAR(40) NOT NULL, |
| 44 | content_hash CHAR(64) NOT NULL, |
| 45 | instrument_id UUID, |
| 46 | company_id UUID, |
| 47 | country VARCHAR(2), |
| 48 | exchange VARCHAR(40), |
| 49 | status VARCHAR(40) NOT NULL, |
| 50 | reliability_level VARCHAR(20) NOT NULL, |
| 51 | entity_resolution_confidence DECIMAL(5, 4) NOT NULL |
| 52 | ); |
| 53 | |
| 54 | CREATE TABLE research_document_sources ( |
| 55 | document_source_id UUID PRIMARY KEY, |
| 56 | document_id UUID NOT NULL, |
| 57 | source_id VARCHAR(120), |
| 58 | source_url VARCHAR(1000) NOT NULL, |
| 59 | discovered_at TIMESTAMP NOT NULL, |
| 60 | CONSTRAINT fk_research_document_sources_document FOREIGN KEY (document_id) REFERENCES research_documents (document_id) |
| 61 | ); |
| 62 | |
| 63 | CREATE TABLE research_events ( |
| 64 | event_id UUID PRIMARY KEY, |
| 65 | instrument_id UUID NOT NULL, |
| 66 | company_id UUID NOT NULL, |
| 67 | event_type VARCHAR(80) NOT NULL, |
| 68 | event_date TIMESTAMP, |
| 69 | detected_at TIMESTAMP NOT NULL, |
| 70 | title VARCHAR(300) NOT NULL, |
| 71 | summary VARCHAR(1200) NOT NULL, |
| 72 | source_document_id UUID NOT NULL, |
| 73 | source_url VARCHAR(1000) NOT NULL, |
| 74 | source_type VARCHAR(60) NOT NULL, |
| 75 | reliability VARCHAR(20) NOT NULL, |
| 76 | confidence DECIMAL(5, 4) NOT NULL, |
| 77 | impact VARCHAR(40) NOT NULL, |
| 78 | time_horizon VARCHAR(40) NOT NULL, |
| 79 | currency VARCHAR(3), |
| 80 | monetary_value DECIMAL(28, 4), |
| 81 | monetary_original VARCHAR(80), |
| 82 | percentage_value DECIMAL(12, 4), |
| 83 | percentage_original VARCHAR(80), |
| 84 | customer VARCHAR(240), |
| 85 | counterparty VARCHAR(240), |
| 86 | location VARCHAR(240), |
| 87 | capacity_value DECIMAL(28, 4), |
| 88 | capacity_unit VARCHAR(40), |
| 89 | status VARCHAR(40) NOT NULL, |
| 90 | CONSTRAINT fk_research_events_document FOREIGN KEY (source_document_id) REFERENCES research_documents (document_id) |
| 91 | ); |
| 92 | |
| 93 | CREATE TABLE research_event_evidence ( |
| 94 | evidence_id UUID PRIMARY KEY, |
| 95 | event_id UUID NOT NULL, |
| 96 | source_document_id UUID NOT NULL, |
| 97 | evidence_reference VARCHAR(600) NOT NULL, |
| 98 | CONSTRAINT fk_research_event_evidence_event FOREIGN KEY (event_id) REFERENCES research_events (event_id) |
| 99 | ); |
| 100 | |
| 101 | CREATE TABLE research_fetch_history ( |
| 102 | fetch_id UUID PRIMARY KEY, |
| 103 | source_id VARCHAR(120), |
| 104 | url VARCHAR(1000) NOT NULL, |
| 105 | attempted_at TIMESTAMP NOT NULL, |
| 106 | status VARCHAR(40) NOT NULL, |
| 107 | http_status INTEGER, |
| 108 | failure_code VARCHAR(120), |
| 109 | bytes_read BIGINT, |
| 110 | latency_ms BIGINT |
| 111 | ); |
| 112 | |
| 113 | CREATE INDEX idx_research_documents_instrument_id ON research_documents (instrument_id); |
| 114 | CREATE INDEX idx_research_documents_company_id ON research_documents (company_id); |
| 115 | CREATE INDEX idx_research_documents_published_at ON research_documents (published_at); |
| 116 | CREATE UNIQUE INDEX idx_research_documents_canonical_url ON research_documents (canonical_url); |
| 117 | CREATE INDEX idx_research_documents_content_hash ON research_documents (content_hash); |
| 118 | CREATE INDEX idx_research_documents_source_type ON research_documents (source_type); |
| 119 | CREATE INDEX idx_research_documents_status ON research_documents (status); |
| 120 | CREATE INDEX idx_research_events_instrument_id ON research_events (instrument_id); |
| 121 | CREATE INDEX idx_research_events_company_id ON research_events (company_id); |
| 122 | CREATE INDEX idx_research_events_event_type ON research_events (event_type); |
| 123 | CREATE INDEX idx_research_events_event_date ON research_events (event_date); |
| 124 | CREATE INDEX idx_research_events_impact ON research_events (impact); |