fix: instance expressions with double quotes not escaped / converted (#…

…709) - like other functions, argument value(s) (instance name) should be allowed to be wrapped in single or double quotes. Previously only singles quotes worked.
XLSForm · Jun 18, 2024 · 59c37e0 · 59c37e0
1 parent 1599d1f
commit 59c37e0
Show file tree

Hide file tree

Showing 2 changed files with 56 additions and 10 deletions.
diff --git a/pyxform/parsing/instance_expression.py b/pyxform/parsing/instance_expression.py
@@ -1,7 +1,7 @@
 import re
 from typing import TYPE_CHECKING
 
-from pyxform.utils import BRACKETED_TAG_REGEX, EXPRESSION_LEXER, ExpLexerToken
+from pyxform.utils import BRACKETED_TAG_REGEX, EXPRESSION_LEXER, ExpLexerToken, node
 
 if TYPE_CHECKING:
     from pyxform.survey import Survey
@@ -116,7 +116,10 @@ def replace_with_output(xml_text: str, context: "SurveyElement", survey: "Survey
                 lambda m: survey._var_repl_function(m, context),
                 old_str,
             )
-            new_strings.append((start, end, old_str, f'<output value="{new_str}" />'))
+            # Generate a node so that character escapes are applied.
+            new_strings.append(
+                (start, end, old_str, node("output", value=new_str).toxml())
+            )
         # Position-based replacement avoids strings which are substrings of other
         # replacements being inserted incorrectly. Offset tracking deals with changing
         # expression positions due to incremental replacement.

diff --git a/tests/test_notes.py b/tests/test_notes.py
@@ -2,7 +2,7 @@
 Test the "note" question type.
 """
 
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 
 from tests.pyxform_test_case import PyxformTestCase
 from tests.xpath_helpers.questions import xpq
@@ -15,8 +15,8 @@ class Case:
     """
 
     label: str
-    xpath: str
     match: set[str]
+    xpath: str = field(default_factory=lambda: xpq.body_input_label_output_value("note"))
 
 
 class TestNotes(PyxformTestCase):
@@ -77,45 +77,88 @@ def test_instance_expression__permutations(self):
         |         | c2        | b    | Big   |
         |         | c2        | s    | Small |
         """
+        # It's a bit confusing, but although double quotes are literally HTML in entity
+        # form (i.e. `&quot;`) in the output, for pyxform test comparisons they get
+        # converted back, so the expected output strings are double quotes not `&quot;`.
         cases = [
             # A pyxform token.
             Case(
                 "${text}",
-                xpq.body_input_label_output_value("note"),
                 {" /test_name/text "},
             ),
             # Instance expression with predicate using pyxform token and equals.
             Case(
                 "instance('c1')/root/item[name = ${q1}]/label",
-                xpq.body_input_label_output_value("note"),
                 {"instance('c1')/root/item[name =  /test_name/q1 ]/label"},
             ),
+            # Instance expression with predicate using pyxform token and equals (double quotes).
+            Case(
+                """instance("c1")/root/item[name = ${q1}]/label""",
+                {"""instance("c1")/root/item[name =  /test_name/q1 ]/label"""},
+            ),
             # Instance expression with predicate using pyxform token and function.
             Case(
                 "instance('c2')/root/item[contains(name, ${q2})]/label",
-                xpq.body_input_label_output_value("note"),
                 {"instance('c2')/root/item[contains(name,  /test_name/q2 )]/label"},
             ),
+            # Instance expression with predicate using pyxform token and function (double quotes).
+            Case(
+                """instance("c2")/root/item[contains("name", ${q2})]/label""",
+                {"""instance("c2")/root/item[contains("name",  /test_name/q2 )]/label"""},
+            ),
+            # Instance expression with predicate using pyxform token and function (mixed quotes).
+            Case(
+                """instance('c2')/root/item[contains("name", ${q2})]/label""",
+                {"""instance('c2')/root/item[contains("name",  /test_name/q2 )]/label"""},
+            ),
             # Instance expression with predicate using pyxform token and equals.
             Case(
                 "instance('c2')/root/item[contains(name, instance('c1')/root/item[name = ${q1}]/label)]/label",
-                xpq.body_input_label_output_value("note"),
                 {
                     "instance('c2')/root/item[contains(name, instance('c1')/root/item[name =  /test_name/q1 ]/label)]/label"
                 },
             ),
+            # Instance expression with predicate using pyxform token and equals (double quotes).
+            Case(
+                """instance("c2")/root/item[contains(name, instance("c1")/root/item[name = ${q1}]/label)]/label""",
+                {
+                    """instance("c2")/root/item[contains(name, instance("c1")/root/item[name =  /test_name/q1 ]/label)]/label"""
+                },
+            ),
+            # Instance expression with predicate using pyxform token and equals (mixed quotes).
+            Case(
+                """instance('c2')/root/item[contains(name, instance("c1")/root/item[name = ${q1}]/label)]/label""",
+                {
+                    """instance('c2')/root/item[contains(name, instance("c1")/root/item[name =  /test_name/q1 ]/label)]/label"""
+                },
+            ),
             # Instance expression with predicate not using a pyxform token.
             Case(
                 "instance('c1')/root/item[name = 'y']/label",
-                xpq.body_input_label_output_value("note"),
                 {"instance('c1')/root/item[name = 'y']/label"},
             ),
+            # Instance expression with predicate not using a pyxform token (double quotes).
+            Case(
+                """instance("c1")/root/item[name = "y"]/label""",
+                {"""instance("c1")/root/item[name = "y"]/label"""},
+            ),
+            # Instance expression with predicate not using a pyxform token (mixed quotes).
+            Case(
+                """instance("c1")/root/item[name = 'y']/label""",
+                {"""instance("c1")/root/item[name = 'y']/label"""},
+            ),
+            # Instance expression with predicate not using a pyxform token (all escaped).
+            Case(
+                """instance("c1")/root/item[name <> 1 and "<>&" = "1"]/label""",
+                {
+                    """instance("c1")/root/item[name &lt;&gt; 1 and "&lt;&gt;&amp;" = "1"]/label"""
+                },
+            ),
         ]
         wrap_scenarios = ("{}", "Text {}", "{} text", "Text {} text")
         # All cases together in one.
         combo_case = Case(
             " ".join(c.label for c in cases),
-            xpq.body_input_label_output_value("note"),
             {m for c in cases for m in c.match},
         )
         cases.append(combo_case)