@taloswork/ose-shared 0.1.11 → 0.1.12

package/skills/document/bin/pdf-render.cjs

@@ -8,6 +8,17 @@ var import_node_path2 = require("node:path");
 // packages/core/dist/document/pdf-extract.js
 var import_node_path = require("node:path");
 var import_node_url = require("node:url");
+
+// packages/core/dist/document/pdf-encrypted.js
+var ENCRYPTED_PDF_MESSAGE = "PDF is password-protected (encrypted); decrypt first: pdf-ops decrypt <in> <password> <out>";
+var PdfEncryptedError = class extends Error {
+  constructor() {
+    super(ENCRYPTED_PDF_MESSAGE);
+    this.name = "PdfEncryptedError";
+  }
+};
+
+// packages/core/dist/document/pdf-extract.js
 var _mupdf = null;
 function loadMupdf() {
   if (!_mupdf) {
@@ -20,21 +31,26 @@ function loadMupdf() {
 async function openDoc(bytes) {
   const mupdf = await loadMupdf();
   const doc = mupdf.Document.openDocument(bytes, "application/pdf");
+  if (doc.needsPassword())
+    throw new PdfEncryptedError();
   if (doc.countPages() === 0) {
     throw new Error("PDF has no readable pages (corrupt, encrypted, or empty)");
   }
   return { mupdf, doc };
 }
-async function renderPdfPages(bytes, pages, scale = 2) {
+async function renderPdfPages(bytes, pages, scale = 2, maxEdge) {
   const { mupdf, doc } = await openDoc(bytes);
   const total = doc.countPages();
   const wanted = pages === "all" ? Array.from({ length: total }, (_, i) => i + 1) : pages;
   const out = [];
-  const matrix = mupdf.Matrix.scale(scale, scale);
   for (const p of wanted) {
     if (p < 1 || p > total)
       continue;
-    const pix = doc.loadPage(p - 1).toPixmap(matrix, mupdf.ColorSpace.DeviceRGB, false, true);
+    const page = doc.loadPage(p - 1);
+    const [x0, y0, x1, y1] = page.getBounds();
+    const longest = Math.max(x1 - x0, y1 - y0);
+    const s = maxEdge && longest > 0 ? Math.min(scale, maxEdge / longest) : scale;
+    const pix = page.toPixmap(mupdf.Matrix.scale(s, s), mupdf.ColorSpace.DeviceRGB, false, true);
     out.push({ page: p, png: pix.asPNG() });
   }
   return out;

package/skills/document/bin/pdf-tables.cjs

@@ -8,6 +8,17 @@ var import_node_path2 = require("node:path");
 // packages/core/dist/document/pdf-extract.js
 var import_node_path = require("node:path");
 var import_node_url = require("node:url");
+
+// packages/core/dist/document/pdf-encrypted.js
+var ENCRYPTED_PDF_MESSAGE = "PDF is password-protected (encrypted); decrypt first: pdf-ops decrypt <in> <password> <out>";
+var PdfEncryptedError = class extends Error {
+  constructor() {
+    super(ENCRYPTED_PDF_MESSAGE);
+    this.name = "PdfEncryptedError";
+  }
+};
+
+// packages/core/dist/document/pdf-extract.js
 var _mupdf = null;
 function loadMupdf() {
   if (!_mupdf) {
@@ -20,6 +31,8 @@ function loadMupdf() {
 async function openDoc(bytes) {
   const mupdf = await loadMupdf();
   const doc = mupdf.Document.openDocument(bytes, "application/pdf");
+  if (doc.needsPassword())
+    throw new PdfEncryptedError();
   if (doc.countPages() === 0) {
     throw new Error("PDF has no readable pages (corrupt, encrypted, or empty)");
   }

package/skills/document/bin/pptx-clone.cjs

@@ -9892,7 +9892,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
     let flat = "";
     for (const m of para.matchAll(textRe)) {
       const text = unescXml(m[2]);
-      segs.push({ el: m[0], attrs: m[1] ?? "", text, start: flat.length });
+      segs.push({ el: m[0], at: m.index, attrs: m[1] ?? "", text, start: flat.length });
       flat += text;
     }
     if (!flat.includes(find))
@@ -9921,7 +9921,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
       return pieces.join("");
     });
     let out = para;
-    for (let i = 0; i < segs.length; i++) {
+    for (let i = segs.length - 1; i >= 0; i--) {
       if (newTexts[i] === segs[i].text)
         continue;
       const escaped = escXml(newTexts[i]);
@@ -9929,7 +9929,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
       if (textTag === "w:t" && /^\s|\s$/.test(newTexts[i]) && !/xml:space="preserve"/.test(attrs)) {
         attrs = `${attrs} xml:space="preserve"`;
       }
-      out = out.replace(segs[i].el, `<${textTag}${attrs}>${escaped}</${textTag}>`);
+      out = out.slice(0, segs[i].at) + `<${textTag}${attrs}>${escaped}</${textTag}>` + out.slice(segs[i].at + segs[i].el.length);
     }
     return out;
   });

package/skills/document/bin/pptx-slide.cjs

@@ -10007,7 +10007,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
     let flat = "";
     for (const m of para.matchAll(textRe)) {
       const text = unescXml(m[2]);
-      segs.push({ el: m[0], attrs: m[1] ?? "", text, start: flat.length });
+      segs.push({ el: m[0], at: m.index, attrs: m[1] ?? "", text, start: flat.length });
       flat += text;
     }
     if (!flat.includes(find))
@@ -10036,7 +10036,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
       return pieces.join("");
     });
     let out2 = para;
-    for (let i = 0; i < segs.length; i++) {
+    for (let i = segs.length - 1; i >= 0; i--) {
       if (newTexts[i] === segs[i].text)
         continue;
       const escaped = escXml(newTexts[i]);
@@ -10044,7 +10044,7 @@ function replaceTextAcrossRuns(xml, find, replacement, opts) {
       if (textTag === "w:t" && /^\s|\s$/.test(newTexts[i]) && !/xml:space="preserve"/.test(attrs)) {
         attrs = `${attrs} xml:space="preserve"`;
       }
-      out2 = out2.replace(segs[i].el, `<${textTag}${attrs}>${escaped}</${textTag}>`);
+      out2 = out2.slice(0, segs[i].at) + `<${textTag}${attrs}>${escaped}</${textTag}>` + out2.slice(segs[i].at + segs[i].el.length);
     }
     return out2;
   });

package/skills/document/bin/xlsx-query.cjs

@@ -12683,7 +12683,7 @@ var require_xlsx = __commonJS({
           var s = o.dense ? [] : {}, n = "Sheet1", next_n = "", sidx = 0;
           var sheets = {}, snames = [], realnames = [];
           var refguess = { s: { r: 0, c: 0 }, e: { r: 0, c: 0 } };
-          var sheetRows2 = o.sheetRows || 0;
+          var sheetRows = o.sheetRows || 0;
           if (d[2] == 0) {
             if (d[3] == 8 || d[3] == 9) {
               if (d.length >= 16 && d[14] == 5 && d[15] === 108) throw new Error("Unsupported Works 3 for Mac file");
@@ -12789,7 +12789,7 @@ var require_xlsx = __commonJS({
                     sidx = val[3];
                     n = "Sheet" + (sidx + 1);
                   }
-                  if (sheetRows2 > 0 && val[0].r >= sheetRows2) break;
+                  if (sheetRows > 0 && val[0].r >= sheetRows) break;
                   if (o.dense) {
                     if (!s[val[0].r]) s[val[0].r] = [];
                     s[val[0].r][val[0].c] = val[1];
@@ -29309,7 +29309,7 @@ var require_xlsx = __commonJS({
           }
         }
         var rows = table.getElementsByTagName("tr");
-        var sheetRows2 = Math.min(opts.sheetRows || 1e7, rows.length);
+        var sheetRows = Math.min(opts.sheetRows || 1e7, rows.length);
         var range = { s: { r: 0, c: 0 }, e: { r: or_R, c: or_C } };
         if (ws["!ref"]) {
           var _range = decode_range(ws["!ref"]);
@@ -29323,7 +29323,7 @@ var require_xlsx = __commonJS({
         var rowinfo = ws["!rows"] || (ws["!rows"] = []);
         var _R = 0, R = 0, _C = 0, C = 0, RS = 0, CS = 0;
         if (!ws["!cols"]) ws["!cols"] = [];
-        for (; _R < rows.length && R < sheetRows2; ++_R) {
+        for (; _R < rows.length && R < sheetRows; ++_R) {
           var row = rows[_R];
           if (is_dom_element_hidden(row)) {
             if (opts.display) continue;
@@ -29379,7 +29379,7 @@ var require_xlsx = __commonJS({
         if (merges.length) ws["!merges"] = (ws["!merges"] || []).concat(merges);
         range.e.r = Math.max(range.e.r, R - 1 + or_R);
         ws["!ref"] = encode_range(range);
-        if (R >= sheetRows2) ws["!fullref"] = encode_range((range.e.r = rows.length - _R + R - 1 + or_R, range));
+        if (R >= sheetRows) ws["!fullref"] = encode_range((range.e.r = rows.length - _R + R - 1 + or_R, range));
         return ws;
       }
       function parse_dom_table(table, _opts) {
@@ -33253,30 +33253,104 @@ var import_promises = require("node:fs/promises");
 
 // packages/core/dist/document/xlsx-query.js
 var XLSX = __toESM(require_xlsx(), 1);
+var AGGREGATE_OPS = ["sum", "avg", "min", "max", "count", "count-nonempty"];
 function parseWorkbook(buffer) {
   const data = buffer instanceof Uint8Array ? buffer : new Uint8Array(buffer);
   return XLSX.read(data, { type: "array" });
 }
-function sheetRows(wb, sheetName) {
+var hasValue = (c) => c != null && c.t !== "z" && c.v != null;
+function sheetGrid(wb, sheetName) {
   const ws = wb.Sheets[sheetName];
-  if (!ws)
-    return [];
-  return XLSX.utils.sheet_to_json(ws, { header: 1, blankrows: false, defval: "" });
+  if (!ws || !ws["!ref"])
+    return { rows: [], width: 0, firstCol: 0 };
+  const range = XLSX.utils.decode_range(ws["!ref"]);
+  const rows = [];
+  for (let r = range.s.r; r <= range.e.r; r++) {
+    const cells = [];
+    for (let c = range.s.c; c <= range.e.c; c++)
+      cells.push(ws[XLSX.utils.encode_cell({ r, c })]);
+    if (cells.some(hasValue))
+      rows.push({ row: r + 1, cells });
+  }
+  return { rows, width: rows.length ? range.e.c - range.s.c + 1 : 0, firstCol: range.s.c };
+}
+function parseNumberText(text) {
+  let t = text.replace(/[¥￥$€£,\s]/g, "").replace(/元$/, "");
+  let sign = 1;
+  const paren = /^\((.*)\)$/.exec(t);
+  if (paren) {
+    sign = -1;
+    t = paren[1];
+  }
+  let scale = 1;
+  if (t.endsWith("%")) {
+    scale = 0.01;
+    t = t.slice(0, -1);
+  }
+  if (t === "" || !/^[+-]?(\d+\.?\d*|\.\d+)(e[+-]?\d+)?$/i.test(t))
+    return null;
+  const n = Number(t);
+  return Number.isFinite(n) ? cleanFloat(sign * n * scale) : null;
+}
+function cellNumber(c) {
+  if (!hasValue(c))
+    return null;
+  if (c.t === "n")
+    return typeof c.v === "number" && Number.isFinite(c.v) ? c.v : null;
+  if (c.t === "s")
+    return parseNumberText(String(c.v));
+  return null;
+}
+var isEmptyCell = (c) => !hasValue(c) || String(c.v).trim() === "";
+var isTextCell = (c) => !isEmptyCell(c) && c.t === "s" && cellNumber(c) == null;
+function cleanFloat(v) {
+  if (!Number.isFinite(v) || v === 0)
+    return v;
+  return Number(v.toPrecision(15));
+}
+var HEADER_SCAN_ROWS = 10;
+function detectHeaderIndex(grid) {
+  const { rows, width } = grid;
+  for (let k = 0; k < Math.min(HEADER_SCAN_ROWS, rows.length - 1); k++) {
+    const cells = rows[k].cells;
+    const filled = cells.filter((c) => !isEmptyCell(c));
+    if (!filled.length || filled.length * 2 <= width)
+      continue;
+    if (cells.filter(isTextCell).length * 2 < filled.length)
+      continue;
+    if (rows[k + 1].cells.some((c) => cellNumber(c) != null))
+      return k;
+  }
+  return 0;
 }
-function headerRow(rows) {
-  const first = rows[0] ?? [];
-  return first.map((c, i) => {
-    const s = c == null ? "" : String(c).trim();
+function layout(grid, opts) {
+  if (!grid.rows.length)
+    return { headerRow: opts.headerRow ?? 1, headers: [], data: [] };
+  let headerCells;
+  let headerRow;
+  let data;
+  if (opts.headerRow != null) {
+    headerRow = opts.headerRow;
+    headerCells = grid.rows.find((r) => r.row === headerRow)?.cells ?? [];
+    data = grid.rows.filter((r) => r.row > headerRow);
+  } else {
+    const k = detectHeaderIndex(grid);
+    headerRow = grid.rows[k].row;
+    headerCells = grid.rows[k].cells;
+    data = grid.rows.slice(k + 1);
+  }
+  const headers = Array.from({ length: grid.width }, (_, i) => {
+    const c = headerCells[i];
+    const s = hasValue(c) ? String(c.v).trim() : "";
     return s.length > 0 ? s : `col${i + 1}`;
   });
+  return { headerRow, headers, data };
 }
-function summarizeWorkbook(wb) {
+function summarizeWorkbook(wb, opts = {}) {
   const sheets = wb.SheetNames.map((name) => {
-    const rows = sheetRows(wb, name);
-    const headers = rows.length > 0 ? headerRow(rows) : [];
-    const dataRows = rows.length > 0 ? rows.length - 1 : 0;
-    const cols = rows.reduce((m, r) => Math.max(m, r.length), 0);
-    return { name, rows: dataRows, cols, headers };
+    const grid = sheetGrid(wb, name);
+    const { headerRow, headers, data } = layout(grid, opts);
+    return { name, rows: data.length, cols: grid.width, headerRow, headers };
   });
   return { sheets };
 }
@@ -33289,54 +33363,95 @@ function resolveColumnIndex(headers, column) {
     return asNum - 1;
   return -1;
 }
-function toNumber(cell) {
-  if (typeof cell === "number" && Number.isFinite(cell))
-    return cell;
-  if (typeof cell === "string") {
-    const cleaned = cell.replace(/[$,\s%]/g, "");
-    if (cleaned === "")
-      return null;
-    const n = Number(cleaned);
-    return Number.isFinite(n) ? n : null;
+var TOTAL_LABEL = /^(合计|总计|小计|total|subtotal|sum)$/i;
+function totalReason(row, colIdx, colLetter) {
+  for (const c of row.cells) {
+    if (!hasValue(c) || c.t !== "s")
+      continue;
+    const label = String(c.v).replace(/\s+/g, "").replace(/[:：]$/, "");
+    if (TOTAL_LABEL.test(label))
+      return `total label "${String(c.v).trim()}"`;
+  }
+  const f = row.cells[colIdx]?.f;
+  if (f) {
+    const m = /^\s*=?\s*SUM\(\s*\$?([A-Z]+)\$?\d+\s*:\s*\$?([A-Z]+)\$?\d+\s*\)\s*$/i.exec(f);
+    if (m && m[1].toUpperCase() === colLetter && m[2].toUpperCase() === colLetter)
+      return `${f.trim().replace(/^=/, "")} over this column`;
   }
   return null;
 }
-function aggregateColumn(wb, sheetName, column, op) {
-  const rows = sheetRows(wb, sheetName);
-  const colIdx = rows.length > 0 ? resolveColumnIndex(headerRow(rows), column) : -1;
-  const base = { sheet: sheetName, column, op, colIdx };
-  if (rows.length === 0)
-    return { ...base, rowsScanned: 0, numericCells: 0, value: null };
-  const dataRows = rows.slice(1);
-  if (colIdx === -1)
-    return { ...base, rowsScanned: dataRows.length, numericCells: 0, value: null };
-  const cells = dataRows.map((r) => r[colIdx]);
+var SKIPPED_CELLS_CAP = 20;
+function aggregateColumn(wb, sheetName, column, op, opts = {}) {
+  const grid = sheetGrid(wb, sheetName);
+  const { headerRow, headers, data } = layout(grid, opts);
+  const colIdx = headers.length > 0 ? resolveColumnIndex(headers, column) : -1;
+  const result = {
+    sheet: sheetName,
+    column,
+    op,
+    headerRow,
+    colIdx,
+    rowsScanned: 0,
+    numericCells: 0,
+    value: null,
+    excludedRows: [],
+    skippedCells: [],
+    skippedCellCount: 0
+  };
+  if (colIdx === -1) {
+    result.rowsScanned = data.length;
+    return result;
+  }
+  const colLetter = XLSX.utils.encode_col(grid.firstCol + colIdx);
+  const kept = [];
+  for (const r of data) {
+    const reason = opts.includeTotals ? null : totalReason(r, colIdx, colLetter);
+    if (reason)
+      result.excludedRows.push({ row: r.row, reason });
+    else
+      kept.push(r);
+  }
+  result.rowsScanned = kept.length;
   if (op === "count") {
-    return { ...base, rowsScanned: dataRows.length, numericCells: 0, value: dataRows.length };
+    result.value = kept.length;
+    return result;
   }
+  const cells = kept.map((r) => ({ row: r.row, cell: r.cells[colIdx] }));
   if (op === "count-nonempty") {
-    const nonEmpty = cells.filter((c) => c != null && String(c).trim() !== "").length;
-    return { ...base, rowsScanned: dataRows.length, numericCells: 0, value: nonEmpty };
+    result.value = cells.filter(({ cell }) => !isEmptyCell(cell)).length;
+    return result;
   }
-  const nums = cells.map(toNumber).filter((n) => n != null);
+  const nums = [];
+  for (const { row, cell } of cells) {
+    if (isEmptyCell(cell))
+      continue;
+    const n = cellNumber(cell);
+    if (n != null) {
+      nums.push(n);
+      continue;
+    }
+    result.skippedCellCount++;
+    if (result.skippedCells.length < SKIPPED_CELLS_CAP)
+      result.skippedCells.push({ row, value: cell.v });
+  }
+  result.numericCells = nums.length;
   if (nums.length === 0)
-    return { ...base, rowsScanned: dataRows.length, numericCells: 0, value: null };
-  let value;
+    return result;
   switch (op) {
     case "sum":
-      value = nums.reduce((a, b) => a + b, 0);
+      result.value = cleanFloat(nums.reduce((a, b) => a + b, 0));
       break;
     case "avg":
-      value = nums.reduce((a, b) => a + b, 0) / nums.length;
+      result.value = cleanFloat(nums.reduce((a, b) => a + b, 0) / nums.length);
       break;
     case "min":
-      value = Math.min(...nums);
+      result.value = Math.min(...nums);
       break;
     case "max":
-      value = Math.max(...nums);
+      result.value = Math.max(...nums);
       break;
   }
-  return { ...base, rowsScanned: dataRows.length, numericCells: nums.length, value };
+  return result;
 }
 var COMPUTE_KEYWORDS_EN = [
   "sum",
@@ -33356,15 +33471,13 @@ var COMPUTE_KEYWORDS_EN = [
   "percentage"
 ];
 var COMPUTE_EN_RE = new RegExp(`\\b(${COMPUTE_KEYWORDS_EN.map((k) => k.replace(/ /g, "\\s+")).join("|")})\\b`, "i");
-function sampleRows(wb, sheetName, limit = 20) {
-  const rows = sheetRows(wb, sheetName);
-  if (rows.length <= 1)
-    return [];
-  const headers = headerRow(rows);
-  return rows.slice(1, 1 + Math.max(0, limit)).map((r) => {
+function sampleRows(wb, sheetName, limit = 20, opts = {}) {
+  const { headers, data } = layout(sheetGrid(wb, sheetName), opts);
+  return data.slice(0, Math.max(0, limit)).map((r) => {
     const obj = {};
     headers.forEach((h, i) => {
-      obj[h] = r[i] ?? "";
+      const c = r.cells[i];
+      obj[h] = hasValue(c) ? c.v : "";
     });
     return obj;
   });
@@ -33379,9 +33492,24 @@ function die(msg, code = 1) {
 function out(obj) {
   process.stdout.write(JSON.stringify(obj, null, 2) + "\n");
 }
+var USAGE = "usage: xlsx-query <file> [--summary | --agg <sheet> <col> <op> | --sample <sheet> [n]] [--header-row N] [--include-totals]";
+function parseFlags(argv) {
+  const opts = {};
+  const positional = [];
+  for (let i = 0; i < argv.length; i++) {
+    if (argv[i] === "--include-totals") opts.includeTotals = true;
+    else if (argv[i] === "--header-row") {
+      const n = Number(argv[++i]);
+      if (!Number.isInteger(n) || n < 1) die("--header-row needs a 1-based row number", 64);
+      opts.headerRow = n;
+    } else positional.push(argv[i]);
+  }
+  return { opts, positional };
+}
 async function main() {
-  const [file, mode, ...rest] = process.argv.slice(2);
-  if (!file) die("usage: xlsx-query <file> [--summary | --agg <sheet> <col> <op> | --sample <sheet> [n]]", 64);
+  const { opts, positional } = parseFlags(process.argv.slice(2));
+  const [file, mode, ...rest] = positional;
+  if (!file) die(USAGE, 64);
   let raw;
   try {
     raw = await (0, import_promises.readFile)(file);
@@ -33395,19 +33523,22 @@ async function main() {
     die(`parse failed: ${e.message}`);
   }
   if (!mode || mode === "--summary") {
-    out(summarizeWorkbook(wb));
+    out(summarizeWorkbook(wb, opts));
     return;
   }
   if (mode === "--agg") {
     const [sheet, col, op] = rest;
     if (!sheet || !col || !op) die("usage: --agg <sheet> <col> <op>", 64);
-    out(aggregateColumn(wb, sheet, col, op));
+    if (!AGGREGATE_OPS.includes(op)) die(`unknown op "${op}" \u2014 use one of: ${AGGREGATE_OPS.join(", ")}`, 64);
+    if (!wb.SheetNames.includes(sheet)) die(`no sheet "${sheet}" \u2014 the workbook has: ${wb.SheetNames.join(", ")}`);
+    out(aggregateColumn(wb, sheet, col, op, opts));
     return;
   }
   if (mode === "--sample") {
     const [sheet, n] = rest;
     if (!sheet) die("usage: --sample <sheet> [n]", 64);
-    out({ sheet, sample: sampleRows(wb, sheet, Math.max(0, Number(n) || 20)) });
+    if (!wb.SheetNames.includes(sheet)) die(`no sheet "${sheet}" \u2014 the workbook has: ${wb.SheetNames.join(", ")}`);
+    out({ sheet, sample: sampleRows(wb, sheet, Math.max(0, Number(n) || 20), opts) });
     return;
   }
   die(`unknown mode: ${mode}`, 64);