@tricoteuses/senat/lib/src/scripts/retrieve_open_data.js

Handle French Sénat's open data

import assert from "assert";
import { execSync } from "child_process";
import commandLineArgs from "command-line-args";
import fs from "fs-extra";
import path from "path";
import StreamZip from "node-stream-zip";
import readline from "readline";
import * as windows1252 from "windows-1252";
import { pipeline } from "stream";
import { promisify } from "util";
import config from "../config";
import { getChosenDatasets, getEnabledDatasets } from "../datasets";
import { commonOptions } from "./shared/cli_helpers";
const badWindows1252CharacterRegex = /[\u0080-\u009f]/g;
const optionsDefinitions = [
    ...commonOptions,
    {
        alias: "a",
        help: "all options: fetch, unzip, repair-encoding, import",
        name: "all",
        type: Boolean,
    },
    {
        alias: "c",
        help: "create TypeScript interfaces from databases schemas into src/raw_types_* directories",
        name: "schema",
        type: Boolean,
    },
    {
        alias: "e",
        help: "repair Windows CP 1252 encoding of SQL dumps",
        name: "repairEncoding",
        type: Boolean,
    },
    {
        alias: "f",
        help: "fetch datasets instead of retrieving them from files",
        name: "fetch",
        type: Boolean,
    },
    {
        alias: "i",
        help: "import SQL dumps into a freshly (re-)created database",
        name: "import",
        type: Boolean,
    },
    {
        alias: "S",
        help: "sudo psql commands with given user",
        name: "sudo",
        type: String,
    },
    {
        alias: "z",
        help: "unzip SQL files",
        name: "unzip",
        type: Boolean,
    },
];
const options = commandLineArgs(optionsDefinitions);
const streamPipeline = promisify(pipeline);
async function downloadFile(url, dest) {
    const response = await fetch(url);
    if (!response.ok) {
        throw new Error(`Download failed ${response.status} ${response.statusText} for ${url}`);
    }
    await streamPipeline(response.body, fs.createWriteStream(dest));
}
/**
 * Copy a dataset database to the main Senat database (overwriting its contents).
 */
async function copyToSenat(dataset, dataDir, options) {
    if (!options["silent"]) {
        console.log(`Copying ${dataset.database} to Senat database...`);
    }
    const sqlFilePath = path.join(dataDir, `${dataset.database}.sql`);
    const schemaDumpFile = path.join(dataDir, `${dataset.database}_schema_dump.sql`);
    // Write the header and then stream the rest of the SQL file
    const schemaSqlWriter = fs.createWriteStream(schemaDumpFile, { encoding: "utf8" });
    // Add CREATE SCHEMA statement at the top
    schemaSqlWriter.write(`DROP SCHEMA IF EXISTS ${dataset.database} CASCADE;\n`);
    schemaSqlWriter.write(`CREATE SCHEMA IF NOT EXISTS ${dataset.database};\n`);
    schemaSqlWriter.write(`GRANT USAGE ON SCHEMA ${dataset.database} TO ${config.db.user};\n`);
    schemaSqlWriter.write(`GRANT SELECT ON ALL TABLES IN SCHEMA ${dataset.database} TO ${config.db.user};\n`);
    schemaSqlWriter.write(`ALTER DEFAULT PRIVILEGES IN SCHEMA ${dataset.database} GRANT SELECT ON TABLES TO ${config.db.user};\n`);
    const lineReader = readline.createInterface({
        input: fs.createReadStream(sqlFilePath, { encoding: "utf8" }),
        crlfDelay: Infinity,
    });
    for await (const line of lineReader) {
        let newLine = line;
        // Replace 'public' schema outside single-quoted strings
        function replacePublicOutsideStrings(line, schema) {
            const parts = line.split(/(')/);
            let inString = false;
            for (let i = 0; i < parts.length; i++) {
                if (parts[i] === "'") {
                    inString = !inString;
                }
                else if (!inString) {
                    // Only replace outside of strings, including before comma
                    parts[i] = parts[i].replace(/\bpublic\b(?=(\s*\.|\s*[,;]|\s|$))/g, schema);
                }
            }
            return parts.join("");
        }
        newLine = replacePublicOutsideStrings(line, dataset.database);
        // Replace SET client_encoding to UTF8
        newLine = newLine.replace(/SET client_encoding = 'LATIN1';/i, "SET client_encoding = 'UTF8';");
        schemaSqlWriter.write(newLine + "\n");
    }
    schemaSqlWriter.end();
    await new Promise((resolve, reject) => {
        schemaSqlWriter.on("finish", () => {
            try {
                execSync(`${options["sudo"] ? `sudo -u ${options["sudo"]} ` : ""}psql --quiet -d senat -f ${schemaDumpFile}`, {
                    env: process.env,
                    encoding: "utf-8",
                    stdio: ["ignore", "pipe", "pipe"],
                });
            }
            catch (error) {
                if (!options["silent"]) {
                    console.error(`Failed to import ${dataset.database} schema:`);
                    if (error.stderr) {
                        console.error(error.stderr);
                    }
                    if (error.stdout) {
                        console.error(error.stdout);
                    }
                }
            }
            resolve();
        });
        schemaSqlWriter.on("error", reject);
    });
}
async function retrieveDataset(dataDir, dataset) {
    const zipFilename = dataset.url.substring(dataset.url.lastIndexOf("/") + 1);
    const zipFilePath = path.join(dataDir, zipFilename);
    if (options["all"] || options["fetch"]) {
        // Fetch & save ZIP file.
        if (!options["silent"]) {
            console.log(`Loading ${dataset.title}: ${zipFilename}…`);
        }
        // Fetch fails with OpenSSL error: dh key too small.
        // (so does "curl").
        // const response = await fetch(dataset.url)
        // if (!response.ok) {
        //   console.error(response.status, response.statusText)
        //   console.error(await response.text())
        //   throw new Error(`Fetch failed: ${dataset.url}`)
        // }
        // await pipeline(response.body!, fs.createWriteStream(zipFilePath))
        fs.removeSync(zipFilePath);
        await downloadFile(dataset.url, zipFilePath);
    }
    const sqlFilename = `${dataset.database}.sql`;
    const sqlFilePath = path.join(dataDir, sqlFilename);
    if (options["all"] || options["unzip"]) {
        if (!options["silent"]) {
            console.log(`Unzipping ${dataset.title}: ${zipFilename}…`);
        }
        fs.removeSync(sqlFilePath);
        const zip = new StreamZip({
            file: zipFilePath,
            storeEntries: true,
        });
        await new Promise((resolve, reject) => {
            zip.on("ready", () => {
                zip.extract(null, dataDir, (err, _count) => {
                    zip.close();
                    if (err) {
                        reject(err);
                    }
                    else {
                        resolve(null);
                    }
                });
            });
        });
        if (dataset.repairZip !== undefined) {
            if (!options["silent"]) {
                console.log(`Repairing Zip path ${dataset.title}: ${sqlFilename}…`);
            }
            dataset.repairZip(dataset, dataDir);
        }
    }
    if ((options["all"] || options["repairEncoding"]) && dataset.repairEncoding) {
        if (!options["silent"]) {
            console.log(`Repairing Windows CP1252 encoding in ${dataset.title}: ${sqlFilename}…`);
        }
        const repairedSqlFilePath = sqlFilePath + ".repaired";
        const repairedSqlWriter = fs.createWriteStream(repairedSqlFilePath, {
            encoding: "utf8",
        });
        // Read the file as latin1 (ISO-8859-1/CP1252) and write as UTF-8
        const lineReader = readline.createInterface({
            input: fs.createReadStream(sqlFilePath, { encoding: "latin1" }),
            crlfDelay: Infinity,
        });
        for await (const line of lineReader) {
            // Optionally repair Windows-1252 control characters
            let repairedLine = line.replace(badWindows1252CharacterRegex, (match) => windows1252.decode(match, { mode: "fatal" }));
            repairedSqlWriter.write(repairedLine + "\n");
        }
        repairedSqlWriter.end();
        await fs.move(repairedSqlFilePath, sqlFilePath, { overwrite: true });
    }
    if (options["all"] || options["import"] || options["schema"]) {
        if (!options["silent"]) {
            console.log(`Importing ${dataset.title}: ${sqlFilename}…`);
        }
        await copyToSenat(dataset, dataDir, options);
        // Create indexes programmatically after import
        if (dataset.indexes) {
            for (const [table, indexes] of Object.entries(dataset.indexes)) {
                for (const index of indexes) {
                    const indexName = index.name;
                    const columns = index.columns.join(", ");
                    const schema = dataset.database;
                    const sql = `CREATE INDEX IF NOT EXISTS ${indexName} ON ${schema}.${table} (${columns});`;
                    try {
                        execSync(`${options["sudo"] ? `sudo -u ${options["sudo"]} ` : ""}psql --quiet -d senat -c "${sql}"`, {
                            env: process.env,
                            encoding: "utf-8",
                            stdio: ["ignore", "ignore", "pipe"],
                        });
                        if (!options["silent"]) {
                            console.log(`Created index: ${indexName} on ${schema}.${table} (${columns})`);
                        }
                    }
                    catch (err) {
                        console.error(`Failed to create index ${indexName} on ${schema}.${table}:`, err);
                    }
                }
            }
        }
    }
    if (options["schema"]) {
        let definitionsDir = path.resolve("src", "raw_types_schemats");
        assert(fs.statSync(definitionsDir).isDirectory());
        if (!options["silent"]) {
            console.log(`Creating TypeScript definitions from schema '${dataset.database}' in database 'senat'…`);
        }
        const dbConnectionString = `postgres://${process.env["PGUSER"]}:${process.env["PGPASSWORD"]}@${process.env["PGHOST"]}:${process.env["PGPORT"]}/senat`;
        let definitionFilePath = path.join(definitionsDir, `${dataset.database}.ts`);
        execSync(`npx schemats generate -c ${dbConnectionString} -s ${dataset.database} -o ${definitionFilePath}`, {
            // cwd: dataDir,
            env: process.env,
            encoding: "utf-8",
            // stdio: ["ignore", "ignore", "pipe"],
        });
        const definition = fs.readFileSync(definitionFilePath, { encoding: "utf8" });
        const definitionRepaired = definition
            .replace(/\r\n/g, "\n")
            .replace(/AUTO-GENERATED FILE @ \d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}/, "AUTO-GENERATED FILE");
        fs.writeFileSync(definitionFilePath, definitionRepaired);
        definitionsDir = path.resolve("src", "raw_types");
        definitionFilePath = path.join(definitionsDir, `${dataset.database}.ts`);
        execSync(`npx kysely-codegen --url '${dbConnectionString}' --default-schema ${dataset.database} --include-pattern '${dataset.database}.*' --out-file ${definitionFilePath}`, {
            env: process.env,
            encoding: "utf-8",
            // stdio: ["ignore", "ignore", "pipe"],
        });
    }
}
async function retrieveOpenData() {
    const dataDir = options["dataDir"];
    assert(dataDir, "Missing argument: data directory");
    process.env = {
        ...process.env,
        PGHOST: process.env["PGHOST"] || config.db.host,
        PGPORT: process.env["PGPORT"] || String(config.db.port),
        PGDATABASE: process.env["PGDATABASE"] || config.db.name,
        PGUSER: process.env["PGUSER"] || config.db.user,
        PGPASSWORD: process.env["PGPASSWORD"] || config.db.password,
    };
    assert(process.env["PGHOST"] && process.env["PGPORT"] && process.env["PGUSER"] && process.env["PGPASSWORD"], "Missing database configuration: environment variables PGHOST, PGPORT, PGUSER and PGPASSWORD or TRICOTEUSES_SENAT_DB_* in .env file");
    console.time("data extraction time");
    // Create role 'opendata' if it does not exist
    execSync(`${options["sudo"] ? `sudo -u ${options["sudo"]} ` : ""}psql --quiet -c "CREATE ROLE opendata" || true`, {
        cwd: dataDir,
        env: process.env,
        encoding: "utf-8",
    });
    execSync(`${options["sudo"] ? `sudo -u ${options["sudo"]} ` : ""}psql --quiet -c "CREATE DATABASE senat WITH OWNER opendata" || true`, {
        cwd: dataDir,
        env: process.env,
        encoding: "utf-8",
    });
    const enabledDatasets = getEnabledDatasets(options["categories"]);
    const chosenDatasets = getChosenDatasets(enabledDatasets);
    for (const dataset of chosenDatasets) {
        await retrieveDataset(dataDir, dataset);
    }
    if (options["schema"]) {
        const dbConnectionString = `postgres://${process.env["PGUSER"]}:${process.env["PGPASSWORD"]}@${process.env["PGHOST"]}:${process.env["PGPORT"]}/senat`;
        const definitionsDir = path.resolve("src", "raw_types");
        const definitionFilePath = path.join(definitionsDir, "senat.ts");
        execSync(`npx kysely-codegen --url '${dbConnectionString}' --out-file ${definitionFilePath}`, {
            env: process.env,
            encoding: "utf-8",
            // stdio: ["ignore", "ignore", "pipe"],
        });
    }
    if (!options["silent"]) {
        console.timeEnd("data extraction time");
    }
}
retrieveOpenData()
    .then(() => process.exit(0))
    .catch((error) => {
    console.log(error);
    process.exit(1);
});