diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8d8bcfbc24..6666a7bf95 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -13,6 +13,7 @@ app.
([#1314](https://github.com/birchill/10ten-ja-reader/issues/1314)).
- Make the `x` key close the popup if it is configured to _both_ close it and
expand it.
+- Fixed recognition of words that end in half-width numerals like 小1.
- Made the options page show up in a new tab on Edge.
- Added a workaround for a
[Firefox bug](https://bugzilla.mozilla.org/show_bug.cgi?id=1860486) that would
diff --git a/src/content/currency.ts b/src/content/currency.ts
index fc834a7c65..9522fc8bae 100644
--- a/src/content/currency.ts
+++ b/src/content/currency.ts
@@ -1,7 +1,7 @@
import {
getCombinedCharRange,
getNegatedCharRange,
- halfWidthNumbers,
+ startsWithNumeral,
} from '../utils/char-range';
import { parseNumber } from './numbers';
@@ -13,25 +13,30 @@ export type CurrencyMeta = {
};
export function lookForCurrency({
+ currentText,
nodeText,
- textDelimiter: originalTextDelimeter,
+ textDelimiter: originalTextDelimiter,
}: {
+ currentText: string;
nodeText: string;
textDelimiter: RegExp;
}): {
textDelimiter: RegExp;
textEnd: number;
} | null {
- // We only need to expand the search range if it starts with a currency
- // symbol. For the 8千円 case, the regular text lookup will find the necessary
- // text.
- if (nodeText.length && nodeText[0] !== '¥' && nodeText[0] !== '￥') {
+ // If the source text might be a currency, expand our text delimiter to allow
+ // extra symbols that would normally be ignored.
+ const sourceText = currentText + nodeText;
+ const mightBeCurrency =
+ sourceText[0] === '¥' ||
+ sourceText[0] === '￥' ||
+ (startsWithNumeral(sourceText) && sourceText.indexOf('円') > 0);
+ if (!mightBeCurrency) {
return null;
}
const japaneseOrPrice = getCombinedCharRange([
- getNegatedCharRange(originalTextDelimeter),
- halfWidthNumbers,
+ getNegatedCharRange(originalTextDelimiter),
/[¥￥\s,、.．。]/,
]);
const textDelimiter = getNegatedCharRange(japaneseOrPrice);
diff --git a/src/content/measure.ts b/src/content/measure.ts
index 9cc242ca73..3a586ff273 100644
--- a/src/content/measure.ts
+++ b/src/content/measure.ts
@@ -1,14 +1,15 @@
import {
getCombinedCharRange,
getNegatedCharRange,
- startsWithNumber,
+ startsWithDigit,
+ startsWithNumeral,
} from '../utils/char-range';
import { parseNumber } from './numbers';
export function lookForMeasure({
nodeText,
- textDelimiter: originalTextDelimeter,
+ textDelimiter: originalTextDelimiter,
}: {
nodeText: string;
textDelimiter: RegExp;
@@ -16,16 +17,14 @@ export function lookForMeasure({
textDelimiter: RegExp;
textEnd: number;
} | null {
- if (!startsWithNumber(nodeText)) {
+ if (!startsWithNumeral(nodeText)) {
return null;
}
- // getTextFromTextNode should already have expanded this range to include
- // half-width numbers and serparators so we just need to add the units and
- // space characters.
+ const includeSeparators = startsWithDigit(nodeText);
const japaneseOrUnit = getCombinedCharRange([
- getNegatedCharRange(originalTextDelimeter),
- /[\sm2㎡²]/,
+ getNegatedCharRange(originalTextDelimiter),
+ includeSeparators ? /[\sm2㎡²,、.．]/ : /[\sm2㎡²]/,
]);
const textDelimiter = getNegatedCharRange(japaneseOrUnit);
diff --git a/src/content/meta.ts b/src/content/meta.ts
index a6e575af77..e210b6a5ff 100644
--- a/src/content/meta.ts
+++ b/src/content/meta.ts
@@ -33,9 +33,9 @@ export function lookForMetadata({
} {
return (
(matchCurrency
- ? lookForCurrency({ nodeText, textDelimiter })
+ ? lookForCurrency({ currentText, nodeText, textDelimiter })
: undefined) ||
- lookForEra({ currentText, nodeText, textEnd }) ||
+ lookForEra({ currentText, nodeText, textEnd, textDelimiter }) ||
lookForShogi({ nodeText, textDelimiter }) ||
lookForMeasure({ nodeText, textDelimiter }) || {
textDelimiter,
diff --git a/src/content/scan-text.ts b/src/content/scan-text.ts
index 4d98a0f924..cffffa5ece 100644
--- a/src/content/scan-text.ts
+++ b/src/content/scan-text.ts
@@ -1,8 +1,4 @@
-import {
- nonJapaneseChar,
- nonJapaneseCharOrNumber,
- startsWithNumber,
-} from '../utils/char-range';
+import { nonJapaneseChar } from '../utils/char-range';
import { CursorPosition } from './get-cursor-position';
import { GetTextAtPointResult } from './get-text';
import { extractGetTextMetadata, lookForMetadata } from './meta';
@@ -92,12 +88,6 @@ export function scanText({
result.text +
nodeText.substring(0, textEnd === -1 ? undefined : textEnd);
- // If the source starts with a number, expand our text delimeter to allow
- // reading the rest of the number since it might be something like 5つ.
- if (!currentText.length && startsWithNumber(nodeText)) {
- textDelimiter = nonJapaneseCharOrNumber;
- }
-
// Check if we should further expand the set of allowed characters in
// order to recognize certain types of metadata-type strings (e.g. years
// or floor space measurements).
diff --git a/src/content/shogi.ts b/src/content/shogi.ts
index c166747776..1a302d6d92 100644
--- a/src/content/shogi.ts
+++ b/src/content/shogi.ts
@@ -66,7 +66,7 @@ export type ShogiMeta = {
export function lookForShogi({
nodeText,
- textDelimiter: originalTextDelimeter,
+ textDelimiter: originalTextDelimiter,
}: {
nodeText: string;
textDelimiter: RegExp;
@@ -79,28 +79,28 @@ export function lookForShogi({
}
// If the test starts with one of the shogi side indicators, then we assume
- // that the text is a shogi move and we can use the shogi delimeter.
+ // that the text is a shogi move and we can use the shogi delimiter.
if (['▲', '△', '☗', '☖'].includes(nodeText[0])) {
return {
- textDelimiter: shogiDelimeter,
- textEnd: nodeText.search(shogiDelimeter),
+ textDelimiter: shogiDelimiter,
+ textEnd: nodeText.search(shogiDelimiter),
};
}
// Otherwise, if it starts with an Arabic number followed by a kanji number
// OR it starts with one of the characters meaning "same position" then
- // expand the delimeter range to include all the shogi characters.
+ // expand the delimiter range to include all the shogi characters.
if (!unprefixedShogiStart.test(nodeText)) {
return null;
}
- const expandedDelimeter = getCombinedCharRange([
- getNegatedCharRange(originalTextDelimeter),
+ const expandedDelimiter = getCombinedCharRange([
+ getNegatedCharRange(originalTextDelimiter),
/[↑]/,
// All the other characters such as 𠔼丶フゝ・○ etc. should already be
// covered by `japaneseChar` so we don't need to add them here.
]);
- const textDelimiter = getNegatedCharRange(expandedDelimeter);
+ const textDelimiter = getNegatedCharRange(expandedDelimiter);
return {
textDelimiter,
@@ -109,7 +109,7 @@ export function lookForShogi({
}
// This needs to be kept in sync with the regexes below.
-const shogiDelimeter =
+const shogiDelimiter =
/[^▲△☗☖1-9１-９一二三四五六七八九同仝－𠔼ド歩兵丶フゝ・香禾キ↑桂土銀ヨ角ク飛ヒ乙金人と成ナ馬マウ龍竜立リ玉王○打引寄上行入右左直行入不生]/u;
const unprefixedShogiStart = /^[1-9１-９][一二三四五六七八九]|[同仝－𠔼ド]/u;
diff --git a/src/content/years.ts b/src/content/years.ts
index 2d1edfb44d..128ead4e88 100644
--- a/src/content/years.ts
+++ b/src/content/years.ts
@@ -1,27 +1,39 @@
+import { getCombinedCharRange, getNegatedCharRange } from '../utils/char-range';
import { parseNumber } from './numbers';
-const nonEraCharacter = /[^\s0-9０-９一二三四五六七八九十百元年]/;
-
export function lookForEra({
currentText,
nodeText,
+ textDelimiter: originalTextDelimiter,
textEnd,
}: {
currentText: string;
nodeText: string;
+ textDelimiter: RegExp;
textEnd: number;
}): {
textDelimiter: RegExp;
textEnd: number;
} | null {
+ // We only want to _extend_ the current range so if `textEnd` is already -1
+ // (i.e. end of the text) then we don't need to do anything.
if (textEnd < 0 || !startsWithEraName(currentText)) {
return null;
}
- const endOfEra = nodeText.substring(textEnd).search(nonEraCharacter);
+ // The original text delimiter should include all the characters needed to
+ // match Japanese years except spaces between the era and the year, and
+ // spaces between the year and the final 年 character, if any.
+ const japaneseOrSpace = getCombinedCharRange([
+ getNegatedCharRange(originalTextDelimiter),
+ /[\s]/,
+ ]);
+ const textDelimiter = getNegatedCharRange(japaneseOrSpace);
+
+ const endOfEra = nodeText.substring(textEnd).search(textDelimiter);
return {
- textDelimiter: nonEraCharacter,
+ textDelimiter,
textEnd: endOfEra === -1 ? -1 : textEnd + endOfEra,
};
}
diff --git a/src/utils/char-range.ts b/src/utils/char-range.ts
index aa83b2530e..90b66a553b 100644
--- a/src/utils/char-range.ts
+++ b/src/utils/char-range.ts
@@ -138,6 +138,8 @@ function isCharacterClassRange(re: RegExp): boolean {
// typically delimit words.
/** @public */
export const japaneseChar = getCombinedCharRange([
+ // We include half-width numbers so we can recognize things like 小1
+ halfWidthNumbers,
fullWidthAlphanumerics,
zeroWidthNonJoiner,
whiteCircle,
@@ -174,18 +176,40 @@ export function getNegatedCharRange(range: RegExp): RegExp {
export const nonJapaneseChar = getNegatedCharRange(japaneseChar);
-export const nonJapaneseCharOrNumber = getNegatedCharRange(
- getCombinedCharRange([japaneseChar, halfWidthNumbers, /[,、.．]/])
-);
-
export function hasKatakana(text: string): boolean {
return katakana.test(text);
}
-const numberStartRegex = /^[0-9０-９一二三四五六七八九十百]/;
+export function startsWithDigit(input: string): boolean {
+ const c = input.length ? input.charCodeAt(0) : 0;
+ return (c >= 48 && c <= 57) || (c >= 65296 && c <= 65305);
+}
-export function startsWithNumber(input: string): boolean {
- return !!input.length && numberStartRegex.test(input);
+const kanjiNumerals = [
+ '〇',
+ '一',
+ '二',
+ '三',
+ '四',
+ '五',
+ '六',
+ '七',
+ '八',
+ '九',
+ '十',
+ '百',
+ '千',
+ '万',
+ '億',
+ '兆',
+ '京',
+];
+
+export function startsWithNumeral(input: string): boolean {
+ return (
+ startsWithDigit(input) ||
+ (!!input.length && kanjiNumerals.includes(input[0]))
+ );
}
const onlyDigits = /^[0-9０-９,，、.．]+$/;
diff --git a/tests/get-text.test.ts b/tests/get-text.test.ts
index 4860e8c223..d84db97aca 100644
--- a/tests/get-text.test.ts
+++ b/tests/get-text.test.ts
@@ -419,6 +419,21 @@ describe('getTextAtPoint', () => {
assertTextResultEqual(result, 'あ\u200cい\u200cう\u200c', [textNode, 0, 6]);
});
+ it('should include trailing half-width numerals', () => {
+ testDiv.append('小1。');
+ const textNode = testDiv.firstChild as Text;
+ const bbox = getBboxForOffset(textNode, 0);
+
+ const result = getTextAtPoint({
+ point: {
+ x: bbox.left,
+ y: bbox.top + bbox.height / 2,
+ },
+ });
+
+ assertTextResultEqual(result, '小1', [textNode, 0, 2]);
+ });
+
it('should include the year when recognizing years', () => {
testDiv.append('昭和56年に');
const textNode = testDiv.firstChild as Text;
@@ -431,7 +446,7 @@ describe('getTextAtPoint', () => {
},
});
- assertTextResultEqual(result, '昭和56年', [textNode, 0, 5]);
+ assertTextResultEqual(result, '昭和56年に', [textNode, 0, 6]);
assert.deepEqual(result!.meta, {
type: 'era',
era: '昭和',
@@ -473,7 +488,7 @@ describe('getTextAtPoint', () => {
},
});
- assertTextResultEqual(result, '昭和５6年', [textNode, 0, 5]);
+ assertTextResultEqual(result, '昭和５6年に', [textNode, 0, 6]);
assert.deepEqual(result!.meta, {
type: 'era',
era: '昭和',
@@ -495,7 +510,7 @@ describe('getTextAtPoint', () => {
},
});
- assertTextResultEqual(result, '昭和 56 年', [textNode, 0, 7]);
+ assertTextResultEqual(result, '昭和 56 年に', [textNode, 0, 8]);
assert.deepEqual(result!.meta, {
type: 'era',
era: '昭和',
@@ -517,7 +532,7 @@ describe('getTextAtPoint', () => {
},
});
- assertTextResultEqual(result, '昭和56', [textNode, 0, 4]);
+ assertTextResultEqual(result, '昭和56に', [textNode, 0, 5]);
assert.deepEqual(result!.meta, {
type: 'era',
era: '昭和',
@@ -542,10 +557,10 @@ describe('getTextAtPoint', () => {
assertTextResultEqual(
result,
- '昭和56年',
+ '昭和56年に',
[firstTextNode, 0, 2],
[middleTextNode, 0, 2],
- [lastTextNode, 0, 1]
+ [lastTextNode, 0, 2]
);
assert.deepEqual(result!.meta, {
type: 'era',
@@ -570,9 +585,9 @@ describe('getTextAtPoint', () => {
assertTextResultEqual(
result,
- '昭和 56年',
+ '昭和 56年に',
[firstTextNode, 0, 3],
- [middleTextNode, 0, 4]
+ [middleTextNode, 0, 5]
);
assert.deepEqual(result!.meta, {
type: 'era',
@@ -586,6 +601,7 @@ describe('getTextAtPoint', () => {
testDiv.innerHTML = '昭和56年に';
const firstTextNode = testDiv.firstChild as Text;
const middleTextNode = testDiv.childNodes[1].firstChild as Text;
+ const finalTextNode = testDiv.lastChild as Text;
const bbox = getBboxForOffset(firstTextNode, 0);
const result = getTextAtPoint({
@@ -597,9 +613,10 @@ describe('getTextAtPoint', () => {
assertTextResultEqual(
result,
- '昭和56年',
+ '昭和56年に',
[firstTextNode, 0, 2],
- [middleTextNode, 0, 3]
+ [middleTextNode, 0, 3],
+ [finalTextNode, 0, 1]
);
assert.deepEqual(result!.meta, {
type: 'era',
@@ -678,7 +695,7 @@ describe('getTextAtPoint', () => {
});
});
- it('should stop at delimeters (even when matching years)', () => {
+ it('should stop at delimiters (even when matching years)', () => {
testDiv.append('昭和三大馬鹿査定」発言に');
const textNode = testDiv.firstChild as Text;
const bbox = getBboxForOffset(textNode, 0);
@@ -719,7 +736,7 @@ describe('getTextAtPoint', () => {
});
});
- it('should recognize Japanese yen values that start with ￥', () => {
+ it('should recognize Japanese yen values that start with ￥ (full-width)', () => {
testDiv.append('価格￥8万8千です');
const textNode = testDiv.firstChild as Text;
const bbox = getBboxForOffset(textNode, 2);
@@ -739,7 +756,7 @@ describe('getTextAtPoint', () => {
});
});
- it('should recognize Japanese yen values that start with ¥', () => {
+ it('should recognize Japanese yen values that start with ¥ (half-width)', () => {
testDiv.append('価格¥ 8万8千です');
const textNode = testDiv.firstChild as Text;
const bbox = getBboxForOffset(textNode, 2);
@@ -759,6 +776,32 @@ describe('getTextAtPoint', () => {
});
});
+ it('should recognize Japanese yen values that start with ¥ in a separate span', () => {
+ testDiv.innerHTML = '¥ 88,000です';
+ const firstTextNode = testDiv.childNodes[0].firstChild as Text;
+ const secondTextNode = testDiv.childNodes[1] as Text;
+ const bbox = getBboxForOffset(firstTextNode, 0);
+
+ const result = getTextAtPoint({
+ point: {
+ x: bbox.left + bbox.width / 2,
+ y: bbox.top + bbox.height / 2,
+ },
+ });
+
+ assertTextResultEqual(
+ result,
+ '¥ 88,000です',
+ [firstTextNode, 0, 1],
+ [secondTextNode, 0, 9]
+ );
+ assert.deepEqual(result!.meta, {
+ type: 'currency',
+ value: 88000,
+ matchLen: 8,
+ });
+ });
+
it('should recognize Japanese yen values that include commas', () => {
testDiv.append('価格8,800円です');
const textNode = testDiv.firstChild as Text;
diff --git a/tests/playground.html b/tests/playground.html
index d53b85f46e..d211022598 100644
--- a/tests/playground.html
+++ b/tests/playground.html
@@ -127,10 +127,11 @@ Measure tests
面積：十二平方米
8万8千平㍍
- Starting number tests
+ Starting/end number tests
Currency tests