fix: parsing history dates

testfiles/dict/cldr_data/collect_date_order.js

@@ -0,0 +1,59 @@
+const fs = require("fs");
+
+const DICT_PATH = "../dictionary.json";
+
+function translateLang(lang) {
+  switch (lang) {
+    case "iw": // Hebrew
+      return "he";
+    case "zh-CN": // Simplified Chinese
+      return "zh-Hans";
+    case "zh-HK":
+      return "zh-Hant-HK";
+    case "zh-TW":
+      return "zh-Hant";
+    default:
+      return lang;
+  }
+}
+
+function isMonthBeforeDay(lang) {
+  const cldrLang = translateLang(lang);
+  const dates = require(`cldr-dates-modern/main/${cldrLang}/ca-gregorian.json`);
+  const dateFields = dates.main[cldrLang].dates.calendars.gregorian;
+
+  const dateFmt = dateFields.dateFormats.short;
+  const mPos = dateFmt.indexOf("M");
+  const dPos = dateFmt.indexOf("d");
+  if (mPos < 0 || dPos < 0) throw new Error(`invalid fmt for ${lang}: ${dateFmt}`);
+  return dPos > mPos;
+}
+
+const dict = JSON.parse(fs.readFileSync(DICT_PATH));
+
+for (const [mainLang, entry] of Object.entries(dict)) {
+  const langs = [mainLang, ...entry["equivalent"]];
+  const dateOrder = entry["date_order"];
+  const mPos = dateOrder.indexOf("M");
+  const dPos = dateOrder.indexOf("D");
+  let expectMbd = mPos < 0 || dPos < 0 ? null : dPos > mPos;
+
+  if (mainLang === "en" || mainLang.startsWith("zh-")) {
+    expectMbd = true;
+  } else if (mainLang === "fr")
+    expectMbd = false;
+  else {
+    for (lang of langs) {
+      const mbd = isMonthBeforeDay(lang)
+      if (expectMbd === null) {
+        expectMbd = mbd;
+      } else if (mbd !== expectMbd) {
+        throw new Error(`unexpected mbd for ${lang}`);
+      }
+    }
+  }
+
+  dict[mainLang]["month_before_day"] = expectMbd;
+}
+
+fs.writeFileSync(DICT_PATH, JSON.stringify(dict, null, 2));

testfiles/dict/cldr_data/collect_day_names.js

@@ -17,27 +17,46 @@ function translateLang(lang) {
   }
 }
 
-function collectMonthNames(lang, by_char, monthNames, weekdayNames) {
+function collectMonthNames(lang, by_char, monthNames, monthShortNames, weekdayNames) {
   const cldrLang = translateLang(lang);
   const dates = require(`cldr-dates-modern/main/${cldrLang}/ca-gregorian.json`);
   const dateFields = dates.main[cldrLang].dates.calendars.gregorian;
 
   const months = dateFields.months["stand-alone"].wide;
 
-  for (const [n, name] of Object.entries(months)) {
-    let name2 = name.toLowerCase();
-    if (name2.includes(n)) {
-      // Some languages dont have named months
-      console.log(`${lang}: month name '${name2}' includes number; skipped`);
-      continue;
+  // Mongolian dates have the extra numbe арван and have to be handled manually
+  if (!["mn"].includes(lang)) {
+    for (const [n, name] of Object.entries(months)) {
+      let name2 = name.toLowerCase();
+      if (name2.includes(n)) {
+        // Some languages dont have named months
+        console.log(`${lang}: month name '${name2}' includes number; skipped`);
+        continue;
+      }
+      if (/\s/g.test(name2)) {
+        throw new Error(`${lang}: month name '${name2}' contains whitespace`);
+      }
+      monthNames[name2] = parseInt(n);
     }
-    if (lang === "mn") {
-      name2 = name2.replace(" сар", "").replace("арван ", "");
+  }
+
+  if (!["bg", "fi", "cs", "iw", "lt", "pt-PT", "sk"].includes(lang)) {
+    const monthsShort = dateFields.months.format.abbreviated;
+    for (const [n, name] of Object.entries(monthsShort)) {
+      let name2 = name.toLowerCase().replaceAll(".", "");
+      if (name2.includes(n)) {
+        // Some languages dont have named months
+        console.log(`${lang}: month name '${name2}' includes number; skipped`);
+        continue;
+      }
+      if (lang === "ca") {
+        name2 = name2.replace("de ", "");
+      }
+      if (/\s/g.test(name2)) {
+        throw new Error(`${lang}: month name '${name2}' contains whitespace`);
+      }
+      monthShortNames[name2] = parseInt(n);
     }
-    if (/\s/g.test(name2)) {
-      throw new Error(`${lang}: month name '${name2}' contains whitespace`);
-    }
-    monthNames[name2.toLowerCase()] = parseInt(n);
   }
 
   const weekdays = dateFields.days["stand-alone"].wide;
@@ -72,12 +91,23 @@ const dict = JSON.parse(fs.readFileSync(DICT_PATH));
 for (const [mainLang, entry] of Object.entries(dict)) {
   const langs = [mainLang, ...entry["equivalent"]];
   let monthNames = {};
+  let monthShortNames = {};
   let weekdayNames = {};
 
   for (lang of langs) {
-    collectMonthNames(lang, entry["by_char"], monthNames, weekdayNames);
+    collectMonthNames(
+      lang,
+      entry["by_char"],
+      monthNames,
+      monthShortNames,
+      weekdayNames
+    );
   }
-  dict[mainLang]["months"] = { ...dict[mainLang]["months"], ...monthNames };
+  dict[mainLang]["months"] = {
+    ...dict[mainLang]["months"],
+    ...monthNames,
+    ...monthShortNames,
+  };
   dict[mainLang]["timeago_nd_tokens"] = {
     ...dict[mainLang]["timeago_nd_tokens"],
     ...weekdayNames,