[New] FR month name little endian parsing

wanasit · Jul 23, 2020 · 072c24c · 072c24c
1 parent 2a6e043
commit 072c24c
Show file tree

Hide file tree

Showing 13 changed files with 522 additions and 125 deletions.
diff --git a/src/common/parsers/SlashDateFormatParser.ts b/src/common/parsers/SlashDateFormatParser.ts
@@ -13,10 +13,7 @@ import { findMostLikelyADYear, findYearClosestToRef } from "../../calculation/ye
 const PATTERN = new RegExp(
     "([^\\d]|^)" +
         "([0-3]{0,1}[0-9]{1})[\\/\\.\\-]([0-3]{0,1}[0-9]{1})" +
-        "(?:" +
-        "[\\/\\.\\-]" +
-        "([0-9]{4}\\s*\\,?\\s*|[0-9]{2}\\s*\\,?\\s*)" +
-        ")?" +
+        "(?:[\\/\\.\\-]([0-9]{4}|[0-9]{2}))?" +
         "(\\W|$)",
     "i"
 );
@@ -52,7 +49,10 @@ export default class SlashDateFormatParser implements Parser {
         }
 
         const index = match.index + match[OPENING_GROUP].length;
-        const text = match[0].substr(match[OPENING_GROUP].length, match[0].length - match[ENDING_GROUP].length);
+        const text = match[0].substr(
+            match[OPENING_GROUP].length,
+            match[0].length - match[OPENING_GROUP].length - match[ENDING_GROUP].length
+        );
 
         // '1.12', '1.12.12' is more like a version numbers
         if (text.match(/^\d\.\d$/) || text.match(/^\d\.\d{1,2}\.\d{1,2}\s*$/)) {

diff --git a/src/locales/fr/constants.ts b/src/locales/fr/constants.ts
@@ -38,6 +38,7 @@ export const MONTH_DICTIONARY: { [word: string]: number } = {
     "juin": 6,
     "jun": 6,
     "juillet": 7,
+    "juil": 7,
     "jul": 7,
     "jul.": 7,
     "août": 8,
@@ -120,31 +121,25 @@ export function parseNumberPattern(match: string): number {
 
 //-----------------------------
 
-export const ORDINAL_NUMBER_PATTERN = `(?:[0-9]{1,2}(?:st|nd|rd|th)?)`;
+export const ORDINAL_NUMBER_PATTERN = `(?:[0-9]{1,2}(?:er)?)`;
 export function parseOrdinalNumberPattern(match: string): number {
     let num = match.toLowerCase();
-    num = num.replace(/(?:st|nd|rd|th)$/i, "");
+    num = num.replace(/(?:er)$/i, "");
     return parseInt(num);
 }
 
 //-----------------------------
-
-export const YEAR_PATTERN = `(?:[1-9][0-9]{0,3}\\s*(?:BE|AD|BC)|[1-2][0-9]{3}|[5-9][0-9])`;
+// 88 p. Chr. n.
+// 234 AC
+export const YEAR_PATTERN = `(?:[1-9][0-9]{0,3}\\s*(?:AC|AD|p\\.\\s*C(?:hr?)?\\.\\s*n\\.)|[1-2][0-9]{3}|[5-9][0-9])`;
 export function parseYear(match: string): number {
-    if (/BE/i.test(match)) {
-        // Buddhist Era
-        match = match.replace(/BE/i, "");
-        return parseInt(match) - 543;
-    }
-
-    if (/BC/i.test(match)) {
-        // Before Christ
+    if (/AC/i.test(match)) {
         match = match.replace(/BC/i, "");
         return -parseInt(match);
     }
 
-    if (/AD/i.test(match)) {
-        match = match.replace(/AD/i, "");
+    if (/AD/i.test(match) || /C/i.test(match)) {
+        match = match.replace(/[^\d]+/i, "");
         return parseInt(match);
     }
 

diff --git a/src/locales/fr/index.ts b/src/locales/fr/index.ts
@@ -9,6 +9,7 @@ import FRMergeDateTimeRefiner from "./refiners/FRMergeDateTimeRefiner";
 import FRMergeDateRangeRefiner from "./refiners/FRMergeDateRangeRefiner";
 import FRWeekdayParser from "./parsers/FRWeekdayParser";
 import FRSpecificTimeExpressionParser from "./parsers/FRSpecificTimeExpressionParser";
+import FRMonthNameLittleEndianParser from "./parsers/FRMonthNameLittleEndianParser";
 
 // Shortcuts
 export const casual = new Chrono(createCasualConfiguration());
@@ -34,6 +35,7 @@ export function createConfiguration(strictMode = true, littleEndian = true): Con
         {
             parsers: [
                 new SlashDateFormatParser(littleEndian),
+                new FRMonthNameLittleEndianParser(),
                 new FRTimeExpressionParser(),
                 new FRSpecificTimeExpressionParser(),
                 new FRWeekdayParser(),

diff --git a/src/locales/fr/parsers/FRMonthNameLittleEndianParser.ts b/src/locales/fr/parsers/FRMonthNameLittleEndianParser.ts
@@ -0,0 +1,64 @@
+import { ParsingContext } from "../../../chrono";
+import { ParsingResult } from "../../../results";
+import { findYearClosestToRef } from "../../../calculation/years";
+import { MONTH_DICTIONARY } from "../constants";
+import { YEAR_PATTERN, parseYear } from "../constants";
+import { ORDINAL_NUMBER_PATTERN, parseOrdinalNumberPattern } from "../constants";
+import { matchAnyPattern } from "../../../utils/pattern";
+import { AbstractParserWithWordBoundaryChecking } from "../../../common/parsers/AbstractParserWithWordBoundary";
+
+const PATTERN = new RegExp(
+    "(?:on\\s*?)?" +
+        `(${ORDINAL_NUMBER_PATTERN})` +
+        `(?:\\s*(?:au|\\-|\\–|jusqu\'au?|\\s)\\s*(${ORDINAL_NUMBER_PATTERN}))?` +
+        `(?:-|/|\\s*(?:de)?\\s*)` +
+        `(${matchAnyPattern(MONTH_DICTIONARY)})` +
+        `(?:(?:-|/|,?\\s*)(${YEAR_PATTERN}(?![^\\s]\\d)))?` +
+        `(?=\\W|$)`,
+    "i"
+);
+
+const DATE_GROUP = 1;
+const DATE_TO_GROUP = 2;
+const MONTH_NAME_GROUP = 3;
+const YEAR_GROUP = 4;
+
+export default class FRMonthNameLittleEndianParser extends AbstractParserWithWordBoundaryChecking {
+    innerPattern(): RegExp {
+        return PATTERN;
+    }
+
+    innerExtract(context: ParsingContext, match: RegExpMatchArray): ParsingResult {
+        //console.log(match)
+
+        const result = context.createParsingResult(match.index, match[0]);
+
+        const month = MONTH_DICTIONARY[match[MONTH_NAME_GROUP].toLowerCase()];
+        const day = parseOrdinalNumberPattern(match[DATE_GROUP]);
+        if (day > 31) {
+            // e.g. "[96 Aug]" => "9[6 Aug]", we need to shift away from the next number
+            match.index = match.index + match[DATE_GROUP].length;
+            return null;
+        }
+
+        result.start.assign("month", month);
+        result.start.assign("day", day);
+
+        if (match[YEAR_GROUP]) {
+            const yearNumber = parseYear(match[YEAR_GROUP]);
+            result.start.assign("year", yearNumber);
+        } else {
+            const year = findYearClosestToRef(context.refDate, day, month);
+            result.start.imply("year", year);
+        }
+
+        if (match[DATE_TO_GROUP]) {
+            const endDate = parseOrdinalNumberPattern(match[DATE_TO_GROUP]);
+
+            result.end = result.start.clone();
+            result.end.assign("day", endDate);
+        }
+
+        return result;
+    }
+}
diff --git a/src/locales/fr/parsers/FRSpecificTimeExpressionParser.ts b/src/locales/fr/parsers/FRSpecificTimeExpressionParser.ts
@@ -41,9 +41,14 @@ export default class FRSpecificTimeExpressionParser implements Parser {
     }
 
     extract(context: ParsingContext, match: RegExpMatchArray): ParsingResult | null {
-        const refDate = dayjs(context.refDate);
         let result = context.createParsingResult(match.index + match[1].length, match[0].substring(match[1].length));
 
+        // This looks more like a year e.g. 2020
+        if (result.text.match(/^\d{4}$/)) {
+            match.index += match[0].length;
+            return null;
+        }
+
         result.start = FRSpecificTimeExpressionParser.extractTimeComponent(result.start.clone(), match);
         if (!result.start) {
             match.index += match[0].length;
@@ -59,10 +64,6 @@ export default class FRSpecificTimeExpressionParser implements Parser {
             }
         }
 
-        if (result.text.match(/^\d+$/)) {
-            return null;
-        }
-
         return result;
     }
 

diff --git a/src/locales/fr/parsers/FRTimeExpressionParser.ts b/src/locales/fr/parsers/FRTimeExpressionParser.ts
@@ -1,4 +1,6 @@
 import { AbstractTimeExpressionParser } from "../../../common/parsers/AbstractTimeExpressionParser";
+import { ParsingComponents, ParsingResult } from "../../../results";
+import { ParsingContext } from "../../../chrono";
 
 export default class FRTimeExpressionParser extends AbstractTimeExpressionParser {
     primaryPrefix(): string {
@@ -8,4 +10,13 @@ export default class FRTimeExpressionParser extends AbstractTimeExpressionParser
     followingPhase(): string {
         return "\\s*(?:\\-|\\–|\\~|\\〜|[àa]|\\?)\\s*";
     }
+
+    extractPrimaryTimeComponents(context: ParsingContext, match: RegExpMatchArray): ParsingComponents | null {
+        // This looks more like a year e.g. 2020
+        if (match[0].match(/^\s*\d{4}\s*$/)) {
+            return null;
+        }
+
+        return super.extractPrimaryTimeComponents(context, match);
+    }
 }
diff --git a/test/en/en_slash.test.ts b/test/en/en_slash.test.ts
@@ -1,6 +1,13 @@
 import * as chrono from "../../src";
 import { testSingleCase, testUnexpectedResult, testWithExpectedDate } from "../test_util";
 
+test("Test - Parsing Offset Expression", function () {
+    testSingleCase(chrono, "    04/2016   ", new Date(2012, 7, 10), (result) => {
+        expect(result.index).toBe(4);
+        expect(result.text).toBe("04/2016");
+    });
+});
+
 test("Test - Single Expression", function () {
     testSingleCase(chrono, "The event is going ahead (04/2016)", new Date(2012, 7, 10), (result) => {
         expect(result.start).not.toBeNull();

diff --git a/test/en/en_time_exp.test.ts b/test/en/en_time_exp.test.ts
@@ -7,6 +7,11 @@ test("Test - Parsing text offset", function () {
         expect(result.index).toBe(2);
         expect(result.text).toBe("11 AM");
     });
+
+    testSingleCase(chrono, "2020 at  11 AM ", new Date(2016, 10 - 1, 1, 8), (result, text) => {
+        expect(result.index).toBe(5);
+        expect(result.text).toBe("at  11 AM");
+    });
 });
 
 test("Test - Time expression", function () {

diff --git a/test/fr/fr_casual.test.ts b/test/fr/fr_casual.test.ts
@@ -162,18 +162,18 @@ test("Test - Combined Expression", function () {
         expect(result.start).toBeDate(new Date(2012, 7, 11, 17));
     });
 
-    // testSingleCase(chrono.fr, "La deadline est demain matin 11h", new Date(2012, 7, 10, 12), (result) => {
-    //     expect(result.index).toBe(16);
-    //     expect(result.text).toBe("demain matin 11h");
-    //
-    //     expect(result.start).not.toBeNull();
-    //     expect(result.start.get("year")).toBe(2012);
-    //     expect(result.start.get("month")).toBe(8);
-    //     expect(result.start.get("day")).toBe(11);
-    //     expect(result.start.get("hour")).toBe(11);
-    //
-    //     expect(result.start).toBeDate(new Date(2012, 7, 11, 11));
-    // });
+    testSingleCase(chrono.fr, "La deadline est demain matin 11h", new Date(2012, 7, 10, 12), (result) => {
+        expect(result.index).toBe(16);
+        expect(result.text).toBe("demain matin 11h");
+
+        expect(result.start).not.toBeNull();
+        expect(result.start.get("year")).toBe(2012);
+        expect(result.start.get("month")).toBe(8);
+        expect(result.start.get("day")).toBe(11);
+        expect(result.start.get("hour")).toBe(11);
+
+        expect(result.start).toBeDate(new Date(2012, 7, 11, 11));
+    });
 });
 
 //