@opencraw/cli 0.1.10 → 0.1.12

package/dist/index.esm.js

@@ -558,6 +558,41 @@ function tryParse(text) {
   }
 }
 
+/**
+ * Whether a cell holds a figure (an amount, a count, a date written in
+ * digits) rather than words: it starts with a number, or its digits are at
+ * least as many as its letters. `2,820`, `$560`, `2659 $560`, `(£0.58m)`,
+ * `-3 days` and `2026` are figures; `FY26 Lodging Rate`,
+ * `FM Classes A, B1 & C3` and `(APRIL 1, 2020)` are words that happen to
+ * contain digits.
+ *
+ * @param cell - The cell's text.
+ * @returns Whether it is a figure.
+ */
+function isFigure(cell) {
+  if (/^[\s(+\-–−<>=~$€£¥]*\d/u.test(cell)) return true;
+  const digits = cell.match(/\d/g)?.length ?? 0;
+  return digits > 0 && digits >= (cell.match(/\p{L}/gu)?.length ?? 0);
+}
+/**
+ * A row that reads like a table's header: at least two cells, none a figure.
+ *
+ * @param cells - The row's non-empty cells.
+ * @returns Whether it could be a header.
+ */
+function isHeaderRow(cells) {
+  return cells.length >= 2 && cells.every(cell => !isFigure(cell));
+}
+/**
+ * A row that reads like a table's data: at least two cells, one a figure.
+ *
+ * @param cells - The row's non-empty cells.
+ * @returns Whether it looks like data.
+ */
+function isDataRow(cells) {
+  return cells.length >= 2 && cells.some(cell => isFigure(cell));
+}
+
 /** Rows shown per sheet in a probe, at most. */
 const ROW_LIMIT$1 = 20;
 /** How far below a header row a probe looks for the first data row. */
@@ -596,23 +631,34 @@ function describeWorkbook(document) {
 function headersOf(sheet) {
   const rows = filledRows(sheet);
   const merged = mergedRows(sheet);
-  return rows.flatMap(({
+  const headers = [];
+  // The table a row belongs to: the header it follows, until an empty row ends the block.
+  let table;
+  for (const [position, {
     index,
     cells
-  }, position) => {
-    if (cells.length < 2 || cells.some(cell => /\d/.test(cell))) return [];
-    if (rows.slice(position + 1, position + 1 + LOOKAHEAD).every(row => !(row.cells.length >= 2 && row.cells.some(cell => /\d/.test(cell))))) return [];
+  }] of rows.entries()) {
+    const previous = rows[position - 1]?.index;
+    if (previous === undefined || index > previous + 1) table = undefined;
+    const secondHeaderRow = table?.merged === true && index === table.header + 1;
+    if (table !== undefined && !secondHeaderRow) continue;
+    if (!isHeaderRow(cells) || rows.slice(position + 1, position + 1 + LOOKAHEAD).every(row => !isDataRow(row.cells))) continue;
     const header = {
       sheet: sheet.name,
       row: index + 1,
       text: cells.join(' | '),
       selector: `^${escape$1(cells[0])}`
     };
-    return [merged.has(index) ? {
+    headers.push(merged.has(index) ? {
       ...header,
       hint: 'merged header cells: try "headerRows": 2'
-    } : header];
-  });
+    } : header);
+    table ??= {
+      header: index,
+      merged: merged.has(index)
+    };
+  }
+  return headers;
 }
 /** The non-empty rows, as their non-empty cells, whitespace collapsed. */
 function filledRows(sheet) {
@@ -826,7 +872,7 @@ function describePdf(document) {
   const headers = rows.filter(({
     row,
     next
-  }) => row.cells.length >= 2 && row.cells.every(cell => !/\d/.test(cell.text)) && next.some(candidate => dataRow(candidate)));
+  }) => isHeaderRow(row.cells.map(cell => cell.text)) && next.some(candidate => isDataRow(candidate.cells.map(cell => cell.text))));
   return {
     pages: document.pages.length,
     rows: rows.slice(0, ROW_LIMIT).map(({
@@ -846,10 +892,6 @@ function describePdf(document) {
     }))
   };
 }
-/** A row of at least two cells, one with a digit: data, not a heading. */
-function dataRow(row) {
-  return row.cells.length >= 2 && row.cells.some(cell => /\d/.test(cell.text));
-}
 function escape(text) {
   return text.replaceAll(/[$()*+.?[\\\]^{|}]/g, String.raw`\$&`);
 }

package/dist/src/probe/table-header.policy.d.ts

@@ -0,0 +1,27 @@
+/**
+ * Whether a cell holds a figure (an amount, a count, a date written in
+ * digits) rather than words: it starts with a number, or its digits are at
+ * least as many as its letters. `2,820`, `$560`, `2659 $560`, `(£0.58m)`,
+ * `-3 days` and `2026` are figures; `FY26 Lodging Rate`,
+ * `FM Classes A, B1 & C3` and `(APRIL 1, 2020)` are words that happen to
+ * contain digits.
+ *
+ * @param cell - The cell's text.
+ * @returns Whether it is a figure.
+ */
+export declare function isFigure(cell: string): boolean;
+/**
+ * A row that reads like a table's header: at least two cells, none a figure.
+ *
+ * @param cells - The row's non-empty cells.
+ * @returns Whether it could be a header.
+ */
+export declare function isHeaderRow(cells: readonly string[]): boolean;
+/**
+ * A row that reads like a table's data: at least two cells, one a figure.
+ *
+ * @param cells - The row's non-empty cells.
+ * @returns Whether it looks like data.
+ */
+export declare function isDataRow(cells: readonly string[]): boolean;
+//# sourceMappingURL=table-header.policy.d.ts.map

package/dist/src/probe/workbook-findings.mapper.d.ts

@@ -19,9 +19,10 @@ export interface WorkbookFindings {
         text: string;
     }[];
     /**
-     * Rows of text cells (no digit) with a data row (a digit) just below: likely
-     * table headers, with the `sheet` and `selector` a `table` extract needs, and
-     * a hint when the header has merged cells (a header over two rows).
+     * Rows of words (no cell a figure) with a data row (a figure) just below:
+     * likely table headers, with the `sheet` and `selector` a `table` extract
+     * needs, and a hint when the header has merged cells (a header over two
+     * rows). The rows right under a header are its table's body, not headers.
      */
     headers: {
         sheet: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@opencraw/cli",
-  "version": "0.1.10",
+  "version": "0.1.12",
   "type": "module",
   "main": "./dist/index.esm.js",
   "module": "./dist/index.esm.js",
@@ -24,7 +24,7 @@
     "!**/*.js.map"
   ],
   "dependencies": {
-    "@opencraw/core": "^0.1.10",
+    "@opencraw/core": "^0.1.12",
     "playwright": "^1.63.0"
   },
   "publishConfig": {