JA: use prefer for ambiguous eras

savoirfairelinux · Jun 15, 2018 · d96b828 · d96b828
1 parent 15426ae
commit d96b828
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 9 deletions.
diff --git a/README.rst b/README.rst
@@ -128,6 +128,10 @@ prefer those in the sequence ``prefer``::
     ななじゅうよん
     >>> num2words(74, lang='ja', reading=True, prefer=['し', 'しち'])
     しちじゅうし
+    >>> num2words(1375, lang='ja', to="year")
+    天授元年
+    >>> num2words(1375, lang='ja', to="year", prefer=['えいわ'])
+    永和元年
 
 **era:** (``year`` only) whether or not to convert the year to the era
 calendar format. Defaults to ``True``::

diff --git a/num2words/lang_JA.py b/num2words/lang_JA.py
@@ -98,7 +98,7 @@ def rendaku_merge_pairs(lpair, rpair):
     (717, ("養老", "ようろう")),
     (724, ("神亀", "じんき")),
     (729, ("天平", "てんぴょう")),
-    # (749, ("天平感宝", "てんぴょうかんぽう")),
+    (749, ("天平感宝", "てんぴょうかんぽう")),
     (749, ("天平勝宝", "てんぴょうしょうほう")),
     (757, ("天平宝字", "てんぴょうじょうじ")),
     (765, ("天平神護", "てんぴょうじんご")),
@@ -205,15 +205,15 @@ def rendaku_merge_pairs(lpair, rpair):
     (1214, ("建保", "けんぽう")),
     (1219, ("承久", "じょうきゅう")),
     (1222, ("貞応", "じょうおう")),
-    # (1225, ("元仁", "げんにん")),
+    (1225, ("元仁", "げんにん")),
     (1225, ("嘉禄", "かろく")),
     (1228, ("安貞", "あんてい")),
     (1229, ("寛喜", "かんき")),
     (1232, ("貞永", "じょうえい")),
     (1233, ("天福", "てんぷく")),
-    # (1235, ("文暦", "ぶんりゃく")),
+    (1235, ("文暦", "ぶんりゃく")),
     (1235, ("嘉禎", "かてい")),
-    # (1239, ("暦仁", "りゃくにん")),
+    (1239, ("暦仁", "りゃくにん")),
     (1239, ("延応", "えんおう")),
     (1240, ("仁治", "にんじ")),
     (1243, ("寛元", "かんげん")),
@@ -230,7 +230,7 @@ def rendaku_merge_pairs(lpair, rpair):
     (1288, ("正応", "しょうおう")),
     (1293, ("永仁", "えいにん")),
     (1299, ("正安", "しょうあん")),
-    # (1303, ("乾元", "けんげん")),
+    (1303, ("乾元", "けんげん")),
     (1303, ("嘉元", "かげん")),
     (1307, ("徳治", "とくじ")),
     (1308, ("延慶", "えんきょう")),
@@ -259,12 +259,12 @@ def rendaku_merge_pairs(lpair, rpair):
     (1368, ("応安", "おうあん")),
     (1370, ("建徳", "けんとく")),
     (1372, ("文中", "ぶんちゅう")),
-    # (1375, ("永和", "えいわ")),
+    (1375, ("永和", "えいわ")),
     (1375, ("天授", "てんじゅ")),
     (1379, ("康暦", "こうりゃく")),
-    # (1381, ("永徳", "えいとく")),
+    (1381, ("永徳", "えいとく")),
     (1381, ("弘和", "こうわ")),
-    # (1384, ("至徳", "しとく")),
+    (1384, ("至徳", "しとく")),
     (1384, ("元中", "げんちゅう")),
     (1387, ("嘉慶", "かけい")),
     (1389, ("康応", "こうおう")),
@@ -437,7 +437,17 @@ def to_year(self, val, suffix=None, longval=True, reading=False,
             if mid == last_era_idx or (ERA_START[mid][0] <= year and
                                        ERA_START[mid + 1][0] > year):
                 era_idx = mid
-
+                # if an era lasting less than a year is preferred, choose it
+                if prefer:
+                    i = mid - 1
+                    while i >= 0 and ERA_START[i][0] == year:
+                        # match kanji or hiragana
+                        if set(ERA_START[i][1]) & set(prefer):
+                            era_idx = i
+                            break
+                        i -= 1
+
+            # ends up at the last index where year >= ERA_START[mid][0]
             if year < ERA_START[mid][0]:
                 last = mid - 1
             else:

diff --git a/tests/test_ja.py b/tests/test_ja.py
@@ -180,3 +180,5 @@ def test_year(self):
         self.assertEqual(n2j(-99, to="year", era=False), "紀元前九十九年")
         self.assertEqual(n2j(-99, to="year", era=False, reading=True),
                          "きげんぜんきゅうじゅうくねん")
+        self.assertEqual(n2j(1375, to="year"), "天授元年")
+        self.assertEqual(n2j(1375, to="year", prefer=["えいわ"]), "永和元年")