@hebcal/geo-sqlite 3.9.0 → 4.1.0

package/bin/build-geonames-sqlite

@@ -2,21 +2,22 @@
 
 const {buildGeonamesSqlite} = require('@hebcal/geo-sqlite');
 
-if (process.argv.length != 8) {
-  const infiles = 'countryInfo.txt cities5000.txt cities-patch.txt admin1CodesASCII.txt IL.txt';
+const argv = process.argv.slice(2);
+if (argv.length !== 7) {
+  const infiles = 'countryInfo.txt cities5000.txt cities-patch.txt admin1CodesASCII.txt IL.txt IL-alternatenames.txt';
   console.error(`Usage: build-geonames-sqlite geonames.sqlite3 ${infiles}`);
   process.exit(1);
 }
 
-const [dbFilename, countryInfo, cities, citiesPatch, admin1CodesASCII, il] = process.argv.slice(2);
-
-buildGeonamesSqlite(
-    dbFilename,
-    countryInfo,
-    cities,
-    citiesPatch,
-    admin1CodesASCII,
-    il,
-).then(() => {
+const filenames = {
+  dbFilename: argv[0],
+  countryInfotxt: argv[1],
+  cities5000txt: argv[2],
+  citiesPatch: argv[3],
+  admin1CodesASCIItxt: argv[4],
+  ILtxt: argv[5],
+  ILalternate: argv[6],
+};
+buildGeonamesSqlite(filenames).then(() => {
   console.log('Done!');
 });

package/bin/download-and-make-dbs

@@ -9,9 +9,13 @@ curl -o $TMPDIR/countryInfo.txt http://download.geonames.org/export/dump/country
 curl -o $TMPDIR/admin1CodesASCII.txt http://download.geonames.org/export/dump/admin1CodesASCII.txt
 curl -o $TMPDIR/cities5000.zip http://download.geonames.org/export/dump/cities5000.zip
 curl -o $TMPDIR/IL.zip http://download.geonames.org/export/dump/IL.zip
+curl -o $TMPDIR/alt-IL.zip http://download.geonames.org/export/dump/alternatenames/IL.zip
 
 cd $TMPDIR
 unzip cities5000.zip
+unzip alt-IL.zip
+mv IL.txt alt-IL.txt
+rm readme.txt
 unzip IL.zip
 
 cd $CURDIR
@@ -23,7 +27,8 @@ $CURDIR/node_modules/.bin/build-geonames-sqlite \
   $TMPDIR/cities5000.txt \
   "$CURDIR/node_modules/@hebcal/geo-sqlite/cities-patch.txt" \
   $TMPDIR/admin1CodesASCII.txt \
-  $TMPDIR/IL.txt
+  $TMPDIR/IL.txt \
+  $TMPDIR/alt-IL.txt
 
 rm -rf $TMPDIR
 

package/dist/index.js

@@ -1,4 +1,4 @@
-/*! @hebcal/geo-sqlite v3.9.0 */
+/*! @hebcal/geo-sqlite v4.1.0 */
 'use strict';
 
 Object.defineProperty(exports, '__esModule', { value: true });
@@ -446,15 +446,17 @@ class GeoDb {
 
 /**
  * Builds `geonames.sqlite3` from files downloaded from geonames.org
- * @param {string} dbFilename
- * @param {string} countryInfotxt
- * @param {string} cities5000txt
- * @param {string} citiesPatch
- * @param {string} admin1CodesASCIItxt
- * @param {string} ILtxt
+ * @param {any} opts
  */
 
-async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, citiesPatch, admin1CodesASCIItxt, ILtxt) {
+async function buildGeonamesSqlite(opts) {
+  const dbFilename = opts.dbFilename;
+  const countryInfotxt = opts.countryInfotxt;
+  const cities5000txt = opts.cities5000txt;
+  const citiesPatch = opts.citiesPatch;
+  const admin1CodesASCIItxt = opts.admin1CodesASCIItxt;
+  const ILtxt = opts.ILtxt;
+  const ILalternate = opts.ILalternate;
   const logger = pino__default["default"]({
     // level: argv.quiet ? 'warn' : 'info',
     transport: {
@@ -509,9 +511,16 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       gtopo30 int,
       timezone nvarchar(40),
       moddate date);`);
-  await doFile(logger, db, cities5000txt, 'geoname', 19);
-  await doFile(logger, db, citiesPatch, 'geoname', 19);
+
+  const truncateAlternateNames = a => {
+    a[3] = '';
+    return true;
+  };
+
+  await doFile(logger, db, cities5000txt, 'geoname', 19, truncateAlternateNames);
+  await doFile(logger, db, citiesPatch, 'geoname', 19, truncateAlternateNames);
   await doFile(logger, db, ILtxt, 'geoname', 19, a => {
+    a[3] = '';
     return a[6] == 'P' && (a[7] == 'PPL' || a[7] == 'STLMT');
   });
   doSql(logger, db, `DROP TABLE IF EXISTS admin1`, `CREATE TABLE admin1 (
@@ -527,8 +536,47 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       WHERE geonameid = 4140963;`, `UPDATE admin1
       SET name = 'Washington, D.C.', asciiname = 'Washington, D.C.'
       WHERE key = 'US.DC';`);
-  doSql(logger, db, `DROP TABLE IF EXISTS geoname_he`, `CREATE TABLE geoname_he AS SELECT * FROM geoname LIMIT 0`);
-  await doFile(logger, db, ILtxt, 'geoname_he', 19, filterPlacesHebrew);
+  doSql(logger, db, `DROP TABLE IF EXISTS alternatenames`, `CREATE TABLE alternatenames (
+    id int PRIMARY KEY,
+    geonameid int NOT NULL,
+    isolanguage varchar(7),
+    name varchar(400),
+    isPreferredName tinyint,
+    isShortName tinyint,
+    isColloquial tinyint,
+    isHistoric tinyint,
+    periodFrom NULL,
+    periodTo NULL
+    );`);
+  await doFile(logger, db, ILalternate, 'alternatenames', 10, a => {
+    const firstchar = a[3][0];
+
+    if (a[2] === 'he' && (firstchar < '\u05D0' || firstchar > '\u05EA')) {
+      a[2] = 'en';
+    }
+
+    if (a[2] === 'he' || a[2] === 'en') {
+      if (a[2] === 'he') {
+        a[3] = a[3].replace(/‘/g, '׳');
+        a[3] = a[3].replace(/’/g, '׳');
+        a[3] = a[3].replace(/\'/g, '׳');
+        a[3] = core.Locale.hebrewStripNikkud(a[3]);
+      } else {
+        a[3] = a[3].replace(/‘/g, '\'');
+        a[3] = a[3].replace(/’/g, '\'');
+      }
+
+      return true;
+    }
+
+    return false;
+  }); // remove duplicates from alternatenames
+
+  doSql(logger, db, `DROP TABLE IF EXISTS altnames`, `CREATE TABLE altnames
+    AS SELECT geonameid, isolanguage, name
+    FROM alternatenames
+    GROUP BY 1, 2, 3
+    `);
   doSql(logger, db, `update admin1 set name='',asciiname='' where key like 'PS.%';`, `update country set country = '' where iso = 'PS';`, `delete from geoname where geonameid = 7303419;`);
   doSql(logger, db, `DROP TABLE IF EXISTS geoname_fulltext`, `CREATE VIRTUAL TABLE geoname_fulltext
       USING fts3(geonameid int, longname text,
@@ -559,47 +607,32 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       AND g.country = c.ISO
       AND g.country||'.'||g.admin1 = a.key
       `, `INSERT INTO geoname_fulltext
-      SELECT g.geonameid, g.name||', ישראל',
-      g.name, '', 'ישראל',
+      SELECT g.geonameid, alt.name||', ישראל',
+      alt.name, '', 'ישראל',
       g.population, g.latitude, g.longitude, g.timezone
-      FROM geoname_he g, admin1 a, country c
-      WHERE g.country = c.ISO
-      AND g.country||'.'||g.admin1 = a.key
+      FROM geoname g, altnames alt
+      WHERE g.country = 'IL'
+      AND alt.isolanguage = 'he'
+      AND g.geonameid = alt.geonameid
+      `, `INSERT INTO geoname_fulltext
+      SELECT g.geonameid, alt.name||', '||a1.asciiname||', Israel',
+      alt.name, a1.asciiname, 'Israel',
+      g.population, g.latitude, g.longitude, g.timezone
+      FROM geoname g, admin1 a1, altnames alt
+      WHERE g.country = 'IL'
+      AND alt.isolanguage = 'en'
+      AND g.geonameid = alt.geonameid
+      AND g.country||'.'||g.admin1 = a1.key
       `);
   db.close();
   return Promise.resolve(true);
 }
-/**
- * @param {string[]} a
- * @return {boolean}
- */
-
-function filterPlacesHebrew(a) {
-  if (a[6] != 'P' || a[7] != 'PPL' && a[7] != 'STLMT') {
-    return false;
-  }
-
-  const alternatenames = a[3].split(',');
-
-  for (const name of alternatenames) {
-    const firstchar = name[0];
-
-    if (firstchar >= '\u05D0' && firstchar <= '\u05EA') {
-      a[1] = core.Locale.hebrewStripNikkud(name); // replace 'name' field with Hebrew
-
-      return true;
-    }
-  }
-
-  return false;
-}
 /**
  * @param {pino.Logger} logger
  * @param {Database} db
  * @param  {...string} sqls
  */
 
-
 function doSql(logger, db, ...sqls) {
   for (let i = 0; i < sqls.length; i++) {
     logger.info(sqls[i]);

package/dist/index.mjs

@@ -1,4 +1,4 @@
-/*! @hebcal/geo-sqlite v3.9.0 */
+/*! @hebcal/geo-sqlite v4.1.0 */
 import Database from 'better-sqlite3';
 import { Location, Locale } from '@hebcal/core';
 import '@hebcal/cities';
@@ -434,15 +434,17 @@ class GeoDb {
 
 /**
  * Builds `geonames.sqlite3` from files downloaded from geonames.org
- * @param {string} dbFilename
- * @param {string} countryInfotxt
- * @param {string} cities5000txt
- * @param {string} citiesPatch
- * @param {string} admin1CodesASCIItxt
- * @param {string} ILtxt
+ * @param {any} opts
  */
 
-async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, citiesPatch, admin1CodesASCIItxt, ILtxt) {
+async function buildGeonamesSqlite(opts) {
+  const dbFilename = opts.dbFilename;
+  const countryInfotxt = opts.countryInfotxt;
+  const cities5000txt = opts.cities5000txt;
+  const citiesPatch = opts.citiesPatch;
+  const admin1CodesASCIItxt = opts.admin1CodesASCIItxt;
+  const ILtxt = opts.ILtxt;
+  const ILalternate = opts.ILalternate;
   const logger = pino({
     // level: argv.quiet ? 'warn' : 'info',
     transport: {
@@ -497,9 +499,16 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       gtopo30 int,
       timezone nvarchar(40),
       moddate date);`);
-  await doFile(logger, db, cities5000txt, 'geoname', 19);
-  await doFile(logger, db, citiesPatch, 'geoname', 19);
+
+  const truncateAlternateNames = a => {
+    a[3] = '';
+    return true;
+  };
+
+  await doFile(logger, db, cities5000txt, 'geoname', 19, truncateAlternateNames);
+  await doFile(logger, db, citiesPatch, 'geoname', 19, truncateAlternateNames);
   await doFile(logger, db, ILtxt, 'geoname', 19, a => {
+    a[3] = '';
     return a[6] == 'P' && (a[7] == 'PPL' || a[7] == 'STLMT');
   });
   doSql(logger, db, `DROP TABLE IF EXISTS admin1`, `CREATE TABLE admin1 (
@@ -515,8 +524,47 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       WHERE geonameid = 4140963;`, `UPDATE admin1
       SET name = 'Washington, D.C.', asciiname = 'Washington, D.C.'
       WHERE key = 'US.DC';`);
-  doSql(logger, db, `DROP TABLE IF EXISTS geoname_he`, `CREATE TABLE geoname_he AS SELECT * FROM geoname LIMIT 0`);
-  await doFile(logger, db, ILtxt, 'geoname_he', 19, filterPlacesHebrew);
+  doSql(logger, db, `DROP TABLE IF EXISTS alternatenames`, `CREATE TABLE alternatenames (
+    id int PRIMARY KEY,
+    geonameid int NOT NULL,
+    isolanguage varchar(7),
+    name varchar(400),
+    isPreferredName tinyint,
+    isShortName tinyint,
+    isColloquial tinyint,
+    isHistoric tinyint,
+    periodFrom NULL,
+    periodTo NULL
+    );`);
+  await doFile(logger, db, ILalternate, 'alternatenames', 10, a => {
+    const firstchar = a[3][0];
+
+    if (a[2] === 'he' && (firstchar < '\u05D0' || firstchar > '\u05EA')) {
+      a[2] = 'en';
+    }
+
+    if (a[2] === 'he' || a[2] === 'en') {
+      if (a[2] === 'he') {
+        a[3] = a[3].replace(/‘/g, '׳');
+        a[3] = a[3].replace(/’/g, '׳');
+        a[3] = a[3].replace(/\'/g, '׳');
+        a[3] = Locale.hebrewStripNikkud(a[3]);
+      } else {
+        a[3] = a[3].replace(/‘/g, '\'');
+        a[3] = a[3].replace(/’/g, '\'');
+      }
+
+      return true;
+    }
+
+    return false;
+  }); // remove duplicates from alternatenames
+
+  doSql(logger, db, `DROP TABLE IF EXISTS altnames`, `CREATE TABLE altnames
+    AS SELECT geonameid, isolanguage, name
+    FROM alternatenames
+    GROUP BY 1, 2, 3
+    `);
   doSql(logger, db, `update admin1 set name='',asciiname='' where key like 'PS.%';`, `update country set country = '' where iso = 'PS';`, `delete from geoname where geonameid = 7303419;`);
   doSql(logger, db, `DROP TABLE IF EXISTS geoname_fulltext`, `CREATE VIRTUAL TABLE geoname_fulltext
       USING fts3(geonameid int, longname text,
@@ -547,47 +595,32 @@ async function buildGeonamesSqlite(dbFilename, countryInfotxt, cities5000txt, ci
       AND g.country = c.ISO
       AND g.country||'.'||g.admin1 = a.key
       `, `INSERT INTO geoname_fulltext
-      SELECT g.geonameid, g.name||', ישראל',
-      g.name, '', 'ישראל',
+      SELECT g.geonameid, alt.name||', ישראל',
+      alt.name, '', 'ישראל',
       g.population, g.latitude, g.longitude, g.timezone
-      FROM geoname_he g, admin1 a, country c
-      WHERE g.country = c.ISO
-      AND g.country||'.'||g.admin1 = a.key
+      FROM geoname g, altnames alt
+      WHERE g.country = 'IL'
+      AND alt.isolanguage = 'he'
+      AND g.geonameid = alt.geonameid
+      `, `INSERT INTO geoname_fulltext
+      SELECT g.geonameid, alt.name||', '||a1.asciiname||', Israel',
+      alt.name, a1.asciiname, 'Israel',
+      g.population, g.latitude, g.longitude, g.timezone
+      FROM geoname g, admin1 a1, altnames alt
+      WHERE g.country = 'IL'
+      AND alt.isolanguage = 'en'
+      AND g.geonameid = alt.geonameid
+      AND g.country||'.'||g.admin1 = a1.key
       `);
   db.close();
   return Promise.resolve(true);
 }
-/**
- * @param {string[]} a
- * @return {boolean}
- */
-
-function filterPlacesHebrew(a) {
-  if (a[6] != 'P' || a[7] != 'PPL' && a[7] != 'STLMT') {
-    return false;
-  }
-
-  const alternatenames = a[3].split(',');
-
-  for (const name of alternatenames) {
-    const firstchar = name[0];
-
-    if (firstchar >= '\u05D0' && firstchar <= '\u05EA') {
-      a[1] = Locale.hebrewStripNikkud(name); // replace 'name' field with Hebrew
-
-      return true;
-    }
-  }
-
-  return false;
-}
 /**
  * @param {pino.Logger} logger
  * @param {Database} db
  * @param  {...string} sqls
  */
 
-
 function doSql(logger, db, ...sqls) {
   for (let i = 0; i < sqls.length; i++) {
     logger.info(sqls[i]);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hebcal/geo-sqlite",
-  "version": "3.9.0",
+  "version": "4.1.0",
   "author": "Michael J. Radwin (https://github.com/mjradwin)",
   "keywords": [
     "hebcal"
@@ -32,7 +32,7 @@
     "@hebcal/core": "^3.33.3",
     "better-sqlite3": "^7.5.0",
     "pino": "^7.8.0",
-    "pino-pretty": "^7.5.1"
+    "pino-pretty": "^7.5.3"
   },
   "scripts": {
     "build": "rollup -c",
@@ -61,10 +61,10 @@
     "@rollup/plugin-json": "^4.1.0",
     "@rollup/plugin-node-resolve": "^13.1.3",
     "ava": "^4.0.1",
-    "eslint": "^8.9.0",
+    "eslint": "^8.10.0",
     "eslint-config-google": "^0.14.0",
     "jsdoc": "^3.6.10",
     "jsdoc-to-markdown": "^7.1.1",
-    "rollup": "^2.68.0"
+    "rollup": "^2.69.1"
   }
 }