twillでの文字エンコーディングを改善する

twill.encoding.patch

--- commands.py	2007-12-28 15:45:16.000000000 +0900
+++ commands.py	2013-02-05 22:22:42.000000000 +0900
@@ -231,8 +231,8 @@
     For explanations of these, please see the Python re module
     documentation.
     """
-    regexp = re.compile(what, _parseFindFlags(flags))
-    page = browser.get_html()
+    regexp = re.compile(unicode(what), _parseFindFlags(flags))
+    page = browser.get_html().decode(browser._browser.encoding(), 'ignore')
 
     m = regexp.search(page)
     if not m:
@@ -252,8 +252,8 @@
     
     Fail if the regular expression is on the page.
     """
-    regexp = re.compile(what, _parseFindFlags(flags))
-    page = browser.get_html()
+    regexp = re.compile(unicode(what), _parseFindFlags(flags))
+    page = browser.get_html().decode(browser._browser.encoding(), 'ignore')
 
     if regexp.search(page):
         raise TwillAssertionError("match to '%s'" % (what,))
@@ -273,7 +273,7 @@
     
     Show the HTML for the current page.
     """
-    html = browser.get_html()
+    html = browser.get_html().decode(browser._browser.encoding(), 'ignore')
     print>>OUT, html
     return html
 
@@ -447,7 +447,7 @@
     if isinstance(control, ClientForm.FileControl):
         raise TwillException('form field is for file upload; use "formfile" instead')
 
-    set_form_control_value(control, value)
+    set_form_control_value(control, value.encode(browser._browser.encoding()))
 
 fv = formvalue
 
@@ -707,8 +707,8 @@
     
     Succeed if the regular expression is in the page title.
     """
-    regexp = re.compile(what)
-    title = browser.get_title()
+    regexp = re.compile(unicode(what))
+    title = browser.get_title().decode(browser._browser.encoding(), 'ignore')
 
     print>>OUT, "title is '%s'." % (title,)
 
@@ -883,7 +883,7 @@
     else:
         print ''
     if check_html:
-        title = browser.get_title()
+        title = browser.get_title().decode(browser._browser.encoding(), 'ignore')
         print >>OUT, '\tPage title:', title
 
         forms = browser.get_all_forms()
--- browser.py	2007-12-28 15:45:16.000000000 +0900
+++ browser.py	2013-02-05 22:42:01.000000000 +0900
@@ -225,7 +225,7 @@
         forms = self.get_all_forms()
         
         for n, f in enumerate(forms):
-            print_form(n, f, OUT)
+            print_form(n, f, OUT, encoding=self._browser.encoding())
 
     def showlinks(self):
         """
@@ -233,7 +233,11 @@
         """
         print>>OUT, 'Links:\n'
         for n, link in enumerate(self._browser.links()):
-            print>>OUT, "%d. %s ==> %s" % (n, link.text, link.url,)
+            if link.text:
+                link_text = link.text.decode(self._browser.encoding(), 'ignore')
+            else:
+                link_text = ""
+            print>>OUT, "%d. %s ==> %s" % (n, link_text, link.url,)
         print>>OUT, ''
 
     def showhistory(self):
--- utils.py	2007-12-28 15:45:16.000000000 +0900
+++ utils.py	2013-02-05 23:15:05.000000000 +0900
@@ -52,7 +52,7 @@
     
     return s
 
-def print_form(n, f, OUT):
+def print_form(n, f, OUT, encoding=None):
     """
     Pretty-print the given form, assigned # n.
     """
@@ -76,11 +76,20 @@
     nonclickies = [c for c in f.controls if c not in clickies]
 
     for n, field in enumerate(f.controls):
+	if encoding is not None:
+            if isinstance(field.value, list):
+                value = "['%s']" % "', '".join(field.value).decode(encoding)
+            else:
+                value = field.value.decode(encoding)
+	else:
+	    value = field.value
         if hasattr(field, 'items'):
             items = [ i.name for i in field.items ]
-            value_displayed = "%s of %s" % (field.value, items)
+            if encoding is not None:
+                items = "['%s']" % "', '".join(items).decode(encoding)
+            value_displayed = "%s of %s" % (value, items)
         else:
-            value_displayed = "%s" % (field.value,)
+            value_displayed = "%s" % (value,)
 
         if field.is_of_kind('clickable'):
             submit_index = "%-2s" % (submit_indices[field],)
@@ -245,7 +254,7 @@
 # stuff to run 'tidy'...
 #
 
-_tidy_cmd = ["tidy", "-q", "-ashtml"]
+_tidy_cmd = ["tidy", "-q", "-ashtml", "-raw"]
 _tidy_exists = True
 
 def run_tidy(html):