feat: Implement WKT validation in data-type.ts (keplergl#3298)

Signed-off-by: bdjulbic <bdjulbic@foursquare.com>

Author: bdjulbic

src/common-utils/src/data-type.ts

@@ -11,6 +11,27 @@ import {h3IsValid} from './h3-utils';
 
 const H3_ANALYZER_TYPE = 'H3';
 
+// Returns true if the value is likely a WKT geometry string (heuristic check).
+const WKT_PREFIX_RE =
+  /^(?:SRID=\d+\s*;\s*)?(?:POINT|LINESTRING|POLYGON|MULTIPOINT|MULTILINESTRING|MULTIPOLYGON|GEOMETRYCOLLECTION)(?:\s+(?:Z|M|ZM))?\s*\(/i;
+
+export function isWkt(value: unknown): boolean {
+  if (typeof value !== 'string') {
+    return false;
+  }
+
+  const s = value.trim();
+  if (s.length < 10) {
+    return false;
+  }
+
+  if (!s.includes('(') || !s.includes(')')) {
+    return false;
+  }
+
+  return WKT_PREFIX_RE.test(s);
+}
+
 export const ACCEPTED_ANALYZER_TYPES = [
   AnalyzerDATA_TYPES.DATE,
   AnalyzerDATA_TYPES.TIME,
@@ -267,6 +288,11 @@ export function getFieldsFromData(data: RowData, fieldOrder: string[]): Field[]
       type = data.some(d => isHexWkb(d[name])) ? AnalyzerDATA_TYPES.GEOMETRY : type;
     }
 
+    // quick check if string is wkt
+    if (type === AnalyzerDATA_TYPES.STRING) {
+      type = data.some(d => isWkt(d[name])) ? AnalyzerDATA_TYPES.GEOMETRY_FROM_STRING : type;
+    }
+
     return {
       name,
       id: name,

test/node/utils/dataset-utils-test.js

@@ -3,6 +3,7 @@
 
 import test from 'tape';
 import {findDefaultColorField, createNewDataEntry} from '@kepler.gl/utils';
+import {isHexWkb, isWkt} from '@kepler.gl/common-utils';
 
 import {processCsvData} from '@kepler.gl/processors';
 
@@ -88,4 +89,42 @@ test('datasetUtils.isHexWkb', t => {
 
   const validEWktNDR = '0020000001000013ff0000000000400000000000000040';
   t.ok(isHexWkb(validEWktNDR), 'A valid hex ewkb in NDR should be valid');
+  t.end();
+});
+
+test('datasetUtils.isWkt', t => {
+  // non-strings
+  t.notOk(isWkt(null), 'null is not a valid WKT');
+  t.notOk(isWkt(undefined), 'undefined is not a valid WKT');
+  t.notOk(isWkt(123), 'number is not a valid WKT');
+  t.notOk(isWkt({}), 'object is not a valid WKT');
+
+  // regular strings / known non-WKT identifiers
+  t.notOk(isWkt(''), 'empty string is not a valid WKT');
+  t.notOk(isWkt('hello world'), 'regular string is not a valid WKT');
+  t.notOk(isWkt('06075'), 'FIPS code should not be a valid WKT');
+  t.notOk(isWkt('8a2a1072b59ffff'), 'H3 code should not be a valid WKT');
+
+  // edge cases (missing coordinates / parentheses)
+  t.notOk(isWkt('POINT'), 'POINT without coordinates should not be a valid WKT');
+  t.notOk(isWkt('POINT 1 2'), 'POINT without parentheses should not be a valid WKT');
+  t.notOk(isWkt('SRID=4326;POINT'), 'SRID prefix without geometry should not be a valid WKT');
+  t.notOk(isWkt('POINT (1 2'), 'missing closing parenthesis should not be a valid WKT');
+  t.notOk(isWkt('POINT ( )'), 'empty coordinates should not be a valid WKT');
+
+  // valid WKT examples (heuristic)
+  t.ok(isWkt('POINT (1 2)'), 'POINT should be recognized as WKT');
+  t.ok(isWkt('POINT(1 2)'), 'POINT without space before parentheses should be recognized as WKT');
+  t.ok(isWkt('LINESTRING (0 0, 1 1)'), 'LINESTRING should be recognized as WKT');
+  t.ok(isWkt('POLYGON ((0 0, 1 0, 1 1, 0 0))'), 'POLYGON should be recognized as WKT');
+  t.ok(
+    isWkt('SRID=4326;POINT(1 2)'),
+    'WKT with SRID prefix should be recognized as WKT'
+  );
+  t.ok(isWkt('POINT Z (1 2 3)'), 'WKT with Z dimension should be recognized as WKT');
+
+  // not WKT but contains parentheses
+  t.notOk(isWkt('HELLO (1 2)'), 'non-WKT string with parentheses should not be valid WKT');
+
+  t.end();
 });