@truto/truto-jsonata 1.0.44 → 1.0.45

package/README.md

@@ -1532,6 +1532,44 @@ Output:
 
 </details>
 
+<details>
+<summary>convertMdToPdf(markdown, options = { title: '', pageSize: 'a4', embedImages: false, pageMargins: [40, 60, 40, 60], defaultStyle: { fontSize: 12, lineHeight: 1.4 } })</summary>
+
+Converts Markdown text to a PDF `Blob` using jsPDF.
+
+**Example:**
+
+```javascript
+import trutoJsonata from '@truto/truto-jsonata'
+
+const markdown = `# Title\n\nThis is a paragraph.`
+const expression = trutoJsonata("$convertMdToPdf(markdown)")
+expression.evaluate({ markdown }).then(blob => {
+  console.log(blob) // Blob { type: 'application/pdf' }
+})
+```
+
+**Options:**
+- `title`: Optional document title
+- `pageSize`: jsPDF page size (e.g., `'a4'`, `'LETTER'`)
+- `embedImages`: Reserved for future use
+- `pageMargins`: `[left, top, right, bottom]` in points
+- `defaultStyle`: `{ fontSize: number, lineHeight: number }`
+
+If your transport requires a JSON-serializable value, convert the PDF Blob to a data URI string and decode it later:
+
+```javascript
+import trutoJsonata from '@truto/truto-jsonata'
+
+const markdown = `# Report\nGenerated content`
+const expr = trutoJsonata("$getDataUri($convertMdToPdf(markdown), 'application/pdf')")
+expr.evaluate({ markdown }).then(dataUri => {
+  // dataUri starts with 'data:application/pdf;base64,'
+  console.log(typeof dataUri, dataUri.startsWith('data:application/pdf;base64,'))
+})
+```
+</details>
+
 ---  
 
 ### Array and Object Utilities (Lodash Enhancements)

package/dist/browser/index.js

@@ -23000,7 +23000,7 @@ var init_buffer = __esm(() => {
       return 1;
     return 0;
   };
-  Buffer4.prototype.includes = function includes(val, byteOffset, encoding) {
+  Buffer4.prototype.includes = function includes2(val, byteOffset, encoding) {
     return this.indexOf(val, byteOffset, encoding) !== -1;
   };
   Buffer4.prototype.indexOf = function indexOf(val, byteOffset, encoding) {
@@ -30337,7 +30337,7 @@ var require_object_inspect = __commonJS((exports, module) => {
       }
       return inspect_(value, opts, depth + 1, seen);
     }
-    if (typeof obj === "function" && !isRegExp(obj)) {
+    if (typeof obj === "function" && !isRegExp2(obj)) {
       var name = nameOf(obj);
       var keys2 = arrObjKeys(obj, inspect2);
       return "[Function" + (name ? ": " + name : " (anonymous)") + "]" + (keys2.length > 0 ? " { " + $join.call(keys2, ", ") + " }" : "");
@@ -30431,7 +30431,7 @@ var require_object_inspect = __commonJS((exports, module) => {
     if (typeof globalThis !== "undefined" && obj === globalThis || typeof global !== "undefined" && obj === global) {
       return "{ [object globalThis] }";
     }
-    if (!isDate2(obj) && !isRegExp(obj)) {
+    if (!isDate2(obj) && !isRegExp2(obj)) {
       var ys = arrObjKeys(obj, inspect2);
       var isPlainObject3 = gPO ? gPO(obj) === Object.prototype : obj instanceof Object || obj.constructor === Object;
       var protoTag = obj instanceof Object ? "" : "null prototype";
@@ -30465,7 +30465,7 @@ var require_object_inspect = __commonJS((exports, module) => {
   function isDate2(obj) {
     return toStr(obj) === "[object Date]" && canTrustToString(obj);
   }
-  function isRegExp(obj) {
+  function isRegExp2(obj) {
     return toStr(obj) === "[object RegExp]" && canTrustToString(obj);
   }
   function isError2(obj) {
@@ -31842,7 +31842,7 @@ var require_utils2 = __commonJS((exports, module) => {
     compactQueue(queue);
     return value;
   };
-  var isRegExp = function isRegExp(obj) {
+  var isRegExp2 = function isRegExp(obj) {
     return Object.prototype.toString.call(obj) === "[object RegExp]";
   };
   var isBuffer3 = function isBuffer(obj) {
@@ -31872,7 +31872,7 @@ var require_utils2 = __commonJS((exports, module) => {
     decode,
     encode,
     isBuffer: isBuffer3,
-    isRegExp,
+    isRegExp: isRegExp2,
     maybeMap,
     merge
   };
@@ -36270,6 +36270,66 @@ function baseSlice(array, start, end) {
 }
 var _baseSlice_default = baseSlice;
 
+// node_modules/lodash-es/_castSlice.js
+function castSlice(array, start, end) {
+  var length = array.length;
+  end = end === undefined ? length : end;
+  return !start && end >= length ? array : _baseSlice_default(array, start, end);
+}
+var _castSlice_default = castSlice;
+
+// node_modules/lodash-es/_hasUnicode.js
+var rsAstralRange = "\\ud800-\\udfff";
+var rsComboMarksRange = "\\u0300-\\u036f";
+var reComboHalfMarksRange = "\\ufe20-\\ufe2f";
+var rsComboSymbolsRange = "\\u20d0-\\u20ff";
+var rsComboRange = rsComboMarksRange + reComboHalfMarksRange + rsComboSymbolsRange;
+var rsVarRange = "\\ufe0e\\ufe0f";
+var rsZWJ = "\\u200d";
+var reHasUnicode = RegExp("[" + rsZWJ + rsAstralRange + rsComboRange + rsVarRange + "]");
+function hasUnicode(string) {
+  return reHasUnicode.test(string);
+}
+var _hasUnicode_default = hasUnicode;
+
+// node_modules/lodash-es/_asciiToArray.js
+function asciiToArray(string) {
+  return string.split("");
+}
+var _asciiToArray_default = asciiToArray;
+
+// node_modules/lodash-es/_unicodeToArray.js
+var rsAstralRange2 = "\\ud800-\\udfff";
+var rsComboMarksRange2 = "\\u0300-\\u036f";
+var reComboHalfMarksRange2 = "\\ufe20-\\ufe2f";
+var rsComboSymbolsRange2 = "\\u20d0-\\u20ff";
+var rsComboRange2 = rsComboMarksRange2 + reComboHalfMarksRange2 + rsComboSymbolsRange2;
+var rsVarRange2 = "\\ufe0e\\ufe0f";
+var rsAstral = "[" + rsAstralRange2 + "]";
+var rsCombo = "[" + rsComboRange2 + "]";
+var rsFitz = "\\ud83c[\\udffb-\\udfff]";
+var rsModifier = "(?:" + rsCombo + "|" + rsFitz + ")";
+var rsNonAstral = "[^" + rsAstralRange2 + "]";
+var rsRegional = "(?:\\ud83c[\\udde6-\\uddff]){2}";
+var rsSurrPair = "[\\ud800-\\udbff][\\udc00-\\udfff]";
+var rsZWJ2 = "\\u200d";
+var reOptMod = rsModifier + "?";
+var rsOptVar = "[" + rsVarRange2 + "]?";
+var rsOptJoin = "(?:" + rsZWJ2 + "(?:" + [rsNonAstral, rsRegional, rsSurrPair].join("|") + ")" + rsOptVar + reOptMod + ")*";
+var rsSeq = rsOptVar + reOptMod + rsOptJoin;
+var rsSymbol = "(?:" + [rsNonAstral + rsCombo + "?", rsCombo, rsRegional, rsSurrPair, rsAstral].join("|") + ")";
+var reUnicode = RegExp(rsFitz + "(?=" + rsFitz + ")|" + rsSymbol + rsSeq, "g");
+function unicodeToArray(string) {
+  return string.match(reUnicode) || [];
+}
+var _unicodeToArray_default = unicodeToArray;
+
+// node_modules/lodash-es/_stringToArray.js
+function stringToArray(string) {
+  return _hasUnicode_default(string) ? _unicodeToArray_default(string) : _asciiToArray_default(string);
+}
+var _stringToArray_default = stringToArray;
+
 // node_modules/lodash-es/_arrayReduce.js
 function arrayReduce(array, iteratee, accumulator, initAccum) {
   var index = -1, length = array == null ? 0 : array.length;
@@ -37503,6 +37563,18 @@ function values(object) {
 }
 var values_default = values;
 
+// node_modules/lodash-es/includes.js
+var nativeMax4 = Math.max;
+function includes(collection, value, fromIndex, guard) {
+  collection = isArrayLike_default(collection) ? collection : values_default(collection);
+  fromIndex = fromIndex && !guard ? toInteger_default(fromIndex) : 0;
+  var length = collection.length;
+  if (fromIndex < 0) {
+    fromIndex = nativeMax4(length + fromIndex, 0);
+  }
+  return isString_default(collection) ? fromIndex <= length && collection.indexOf(value, fromIndex) > -1 : !!length && _baseIndexOf_default(collection, value, fromIndex) > -1;
+}
+var includes_default = includes;
 // node_modules/lodash-es/_parent.js
 function parent(object, path) {
   return path.length < 2 ? object : _baseGet_default(object, _baseSlice_default(path, 0, -1));
@@ -37545,6 +37617,17 @@ function isNil(value) {
   return value == null;
 }
 var isNil_default = isNil;
+// node_modules/lodash-es/_baseIsRegExp.js
+var regexpTag5 = "[object RegExp]";
+function baseIsRegExp(value) {
+  return isObjectLike_default(value) && _baseGetTag_default(value) == regexpTag5;
+}
+var _baseIsRegExp_default = baseIsRegExp;
+
+// node_modules/lodash-es/isRegExp.js
+var nodeIsRegExp = _nodeUtil_default && _nodeUtil_default.isRegExp;
+var isRegExp = nodeIsRegExp ? _baseUnary_default(nodeIsRegExp) : _baseIsRegExp_default;
+var isRegExp_default = isRegExp;
 // node_modules/lodash-es/join.js
 var arrayProto2 = Array.prototype;
 var nativeJoin = arrayProto2.join;
@@ -37830,6 +37913,55 @@ var sortBy = _baseRest_default(function(collection, iteratees) {
   return _baseOrderBy_default(collection, _baseFlatten_default(iteratees, 1), []);
 });
 var sortBy_default = sortBy;
+// node_modules/lodash-es/split.js
+var MAX_ARRAY_LENGTH = 4294967295;
+function split(string, separator, limit) {
+  if (limit && typeof limit != "number" && _isIterateeCall_default(string, separator, limit)) {
+    separator = limit = undefined;
+  }
+  limit = limit === undefined ? MAX_ARRAY_LENGTH : limit >>> 0;
+  if (!limit) {
+    return [];
+  }
+  string = toString_default(string);
+  if (string && (typeof separator == "string" || separator != null && !isRegExp_default(separator))) {
+    separator = _baseToString_default(separator);
+    if (!separator && _hasUnicode_default(string)) {
+      return _castSlice_default(_stringToArray_default(string), 0, limit);
+    }
+  }
+  return string.split(separator, limit);
+}
+var split_default = split;
+// node_modules/lodash-es/_charsEndIndex.js
+function charsEndIndex(strSymbols, chrSymbols) {
+  var index = strSymbols.length;
+  while (index-- && _baseIndexOf_default(chrSymbols, strSymbols[index], 0) > -1) {}
+  return index;
+}
+var _charsEndIndex_default = charsEndIndex;
+
+// node_modules/lodash-es/_charsStartIndex.js
+function charsStartIndex(strSymbols, chrSymbols) {
+  var index = -1, length = strSymbols.length;
+  while (++index < length && _baseIndexOf_default(chrSymbols, strSymbols[index], 0) > -1) {}
+  return index;
+}
+var _charsStartIndex_default = charsStartIndex;
+
+// node_modules/lodash-es/trim.js
+function trim(string, chars, guard) {
+  string = toString_default(string);
+  if (string && (guard || chars === undefined)) {
+    return _baseTrim_default(string);
+  }
+  if (!string || !(chars = _baseToString_default(chars))) {
+    return string;
+  }
+  var strSymbols = _stringToArray_default(string), chrSymbols = _stringToArray_default(chars), start = _charsStartIndex_default(strSymbols, chrSymbols), end = _charsEndIndex_default(strSymbols, chrSymbols) + 1;
+  return _castSlice_default(strSymbols, start, end).join("");
+}
+var trim_default = trim;
 // src/functions/base64decode.ts
 function base64decode(base64, urlSafe = false) {
   if (urlSafe) {
@@ -45468,7 +45600,7 @@ var formatPlainText = (x) => {
     return textToReturn;
   }
 };
-var convertNotionToMd = function(block3, level = 1) {
+var convertNotionToMd = function(block3, level = 1, linkChildPages = false) {
   const n = `
 
 `;
@@ -45480,16 +45612,16 @@ var convertNotionToMd = function(block3, level = 1) {
     case "bookmark":
       return `[${data.url}](data.url)${n}`;
     case "bulleted_list_item":
-      childData = map_default(block3.children, (child) => convertNotionToMd(child, level + 1)).join(repeat_default("\t", level));
-      return `- ${plainText.join("")}
+      childData = join_default(map_default(block3.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat_default("\t", level));
+      return `- ${join_default(plainText, "")}
 ` + (childData ? `${repeat_default("\t", level)}${childData}` : "");
     case "numbered_list_item":
-      childData = map_default(block3.children, (child) => convertNotionToMd(child, level + 1)).join(repeat_default("\t", level));
-      return `${block3.number}. ${plainText.join("")}
+      childData = join_default(map_default(block3.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat_default("\t", level));
+      return `${block3.number}. ${join_default(plainText, "")}
 ` + (childData ? `${repeat_default("\t", level)}${childData}` : "");
     case "quote":
     case "callout":
-      return `> ${plainText.join("")}${n}`;
+      return `> ${join_default(plainText, "")}${n}`;
     case "code":
       return `\`\`\`${data.language}
 ${plainText}
@@ -45501,51 +45633,60 @@ ${plainText}
     case "equation":
       return data.expression;
     case "paragraph":
-      return plainText.join("") + n;
+      return join_default(plainText, "") + n;
     case "video":
     case "pdf":
     case "file":
       return `[${caption.length ? caption : "File"}](${data.file ? data.file.url : data.external ? data.external.url : ""})${n}`;
     case "heading_1":
-      return `# ${plainText.join("")}${n}`;
+      return `# ${join_default(plainText, "")}${n}`;
     case "heading_2":
-      return `## ${plainText.join("")}${n}`;
+      return `## ${join_default(plainText, "")}${n}`;
     case "heading_3":
-      return `### ${plainText.join("")}${n}`;
+      return `### ${join_default(plainText, "")}${n}`;
     case "heading_4":
-      return `#### ${plainText.join("")}${n}`;
+      return `#### ${join_default(plainText, "")}${n}`;
     case "heading_5":
-      return `##### ${plainText.join("")}${n}`;
+      return `##### ${join_default(plainText, "")}${n}`;
     case "heading_6":
-      return `###### ${plainText.join("")}${n}`;
+      return `###### ${join_default(plainText, "")}${n}`;
     case "image":
       return `![${caption.length ? caption : "Image"}](${data.file ? data.file.url : data.external ? data.external.url : ""})${n}`;
+    case "child_page":
+      if (linkChildPages) {
+        const pageTitle = data.title || "Untitled Page";
+        const pageUrl = `https://www.notion.so/${block3.id.replace(/-/g, "")}`;
+        return `[${pageTitle}](${pageUrl})${n}`;
+      } else {
+        childData = join_default(map_default(block3.children, (child) => convertNotionToMd(child, level, linkChildPages)), "");
+        return (data.title || "Untitled Page") + n + childData;
+      }
     case "table":
       if (block3.children) {
         const firstChild = block3.children[0];
         const remainingChildren = block3.children.slice(1);
-        const header = `| ${firstChild.table_row.cells.map((x) => x.map((y) => formatPlainText(y)).join("")).join(" | ")} |
+        const header = `| ${join_default(firstChild.table_row.cells.map((x) => join_default(x.map((y) => formatPlainText(y)), "")), " | ")} |
 `;
         const divider = `|${repeat_default("---|", data.table_width)}
 `;
-        const rows = remainingChildren.map((row) => {
-          return `| ${row.table_row.cells.map((x) => x.map((y) => formatPlainText(y)).join("")).join(" | ")} |`;
-        }).join(`
+        const rows = join_default(remainingChildren.map((row) => {
+          return `| ${join_default(row.table_row.cells.map((x) => join_default(x.map((y) => formatPlainText(y)), "")), " | ")} |`;
+        }), `
 `);
         return `${header}${divider}${rows}${n}`;
       }
       return `Table as CSV
 `;
     case "table_row":
-      return `| ${data.cells.map((x) => x.map((y) => formatPlainText(y)).join("")).join(" | ")} |
+      return `| ${join_default(data.cells.map((x) => join_default(x.map((y) => formatPlainText(y)), "")), " | ")} |
 `;
     case "to_do":
-      childData = map_default(block3.children, (child) => convertNotionToMd(child, level + 1)).join(repeat_default("\t", level));
-      return `- [${data.checked ? "X" : " "}] ${plainText.join("")}
+      childData = join_default(map_default(block3.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat_default("\t", level));
+      return `- [${data.checked ? "X" : " "}] ${join_default(plainText, "")}
 ` + (childData ? `${repeat_default("\t", level)}${childData}` : "");
     default:
-      childData = map_default(block3.children, (child) => convertNotionToMd(child)).join("");
-      return plainText.join("") + n + childData;
+      childData = join_default(map_default(block3.children, (child) => convertNotionToMd(child, level, linkChildPages)), "");
+      return join_default(plainText, "") + n + childData;
   }
 };
 var resolveChildren = function(block3, blocks) {
@@ -45595,13 +45736,13 @@ var insertNewLinesBetweenLists = function(blocks) {
     return block3;
   }));
 };
-var convertNotionToMarkdown = function(blocks) {
+var convertNotionToMarkdown = function(blocks, linkChildPages = false) {
   const arrayBlocks = insertNewLinesBetweenLists(castArray_default(blocks));
   const parentBlocks = reject_default(arrayBlocks, (block3) => block3.parent?.type === "block_id");
   const blocksWithChildren = numberOrderedLists(parentBlocks.map((block3) => {
     return resolveChildren(block3, arrayBlocks);
   }));
-  return blocksWithChildren.map((block3) => convertNotionToMd(block3)).join("");
+  return join_default(blocksWithChildren.map((block3) => convertNotionToMd(block3, 1, linkChildPages)), "");
 };
 var convertNotionToMarkdown_default = convertNotionToMarkdown;
 
@@ -56894,7 +57035,7 @@ Message: ${result.detail.join(`
     const result = await response.json();
     return result["examples"];
   }
-  async createExample(inputs, outputs, { datasetId, datasetName, createdAt, exampleId, metadata, split, sourceRunId }) {
+  async createExample(inputs, outputs, { datasetId, datasetName, createdAt, exampleId, metadata, split: split2, sourceRunId }) {
     let datasetId_ = datasetId;
     if (datasetId_ === undefined && datasetName === undefined) {
       throw new Error("Must provide either datasetName or datasetId");
@@ -56912,7 +57053,7 @@ Message: ${result.detail.join(`
       created_at: createdAt_?.toISOString(),
       id: exampleId,
       metadata,
-      split,
+      split: split2,
       source_run_id: sourceRunId
     };
     const response = await this.caller.call(_getFetchImplementation(), `${this.apiUrl}/examples`, {
@@ -57014,8 +57155,8 @@ Message: ${result.detail.join(`
       }
     }
     if (splits !== undefined) {
-      for (const split of splits) {
-        params.append("splits", split);
+      for (const split2 of splits) {
+        params.append("splits", split2);
       }
     }
     if (metadata !== undefined) {
@@ -67177,6 +67318,94 @@ function zipSqlResponse(columns, data, key) {
 }
 var zipSqlResponse_default = zipSqlResponse;
 
+// src/functions/convertMdToPdf.ts
+function getFilenameFromHeaders(resp, fallback) {
+  const cd = resp.headers.get("content-disposition") || "";
+  const matchStar = cd.match(/filename\*=([^;]+)/i);
+  if (matchStar) {
+    const value = trim_default(matchStar[1]);
+    const parts = split_default(value, "''");
+    if (parts.length === 2) {
+      try {
+        return decodeURIComponent(parts[1]);
+      } catch {
+        return parts[1];
+      }
+    }
+    return trim_default(value, `'"`);
+  }
+  const match2 = cd.match(/filename="?([^";]+)"?/i);
+  if (match2)
+    return match2[1];
+  return fallback || "document.pdf";
+}
+async function convertMdToPdf(markdown, options3 = {}, assetHeaders = {}) {
+  const documentParserApiUrl = this.environment.lookup("documentParserApiUrl");
+  const documentParserApiKey = this.environment.lookup("documentParserApiKey");
+  if (!documentParserApiKey) {
+    throw new Error("Document parser API key not found in environment");
+  }
+  if (!documentParserApiUrl) {
+    throw new Error("Document parser API URL not found in environment");
+  }
+  if (!markdown || typeof markdown !== "string") {
+    throw new AbortError("Markdown content is required and must be a string");
+  }
+  const requestBody = {
+    markdown,
+    options: options3,
+    assetHeaders
+  };
+  return await pRetry(async () => {
+    const response = await fetch(`${documentParserApiUrl}/md-to-pdf`, {
+      method: "POST",
+      headers: {
+        Authorization: `Bearer ${documentParserApiKey}`,
+        "Content-Type": "application/json",
+        Accept: "application/pdf",
+        "User-Agent": "truto"
+      },
+      body: JSON.stringify(requestBody)
+    });
+    if (!response.ok) {
+      if (response.status === 429) {
+        throw new Error("Rate limit exceeded");
+      }
+      if (response.status >= 500) {
+        throw new Error(`Server error: ${response.status}`);
+      }
+      const errorText = await response.text();
+      let errorMessage = `PDF conversion failed (${response.status})`;
+      try {
+        const errorData = JSON.parse(errorText);
+        errorMessage = errorData.error || errorMessage;
+        if (errorData.details) {
+          errorMessage += `: ${errorData.details}`;
+        }
+      } catch {
+        errorMessage += `: ${errorText}`;
+      }
+      throw new AbortError(errorMessage);
+    }
+    const contentType = response.headers.get("content-type");
+    if (!contentType || !includes_default(contentType, "application/pdf")) {
+      throw new AbortError(`Expected PDF but received: ${contentType}`);
+    }
+    const arrayBuffer = await response.arrayBuffer();
+    const filename = getFilenameFromHeaders(response, options3.filename);
+    const blob2 = new Blob([arrayBuffer], { type: "application/pdf" });
+    blob2.name = filename;
+    return blob2;
+  }, {
+    retries: 5,
+    maxTimeout: 30000,
+    minTimeout: 2500,
+    onFailedAttempt: (error) => {
+      console.warn(`PDF conversion attempt ${error.attemptNumber} failed:`, error.message);
+    }
+  });
+}
+
 // src/registerJsonataExtensions.ts
 function registerJsonataExtensions(expression) {
   expression.registerFunction("dtFromIso", dtFromIso_default);
@@ -67202,6 +67431,7 @@ function registerJsonataExtensions(expression) {
   expression.registerFunction("difference", function(arr1, arr2) {
     return difference_default(arr1, arr2);
   });
+  expression.registerFunction("convertMdToPdf", convertMdToPdf);
   expression.registerFunction("toNumber", toNumber_default2);
   expression.registerFunction("jsonParse", jsonParse_default);
   expression.registerFunction("getMimeType", getMimeType_default);
@@ -67280,4 +67510,4 @@ export {
   trutoJsonata as default
 };
 
-//# debugId=C7DE5E1DF9429C0D64756E2164756E21
+//# debugId=B3F4257EBAFAE85064756E2164756E21