round-4 bisection: no-validate / shared-hf-parts / mini-28fields probes

tests/make_bisect_files.py

@@ -16,6 +16,10 @@ feature family:
   05 first8       - body truncated after the 8th section boundary
   06 no_images    - drawing paragraphs removed
 
+  19 no_validate    - +w:doNotValidateAgainstSchema in settings.xml
+  20 shared_hf_parts - every section points at section 1's 6 header/footer parts
+  21 mini_28fields  - the MINI book with the TOC entry replicated to 28 fields
+
 Open each in Word; whichever show the error dialog pinpoint the feature
 family at fault.
 """
@@ -54,6 +58,9 @@ NAMES = [
     'bisect_16_no_titles.docx',
     'bisect_17_page_fields.docx',
     'bisect_18_no_dirty.docx',
+    'bisect_19_no_validate.docx',
+    'bisect_20_shared_hf_parts.docx',
+    'bisect_21_mini_28fields.docx',
 ]
 
 
@@ -285,6 +292,72 @@ def make_no_dirty(doc):
                 rpr.remove(d)
 
 
+def make_no_validate(doc):
+    """Add <w:doNotValidateAgainstSchema/> to settings: Word skips schema
+    validation on open. If the full book then opens, the failure is a
+    schema-level violation hit during the dirty-field update/repagination
+    pass, not a content issue."""
+    settings = doc.settings.element
+    el = etree.Element(qn('w:doNotValidateAgainstSchema'))
+    # canonical CT_Settings order: ... savePreviewPicture,
+    # doNotValidateAgainstSchema, saveInvalidXml, ... compat, rsids, mathPr
+    anchor = None
+    for tag in ('w:compat', 'w:rsids', 'w:mathPr'):
+        anchor = settings.find(qn(tag))
+        if anchor is not None:
+            break
+    if anchor is not None:
+        anchor.addprevious(el)
+    else:
+        settings.append(el)
+
+
+def make_shared_hf_parts(doc):
+    """Point every section at the first section's 6 header/footer parts, so
+    the 216 part *package* collapses to 6 referenced parts while keeping the
+    36-section body structure and dirty fields."""
+    first = None
+    for sectPr in doc.element.body.iter(qn('w:sectPr')):
+        if sectPr.getparent().tag == qn('w:pPr'):
+            first = sectPr
+            break
+    if first is None:
+        return
+    mapping = {}
+    for ref in first:
+        if ref.tag in (qn('w:headerReference'), qn('w:footerReference')):
+            mapping[(ref.tag, ref.get(qn('w:type')))] = ref.get(qn('r:id'))
+    for sectPr in doc.element.body.iter(qn('w:sectPr')):
+        if sectPr is first:
+            continue
+        for ref in list(sectPr):
+            if ref.tag in (qn('w:headerReference'), qn('w:footerReference')):
+                key = (ref.tag, ref.get(qn('w:type')))
+                if key in mapping:
+                    ref.set(qn('r:id'), mapping[key])
+
+
+def make_mini_28fields(doc):
+    """Return a copy of the MINI book (which opens in Word with 3 dirty
+    PAGEREF fields) with the first TOC entry paragraph replicated 25 more
+    times -> 28 dirty fields in the mini shell. Tests pure field density:
+    if this fails while the 3-field mini opens, the trigger is field count
+    interacting with something, not the full shell."""
+    mini = Document(os.path.join(HERE, 'word_test_output.docx'))
+    body = mini.element.body
+    first_entry = None
+    for p in body.iter(qn('w:p')):
+        if p.find(".//" + qn('w:instrText')) is not None:
+            first_entry = p
+            break
+    if first_entry is None:
+        return mini
+    idx = list(body).index(first_entry)
+    for k in range(25):
+        body.insert(idx + 1 + k, deepcopy(first_entry))
+    return mini
+
+
 MAKERS = [
     make_no_running,
     make_no_oddpage,
@@ -304,6 +377,9 @@ MAKERS = [
     make_no_titles,
     make_page_fields,
     make_no_dirty,
+    make_no_validate,
+    make_shared_hf_parts,
+    make_mini_28fields,
 ]
 
 
@@ -314,7 +390,7 @@ def main():
                  "Regenerate it via the API first, or pass its path as argv[1].")
     for maker, name in zip(MAKERS, NAMES):
         doc = Document(base)
-        maker(doc)
+        doc = maker(doc) or doc
         path = _save(doc, name)
         # sanity: it must still open via python-docx
         Document(path)