From 1f3fdd85cf1232c74f714e57ec19916eea2dffcb Mon Sep 17 00:00:00 2001
From: Benjamin Dauvergne <bdauvergne@entrouvert.com>
Date: Fri, 25 Mar 2016 23:25:39 +0100
Subject: [PATCH 2/2] prevent non utf-8 strings from breaking /inspect (#10447)

---
 tests/test_backoffice_pages.py | 34 +++++++++++++++++++++++++++++++---
 wcs/backoffice/management.py   | 19 +++++++++++++++++--
 2 files changed, 48 insertions(+), 5 deletions(-)

diff --git a/tests/test_backoffice_pages.py b/tests/test_backoffice_pages.py
index 2f99437..e441817 100644
--- a/tests/test_backoffice_pages.py
+++ b/tests/test_backoffice_pages.py
@@ -1,3 +1,4 @@
+# -*- coding: utf-8 -*-
 import datetime
 import json
 import os
@@ -93,7 +94,7 @@ def create_environment(pub, set_receiver=True):
             fields.ItemField(id='2', label='2nd field', type='item',
                 items=['foo', 'bar', 'baz']),
             fields.ItemField(id='3', label='3rd field', type='item',
-                data_source=datasource, in_listing=False),
+                data_source=datasource, in_listing=False, varname='foo'),
         ]
 
     formdef.store()
@@ -1991,14 +1992,41 @@ def test_backoffice_criticality_formdata_view(pub):
     resp = app.get(formdata.get_url(backoffice=True))
     assert 'Criticality Level: red' in resp.body
 
+
+class IHateUnicode(object):
+    def __unicode__(self):
+        raise Exception('HATE!!')
+
+    def __repr__(self):
+        return 'ok'
+
+
 def test_inspect_page(pub):
-    user = create_user(pub)
+    create_user(pub)
     create_environment(pub)
 
     formdef = FormDef.get_by_urlname('form-title')
     formdata = [x for x in formdef.data_class().select() if x.status == 'wf-new'][0]
+    # temper with field 3 structured values
+
+    formdata.data['3_structured'] = {
+        'unicode': u'uné',
+        'str_but_non_utf8': '\xed\xa0\x00',
+        'non_unicode_convertible': IHateUnicode(),
+    }
+    formdata.store()
 
     resp = login(get_app(pub)).get('%sinspect' % formdata.get_url(backoffice=True), status=403)
 
-    user = create_user(pub, is_admin=True)
+    create_user(pub, is_admin=True)
     resp = login(get_app(pub)).get('%sinspect' % formdata.get_url(backoffice=True), status=200)
+    pq = resp.pyquery.remove_namespaces()
+    assert (pq('[title="form_var_foo_unicode"]')
+            .parents('li').children('div.value span')
+            .text() == u'uné')
+    assert (pq('[title="form_var_foo_non_unicode_convertible"]')
+            .parents('li').children('div.value span')
+            .text().startswith('ok '))
+    assert (pq('[title="form_var_foo_str_but_non_utf8"]')
+            .parents('li').children('div.value span')
+            .text() == '\'\\xed\\xa0\\x00\'')
diff --git a/wcs/backoffice/management.py b/wcs/backoffice/management.py
index 0d4735e..081bcaa 100644
--- a/wcs/backoffice/management.py
+++ b/wcs/backoffice/management.py
@@ -1805,6 +1805,7 @@ class FormBackOfficeStatusPage(FormStatusPage):
         if not (get_publisher().get_backoffice_root().is_accessible('forms') or
                 get_publisher().get_backoffice_root().is_accessible('workflows')):
             raise errors.AccessForbiddenError()
+        charset = get_publisher().site_charset
         get_response().breadcrumb.append(('inspect', _('Form Inspector')))
         self.html_top(self.formdef.name)
         r = TemplateIO(html=True)
@@ -1814,11 +1815,25 @@ class FormBackOfficeStatusPage(FormStatusPage):
         r += htmltext(' <li><h3>%s</h3></li>') % _('Substitution variables')
         substvars = self.filled.get_substitution_variables()
         substvars.update(self.filled.formdef.get_substitution_variables())
+
+        def safe(v):
+            if isinstance(v, str):
+                try:
+                    unicode(v, charset)
+                except UnicodeDecodeError:
+                    v = repr(v)
+            else:
+                try:
+                    v = unicode(v).encode(charset)
+                except:
+                    v = repr(v)
+            return v
         for k, v in sorted(substvars.items()):
+            k = safe(k)
             r += htmltext('<li><code title="%s">%s</code>') % (k, k)
-            r += htmltext('  <div class="value"><span>%s</span>') % v
+            r += htmltext('  <div class="value"><span>%s</span>') % safe(v)
             if not isinstance(v, basestring):
-                r += htmltext(' <span class="type">(%s)</span>') % type(v)
+                r += htmltext(' <span class="type">(%r)</span>') % type(v)
             r += htmltext('</div>')
         r += htmltext('</ul>')
         return r.getvalue()
-- 
2.1.4