From 237d861c147fd6027b68c910425ebb361dd7b0e2 Mon Sep 17 00:00:00 2001
From: Michael Williamson <mike@zwobble.org>
Date: Sat, 30 Nov 2024 14:03:35 +0000
Subject: [PATCH] Read whether complex field checkbox is checked

---
 mammoth/docx/body_xml.py     | 17 ++++++-
 tests/docx/body_xml_tests.py | 93 ++++++++++++++++++++++++++++++++++++
 2 files changed, 108 insertions(+), 2 deletions(-)

diff --git a/mammoth/docx/body_xml.py b/mammoth/docx/body_xml.py
index 5e891fd5..134917fb 100644
--- a/mammoth/docx/body_xml.py
+++ b/mammoth/docx/body_xml.py
@@ -134,7 +134,10 @@ def _read_run_style(properties):
         return _read_style(properties, "w:rStyle", "Run", styles.find_character_style_by_id)
 
     def read_boolean_element(element):
-        return element and element.attributes.get("w:val") not in ["false", "0"]
+        if element is None:
+            return False
+        else:
+            return element.attributes.get("w:val") not in ["false", "0"]
 
     def read_underline_element(element):
         return element and element.attributes.get("w:val") not in [None, "false", "0", "none"]
@@ -231,7 +234,17 @@ def parse_instr_text(instr_text, *, fld_char):
 
         checkbox_result = re.match(r'\s*FORMCHECKBOX\s*', instr_text)
         if checkbox_result is not None:
-            return complex_fields.checkbox(checked=False)
+            checkbox_element = fld_char \
+                .find_child_or_null("w:ffData") \
+                .find_child_or_null("w:checkBox")
+            checked_element = checkbox_element.find_child("w:checked")
+
+            if checked_element is None:
+                checked = read_boolean_element(checkbox_element.find_child("w:default"))
+            else:
+                checked = read_boolean_element(checked_element)
+
+            return complex_fields.checkbox(checked=checked)
 
         return None
 
diff --git a/tests/docx/body_xml_tests.py b/tests/docx/body_xml_tests.py
index cb5a9b5b..8c1fdf08 100644
--- a/tests/docx/body_xml_tests.py
+++ b/tests/docx/body_xml_tests.py
@@ -676,6 +676,99 @@ def test_complex_field_checkbox_with_separate_is_read(self):
             is_run(children=is_sequence(is_checkbox())),
         )))
 
+    def test_complex_field_checkbox_without_default_nor_checked_is_unchecked(self):
+        element = self._complex_field_checkbox_paragraph([
+            xml_element("w:checkBox"),
+        ])
+
+        paragraph = _read_and_get_document_xml_element(element);
+
+        assert_that(paragraph, is_paragraph(children=is_sequence(
+            is_empty_run,
+            is_empty_run,
+            is_run(children=is_sequence(is_checkbox(checked=False))),
+        )))
+
+    def test_complex_field_checkbox_with_default_0_and_without_checked_is_unchecked(self):
+        element = self._complex_field_checkbox_paragraph([
+            xml_element("w:checkBox", {}, [
+                xml_element("w:default", {"w:val": "0"}),
+            ]),
+        ])
+
+        paragraph = _read_and_get_document_xml_element(element);
+
+        assert_that(paragraph, is_paragraph(children=is_sequence(
+            is_empty_run,
+            is_empty_run,
+            is_run(children=is_sequence(is_checkbox(checked=False))),
+        )))
+
+    def test_complex_field_checkbox_with_default_1_and_without_checked_is_checked(self):
+        element = self._complex_field_checkbox_paragraph([
+            xml_element("w:checkBox", {}, [
+                xml_element("w:default", {"w:val": "1"}),
+            ]),
+        ])
+
+        paragraph = _read_and_get_document_xml_element(element);
+
+        assert_that(paragraph, is_paragraph(children=is_sequence(
+            is_empty_run,
+            is_empty_run,
+            is_run(children=is_sequence(is_checkbox(checked=True))),
+        )))
+
+    def test_complex_field_checkbox_with_default_1_and_checked_0_is_unchecked(self):
+        element = self._complex_field_checkbox_paragraph([
+            xml_element("w:checkBox", {}, [
+                xml_element("w:default", {"w:val": "1"}),
+                xml_element("w:checked", {"w:val": "0"}),
+            ]),
+        ])
+
+        paragraph = _read_and_get_document_xml_element(element);
+
+        assert_that(paragraph, is_paragraph(children=is_sequence(
+            is_empty_run,
+            is_empty_run,
+            is_run(children=is_sequence(is_checkbox(checked=False))),
+        )))
+
+    def test_complex_field_checkbox_with_default_0_and_checked_1_is_checked(self):
+        element = self._complex_field_checkbox_paragraph([
+            xml_element("w:checkBox", {}, [
+                xml_element("w:default", {"w:val": "0"}),
+                xml_element("w:checked", {"w:val": "1"}),
+            ]),
+        ])
+
+        paragraph = _read_and_get_document_xml_element(element);
+
+        assert_that(paragraph, is_paragraph(children=is_sequence(
+            is_empty_run,
+            is_empty_run,
+            is_run(children=is_sequence(is_checkbox(checked=True))),
+        )))
+
+    def _complex_field_checkbox_paragraph(self, ff_data_children):
+        return xml_element("w:p", {}, [
+            xml_element("w:r", {}, [
+                xml_element("w:fldChar", {"w:fldCharType": "begin"}, [
+                    xml_element("w:ffData", {}, ff_data_children)
+                ]),
+            ]),
+            xml_element("w:instrText", {}, [
+                xml_text(' FORMCHECKBOX ')
+            ]),
+            xml_element("w:r", {}, [
+                xml_element("w:fldChar", {"w:fldCharType": "separate"})
+            ]),
+            xml_element("w:r", {}, [
+                xml_element("w:fldChar", {"w:fldCharType": "end"})
+            ]),
+        ])
+
 
 def test_can_read_tab_element():
     element = xml_element("w:tab")