修复有的词获取不到 defination 的问题 (#75)

* 修复有的词获取不到 defination 的问题

* fix lost spaces in sentences

addon/queryApi/eudict.py

@@ -16,11 +16,29 @@ def __init__(self, html, term):
 
     @property
     def definition(self) -> list:
-        els = self._soap.select('div #ExpFCChild li')  # 多词性
-        els = self._soap.select('div #ExpFCChild .exp') if not els else els  # 单一词性
         ret = []
+        div = self._soap.select('div #ExpFCChild')
+        if not div:
+            return ret
+
+        div = div[0]
+        els = div.select('li') # 多词性
+        if not els: # 单一词性
+            els = div.select('.exp')
+        if not els: # 还有一奇怪的情况，不在任何的标签里面
+            trans = div.find(id='trans')
+            trans.replace_with('') if trans else ''
+
+            script = div.find('script')
+            script.replace_with('') if script else ''
+
+            for atag in div.find_all('a'): # 赞踩这些字样
+                atag.replace_with('')
+            els = [div]
+
         for el in els:
             ret.append(el.get_text(strip=True))
+
         return ret
 
     @property
@@ -88,7 +106,7 @@ def sentence(self) -> list:
         for el in els:
             try:
                 line = el.select('p')
-                sentence = line[0].get_text(strip=True)
+                sentence = "".join([ str(c) for c in line[0].contents])
                 sentence_translation = line[1].get_text(strip=True)
                 ret.append((sentence, sentence_translation))
             except KeyError as e:
@@ -152,7 +170,7 @@ def query(cls, word) -> dict:
         queryResult = None
         try:
             rsp = cls.session.get(cls.url.format(word), timeout=cls.timeout)
-            logger.debug(f'code:{rsp.status_code}- word:{word} text:{rsp.text}')
+            logger.debug(f'code:{rsp.status_code}- word:{word} text:{rsp.text[:100]}')
             queryResult = cls.parser(rsp.text, word).result
         except Exception as e:
             logger.exception(e)

test/test_queryapi.py

@@ -25,7 +25,20 @@ def test_eudict_with_none():
     ret = check_result(res)
     assert set(ret) - set(keys) == set()
 
-def test_eudict_implication_all():
+def test_eudict_implication():
+    # 不包含图片，定义不在正常规则内，包含 trans
     res = api.query('implication')
     ret = check_result(res)
     assert set(ret) - set(['image']) == set()
+
+def test_eudict_epitomize():
+    # 不包含图片，定义不在正常规则内
+    res = api.query('epitomize')
+    ret = check_result(res)
+    assert set(ret) - set(['image', 'phrase']) == set()
+
+def test_eudict_periodical():
+    # 包含图片，定义不在正常规则内
+    res = api.query('periodical')
+    ret = check_result(res)
+    assert set(ret) - set(['image', 'phrase']) == set()