Files
gupi-osint-board/migrations/027_evidence_text_matching.sql
T

80 lines
4.3 KiB
SQL
Raw Normal View History

-- Data-defined evidence recognition for player-uploaded sources. Text extraction
-- belongs to the immutable asset; matching rules belong to a clonable board;
-- evaluations and awarded flags belong to the mutable level instance.
CREATE TABLE osint.asset_text_extractions (
id UUID PRIMARY KEY,
asset_id UUID NOT NULL REFERENCES osint.assets(id) ON DELETE CASCADE,
extractor TEXT NOT NULL,
extractor_version TEXT NOT NULL,
language TEXT NOT NULL,
status TEXT NOT NULL CHECK (status IN ('succeeded','unsupported','failed')),
extracted_text TEXT NOT NULL DEFAULT '',
error_message TEXT,
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
updated_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
UNIQUE (asset_id,extractor,extractor_version,language)
);
CREATE INDEX asset_text_extractions_asset_idx ON osint.asset_text_extractions (asset_id,created_at DESC);
CREATE TABLE osint.evidence_match_rules (
id UUID PRIMARY KEY,
board_id UUID NOT NULL REFERENCES osint.boards(id) ON DELETE CASCADE,
origin_rule_id UUID REFERENCES osint.evidence_match_rules(id) ON DELETE SET NULL,
name TEXT NOT NULL,
flag_key TEXT NOT NULL CHECK (flag_key ~ '^[a-z][a-z0-9_.-]{0,63}$'),
matcher_version TEXT NOT NULL DEFAULT 'char_trigram_v1' CHECK (matcher_version = 'char_trigram_v1'),
minimum_anchor_matches SMALLINT NOT NULL DEFAULT 1 CHECK (minimum_anchor_matches > 0),
enabled BOOLEAN NOT NULL DEFAULT TRUE,
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
updated_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
UNIQUE (board_id,id),
UNIQUE (board_id,name)
);
CREATE INDEX evidence_match_rules_board_flag_idx ON osint.evidence_match_rules (board_id,flag_key) WHERE enabled;
CREATE TABLE osint.evidence_match_anchors (
id UUID PRIMARY KEY,
rule_id UUID NOT NULL REFERENCES osint.evidence_match_rules(id) ON DELETE CASCADE,
phrase_text TEXT NOT NULL CHECK (char_length(btrim(phrase_text)) >= 12),
minimum_similarity NUMERIC(4,3) NOT NULL DEFAULT 0.720 CHECK (minimum_similarity >= 0.500 AND minimum_similarity <= 1),
sort_order SMALLINT NOT NULL DEFAULT 0 CHECK (sort_order >= 0),
UNIQUE (rule_id,sort_order)
);
CREATE TABLE osint.evidence_match_evaluations (
id UUID PRIMARY KEY,
level_id UUID NOT NULL,
board_id UUID NOT NULL,
document_exhibit_id UUID NOT NULL REFERENCES osint.document_exhibits(exhibit_id) ON DELETE CASCADE,
extraction_id UUID NOT NULL REFERENCES osint.asset_text_extractions(id) ON DELETE CASCADE,
rule_id UUID NOT NULL,
matched BOOLEAN NOT NULL,
matched_anchor_count SMALLINT NOT NULL CHECK (matched_anchor_count >= 0),
score NUMERIC(4,3) NOT NULL CHECK (score >= 0 AND score <= 1),
evaluated_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
FOREIGN KEY (level_id,board_id) REFERENCES osint.levels(id,board_id) ON DELETE CASCADE,
FOREIGN KEY (board_id,document_exhibit_id) REFERENCES osint.exhibits(board_id,id) ON DELETE CASCADE,
FOREIGN KEY (board_id,rule_id) REFERENCES osint.evidence_match_rules(board_id,id) ON DELETE CASCADE,
UNIQUE (level_id,document_exhibit_id,rule_id)
);
CREATE INDEX evidence_match_evaluations_level_idx ON osint.evidence_match_evaluations (level_id,evaluated_at DESC);
CREATE TABLE osint.evidence_match_anchor_evaluations (
evaluation_id UUID NOT NULL REFERENCES osint.evidence_match_evaluations(id) ON DELETE CASCADE,
anchor_id UUID NOT NULL REFERENCES osint.evidence_match_anchors(id) ON DELETE CASCADE,
similarity NUMERIC(4,3) NOT NULL CHECK (similarity >= 0 AND similarity <= 1),
matched BOOLEAN NOT NULL,
matched_text TEXT NOT NULL DEFAULT '',
PRIMARY KEY (evaluation_id,anchor_id)
);
ALTER TABLE osint.level_flags
ADD COLUMN awarded_by_evidence_match_id UUID REFERENCES osint.evidence_match_evaluations(id) ON DELETE SET NULL;
COMMENT ON TABLE osint.asset_text_extractions IS 'Cached OCR or direct-text extraction for one immutable uploaded asset.';
COMMENT ON TABLE osint.evidence_match_rules IS 'Clonable level-authored rule that awards a flag when enough distinctive text anchors match an uploaded source.';
COMMENT ON TABLE osint.evidence_match_anchors IS 'Alternative or cumulative reference passages used by one evidence match rule.';
COMMENT ON TABLE osint.evidence_match_evaluations IS 'Auditable result of applying one board rule to one player-uploaded document.';
COMMENT ON TABLE osint.evidence_match_anchor_evaluations IS 'Per-anchor fuzzy score and best normalized text window for an evidence evaluation.';