Skip to content

World's first open-source enterprise world model.

A sheet's merged cells fill every row and column they cover (#1039)

Coderfu9734d1806862
diff --git a/crates/utopia-ingest/src/parsers.rs b/crates/utopia-ingest/src/parsers.rs
index e3fbc828..38c8c88d 100644
--- a/crates/utopia-ingest/src/parsers.rs
+++ b/crates/utopia-ingest/src/parsers.rs
@@ -734,7 +734,7 @@ fn pptx_order(
/// xlsx / xls / ods: calamine 全格式读取,每 sheet 输出制表符表格(限前 2000 行)。
pub fn spreadsheet(bytes: &[u8]) -> anyhow::Result<String> {
- use calamine::{Data, Reader as _};
+ use calamine::Reader as _;
let mut workbook = calamine::open_workbook_auto_from_rs(Cursor::new(bytes.to_vec()))
.context("Failed to open spreadsheet")?;
let mut out = String::new();
@@ -745,27 +745,20 @@ pub fn spreadsheet(bytes: &[u8]) -> anyhow::Result<String> {
if range.is_empty() {
continue;
}
+ // xlsx/xls 把 merge 单独给出;xlsb/ods 的 calamine reader 没有这一步,按无 merge 读。
+ let merges = match &mut workbook {
+ calamine::Sheets::Xlsx(book) => book
+ .merge_cells_by_sheet_name(&sheet_name)
+ .unwrap_or_default(),
+ calamine::Sheets::Xls(book) => book
+ .merge_cells_by_sheet_name(&sheet_name)
+ .unwrap_or_default(),
+ _ => Vec::new(),
+ };
out.push_str(&format!("\n# Sheet: {sheet_name}\n\n"));
// 每张表按网格渲染成带列头的 Markdown 表;一格字都没有的表退回制表符分隔
- let rows: Vec<GridRow> = range
- .rows()
- .take(2000)
- .map(|row| {
- row.iter()
- .map(|c| {
- let text = match c {
- Data::Empty => String::new(),
- Data::DateTime(d) => excel_date(d),
- Data::Float(f) => excel_number(*f),
- other => other.to_string(),
- };
- (text, 1, 0)
- })
- .collect::<GridRow>()
- })
- .filter(|line| line.iter().any(|(s, _, _)| !s.is_empty()))
- .collect();
- match crate::table::render_grid(&rows, true) {
+ let (rows, headers) = spreadsheet_grid(&range, &merges);
+ match crate::table::render_grid_with_headers(&rows, headers) {
Some(md) => {
out.push_str(&md);
out.push('\n');
@@ -782,6 +775,117 @@ pub fn spreadsheet(bytes: &[u8]) -> anyhow::Result<String> {
Ok(out)
}
+#[derive(Clone)]
+struct GridMerge {
+ start: (usize, usize),
+ end: (usize, usize),
+ text: String,
+}
+
+/// 把 calamine 的矩形 range 和 merge 信息铺成 [`GridRow`]。横向 merge 变成一格的
+/// span,纵向 merge 把锚点的值带到覆盖的每一行;anchor 之外的值不再单独成格。
+fn spreadsheet_grid(
+ range: &calamine::Range<calamine::Data>,
+ merges: &[calamine::Dimensions],
+) -> (Vec<GridRow>, Option<std::ops::Range<usize>>) {
+ let Some((start_row, start_col)) = range.start() else {
+ return (Vec::new(), None);
+ };
+ let height = range.height().min(2000);
+ let width = range.width();
+ if height == 0 || width == 0 {
+ return (Vec::new(), None);
+ }
+ let end_row = start_row + height as u32 - 1;
+ let end_col = start_col + width as u32 - 1;
+
+ let mut regions = Vec::new();
+ for merge in merges {
+ if merge.start.0 > merge.end.0
+ || merge.start.1 > merge.end.1
+ || merge.end.0 < start_row
+ || merge.end.1 < start_col
+ || merge.start.0 > end_row
+ || merge.start.1 > end_col
+ {
+ continue;
+ }
+ let start = (
+ (merge.start.0.max(start_row) - start_row) as usize,
+ (merge.start.1.max(start_col) - start_col) as usize,
+ );
+ let end = (
+ (merge.end.0.min(end_row) - start_row) as usize,
+ (merge.end.1.min(end_col) - start_col) as usize,
+ );
+ regions.push(GridMerge {
+ start,
+ end,
+ text: spreadsheet_cell_text(range.get(start)),
+ });
+ }
+
+ let mut owner = vec![usize::MAX; height * width];
+ for (index, region) in regions.iter().enumerate() {
+ for row in region.start.0..=region.end.0 {
+ for col in region.start.1..=region.end.1 {
+ owner[row * width + col] = index;
+ }
+ }
+ }
+
+ let mut rows: Vec<GridRow> = Vec::with_capacity(height);
+ for row in 0..height {
+ let mut cells = GridRow::new();
+ let mut col = 0usize;
+ while col < width {
+ let index = owner[row * width + col];
+ if index == usize::MAX {
+ cells.push((spreadsheet_cell_text(range.get((row, col))), 1, 0));
+ col += 1;
+ continue;
+ }
+ let region = &regions[index];
+ if col == region.start.1 {
+ cells.push((region.text.clone(), region.end.1 - region.start.1 + 1, 0));
+ col = region.end.1 + 1;
+ } else {
+ col += 1;
+ }
+ }
+ rows.push(cells);
+ }
+
+ let headers = spreadsheet_headers(&regions, &rows);
+ (rows, headers)
+}
+
+fn spreadsheet_cell_text(cell: Option<&calamine::Data>) -> String {
+ match cell {
+ None | Some(calamine::Data::Empty) => String::new(),
+ Some(calamine::Data::DateTime(d)) => excel_date(d),
+ Some(calamine::Data::Float(f)) => excel_number(*f),
+ Some(other) => other.to_string(),
+ }
+}
+
+/// 表头从第一排至少两格有字的行开始;这一块里纵向 merge 伸到的行也是表头。
+fn spreadsheet_headers(regions: &[GridMerge], rows: &[GridRow]) -> Option<std::ops::Range<usize>> {
+ let start = rows
+ .iter()
+ .position(|row| row.iter().filter(|(text, _, _)| !text.is_empty()).count() >= 2)?;
+ let mut end = start + 1;
+ while let Some(next) = regions
+ .iter()
+ .filter(|region| region.start.0 >= start && region.start.0 < end && region.end.0 >= end)
+ .map(|region| region.end.0 + 1)
+ .max()
+ {
+ end = next;
+ }
+ Some(start..end.min(rows.len()))
+}
+
/// 日期格按它显示的样子写。
///
/// calamine 按格子挂的数字格式认出了日期,交来的却还是 Excel 存的那个数:从 1899-12-30
diff --git a/crates/utopia-ingest/src/table.rs b/crates/utopia-ingest/src/table.rs
index dc03b3ab..3640241f 100644
--- a/crates/utopia-ingest/src/table.rs
+++ b/crates/utopia-ingest/src/table.rs
@@ -330,8 +330,35 @@ pub(crate) fn render_grid(
rows: &[Vec<(String, usize, u32)>],
first_is_header: bool,
) -> Option<String> {
- let rows: Vec<Row> = rows
- .iter()
+ let rows = grid_rows(rows);
+ let forced = first_is_header
+ .then(|| {
+ rows.iter()
+ .position(|r| r.cells.iter().filter(|c| !c.text.is_empty()).count() >= 2)
+ })
+ .flatten()
+ .map(|row| row..row + 1);
+ render_rows(rows, &[], forced).map(|(md, _)| md)
+}
+
+/// 电子表格的 merged cells 可以给出不止一行的表头。调用方已经把行投影成网格,
+/// 这里只负责把明确的表头区间按表头处理。
+pub(crate) fn render_grid_with_headers(
+ rows: &[Vec<(String, usize, u32)>],
+ headers: Option<std::ops::Range<usize>>,
+) -> Option<String> {
+ let rows = grid_rows(rows);
+ let forced = headers
+ .map(|mut range| {
+ range.end = range.end.min(rows.len());
+ range
+ })
+ .filter(|range| !range.is_empty());
+ render_rows(rows, &[], forced).map(|(md, _)| md)
+}
+
+fn grid_rows(rows: &[Vec<(String, usize, u32)>]) -> Vec<Row> {
+ rows.iter()
.map(|cells| {
let mut col = 0usize;
let mut out = Vec::with_capacity(cells.len());
@@ -350,20 +377,13 @@ pub(crate) fn render_grid(
width: col,
}
})
- .collect();
- let forced = first_is_header
- .then(|| {
- rows.iter()
- .position(|r| r.cells.iter().filter(|c| !c.text.is_empty()).count() >= 2)
- })
- .flatten();
- render_rows(rows, &[], forced).map(|(md, _)| md)
+ .collect()
}
fn render_rows(
mut rows: Vec<Row>,
inherited: &[String],
- forced_header: Option<usize>,
+ forced_header: Option<std::ops::Range<usize>>,
) -> Option<(String, Vec<String>)> {
let width = rows.iter().map(|r| r.width).max().unwrap_or(0);
if width == 0 {
@@ -382,12 +402,12 @@ fn render_rows(
let (mut headers_seen, mut data_seen) = (false, false);
for (i, row) in rows.iter_mut().enumerate() {
let filled = row.cells.iter().filter(|c| !c.text.is_empty()).count();
- let k = match forced_header {
- Some(h) if i == h => Kind::Header,
+ let k = match forced_header.as_ref() {
+ Some(header) if header.contains(&i) => Kind::Header,
// 列头之后的行是记录,哪怕一格数字都没有;只填了一个数字的行(序号列)也是记录,
// 只填了一个词的行才按小节算
- Some(h)
- if i > h
+ Some(header)
+ if i >= header.end
&& (filled >= 2
|| row
.cells
diff --git a/crates/utopia-ingest/tests/spreadsheet_merges.rs b/crates/utopia-ingest/tests/spreadsheet_merges.rs
new file mode 100644
index 00000000..8ca944ab
--- /dev/null
+++ b/crates/utopia-ingest/tests/spreadsheet_merges.rs
@@ -0,0 +1,240 @@
+//! 电子表格里的合并区域投影到它覆盖的所有格,和 HTML 的 colspan/rowspan 一样。
+
+use std::io::{Cursor, Write};
+
+fn workbook(sheet_data: &str, merges: &[&str]) -> Vec<u8> {
+ let merge_cells = (!merges.is_empty()).then(|| {
+ let cells = merges
+ .iter()
+ .map(|range| format!(r#"<mergeCell ref="{range}"/>"#))
+ .collect::<String>();
+ format!(
+ r#"<mergeCells count="{}">{cells}</mergeCells>"#,
+ merges.len()
+ )
+ });
+ let worksheet = format!(
+ r#"<worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main"><sheetData>{sheet_data}</sheetData>{}</worksheet>"#,
+ merge_cells.unwrap_or_default()
+ );
+ let mut zip = zip::ZipWriter::new(Cursor::new(Vec::new()));
+ for (part, xml) in [
+ ("[Content_Types].xml", r#"<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"><Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/><Default Extension="xml" ContentType="application/xml"/><Override PartName="/xl/workbook.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet.main+xml"/><Override PartName="/xl/worksheets/sheet1.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.worksheet+xml"/></Types>"#.to_string()),
+ ("_rels/.rels", r#"<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument" Target="xl/workbook.xml"/></Relationships>"#.to_string()),
+ ("xl/workbook.xml", r#"<workbook xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships"><sheets><sheet name="Ledger" sheetId="1" r:id="rId1"/></sheets></workbook>"#.to_string()),
+ ("xl/_rels/workbook.xml.rels", r#"<Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships"><Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet" Target="worksheets/sheet1.xml"/></Relationships>"#.to_string()),
+ ("xl/worksheets/sheet1.xml", worksheet),
+ ] {
+ zip.start_file(part, zip::write::SimpleFileOptions::default())
+ .unwrap();
+ zip.write_all(xml.as_bytes()).unwrap();
+ }
+ zip.finish().unwrap().into_inner()
+}
+
+fn words(cell: &str, text: &str) -> String {
+ format!(r#"<c r="{cell}" t="inlineStr"><is><t>{text}</t></is></c>"#)
+}
+
+fn number(cell: &str, value: i32) -> String {
+ format!(r#"<c r="{cell}"><v>{value}</v></c>"#)
+}
+
+fn formula(cell: &str, expression: &str, value: i32) -> String {
+ format!(r#"<c r="{cell}"><f>{expression}</f><v>{value}</v></c>"#)
+}
+
+fn read(sheet_data: &str, merges: &[&str]) -> String {
+ utopia_ingest::parse("ledger.xlsx", &workbook(sheet_data, merges))
+ .unwrap()
+ .text
+}
+
+#[test]
+fn a_two_row_merged_header_stacks_each_column() {
+ let sheet_data = [
+ format!(
+ "<row r=\"1\">{}{}{}</row>",
+ words("A1", "区域"),
+ words("B1", "2024年"),
+ words("F1", "2025年")
+ ),
+ format!(
+ "<row r=\"2\">{}{}{}{}{}{}{}{}{}</row>",
+ words("B2", "Q1"),
+ words("C2", "Q2"),
+ words("D2", "Q3"),
+ words("E2", "Q4"),
+ words("F2", "Q1"),
+ words("G2", "Q2"),
+ words("H2", "Q3"),
+ words("I2", "Q4"),
+ words("A2", "")
+ ),
+ format!(
+ "<row r=\"3\">{}{}{}{}{}{}{}{}{}</row>",
+ words("A3", "华东"),
+ number("B3", 120),
+ number("C3", 135),
+ number("D3", 141),
+ number("E3", 150),
+ number("F3", 160),
+ number("G3", 171),
+ number("H3", 180),
+ number("I3", 190)
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &["A1:A2", "B1:E1", "F1:I1"]);
+ assert!(
+ text.contains("| 区域 | 2024年 Q1 | 2024年 Q2 | 2024年 Q3 | 2024年 Q4 | 2025年 Q1 | 2025年 Q2 | 2025年 Q3 | 2025年 Q4 |"),
+ "{text}"
+ );
+ assert!(
+ text.contains("| 华东 | 120 | 135 | 141 | 150 | 160 | 171 | 180 | 190 |"),
+ "{text}"
+ );
+}
+
+#[test]
+fn a_vertical_merge_carries_its_anchor_down_every_covered_row() {
+ let sheet_data = [
+ format!(
+ "<row r=\"1\">{}{}{}</row>",
+ words("A1", "大区 城市"),
+ words("B1", "门店数"),
+ words("C1", "营收(万元)")
+ ),
+ format!(
+ "<row r=\"2\">{}{}{}</row>",
+ words("A2", "华东"),
+ number("B2", 12),
+ number("C2", 860)
+ ),
+ format!(
+ "<row r=\"3\">{}{}{}</row>",
+ words("A3", "上海"),
+ number("B3", 8),
+ number("C3", 540)
+ ),
+ format!(
+ "<row r=\"4\">{}{}{}</row>",
+ words("A4", "杭州"),
+ number("B4", 6),
+ number("C4", 410)
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &["A2:A4"]);
+ assert!(
+ text.contains("| 大区 城市 | 门店数 | 营收(万元) |"),
+ "{text}"
+ );
+ assert!(text.contains("| 华东 | 12 | 860 |"), "{text}");
+ assert!(text.contains("| 华东 | 8 | 540 |"), "{text}");
+ assert!(text.contains("| 华东 | 6 | 410 |"), "{text}");
+}
+
+#[test]
+fn a_merge_uses_the_top_left_value_not_a_stale_covered_cell() {
+ let sheet_data = [
+ format!(
+ "<row r=\"1\">{}{}{}</row>",
+ words("A1", "项目"),
+ words("B1", "金额"),
+ words("C1", "备注")
+ ),
+ format!(
+ "<row r=\"2\">{}{}{}</row>",
+ words("A2", "合计"),
+ number("B2", 7),
+ words("C2", "旧值")
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &["B2:C2"]);
+ assert!(text.contains("| 合计 | 7 |"), "{text}");
+ assert!(!text.contains("旧值"), "{text}");
+}
+
+#[test]
+fn a_formula_value_is_projected_and_an_empty_merge_does_not_leak_stale_cells() {
+ let sheet_data = [
+ format!(
+ "<row r=\"1\">{}{}{}</row>",
+ words("A1", "项目"),
+ words("B1", "金额"),
+ words("C1", "备注")
+ ),
+ format!(
+ "<row r=\"2\">{}{}</row>",
+ words("A2", "公式"),
+ formula("B2", "SUM(B4:B4)", 7)
+ ),
+ format!(
+ "<row r=\"3\">{}{}{}</row>",
+ words("A3", "普通"),
+ number("B3", 99),
+ words("C3", "旧值")
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &["B2:B3", "C2:C3"]);
+ assert!(text.contains("| 公式 | 7 |"), "{text}");
+ assert!(text.contains("| 普通 | 7 |"), "{text}");
+ assert!(!text.contains("99"), "{text}");
+ assert!(!text.contains("旧值"), "{text}");
+}
+
+#[test]
+fn a_merge_past_the_used_range_is_clipped_without_moving_later_columns() {
+ let sheet_data = [
+ format!("<row r=\"1\">{}</row>", words("A1", "标题")),
+ format!(
+ "<row r=\"2\">{}{}{}</row>",
+ words("A2", "项目"),
+ words("B2", "数量"),
+ words("C2", "备注")
+ ),
+ format!(
+ "<row r=\"3\">{}{}{}</row>",
+ words("A3", "合计"),
+ number("B3", 3),
+ number("C3", 4)
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &["A1:Z1"]);
+ assert!(text.contains("\n标题:\n"), "{text}");
+ assert!(text.contains("| 项目 | 数量 | 备注 |"), "{text}");
+ assert!(text.contains("| 合计 | 3 | 4 |"), "{text}");
+}
+
+#[test]
+fn non_merged_cells_keep_their_order() {
+ let sheet_data = [
+ format!(
+ "<row r=\"1\">{}{}{}</row>",
+ words("A1", "项目"),
+ words("B1", "年份"),
+ words("C1", "金额")
+ ),
+ format!(
+ "<row r=\"2\">{}{}{}</row>",
+ words("A2", "甲"),
+ number("B2", 2024),
+ number("C2", 1)
+ ),
+ format!(
+ "<row r=\"3\">{}{}{}</row>",
+ words("A3", "乙"),
+ number("B3", 2025),
+ number("C3", 2)
+ ),
+ ]
+ .concat();
+ let text = read(&sheet_data, &[]);
+ assert!(text.contains("| 项目 | 年份 | 金额 |"), "{text}");
+ assert!(text.contains("| 甲 | 2024 | 1 |"), "{text}");
+ assert!(text.contains("| 乙 | 2025 | 2 |"), "{text}");
+}