feat(docx): give a word document the page it was written on

src/odr/internal/oldms/text/AGENTS.md

@@ -193,6 +193,11 @@ WordDocument stream
   (`PlcBtePapx` → `PapxFkp`) belongs here too.
 - **Fields show only the cached result** (§2.8.25) — never evaluated. Acceptable
   for "visible text".
+- **No page layout** — `text_root_page_layout` returns `{}`, so the paginated
+  HTML renders a page box that hugs the text instead of a sheet of paper with
+  margins. Section properties (`PlcfSed` → `Sepx`, `sprmSXaPage`/`sprmSYaPage`
+  for the size, `sprmSDxaLeft`/`Right`, `sprmSDyaTop`/`sprmSDyaBottom` for the
+  margins, §2.6.4) are unparsed.
 - **Images / OLE / drawn objects** — anchor chars dropped; would need `PlcfSpa` /
   Office Art (`dggInfo`).
 - **Encrypted / obfuscated** — `fEncrypted`/`fObfuscated` parsed but not acted on;

src/odr/internal/ooxml/text/AGENTS.md

@@ -38,6 +38,13 @@ interleaved list. `w:lvlText` is the template, `%N` naming a level's counter;
 the shared expansion and the number formats are in `common/list_numbering.*`,
 where ODF lowers to the same shape.
 
+**Page layout comes from the first `w:sectPr` in document order.** The model
+carries one `PageLayout` per text root, while Word carries one per section —
+and `w:body/w:sectPr` describes the *last* section, a section being closed by
+the `w:pPr/w:sectPr` of its final paragraph. Taking the first one in document
+order therefore gives the layout the document opens with, and collapses to the
+body's own for the single-section documents that are the norm.
+
 **Style resolution mixes a static hierarchy with a runtime cascade.**
 `StyleRegistry` indexes `w:style` by `w:styleId` and pre-flattens the
 `w:basedOn` chain: each `Style` recursively resolves its parent, copies the

src/odr/internal/ooxml/text/README.md

@@ -60,7 +60,8 @@ Roughly ordered by importance.
   - [x] cell vertical alignment, borders
   - [ ] cell width (parsed but not applied)
   - [ ] table row styles
-- [x] page layout (via master page)
+- [x] page layout (`w:sectPr`: size, orientation, margins)
+  - [ ] one layout per section; the first section's applies to the document
 - [ ] graphic / drawing styles
 
 ## References

src/odr/internal/ooxml/text/ooxml_text_document.cpp

@@ -23,8 +23,45 @@ namespace odr::internal::ooxml::text {
 namespace {
 std::unique_ptr<abstract::ElementAdapter>
 create_element_adapter(const Document &document, ElementRegistry &registry);
+
+/// `w:top` and `w:bottom` are signed: a negative margin lets the header flow
+/// into the body, which as a CSS margin would push the text off the page.
+std::optional<Measure> read_page_margin(const pugi::xml_attribute attribute) {
+  const std::optional<Measure> margin = read_twips_attribute(attribute);
+  if (margin.has_value() && margin->magnitude() < 0) {
+    return Measure(0, margin->unit());
+  }
+  return margin;
 }
 
+/// [ECMA-376] 17.6.17 `w:sectPr`. The body's own are the *last* section's, so
+/// the first in document order is the one the document opens with.
+PageLayout read_page_layout(const pugi::xml_node body) {
+  const pugi::xml_node section_properties =
+      body.select_node(".//w:sectPr").node();
+
+  PageLayout result;
+
+  const pugi::xml_node page_size = section_properties.child("w:pgSz");
+  result.width = read_twips_attribute(page_size.attribute("w:w"));
+  result.height = read_twips_attribute(page_size.attribute("w:h"));
+  if (const pugi::xml_attribute orientation = page_size.attribute("w:orient")) {
+    result.print_orientation =
+        std::strcmp("landscape", orientation.value()) == 0
+            ? PrintOrientation::landscape
+            : PrintOrientation::portrait;
+  }
+
+  const pugi::xml_node page_margin = section_properties.child("w:pgMar");
+  result.margin.right = read_page_margin(page_margin.attribute("w:right"));
+  result.margin.top = read_page_margin(page_margin.attribute("w:top"));
+  result.margin.left = read_page_margin(page_margin.attribute("w:left"));
+  result.margin.bottom = read_page_margin(page_margin.attribute("w:bottom"));
+
+  return result;
+}
+} // namespace
+
 Document::Document(std::shared_ptr<abstract::ReadableFilesystem> files)
     : internal::Document(FileType::office_open_xml_document, DocumentType::text,
                          std::move(files)) {
@@ -40,6 +77,9 @@ Document::Document(std::shared_ptr<abstract::ReadableFilesystem> files)
   m_document_relations =
       parse_relationships(*m_files, AbsPath("/word/document.xml"));
 
+  m_page_layout =
+      read_page_layout(m_document_xml.document_element().child("w:body"));
+
   m_root_element = parse_tree(
       m_element_registry, m_document_xml.document_element().child("w:body"));
 
@@ -69,6 +109,8 @@ const Relations &Document::document_relations() const {
   return m_document_relations;
 }
 
+const PageLayout &Document::page_layout() const { return m_page_layout; }
+
 bool Document::is_editable() const noexcept { return true; }
 
 bool Document::is_savable(const bool encrypted) const noexcept {
@@ -242,7 +284,7 @@ class ElementAdapter final : public abstract::ElementAdapter,
 
   [[nodiscard]] PageLayout text_root_page_layout(
       [[maybe_unused]] const ElementIdentifier element_id) const override {
-    return {};
+    return m_document->page_layout();
   }
   [[nodiscard]] ElementIdentifier text_root_first_master_page(
       [[maybe_unused]] const ElementIdentifier element_id) const override {

src/odr/internal/ooxml/text/ooxml_text_document.hpp

@@ -1,5 +1,7 @@
 #pragma once
 
+#include <odr/style.hpp>
+
 #include <odr/internal/common/document.hpp>
 #include <odr/internal/ooxml/ooxml_util.hpp>
 #include <odr/internal/ooxml/text/ooxml_text_element_registry.hpp>
@@ -23,6 +25,7 @@ class Document final : public internal::Document {
   [[nodiscard]] const ElementRegistry &element_registry() const;
   [[nodiscard]] const StyleRegistry &style_registry() const;
   [[nodiscard]] const Relations &document_relations() const;
+  [[nodiscard]] const PageLayout &page_layout() const;
 
   [[nodiscard]] bool is_editable() const noexcept override;
   [[nodiscard]] bool is_savable(bool encrypted) const noexcept override;
@@ -37,6 +40,8 @@ class Document final : public internal::Document {
 
   Relations m_document_relations;
 
+  PageLayout m_page_layout;
+
   ElementRegistry m_element_registry;
   StyleRegistry m_style_registry;
   NumberingRegistry m_numbering_registry;

test/data.cmake

@@ -17,9 +17,9 @@ odr_test_data(
 odr_test_data(
         PATH "reference-output/odr-public"
         URL "https://github.com/opendocument-app/OpenDocument.test.output.git"
-        REVISION "0a1114a9a50417db31cb91193b356cf56b745b8e")
+        REVISION "40c457831835a8e9437bfe6494cb96b4c6159fab")
 
 odr_test_data(
         PATH "reference-output/odr-private"
         URL "https://github.com/opendocument-app/OpenDocument.test-private.output.git"
-        REVISION "20a86c3be30a879ff6c4be21a246c4a17f95ff2e")
+        REVISION "8f7383f64c795602793a8c20a9c76bea197ad9c1")

test/src/document_test.cpp

@@ -108,6 +108,26 @@ TEST(Document, odt) {
   EXPECT_EQ(Measure("0.7874in"), page_layout.margin.top);
 }
 
+TEST(Document, docx_page_layout) {
+  const Logger logger = Logger::create_stdio("odr-test", LogLevel::verbose);
+
+  const DocumentFile document_file(
+      TestData::test_file_path("odr-public/docx/sample3.docx"), logger);
+
+  const Document document = document_file.document();
+
+  const PageLayout page_layout =
+      document.root_element().as_text_root().page_layout();
+  // Two sections; the 0.5in bottom margin is the first one's, the body's own
+  // `w:sectPr` says 1in.
+  EXPECT_EQ(Measure("8.5in"), page_layout.width);
+  EXPECT_EQ(Measure("11in"), page_layout.height);
+  EXPECT_EQ(Measure("1in"), page_layout.margin.top);
+  EXPECT_EQ(Measure("1in"), page_layout.margin.right);
+  EXPECT_EQ(Measure("0.5in"), page_layout.margin.bottom);
+  EXPECT_EQ(Measure("1in"), page_layout.margin.left);
+}
+
 TEST(Document, xlsx_sheet_names) {
   const Logger logger = Logger::create_stdio("odr-test", LogLevel::verbose);