Fix: Apply lazy loading for UK patterns

wanasit · Nov 19, 2023 · 4c4d6d1 · 4c4d6d1
1 parent 30128bc
commit 4c4d6d1
Show file tree

Hide file tree

Showing 10 changed files with 98 additions and 142 deletions.
diff --git a/src/locales/uk/parsers/AbstractParserWithWordBoundaryChecking.ts b/src/locales/uk/parsers/AbstractParserWithWordBoundaryChecking.ts
@@ -0,0 +1,27 @@
+// noinspection DuplicatedCode
+
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+import { REGEX_PARTS } from "../constants";
+import { ParsingContext } from "../../../chrono";
+
+export abstract class AbstractParserWithLeftBoundaryChecking extends AbstractParserWithWordBoundaryChecking {
+    abstract innerPatternString(context: ParsingContext): string;
+
+    patternLeftBoundary(): string {
+        return REGEX_PARTS.leftBoundary;
+    }
+
+    innerPattern(context: ParsingContext): RegExp {
+        return new RegExp(this.innerPatternString(context), REGEX_PARTS.flags);
+    }
+
+    innerPatternHasChange(context: ParsingContext, currentInnerPattern: RegExp): boolean {
+        return false;
+    }
+}
+
+export abstract class AbstractParserWithLeftRightBoundaryChecking extends AbstractParserWithLeftBoundaryChecking {
+    innerPattern(context: ParsingContext): RegExp {
+        return new RegExp(`${this.innerPatternString(context)}${REGEX_PARTS.rightBoundary}`, REGEX_PARTS.flags);
+    }
+}
diff --git a/src/locales/uk/parsers/UKCasualDateParser.ts b/src/locales/uk/parsers/UKCasualDateParser.ts
@@ -1,21 +1,11 @@
 import { ParsingContext } from "../../../chrono";
 import { ParsingComponents, ParsingResult } from "../../../results";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import * as references from "../../../common/casualReferences";
-import { REGEX_PARTS } from "../constants";
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
-const PATTERN = new RegExp(
-    `(?:з|із|від)?\\s*(сьогодні|вчора|завтра|післязавтра|післяпіслязавтра|позапозавчора|позавчора)${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
-
-export default class UKCasualDateParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(context: ParsingContext): RegExp {
-        return PATTERN;
+export default class UKCasualDateParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return `(?:з|із|від)?\\s*(сьогодні|вчора|завтра|післязавтра|післяпіслязавтра|позапозавчора|позавчора)`;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | ParsingResult {

diff --git a/src/locales/uk/parsers/UKCasualTimeParser.ts b/src/locales/uk/parsers/UKCasualTimeParser.ts
@@ -1,22 +1,12 @@
 import { ParsingContext } from "../../../chrono";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import * as references from "../../../common/casualReferences";
 import { assignSimilarDate } from "../../../utils/dayjs";
 import dayjs from "dayjs";
-import { REGEX_PARTS } from "../constants";
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
-const PATTERN = new RegExp(
-    `(зараз|минулого\\s*вечора|минулої\\s*ночі|наступної\\s*ночі|сьогодні\\s*вночі|цієї\\s*ночі|цього ранку|вранці|ранку|зранку|опівдні|ввечері|вечора|опівночі|вночі)` +
-        `${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
-export default class UKCasualTimeParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern() {
-        return PATTERN;
+export default class UKCasualTimeParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return `(зараз|минулого\\s*вечора|минулої\\s*ночі|наступної\\s*ночі|сьогодні\\s*вночі|цієї\\s*ночі|цього ранку|вранці|ранку|зранку|опівдні|ввечері|вечора|опівночі|вночі)`;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray) {

diff --git a/src/locales/uk/parsers/UKMonthNameLittleEndianParser.ts b/src/locales/uk/parsers/UKMonthNameLittleEndianParser.ts
@@ -1,41 +1,32 @@
 import { ParsingContext } from "../../../chrono";
 import { ParsingResult } from "../../../results";
 import { findYearClosestToRef } from "../../../calculation/years";
-import { MONTH_DICTIONARY, REGEX_PARTS } from "../constants";
+import { MONTH_DICTIONARY } from "../constants";
 import { YEAR_PATTERN, parseYearPattern } from "../constants";
 import { ORDINAL_NUMBER_PATTERN, parseOrdinalNumberPattern } from "../constants";
 import { matchAnyPattern } from "../../../utils/pattern";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
-
-// prettier-ignore
-const PATTERN = new RegExp(
-    `(?:з|із)?\\s*(${ORDINAL_NUMBER_PATTERN})` +
-    `(?:` +
-    `\\s{0,3}(?:по|-|–|до)?\\s{0,3}` +
-    `(${ORDINAL_NUMBER_PATTERN})` +
-    `)?` +
-    `(?:-|\\/|\\s{0,3}(?:of)?\\s{0,3})` +
-    `(${matchAnyPattern(MONTH_DICTIONARY)})` +
-    `(?:` +
-    `(?:-|\\/|,?\\s{0,3})` +
-    `(${YEAR_PATTERN}(?![^\\s]\\d))` +
-    `)?` +
-    `${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
 const DATE_GROUP = 1;
 const DATE_TO_GROUP = 2;
 const MONTH_NAME_GROUP = 3;
 const YEAR_GROUP = 4;
 
-export default class UKMonthNameLittleEndianParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(): RegExp {
-        return PATTERN;
+export default class UKMonthNameLittleEndianParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return (
+            `(?:з|із)?\\s*(${ORDINAL_NUMBER_PATTERN})` +
+            `(?:` +
+            `\\s{0,3}(?:по|-|–|до)?\\s{0,3}` +
+            `(${ORDINAL_NUMBER_PATTERN})` +
+            `)?` +
+            `(?:-|\\/|\\s{0,3}(?:of)?\\s{0,3})` +
+            `(${matchAnyPattern(MONTH_DICTIONARY)})` +
+            `(?:` +
+            `(?:-|\\/|,?\\s{0,3})` +
+            `(${YEAR_PATTERN}(?![^\\s]\\d))` +
+            `)?`
+        );
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingResult {

diff --git a/src/locales/uk/parsers/UKMonthNameParser.ts b/src/locales/uk/parsers/UKMonthNameParser.ts
@@ -1,20 +1,9 @@
-import { FULL_MONTH_NAME_DICTIONARY, MONTH_DICTIONARY, REGEX_PARTS } from "../constants";
+import { FULL_MONTH_NAME_DICTIONARY, MONTH_DICTIONARY } from "../constants";
 import { ParsingContext } from "../../../chrono";
 import { findYearClosestToRef } from "../../../calculation/years";
 import { matchAnyPattern } from "../../../utils/pattern";
 import { YEAR_PATTERN, parseYearPattern } from "../constants";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
-
-const PATTERN = new RegExp(
-    `((?:в|у)\\s*)?` +
-        `(${matchAnyPattern(MONTH_DICTIONARY)})` +
-        `\\s*` +
-        `(?:` +
-        `[,-]?\\s*(${YEAR_PATTERN})?` +
-        `)?` +
-        `(?=[^\\s\\w]|\\s+[^0-9]|\\s+$|$)`,
-    REGEX_PARTS.flags
-);
+import { AbstractParserWithLeftBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
 const MONTH_NAME_GROUP = 2;
 const YEAR_GROUP = 3;
@@ -25,13 +14,17 @@ const YEAR_GROUP = 3;
  * - Січень 2012
  * - Січень
  */
-export default class UkMonthNameParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(): RegExp {
-        return PATTERN;
+export default class UkMonthNameParser extends AbstractParserWithLeftBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return (
+            `((?:в|у)\\s*)?` +
+            `(${matchAnyPattern(MONTH_DICTIONARY)})` +
+            `\\s*` +
+            `(?:` +
+            `[,-]?\\s*(${YEAR_PATTERN})?` +
+            `)?` +
+            `(?=[^\\s\\w]|\\s+[^0-9]|\\s+$|$)`
+        );
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray) {

diff --git a/src/locales/uk/parsers/UKRelativeDateFormatParser.ts b/src/locales/uk/parsers/UKRelativeDateFormatParser.ts
@@ -1,27 +1,19 @@
-import { REGEX_PARTS, TIME_UNIT_DICTIONARY } from "../constants";
+import { TIME_UNIT_DICTIONARY } from "../constants";
 import { ParsingContext } from "../../../chrono";
 import { ParsingComponents } from "../../../results";
 import dayjs from "dayjs";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import { matchAnyPattern } from "../../../utils/pattern";
-
-const PATTERN = new RegExp(
-    `(в минулому|у минулому|на минулому|минулого|на наступному|в наступному|у наступному|наступного|на цьому|в цьому|у цьому|цього)\\s*(${matchAnyPattern(
-        TIME_UNIT_DICTIONARY
-    )})(?=\\s*)${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
 const MODIFIER_WORD_GROUP = 1;
 const RELATIVE_WORD_GROUP = 2;
 
-export default class UKRelativeDateFormatParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(): RegExp {
-        return PATTERN;
+export default class UKRelativeDateFormatParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return (
+            `(в минулому|у минулому|на минулому|минулого|на наступному|в наступному|у наступному|наступного|на цьому|в цьому|у цьому|цього)\\s*` +
+            `(${matchAnyPattern(TIME_UNIT_DICTIONARY)})(?=\\s*)`
+        );
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {

diff --git a/src/locales/uk/parsers/UKTimeUnitAgoFormatParser.ts b/src/locales/uk/parsers/UKTimeUnitAgoFormatParser.ts
@@ -1,18 +1,12 @@
 import { ParsingContext } from "../../../chrono";
-import { parseTimeUnits, REGEX_PARTS, TIME_UNITS_PATTERN } from "../constants";
+import { parseTimeUnits, TIME_UNITS_PATTERN } from "../constants";
 import { ParsingComponents } from "../../../results";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import { reverseTimeUnits } from "../../../utils/timeunits";
+import { AbstractParserWithLeftBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
-const PATTERN = new RegExp(`(${TIME_UNITS_PATTERN})\\s{0,5}тому(?=(?:\\W|$))`, REGEX_PARTS.flags);
-
-export default class UKTimeUnitAgoFormatParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(): RegExp {
-        return PATTERN;
+export default class UKTimeUnitAgoFormatParser extends AbstractParserWithLeftBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return `(${TIME_UNITS_PATTERN})\\s{0,5}тому(?=(?:\\W|$))`;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray) {

diff --git a/src/locales/uk/parsers/UKTimeUnitCasualRelativeFormatParser.ts b/src/locales/uk/parsers/UKTimeUnitCasualRelativeFormatParser.ts
@@ -1,21 +1,12 @@
 import { TIME_UNITS_PATTERN, parseTimeUnits, REGEX_PARTS } from "../constants";
 import { ParsingContext } from "../../../chrono";
 import { ParsingComponents } from "../../../results";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import { reverseTimeUnits } from "../../../utils/timeunits";
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
-const PATTERN = new RegExp(
-    `(ці|останні|минулі|майбутні|наступні|після|через|\\+|-)\\s*(${TIME_UNITS_PATTERN})${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
-
-export default class UKTimeUnitCasualRelativeFormatParser extends AbstractParserWithWordBoundaryChecking {
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
-    }
-
-    innerPattern(): RegExp {
-        return PATTERN;
+export default class UKTimeUnitCasualRelativeFormatParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return `(ці|останні|минулі|майбутні|наступні|після|через|\\+|-)\\s*(${TIME_UNITS_PATTERN})`;
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {

diff --git a/src/locales/uk/parsers/UKTimeUnitWithinFormatParser.ts b/src/locales/uk/parsers/UKTimeUnitWithinFormatParser.ts
@@ -4,20 +4,16 @@ import { ParsingComponents } from "../../../results";
 import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 
 const PATTERN = `(?:(?:приблизно|орієнтовно)\\s*(?:~\\s*)?)?(${TIME_UNITS_PATTERN})${REGEX_PARTS.rightBoundary}`;
-const PATTERN_WITH_PREFIX = new RegExp(
-    `(?:протягом|на протязі|протягом|упродовж|впродовж)\\s*${PATTERN}`,
-    REGEX_PARTS.flags
-);
-
-const PATTERN_WITHOUT_PREFIX = new RegExp(PATTERN, "i");
 
 export default class UKTimeUnitWithinFormatParser extends AbstractParserWithWordBoundaryChecking {
     patternLeftBoundary(): string {
         return REGEX_PARTS.leftBoundary;
     }
 
     innerPattern(context: ParsingContext): RegExp {
-        return context.option.forwardDate ? PATTERN_WITHOUT_PREFIX : PATTERN_WITH_PREFIX;
+        return context.option.forwardDate
+            ? new RegExp(PATTERN, "i")
+            : new RegExp(`(?:протягом|на протязі|протягом|упродовж|впродовж)\\s*${PATTERN}`, REGEX_PARTS.flags);
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {

diff --git a/src/locales/uk/parsers/UKWeekdayParser.ts b/src/locales/uk/parsers/UKWeekdayParser.ts
@@ -1,34 +1,26 @@
 import { ParsingContext } from "../../../chrono";
 import { ParsingComponents } from "../../../results";
 // TODO: ADD REGEX_PARTS below
-import { REGEX_PARTS, WEEKDAY_DICTIONARY } from "../constants";
+import { WEEKDAY_DICTIONARY } from "../constants";
 import { matchAnyPattern } from "../../../utils/pattern";
-import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
 import { createParsingComponentsAtWeekday } from "../../../common/calculation/weekdays";
-
-const PATTERN = new RegExp(
-    `(?:(?:,|\\(|（)\\s*)?` +
-        `(?:в\\s*?)?` +
-        `(?:у\\s*?)?` +
-        `(?:(цей|минулого|минулий|попередній|попереднього|наступного|наступний|наступному)\\s*)?` +
-        `(${matchAnyPattern(WEEKDAY_DICTIONARY)})` +
-        `(?:\\s*(?:,|\\)|）))?` +
-        `(?:\\s*(на|у|в)\\s*(цьому|минулому|наступному)\\s*тижні)?` +
-        `${REGEX_PARTS.rightBoundary}`,
-    REGEX_PARTS.flags
-);
+import { AbstractParserWithLeftRightBoundaryChecking } from "./AbstractParserWithWordBoundaryChecking";
 
 const PREFIX_GROUP = 1;
 const WEEKDAY_GROUP = 2;
 const POSTFIX_GROUP = 3;
 
-export default class UKWeekdayParser extends AbstractParserWithWordBoundaryChecking {
-    innerPattern(): RegExp {
-        return PATTERN;
-    }
-
-    patternLeftBoundary(): string {
-        return REGEX_PARTS.leftBoundary;
+export default class UKWeekdayParser extends AbstractParserWithLeftRightBoundaryChecking {
+    innerPatternString(context: ParsingContext): string {
+        return (
+            `(?:(?:,|\\(|（)\\s*)?` +
+            `(?:в\\s*?)?` +
+            `(?:у\\s*?)?` +
+            `(?:(цей|минулого|минулий|попередній|попереднього|наступного|наступний|наступному)\\s*)?` +
+            `(${matchAnyPattern(WEEKDAY_DICTIONARY)})` +
+            `(?:\\s*(?:,|\\)|）))?` +
+            `(?:\\s*(на|у|в)\\s*(цьому|минулому|наступному)\\s*тижні)?`
+        );
     }
 
     innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingComponents {