Make sure strings are XML compatible

mkoura · Sep 4, 2019 · 6bf17b3 · 6bf17b3
1 parent 9335aa7
commit 6bf17b3
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 11 deletions.
diff --git a/dump2polarion/exporters/requirements_exporter.py b/dump2polarion/exporters/requirements_exporter.py
@@ -222,12 +222,12 @@ def _requirement_element(self, parent_element, req_data):
         requirement = etree.SubElement(parent_element, "requirement", attrs)
 
         title_el = etree.SubElement(requirement, "title")
-        title_el.text = title
+        title_el.text = utils.get_unicode_str(title)
 
         description = req_data.get("description")
         if description:
             description_el = etree.SubElement(requirement, "description")
-            description_el.text = description
+            description_el.text = utils.get_unicode_str(description)
 
         self._fill_custom_fields(requirement, custom_fields)
 

diff --git a/dump2polarion/exporters/testcases_exporter.py b/dump2polarion/exporters/testcases_exporter.py
@@ -267,11 +267,11 @@ def _add_test_steps(parent, testcase_data):
             for index, step in enumerate(steps):
                 test_step = etree.SubElement(test_steps, "test-step")
                 test_step_col = etree.SubElement(test_step, "test-step-column", id="step")
-                test_step_col.text = step
+                test_step_col.text = utils.get_unicode_str(step)
 
                 test_res_col = etree.SubElement(test_step, "test-step-column", id="expectedResult")
                 try:
-                    test_res_col.text = results[index]
+                    test_res_col.text = utils.get_unicode_str(results[index])
                 except IndexError:
                     test_res_col.text = ""
         else:
@@ -365,12 +365,12 @@ def _testcase_element(self, parent_element, testcase_data):
         testcase = etree.SubElement(parent_element, "testcase", attrs)
 
         title_el = etree.SubElement(testcase, "title")
-        title_el.text = testcase_title
+        title_el.text = utils.get_unicode_str(testcase_title)
 
         description = testcase_data.get("description")
         if description:
             description_el = etree.SubElement(testcase, "description")
-            description_el.text = description
+            description_el.text = utils.get_unicode_str(description)
 
         self._add_test_steps(testcase, testcase_data)
         self._fill_custom_fields(testcase, custom_fields)

diff --git a/dump2polarion/utils.py b/dump2polarion/utils.py
@@ -10,6 +10,7 @@
 import logging
 import os
 import random
+import re
 import string
 from collections import OrderedDict
 
@@ -26,15 +27,19 @@
 logger = logging.getLogger(__name__)
 
 NO_BLANKS_PARSER = etree.XMLParser(remove_blank_text=True)
+# from https://stackoverflow.com/a/25920392
+VALID_XML_RE = re.compile("[^\u0020-\uD7FF\u0009\u000A\u000D\uE000-\uFFFD\U00010000-\U0010FFFF]+")
 
 
 def get_unicode_str(obj):
-    """Makes sure obj is a unicode string."""
+    """Makes sure obj is a valid XML unicode string."""
     if isinstance(obj, six.text_type):
-        return obj
-    if isinstance(obj, six.binary_type):
-        return obj.decode("utf-8", errors="ignore")
-    return six.text_type(obj)
+        text = obj
+    elif isinstance(obj, six.binary_type):
+        text = obj.decode("utf-8", errors="ignore")
+    else:
+        text = six.text_type(obj)
+    return VALID_XML_RE.sub("", text)
 
 
 def init_log(log_level):