tools/build_sqlite.py

"""Build the public SQLite projection from graph.json using only the standard library."""
from pathlib import Path
import json, sqlite3, hashlib

root = Path(__file__).resolve().parent.parent
graph = json.loads((root / "graph.json").read_text(encoding="utf-8"))
target = root / "research.sqlite"
temporary = root / "research.sqlite.build"
if temporary.exists():
    temporary.unlink()
con = sqlite3.connect(temporary)
con.execute("PRAGMA foreign_keys=ON")
con.executescript("""
CREATE TABLE release (version TEXT PRIMARY KEY, schema_version TEXT, graph_sha256 TEXT);
CREATE TABLE problems (id TEXT PRIMARY KEY, title TEXT, question TEXT, formulation TEXT, json TEXT NOT NULL);
CREATE TABLE relations (id TEXT PRIMARY KEY, from_id TEXT REFERENCES problems(id), to_id TEXT REFERENCES problems(id), relation TEXT, epistemic_kind TEXT, json TEXT NOT NULL);
CREATE TABLE evidence (id TEXT PRIMARY KEY, statement TEXT, json TEXT NOT NULL);
CREATE TABLE documents (id TEXT PRIMARY KEY, title TEXT, language TEXT, epistemic_kind TEXT, markdown TEXT, sha256 TEXT, json TEXT NOT NULL);
CREATE TABLE document_sections (id TEXT PRIMARY KEY, document_id TEXT REFERENCES documents(id), section_order INTEGER, title TEXT, markdown TEXT, epistemic_kind TEXT, json TEXT NOT NULL);
CREATE TABLE document_links (id TEXT PRIMARY KEY, document_id TEXT REFERENCES documents(id), problem_id TEXT REFERENCES problems(id), relation TEXT, epistemic_kind TEXT, warrant TEXT);
CREATE VIRTUAL TABLE section_search USING fts5(document_id UNINDEXED, section_id UNINDEXED, title, body);
CREATE INDEX section_parent ON document_sections(document_id, section_order);
CREATE INDEX document_problem ON document_links(problem_id);
""")
j = lambda x: json.dumps(x, ensure_ascii=False, separators=(",", ":"))
con.execute("INSERT INTO release VALUES (?,?,?)", (graph["version"], graph["schema_version"], hashlib.sha256((root/"graph.json").read_bytes()).hexdigest()))
for n in graph["nodes"]:
    con.execute("INSERT INTO problems VALUES (?,?,?,?,?)", (n["id"],n["title"],n["question"],n["formulation"],j(n)))
for e in graph["edges"]:
    con.execute("INSERT INTO relations VALUES (?,?,?,?,?,?)",(e["id"],e["from"],e["to"],e["relation"],e["epistemic_kind"],j(e)))
for e in graph["evidence"]:
    con.execute("INSERT INTO evidence VALUES (?,?,?)",(e["id"],e["statement"],j(e)))
for d in graph["documents"]:
    con.execute("INSERT INTO documents VALUES (?,?,?,?,?,?,?)",(d["id"],d["title"],d["language"],d["epistemic_kind"],d["markdown"],d["sha256"],j(d)))
for s in graph["document_sections"]:
    con.execute("INSERT INTO document_sections VALUES (?,?,?,?,?,?,?)",(s["id"],s["document"],s["order"],s["title"],s["markdown"],s["epistemic_kind"],j(s)))
    con.execute("INSERT INTO section_search VALUES (?,?,?,?)",(s["document"],s["id"],s["title"],s["markdown"]))
for l in graph["document_links"]:
    con.execute("INSERT INTO document_links VALUES (?,?,?,?,?,?)",(l["id"],l["document"],l["node"],l["relation"],l["epistemic_kind"],l["warrant"]))
assert con.execute("PRAGMA integrity_check").fetchone()[0] == "ok"
assert not con.execute("PRAGMA foreign_key_check").fetchall()
assert con.execute("SELECT count(*) FROM section_search WHERE section_search MATCH ?",("plasticity",)).fetchone()[0] > 0
con.commit()
con.execute("VACUUM")
con.close()
temporary.replace(target)
print(json.dumps({"database":target.name,"version":graph["version"],"documents":len(graph["documents"]),"sections":len(graph["document_sections"]),"integrity":"ok","foreign_keys":"ok","full_text_search":"ok"}))