feat: add support for metric suffixes on currency values

Fixes #2000.

CHANGELOG.md

@@ -13,6 +13,8 @@ app.
 - Made tap detection more reliable
   thanks to [@maiself](https://github.com/maiself)
   ([#2014](https://github.com/birchill/10ten-ja-reader/pull/2014)).
+- Added handling for metric suffixes on currencies (e.g. 40k円)
+  ([#2000](https://github.com/birchill/10ten-ja-reader/pull/2000)).
 - Added handling for 戶 and 內 kyūjitai.
 - Added deinflection for additional forms of -sugiru
   ([#2033](https://github.com/birchill/10ten-ja-reader/pull/2033)).

src/content/currency.test.ts

@@ -95,5 +95,39 @@ describe('extractCurrencyMetadata', () => {
       value: 10000000000,
       matchLen: 5,
     });
+
+    // Metric suffixes
+    expect(extractCurrencyMetadata('200k円')).toEqual({
+      type: 'currency',
+      value: 200_000,
+      matchLen: 5,
+    });
+    expect(extractCurrencyMetadata('1k 円')).toEqual({
+      type: 'currency',
+      value: 1_000,
+      matchLen: 4,
+    });
+    expect(extractCurrencyMetadata('1M円')).toEqual({
+      type: 'currency',
+      value: 1_000_000,
+      matchLen: 3,
+    });
+    expect(extractCurrencyMetadata('￥40k')).toEqual({
+      type: 'currency',
+      value: 40_000,
+      matchLen: 4,
+    });
+    // We don't allow white space before the suffix
+    expect(extractCurrencyMetadata('￥40 k')).toEqual({
+      type: 'currency',
+      value: 40,
+      matchLen: 3,
+    });
+    // Should ignore the suffix since it's not followed by a word boundary
+    expect(extractCurrencyMetadata('￥40billion')).toEqual({
+      type: 'currency',
+      value: 40,
+      matchLen: 3,
+    });
   });
 });

src/content/currency.ts

@@ -37,7 +37,7 @@ export function lookForCurrency({
 
   const japaneseOrPrice = getCombinedCharRange([
     getNegatedCharRange(originalTextDelimiter),
-    /[¥￥\s,、.．。]/,
+    /[¥￥\s,、.．。kKmMbBtT]/,
   ]);
   const textDelimiter = getNegatedCharRange(japaneseOrPrice);
 
@@ -48,7 +48,7 @@ export function lookForCurrency({
 }
 
 const currencyRegex =
-  /([￥¥]\s*([0-9.,０-９。．、〇一二三四五六七八九十百千万億兆京]+))|(([0-9.,０-９。．、〇一二三四五六七八九十百千万億兆京]+)\s*円)/;
+  /([￥¥]\s*([0-9.,０-９。．、〇一二三四五六七八九十百千万億兆京]+)([kKmMbBtT]\b)?)|(([0-9.,０-９。．、〇一二三四五六七八九十百千万億兆京]+)([kKmMbBtT])?\s*円)/;
 
 export function extractCurrencyMetadata(
   text: string
@@ -58,16 +58,38 @@ export function extractCurrencyMetadata(
     return undefined;
   }
 
-  const valueStr = matches[2] ?? matches[4];
+  const valueStr = matches[2] ?? matches[5];
 
   if (!valueStr) {
     return undefined;
   }
 
-  const value = parseNumber(valueStr);
-  if (!value) {
+  let value = parseNumber(valueStr);
+  if (value === null) {
     return undefined;
   }
 
+  // Handle metric suffixes---we handle them here instead of in parseNumber
+  // because we only support them when they are part of a currency.
+  const metricSuffix = matches[2] ? matches[3] : matches[6];
+  switch (metricSuffix) {
+    case 'k':
+    case 'K':
+      value *= 1_000;
+      break;
+    case 'm':
+    case 'M':
+      value *= 1_000_000;
+      break;
+    case 'b':
+    case 'B':
+      value *= 1_000_000_000;
+      break;
+    case 't':
+    case 'T':
+      value *= 1_000_000_000_000;
+      break;
+  }
+
   return { type: 'currency', value, matchLen: matches[0].length };
 }

src/content/numbers.test.ts

@@ -58,6 +58,10 @@ describe('parseNumber', () => {
     // Putting the powers of ten in the wrong order
     expect(parseNumber('七十二百一')).toStrictEqual(null);
 
+    // Don't handle metric suffixes for bare numbers--we only allow them when
+    // part of a currency.
+    expect(parseNumber('40k')).toStrictEqual(null);
+
     // Completely invalid inputs
     expect(parseNumber('abc')).toStrictEqual(null);
     expect(parseNumber('')).toStrictEqual(null);
@@ -79,6 +83,7 @@ describe('extractNumberMetadata', () => {
     expect(extractNumberMetadata('43.2')).toStrictEqual(undefined);
     expect(extractNumberMetadata('４３．２')).toStrictEqual(undefined);
     expect(extractNumberMetadata('４３。２')).toStrictEqual(undefined);
+    expect(extractNumberMetadata('40k')).toStrictEqual(undefined);
 
     // Shouldn't be zero
     expect(extractNumberMetadata('〇〇〇')).toStrictEqual(undefined);

tests/get-text.test.ts

@@ -899,6 +899,26 @@ describe('getTextAtPoint', () => {
     });
   });
 
+  it('should recognize Japanese yen values with metric suffixes', () => {
+    testDiv.append('1k 円 for 240 blank cards...');
+    const textNode = testDiv.firstChild as Text;
+    const bbox = getBboxForOffset(textNode, 0);
+
+    const result = getTextAtPoint({
+      point: {
+        x: bbox.left + bbox.width / 2,
+        y: bbox.top + bbox.height / 2,
+      },
+    });
+
+    assertTextResultEqual(result, '1k 円 ', [textNode, 0, 5]);
+    assert.deepEqual(result?.meta, {
+      type: 'currency',
+      value: 1_000,
+      matchLen: 4,
+    });
+  });
+
   it('should recognize 畳 measurements', () => {
     testDiv.append('面積：6畳です');
     const textNode = testDiv.firstChild as Text;

tests/playground.html

@@ -145,6 +145,7 @@ <h3>Currency tests</h3>
     <li>100億円</li>
     <li>25.1万円</li>
     <li><span>39,800</span><span>万円</span></li>
+    <li>1,000,000円＝1,000k円＝1M円です</li>
   </ul>
 
   <h3>Number tests</h3>