From 9324d6eaafd7d2ebe495a8f9cbe4501997c40c8a Mon Sep 17 00:00:00 2001
From: NateScarlet <NateScarlet@Gmail.com>
Date: Sat, 9 Mar 2019 01:10:53 +0800
Subject: [PATCH] Handle \u3000 used in 2016

---
 2016.json         | Bin 132 -> 7244 bytes
 fetch_holidays.py |   8 ++++++--
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/2016.json b/2016.json
index adbec9de269420f7e46cdd2a4682f723ffd8156b..ba229abd2910c35007091da4f2371ad485f1d824 100644
GIT binary patch
literal 7244
zcmd^^%Sr<=6o&te8$q8z0`BZ(rWdVSUAR)L;6|#H+FC)|Qb(ngZo2SU-1`{5hg+Y(
z%b!d~r}Z+J2}8;dNJu!GoG+P^a}M+Vc84V_Vr9|<256#=J_I(o<{}RVWvtCe^(fP$
zWFJ=($)hQWFMVxzY&~krvE8bc#TmM|K^6^kxH?3ex;k8KaLlpqa^|Tuk8|IO5;C;K
z<=mlFvr;E&wSWqW)GK$rq-9~Y+Bb?*QVb#prq3tf3PwRY>)fe1IR$~O{pY5b*(s0?
zMgh6A<nvEie4cYgowwgdYdNRwduXw3D-vIu$6L9k?l60>^XM1L{!6*q8jqvx`-{@$
zZRYu9MQiz)vPI$x^3Mn7x;E!L-R4FrC@U$TlrOBVXU@+T>o8`dfznL{Wg!)mMQzGt
zBIZJ8Id*PPRJnmZX%a8cWA&jLHwTg#t<0}&euK<$+WiW>40jIV-e+0er``F1DNfrn
z1aq8r`=>cheg8}*vSfCj25~*KLJxINviij}$EjakClNu-aoSy7o8#24uaoi6*Tehc
zcpfUFiv2l}IZpjIHwh0l$7y$MW{%VDn#CNa-4WCrr`-|M9H-q8)EsAeBB*{98kH-A
N|0~thU#5}jKLE~(<b(hK

literal 132
zcmb>CQUC&_%GAUnB`XCZ14A<%h)6+VL26MkP%IkE2dl~`DJihh*Do(G*GtbY(@W0N
zugXZxOHWSKPtMOPNzE(K2O40iYiOizXaEz6H#9J|G||f_$yEXy6bm&!C9x80dMsBh
E00LMeTmS$7

diff --git a/fetch_holidays.py b/fetch_holidays.py
index 47b2e26..5739c9b 100755
--- a/fetch_holidays.py
+++ b/fetch_holidays.py
@@ -44,18 +44,23 @@ def get_paper(url):
     soup = bs4.BeautifulSoup(response.text, features='html.parser')
     container = soup.find('td', class_='b12c')
     assert container, f'Can not get paper container from url: {url}'
-    ret = container.get_text()
+    ret = container.get_text().replace('\u3000', '\n')
     assert ret, f'Can not get paper context from url: {url}'
     return ret
 
 
 def get_rules(paper: str):
     lines: list = paper.splitlines()
+    count = 0
     for i in sorted(set(lines), key=lines.index):
         match = re.match(r'[一二三四五六七八九十]、(.+?)：(.+)', i)
         if match:
+            count += 1
             yield match.groups()
 
+    if not count:
+        raise NotImplementedError(lines)
+
 
 def _cast_int(value):
     return int(value) if value else None
@@ -227,7 +232,6 @@ def fetch_holiday(year: int):
                 'name': name,
                 **j
             } for j in DescriptionParser(description).parse(year))
-
     return {
         'year': year,
         'papers': papers,