Merge pull request #228 from advarisk/fix-select-option-values

fix html.SelectElement stripping whitespace from <option> values

2 files changed, 31 insertions, 20 deletions

diff --git a/src/lxml/html/__init__.py b/src/lxml/html/__init__.py
index b0ab2c1e..c0297d67 100644
--- a/src/lxml/html/__init__.py
+++ b/src/lxml/html/__init__.py
@@ -1327,9 +1327,7 @@ class SelectElement(InputMixin, HtmlElement):
             if el.get('selected') is not None:
                 value = el.get('value')
                 if value is None:
-                    value = el.text or ''
-                if value:
-                    value = value.strip()
+                    value = (el.text or '').strip()
                 return value
         return None
 
@@ -1344,13 +1342,10 @@ class SelectElement(InputMixin, HtmlElement):
             return
         checked_option = None
         if value is not None:
-            value = value.strip()
             for el in _options_xpath(self):
                 opt_value = el.get('value')
                 if opt_value is None:
-                    opt_value = el.text or ''
-                if opt_value:
-                    opt_value = opt_value.strip()
+                    opt_value = (el.text or '').strip()
                 if opt_value == value:
                     checked_option = el
                     break
@@ -1381,9 +1376,7 @@ class SelectElement(InputMixin, HtmlElement):
         for el in _options_xpath(self):
             value = el.get('value')
             if value is None:
-                value = el.text or ''
-            if value:
-                value = value.strip()
+                value = (el.text or '').strip()
             options.append(value)
         return options
 
@@ -1428,18 +1421,14 @@ class MultipleSelectOptions(SetMixin):
             if 'selected' in option.attrib:
                 opt_value = option.get('value')
                 if opt_value is None:
-                    opt_value = option.text or ''
-                if opt_value:
-                    opt_value = opt_value.strip()
+                    opt_value = (option.text or '').strip()
                 yield opt_value
 
     def add(self, item):
         for option in self.options:
             opt_value = option.get('value')
             if opt_value is None:
-                opt_value = option.text or ''
-            if opt_value:
-                opt_value = opt_value.strip()
+                opt_value = (option.text or '').strip()
             if opt_value == item:
                 option.set('selected', '')
                 break
@@ -1451,9 +1440,7 @@ class MultipleSelectOptions(SetMixin):
         for option in self.options:
             opt_value = option.get('value')
             if opt_value is None:
-                opt_value = option.text or ''
-            if opt_value:
-                opt_value = opt_value.strip()
+                opt_value = (option.text or '').strip()
             if opt_value == item:
                 if 'selected' in option.attrib:
                     del option.attrib['selected']
diff --git a/src/lxml/html/tests/test_forms.txt b/src/lxml/html/tests/test_forms.txt
index 25773013..e475587b 100644
--- a/src/lxml/html/tests/test_forms.txt
+++ b/src/lxml/html/tests/test_forms.txt
@@ -28,6 +28,14 @@
 ...     <option value="3">number 3</option>
 ...     <option>number 4</option>
 ...   </select>
+...   <select name="select3">
+...     <option value="01 " selected>text 1</option>
+...     <option value=" 02">text 2</option>
+...   </select>
+...   <select name="select4" multiple>
+...     <option value="01 " selected>text 1</option>
+...     <option value=" 02">text 2</option>
+...   </select>
 ...   <input type="file" name="file_field" value="nonsense_value">
 ...   <input type="submit" name="submit1" value="submit">
 ...   <input type="submit" name="submit2" value="submit">
@@ -133,10 +141,24 @@ ValueError: There is no option with the value 'asdf'
 >>> select.value.remove('number 4')
 >>> select.value_options
 ['1', '2', '3', 'number 4']
+>>> select = f.inputs['select3']
+>>> select.value
+'01 '
+>>> select.value_options
+['01 ', ' 02']
+>>> select.value = " 02"
+>>> select.value
+' 02'
+>>> select = f.inputs['select4']
+>>> select.value # doctest:+NOPARSE_MARKUP
+<MultipleSelectOptions {'01 '} for select name='select4'>
+>>> select.value.add(' 02')
+>>> select.value # doctest:+NOPARSE_MARKUP
+<MultipleSelectOptions {'01 ', ' 02'} for select name='select4'>
 >>> try: from urllib import urlencode
 ... except ImportError: from urllib.parse import urlencode
 >>> print(urlencode(f.form_values()))
-hidden_field=new+value&text_field=text_value&single_checkbox=on&single_checkbox2=good&check_group=1&check_group=2&check_group=3&textarea_field=some+text&select1=No+value&select2=2
+hidden_field=new+value&text_field=text_value&single_checkbox=on&single_checkbox2=good&check_group=1&check_group=2&check_group=3&textarea_field=some+text&select1=No+value&select2=2&select3=+02&select4=01+&select4=+02
 >>> fields = f.fields
 >>> fields # doctest:+NOPARSE_MARKUP
 <FieldsDict for form 0>
@@ -149,6 +171,8 @@ radios: None
 reset1: None
 select1: 'No value'
 select2: <MultipleSelectOptions {'2'} for select name='select2'>
+select3: ' 02'
+select4: <MultipleSelectOptions {'01 ', ' 02'} for select name='select4'>
 single_checkbox: 'on'
 single_checkbox2: 'good'
 submit1: 'submit'