Extract ASSET_TYPE_MAP into its own module; add Vitest + parser test suite

lib/__tests__/parser.test.ts

@@ -0,0 +1,144 @@
+import { describe, expect, it } from 'vitest';
+import { parseQuery, validateQuery } from '../parser';
+
+describe('parseQuery — structured queries', () => {
+  it('parses type, operator, region, radius', () => {
+    const parsed = parseQuery('type:data_center operator:google region:bavaria radius:100');
+    expect(parsed.type).toBe('data_center');
+    expect(parsed.operator).toBe('google');
+    expect(parsed.region).toBe('bavaria');
+    expect(parsed.radius).toBe(100);
+  });
+
+  it('parses country and near', () => {
+    const parsed = parseQuery('type:airport country:france');
+    expect(parsed.type).toBe('airport');
+    expect(parsed.country).toBe('france');
+  });
+
+  it('ignores unknown type keys', () => {
+    const parsed = parseQuery('type:not_a_real_type region:karnataka');
+    expect(parsed.type).toBeNull();
+  });
+
+  it('caps radius at 500 km', () => {
+    const parsed = parseQuery('type:telecom near:delhi radius:9999');
+    expect(parsed.radius).toBe(500);
+  });
+
+  it('decodes underscores in operator and region values', () => {
+    const parsed = parseQuery('type:power_plant operator:tata_power region:tamil_nadu');
+    expect(parsed.operator).toBe('tata power');
+    expect(parsed.region).toBe('tamil nadu');
+  });
+
+  it('defaults radius to 50 when unspecified', () => {
+    const parsed = parseQuery('type:telecom region:karnataka');
+    expect(parsed.radius).toBe(50);
+  });
+});
+
+describe('parseQuery — natural language', () => {
+  it('extracts telecom towers in a region', () => {
+    const parsed = parseQuery('telecom towers in karnataka');
+    expect(parsed.type).toBe('telecom');
+    expect(parsed.region).toBe('karnataka');
+    expect(parsed.near).toBeNull();
+  });
+
+  it('extracts "near X" as a proximity search', () => {
+    const parsed = parseQuery('power plants near mumbai');
+    expect(parsed.type).toBe('power_plant');
+    expect(parsed.near).toBe('mumbai');
+    expect(parsed.region).toBeNull();
+  });
+
+  it('extracts data centers', () => {
+    const parsed = parseQuery('data centers in california');
+    expect(parsed.type).toBe('data_center');
+    expect(parsed.region).toBe('california');
+  });
+
+  it('extracts airports', () => {
+    expect(parseQuery('airports in germany').type).toBe('airport');
+  });
+
+  it('extracts substations', () => {
+    expect(parseQuery('substations in texas').type).toBe('substation');
+  });
+
+  it('normalizes an operator alias to its canonical form', () => {
+    const parsed = parseQuery('aws data centers in oregon');
+    expect(parsed.operator).toBe('amazon');
+  });
+
+  it('parses an explicit radius from "within"', () => {
+    const parsed = parseQuery('cell towers near delhi within 25 km');
+    expect(parsed.near).toBe('delhi');
+    expect(parsed.radius).toBe(25);
+  });
+
+  it('strips stopwords like "show all" from the region', () => {
+    const parsed = parseQuery('show all hospitals in mumbai');
+    expect(parsed.type).toBe('hospital');
+    expect(parsed.region?.toLowerCase()).toContain('mumbai');
+  });
+
+  it('recognises cell vs broadcast vs radio tower variants', () => {
+    expect(parseQuery('cell towers in chennai').type).toBe('cell_tower');
+    expect(parseQuery('broadcast towers in chennai').type).toBe('broadcast_tower');
+    expect(parseQuery('radio towers in chennai').type).toBe('radio_tower');
+  });
+
+  it('recognises surveillance / CCTV', () => {
+    expect(parseQuery('surveillance cameras in london').type).toBe('surveillance_camera');
+    expect(parseQuery('cctv in london').type).toBe('surveillance_camera');
+  });
+});
+
+describe('parseQuery — empty / whitespace', () => {
+  it('returns an empty ParsedQuery for an empty string', () => {
+    const parsed = parseQuery('');
+    expect(parsed.type).toBeNull();
+    expect(parsed.region).toBeNull();
+    expect(parsed.near).toBeNull();
+    expect(parsed.raw).toBe('');
+  });
+
+  it('trims whitespace', () => {
+    const parsed = parseQuery('   telecom towers in karnataka   ');
+    expect(parsed.type).toBe('telecom');
+    expect(parsed.region).toBe('karnataka');
+  });
+});
+
+describe('validateQuery', () => {
+  it('rejects queries with no type and no operator', () => {
+    const parsed = parseQuery('in karnataka');
+    const result = validateQuery(parsed);
+    expect(result.valid).toBe(false);
+    expect(result.error).toMatch(/asset type or operator/i);
+  });
+
+  it('rejects queries with no geographic scope', () => {
+    const parsed = parseQuery('type:data_center');
+    const result = validateQuery(parsed);
+    expect(result.valid).toBe(false);
+    expect(result.error).toMatch(/geographic scope/i);
+  });
+
+  it('accepts a complete structured query', () => {
+    const parsed = parseQuery('type:telecom region:karnataka');
+    expect(validateQuery(parsed).valid).toBe(true);
+  });
+
+  it('accepts a complete natural query', () => {
+    const parsed = parseQuery('power plants in california');
+    expect(validateQuery(parsed).valid).toBe(true);
+  });
+
+  it('accepts a "near" proximity query', () => {
+    const parsed = parseQuery('data centers near london');
+    expect(validateQuery(parsed).valid).toBe(true);
+  });
+});