feat(xlsx): print worksheet gridlines when printOptions asks for them

crates/office2pdf/src/ir/elements.rs

@@ -535,6 +535,13 @@ pub struct Table {
     /// PowerPoint's border-painting conventions are unmeasured against
     /// their native GT, so they keep the centred-stroke path.
     pub paints_borders_inside_boundary: bool,
+    /// When true, `<printOptions gridLines="1"/>` asks Excel to print its
+    /// gridline hairline on every cell boundary of the printed range, under
+    /// any explicit border styling (issue #622). Only spreadsheet tables set
+    /// this, and it is honoured only together with
+    /// `paints_borders_inside_boundary`, whose boundary-band machinery the
+    /// gridlines reuse; Word/PowerPoint tables never print gridlines.
+    pub prints_gridlines: bool,
 }
 
 /// A table row.

crates/office2pdf/src/parser/docx_tables.rs

@@ -180,6 +180,7 @@ pub(super) fn convert_table(
         // Word GT has not verified descender seating for bottom cells (#618).
         seats_bottom_aligned_text_on_descender: false,
         paints_borders_inside_boundary: false,
+        prints_gridlines: false,
     }
 }
 

crates/office2pdf/src/parser/pptx_table_style_tests.rs

@@ -200,6 +200,7 @@ fn test_apply_table_style_first_row_gets_header_fill_and_text_color() {
         default_vertical_align: None,
         seats_bottom_aligned_text_on_descender: false,
         paints_borders_inside_boundary: false,
+        prints_gridlines: false,
     };
 
     table_styles::apply_table_style(&mut table, &props, &styles);
@@ -284,6 +285,7 @@ fn test_apply_table_style_banded_rows_skip_first_row() {
         default_vertical_align: None,
         seats_bottom_aligned_text_on_descender: false,
         paints_borders_inside_boundary: false,
+        prints_gridlines: false,
     };
 
     table_styles::apply_table_style(&mut table, &props, &styles);
@@ -358,6 +360,7 @@ fn test_apply_table_style_explicit_cell_fill_not_overridden() {
         default_vertical_align: None,
         seats_bottom_aligned_text_on_descender: false,
         paints_borders_inside_boundary: false,
+        prints_gridlines: false,
     };
 
     table_styles::apply_table_style(&mut table, &props, &styles);
@@ -403,6 +406,7 @@ fn test_apply_table_style_missing_style_id_is_noop() {
         default_vertical_align: None,
         seats_bottom_aligned_text_on_descender: false,
         paints_borders_inside_boundary: false,
+        prints_gridlines: false,
     };
 
     table_styles::apply_table_style(&mut table, &props, &styles);

crates/office2pdf/src/parser/pptx_tables.rs

@@ -408,6 +408,7 @@ impl<'a> PptxTableParser<'a> {
             // cells (#618).
             seats_bottom_aligned_text_on_descender: false,
             paints_borders_inside_boundary: false,
+            prints_gridlines: false,
         };
         table_styles::apply_table_style(&mut table, &self.table_props, self.table_styles);
         table

crates/office2pdf/src/parser/xlsx.rs

@@ -13,6 +13,8 @@ pub(crate) mod cond_fmt_raw;
 
 #[path = "xlsx_fit_to_page.rs"]
 mod fit_to_page;
+#[path = "xlsx_print_options.rs"]
+mod print_options;
 #[path = "xlsx_tables.rs"]
 mod tables;
 #[path = "xlsx_cells.rs"]
@@ -340,6 +342,7 @@ impl XlsxParser {
         let metadata = extract_xlsx_metadata(&book);
         let cond_fmt_hints = cond_fmt_raw::extract_cond_fmt_hints(data);
         let fitting_sheets = fit_to_page::sheets_fitting_to_page(data);
+        let gridline_sheets = print_options::sheets_printing_gridlines(data);
         let mut row_stripes = tables::extract_row_stripes(data);
         let normal_font = extract_normal_font(data);
 
@@ -405,6 +408,7 @@ impl XlsxParser {
             };
 
             let sheet_name = sheet.get_name().to_string();
+            let sheet_prints_gridlines: bool = gridline_sheets.contains(&sheet_name);
 
             // Extract sheet header/footer
             let hf = sheet.get_header_footer();
@@ -488,6 +492,7 @@ impl XlsxParser {
                                 default_vertical_align: Some(crate::ir::CellVerticalAlign::Bottom),
                                 seats_bottom_aligned_text_on_descender: true,
                                 paints_borders_inside_boundary: true,
+                                prints_gridlines: sheet_prints_gridlines,
                             },
                             header: sheet_header.clone(),
                             footer: sheet_footer.clone(),
@@ -551,6 +556,7 @@ impl Parser for XlsxParser {
         let metadata = extract_xlsx_metadata(&book);
         let cond_fmt_hints = cond_fmt_raw::extract_cond_fmt_hints(data);
         let fitting_sheets = fit_to_page::sheets_fitting_to_page(data);
+        let gridline_sheets = print_options::sheets_printing_gridlines(data);
         let mut row_stripes = tables::extract_row_stripes(data);
         let normal_font = extract_normal_font(data);
 
@@ -637,6 +643,7 @@ impl Parser for XlsxParser {
             // Collect row page breaks and split rows into page segments
             let row_breaks = collect_row_breaks(sheet);
             let sheet_name = sheet.get_name().to_string();
+            let sheet_prints_gridlines: bool = gridline_sheets.contains(&sheet_name);
 
             // Extract sheet header/footer
             let hf = sheet.get_header_footer();
@@ -689,6 +696,7 @@ impl Parser for XlsxParser {
                                 default_vertical_align: Some(crate::ir::CellVerticalAlign::Bottom),
                                 seats_bottom_aligned_text_on_descender: true,
                                 paints_borders_inside_boundary: true,
+                                prints_gridlines: sheet_prints_gridlines,
                             },
                             header: sheet_header.clone(),
                             footer: sheet_footer.clone(),
@@ -766,6 +774,7 @@ impl Parser for XlsxParser {
                                     ),
                                     seats_bottom_aligned_text_on_descender: true,
                                     paints_borders_inside_boundary: true,
+                                    prints_gridlines: sheet_prints_gridlines,
                                 },
                                 header: sheet_header.clone(),
                                 footer: sheet_footer.clone(),

crates/office2pdf/src/parser/xlsx_pagination.rs

@@ -292,6 +292,7 @@ fn slice_table_columns(table: &Table, start: usize, end: usize) -> Table {
         default_vertical_align: table.default_vertical_align,
         seats_bottom_aligned_text_on_descender: table.seats_bottom_aligned_text_on_descender,
         paints_borders_inside_boundary: table.paints_borders_inside_boundary,
+        prints_gridlines: table.prints_gridlines,
     }
 }
 

crates/office2pdf/src/parser/xlsx_pagination_tests.rs

@@ -67,6 +67,7 @@ fn make_page(column_widths: Vec<f64>, rows: Vec<TableRow>) -> SheetPage {
             default_vertical_align: None,
             seats_bottom_aligned_text_on_descender: false,
             paints_borders_inside_boundary: false,
+            prints_gridlines: false,
         },
         header: None,
         footer: None,

crates/office2pdf/src/parser/xlsx_print_options.rs

@@ -0,0 +1,101 @@
+use std::collections::HashSet;
+
+use quick_xml::Reader;
+use quick_xml::events::{BytesStart, Event};
+
+use super::cond_fmt_raw::{
+    parse_relationships, parse_sheet_relationships, read_zip_text, worksheet_path,
+};
+
+/// Names of the sheets whose `<printOptions gridLines="1"/>` is set.
+///
+/// The flag strictly gates printed gridlines: native Excel GT of a sheet
+/// without it contains zero gridline primitives, while a flagged sheet rules
+/// every cell boundary of the printed range (issue #622). umya-spreadsheet's
+/// `PrintOptions` models only the centering attributes, so the flag is read
+/// from the archive directly, like `sheets_fitting_to_page`.
+pub(crate) fn sheets_printing_gridlines(data: &[u8]) -> HashSet<String> {
+    let mut printing: HashSet<String> = HashSet::new();
+    let Ok(mut archive) = crate::parser::open_zip(data) else {
+        return printing;
+    };
+    let Some(workbook_xml) = read_zip_text(&mut archive, "xl/workbook.xml") else {
+        return printing;
+    };
+    let Some(relationships_xml) = read_zip_text(&mut archive, "xl/_rels/workbook.xml.rels") else {
+        return printing;
+    };
+
+    let relationships = parse_relationships(&relationships_xml);
+    for (sheet_name, relationship_id) in parse_sheet_relationships(&workbook_xml) {
+        let Some(target) = relationships.get(&relationship_id) else {
+            continue;
+        };
+        let Some(worksheet_xml) = read_zip_text(&mut archive, &worksheet_path(target)) else {
+            continue;
+        };
+        if worksheet_prints_gridlines(&worksheet_xml) {
+            printing.insert(sheet_name);
+        }
+    }
+    printing
+}
+
+/// Whether the worksheet's `<printOptions>` asks for printed gridlines.
+///
+/// Unlike `<pageSetUpPr>`, `<printOptions>` follows `<sheetData>` in
+/// CT_Worksheet, so the scan must run to the end of the document. The
+/// `<customSheetViews>` subtree is skipped: CT_Worksheet places it before the
+/// sheet-level `<printOptions>`, and each CT_CustomSheetView nests its own
+/// `<printOptions>`, which describes that saved view only — a first-match
+/// scan would read the view's options and shadow the sheet's.
+fn worksheet_prints_gridlines(worksheet_xml: &str) -> bool {
+    let mut reader = Reader::from_str(worksheet_xml);
+    // Element depth inside the skipped `<customSheetViews>` subtree; 0 means
+    // the scan is at sheet level.
+    let mut skipped_subtree_depth: usize = 0;
+    loop {
+        match reader.read_event() {
+            Ok(Event::Start(ref element)) => {
+                if skipped_subtree_depth > 0 {
+                    skipped_subtree_depth += 1;
+                } else if element.local_name().as_ref() == b"customSheetViews" {
+                    skipped_subtree_depth = 1;
+                } else if element.local_name().as_ref() == b"printOptions" {
+                    return print_options_request_gridlines(element);
+                }
+            }
+            Ok(Event::Empty(ref element)) => {
+                if skipped_subtree_depth == 0 && element.local_name().as_ref() == b"printOptions" {
+                    return print_options_request_gridlines(element);
+                }
+            }
+            Ok(Event::End(_)) => {
+                skipped_subtree_depth = skipped_subtree_depth.saturating_sub(1);
+            }
+            Ok(Event::Eof) | Err(_) => return false,
+            _ => {}
+        }
+    }
+}
+
+/// ECMA-376 §18.3.1.70: gridlines print only when `gridLines` and
+/// `gridLinesSet` (default true) are both true — an explicit
+/// `gridLinesSet="0"` vetoes `gridLines="1"`.
+fn print_options_request_gridlines(element: &BytesStart<'_>) -> bool {
+    let mut grid_lines: bool = false;
+    let mut grid_lines_set: bool = true;
+    for attribute in element.attributes().flatten() {
+        let is_on: bool = matches!(attribute.value.as_ref(), b"1" | b"true");
+        match attribute.key.local_name().as_ref() {
+            b"gridLines" => grid_lines = is_on,
+            b"gridLinesSet" => grid_lines_set = is_on,
+            _ => {}
+        }
+    }
+    grid_lines && grid_lines_set
+}
+
+#[cfg(test)]
+#[path = "xlsx_print_options_tests.rs"]
+mod tests;

crates/office2pdf/src/parser/xlsx_print_options_tests.rs

@@ -0,0 +1,95 @@
+use super::*;
+
+// `printOptions` follows `sheetData` in CT_Worksheet — the real
+// NumberFormatTests fixture writes `<printOptions headings="1"
+// gridLines="1"/>` between the sheet data and the page margins.
+const PRINTS_GRIDLINES: &str = r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+  <sheetData><row r="1"><c r="A1"><v>1</v></c></row></sheetData>
+  <printOptions headings="1" gridLines="1"/>
+  <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+</worksheet>"#;
+
+const NO_PRINT_OPTIONS: &str = r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+  <sheetData><row r="1"><c r="A1"><v>1</v></c></row></sheetData>
+  <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+</worksheet>"#;
+
+// Excel also writes `printOptions` for centering alone; that must not turn
+// gridlines on.
+const CENTERED_ONLY: &str = r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+  <sheetData/>
+  <printOptions horizontalCentered="1"/>
+</worksheet>"#;
+
+// CT_Worksheet orders `customSheetViews` before the sheet-level
+// `printOptions`, and each custom view nests its own `<printOptions>`
+// (CT_CustomSheetView). A first-match scan reads the view's options instead
+// of the sheet's.
+const CUSTOM_VIEW_THEN_SHEET_LEVEL: &str = r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+  <sheetData><row r="1"><c r="A1"><v>1</v></c></row></sheetData>
+  <customSheetViews>
+    <customSheetView guid="{3C29A897-4F3B-4A0B-9A5C-2D53E1F1F001}" scale="85">
+      <selection activeCell="A1" sqref="A1"/>
+      <printOptions horizontalCentered="1"/>
+      <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+    </customSheetView>
+  </customSheetViews>
+  <printOptions headings="1" gridLines="1"/>
+  <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+</worksheet>"#;
+
+const CUSTOM_VIEW_GRIDLINES_ONLY: &str = r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+  <sheetData><row r="1"><c r="A1"><v>1</v></c></row></sheetData>
+  <customSheetViews>
+    <customSheetView guid="{3C29A897-4F3B-4A0B-9A5C-2D53E1F1F001}" printArea="1">
+      <printOptions gridLines="1"/>
+      <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+    </customSheetView>
+  </customSheetViews>
+  <pageMargins left="0.7" right="0.7" top="0.75" bottom="0.75" header="0.3" footer="0.3"/>
+</worksheet>"#;
+
+#[test]
+fn reads_grid_lines_after_sheet_data() {
+    assert!(worksheet_prints_gridlines(PRINTS_GRIDLINES));
+}
+
+#[test]
+fn custom_view_print_options_do_not_shadow_the_sheet_level_element() {
+    assert!(worksheet_prints_gridlines(CUSTOM_VIEW_THEN_SHEET_LEVEL));
+}
+
+#[test]
+fn custom_view_grid_lines_do_not_leak_to_the_sheet() {
+    assert!(!worksheet_prints_gridlines(CUSTOM_VIEW_GRIDLINES_ONLY));
+}
+
+#[test]
+fn explicit_grid_lines_set_false_vetoes_grid_lines() {
+    // ECMA-376 §18.3.1.70: gridlines print only when `gridLines` AND
+    // `gridLinesSet` (default true) are both true.
+    let vetoed = PRINTS_GRIDLINES.replace(r#"gridLines="1""#, r#"gridLines="1" gridLinesSet="0""#);
+    assert!(!worksheet_prints_gridlines(&vetoed));
+    let vetoed_spelt = PRINTS_GRIDLINES.replace(
+        r#"gridLines="1""#,
+        r#"gridLines="true" gridLinesSet="false""#,
+    );
+    assert!(!worksheet_prints_gridlines(&vetoed_spelt));
+    let confirmed =
+        PRINTS_GRIDLINES.replace(r#"gridLines="1""#, r#"gridLines="1" gridLinesSet="1""#);
+    assert!(worksheet_prints_gridlines(&confirmed));
+}
+
+#[test]
+fn absent_or_unrelated_print_options_stay_off() {
+    assert!(!worksheet_prints_gridlines(NO_PRINT_OPTIONS));
+    assert!(!worksheet_prints_gridlines(CENTERED_ONLY));
+}
+
+#[test]
+fn accepts_the_boolean_spelt_out() {
+    let spelt = PRINTS_GRIDLINES.replace(r#"gridLines="1""#, r#"gridLines="true""#);
+    assert!(worksheet_prints_gridlines(&spelt));
+    let off = PRINTS_GRIDLINES.replace(r#"gridLines="1""#, r#"gridLines="0""#);
+    assert!(!worksheet_prints_gridlines(&off));
+}

crates/office2pdf/src/parser/xlsx_tests.rs

@@ -946,3 +946,53 @@ fn test_empty_sheet_context_derives_metric_from_normal_font() {
     assert_eq!(fallback_ctx.default_column_width_pt, 8.0 * 5.25 + 5.0);
     assert_eq!(fallback_ctx.normal_font, None);
 }
+
+/// Rewrite the workbook's worksheet parts, inserting `insertion` before each
+/// closing `</worksheet>` tag. umya's writer does not model
+/// `printOptions@gridLines`, so the attribute is injected into the archive
+/// the way Excel writes it — after `sheetData`.
+fn inject_before_worksheet_close(xlsx: &[u8], insertion: &str) -> Vec<u8> {
+    let mut archive = zip::ZipArchive::new(Cursor::new(xlsx.to_vec())).unwrap();
+    let mut writer = zip::ZipWriter::new(Cursor::new(Vec::new()));
+    for index in 0..archive.len() {
+        let mut file = archive.by_index(index).unwrap();
+        let name: String = file.name().to_string();
+        let mut contents: Vec<u8> = Vec::new();
+        std::io::Read::read_to_end(&mut file, &mut contents).unwrap();
+        if name.starts_with("xl/worksheets/") && name.ends_with(".xml") {
+            let text: String = String::from_utf8(contents).unwrap();
+            contents = text
+                .replace("</worksheet>", &format!("{insertion}</worksheet>"))
+                .into_bytes();
+        }
+        writer
+            .start_file(name, zip::write::FileOptions::default())
+            .unwrap();
+        std::io::Write::write_all(&mut writer, &contents).unwrap();
+    }
+    writer.finish().unwrap().into_inner()
+}
+
+#[test]
+fn test_print_options_grid_lines_flags_the_sheet_table() {
+    let plain = build_xlsx_bytes("Sheet1", &[("A1", "x"), ("B2", "y")]);
+    let flagged = inject_before_worksheet_close(&plain, r#"<printOptions gridLines="1"/>"#);
+
+    let parser = XlsxParser;
+    let (doc, _warnings) = parser.parse(&flagged, &ConvertOptions::default()).unwrap();
+    let table = &get_sheet_page(&doc, 0).table;
+    assert!(
+        table.prints_gridlines,
+        "printOptions gridLines must set the table's gridline flag"
+    );
+    assert!(
+        table.paints_borders_inside_boundary,
+        "the gridline flag rides on the boundary-band regime"
+    );
+
+    let (doc, _warnings) = parser.parse(&plain, &ConvertOptions::default()).unwrap();
+    assert!(
+        !get_sheet_page(&doc, 0).table.prints_gridlines,
+        "a sheet without printOptions must not print gridlines"
+    );
+}