@truto/truto-jsonata 1.0.44 → 1.0.46

package/dist/esm/index.js

@@ -3,9 +3,9 @@ import jsonata from "jsonata";
 
 // src/registerJsonataExtensions.ts
 import {
-  castArray as castArray5,
+  castArray as castArray6,
   chunk as chunk2,
-  compact as compact2,
+  compact as compact3,
   difference,
   filter as filter2,
   find as find2,
@@ -13,7 +13,7 @@ import {
   flattenDeep as flattenDeep3,
   flattenDepth,
   groupBy as groupBy3,
-  join as join2,
+  join as join3,
   keyBy,
   omit as omit2,
   orderBy,
@@ -2192,8 +2192,98 @@ var convertMarkdownToSlack = (text) => {
 };
 var convertMarkdownToSlack_default = convertMarkdownToSlack;
 
+// src/functions/convertMdToPdf.ts
+import pRetry, { AbortError } from "p-retry";
+import { includes, split, trim } from "lodash-es";
+function getFilenameFromHeaders(resp, fallback) {
+  const cd = resp.headers.get("content-disposition") || "";
+  const matchStar = cd.match(/filename\*=([^;]+)/i);
+  if (matchStar) {
+    const value = trim(matchStar[1]);
+    const parts = split(value, "''");
+    if (parts.length === 2) {
+      try {
+        return decodeURIComponent(parts[1]);
+      } catch {
+        return parts[1];
+      }
+    }
+    return trim(value, `'"`);
+  }
+  const match = cd.match(/filename="?([^";]+)"?/i);
+  if (match)
+    return match[1];
+  return fallback || "document.pdf";
+}
+async function convertMdToPdf(markdown, options = {}, 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,
+    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(contentType, "application/pdf")) {
+      throw new AbortError(`Expected PDF but received: ${contentType}`);
+    }
+    const arrayBuffer = await response.arrayBuffer();
+    const filename = getFilenameFromHeaders(response, options.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/functions/convertNotionToMarkdown.ts
-import { castArray as castArray3, flatten, map as map3, reject, repeat } from "lodash-es";
+import { castArray as castArray3, flatten, join as join2, map as map3, reject, repeat } from "lodash-es";
 var formatPlainText = (x) => {
   if (x) {
     if (x.href) {
@@ -2223,7 +2313,7 @@ var formatPlainText = (x) => {
     return textToReturn;
   }
 };
-var convertNotionToMd = function(block, level = 1) {
+var convertNotionToMd = function(block, level = 1, linkChildPages = false) {
   const n = `
 
 `;
@@ -2235,16 +2325,16 @@ var convertNotionToMd = function(block, level = 1) {
     case "bookmark":
       return `[${data.url}](data.url)${n}`;
     case "bulleted_list_item":
-      childData = map3(block.children, (child) => convertNotionToMd(child, level + 1)).join(repeat("\t", level));
-      return `- ${plainText.join("")}
+      childData = join2(map3(block.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat("\t", level));
+      return `- ${join2(plainText, "")}
 ` + (childData ? `${repeat("\t", level)}${childData}` : "");
     case "numbered_list_item":
-      childData = map3(block.children, (child) => convertNotionToMd(child, level + 1)).join(repeat("\t", level));
-      return `${block.number}. ${plainText.join("")}
+      childData = join2(map3(block.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat("\t", level));
+      return `${block.number}. ${join2(plainText, "")}
 ` + (childData ? `${repeat("\t", level)}${childData}` : "");
     case "quote":
     case "callout":
-      return `> ${plainText.join("")}${n}`;
+      return `> ${join2(plainText, "")}${n}`;
     case "code":
       return `\`\`\`${data.language}
 ${plainText}
@@ -2256,51 +2346,60 @@ ${plainText}
     case "equation":
       return data.expression;
     case "paragraph":
-      return plainText.join("") + n;
+      return join2(plainText, "") + n;
     case "video":
     case "pdf":
     case "file":
       return `[${caption.length ? caption : "File"}](${data.file ? data.file.url : data.external ? data.external.url : ""})${n}`;
     case "heading_1":
-      return `# ${plainText.join("")}${n}`;
+      return `# ${join2(plainText, "")}${n}`;
     case "heading_2":
-      return `## ${plainText.join("")}${n}`;
+      return `## ${join2(plainText, "")}${n}`;
     case "heading_3":
-      return `### ${plainText.join("")}${n}`;
+      return `### ${join2(plainText, "")}${n}`;
     case "heading_4":
-      return `#### ${plainText.join("")}${n}`;
+      return `#### ${join2(plainText, "")}${n}`;
     case "heading_5":
-      return `##### ${plainText.join("")}${n}`;
+      return `##### ${join2(plainText, "")}${n}`;
     case "heading_6":
-      return `###### ${plainText.join("")}${n}`;
+      return `###### ${join2(plainText, "")}${n}`;
     case "image":
       return `![${caption.length ? caption : "Image"}](${data.file ? data.file.url : data.external ? data.external.url : ""})${n}`;
+    case "child_page":
+      if (linkChildPages) {
+        const pageTitle = data.title || "Untitled Page";
+        const pageUrl = `https://www.notion.so/${block.id.replace(/-/g, "")}`;
+        return `[${pageTitle}](${pageUrl})${n}`;
+      } else {
+        childData = join2(map3(block.children, (child) => convertNotionToMd(child, level, linkChildPages)), "");
+        return (data.title || "Untitled Page") + n + childData;
+      }
     case "table":
       if (block.children) {
         const firstChild = block.children[0];
         const remainingChildren = block.children.slice(1);
-        const header = `| ${firstChild.table_row.cells.map((x) => x.map((y) => formatPlainText(y)).join("")).join(" | ")} |
+        const header = `| ${join2(firstChild.table_row.cells.map((x) => join2(x.map((y) => formatPlainText(y)), "")), " | ")} |
 `;
         const divider = `|${repeat("---|", 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 = join2(remainingChildren.map((row) => {
+          return `| ${join2(row.table_row.cells.map((x) => join2(x.map((y) => formatPlainText(y)), "")), " | ")} |`;
+        }), `
 `);
         return `${header}${divider}${rows}${n}`;
       }
       return `Table as CSV
 `;
     case "table_row":
-      return `| ${data.cells.map((x) => x.map((y) => formatPlainText(y)).join("")).join(" | ")} |
+      return `| ${join2(data.cells.map((x) => join2(x.map((y) => formatPlainText(y)), "")), " | ")} |
 `;
     case "to_do":
-      childData = map3(block.children, (child) => convertNotionToMd(child, level + 1)).join(repeat("\t", level));
-      return `- [${data.checked ? "X" : " "}] ${plainText.join("")}
+      childData = join2(map3(block.children, (child) => convertNotionToMd(child, level + 1, linkChildPages)), repeat("\t", level));
+      return `- [${data.checked ? "X" : " "}] ${join2(plainText, "")}
 ` + (childData ? `${repeat("\t", level)}${childData}` : "");
     default:
-      childData = map3(block.children, (child) => convertNotionToMd(child)).join("");
-      return plainText.join("") + n + childData;
+      childData = join2(map3(block.children, (child) => convertNotionToMd(child, level, linkChildPages)), "");
+      return join2(plainText, "") + n + childData;
   }
 };
 var resolveChildren = function(block, blocks) {
@@ -2350,13 +2449,13 @@ var insertNewLinesBetweenLists = function(blocks) {
     return block;
   }));
 };
-var convertNotionToMarkdown = function(blocks) {
+var convertNotionToMarkdown = function(blocks, linkChildPages = false) {
   const arrayBlocks = insertNewLinesBetweenLists(castArray3(blocks));
   const parentBlocks = reject(arrayBlocks, (block) => block.parent?.type === "block_id");
   const blocksWithChildren = numberOrderedLists(parentBlocks.map((block) => {
     return resolveChildren(block, arrayBlocks);
   }));
-  return blocksWithChildren.map((block) => convertNotionToMd(block)).join("");
+  return join2(blocksWithChildren.map((block) => convertNotionToMd(block, 1, linkChildPages)), "");
 };
 var convertNotionToMarkdown_default = convertNotionToMarkdown;
 
@@ -2716,12 +2815,12 @@ var firstNonEmpty_default = firstNonEmpty;
 // src/functions/generateEmbeddingsCohere.ts
 import { castArray as castArray4, chunk, isEmpty as isEmpty4 } from "lodash-es";
 import pMap from "p-map";
-import pRetry, { AbortError } from "p-retry";
+import pRetry2, { AbortError as AbortError2 } from "p-retry";
 async function generateEmbeddingsCohere(body, api_key) {
   if (!isEmpty4(body.texts)) {
     const chunks = chunk(castArray4(body.texts), 20);
     return await pMap(chunks, async (chunk2) => {
-      return await pRetry(async () => {
+      return await pRetry2(async () => {
         const response = await fetch("https://api.cohere.com/v2/embed", {
           method: "POST",
           headers: {
@@ -2739,7 +2838,7 @@ async function generateEmbeddingsCohere(body, api_key) {
           if (response.status >= 500) {
             throw new Error("Server error");
           }
-          throw new AbortError(await response.text());
+          throw new AbortError2(await response.text());
         }
         return await response.json();
       }, { retries: 5, maxTimeout: 5000, minTimeout: 2500 });
@@ -2747,7 +2846,7 @@ async function generateEmbeddingsCohere(body, api_key) {
       concurrency: 1
     });
   } else if (!isEmpty4(body.images)) {
-    return await pRetry(async () => {
+    return await pRetry2(async () => {
       const response = await fetch("https://api.cohere.com/v2/embed", {
         method: "POST",
         headers: {
@@ -2765,7 +2864,7 @@ async function generateEmbeddingsCohere(body, api_key) {
         if (response.status >= 500) {
           throw new Error("Server error");
         }
-        throw new AbortError(await response.text());
+        throw new AbortError2(await response.text());
       }
       return await response.json();
     }, { retries: 5, maxTimeout: 5000, minTimeout: 2500 });
@@ -2822,6 +2921,18 @@ function jsonParse(str) {
 }
 var jsonParse_default = jsonParse;
 
+// src/functions/jsonToCsv.ts
+import { Parser } from "@json2csv/plainjs";
+import { castArray as castArray5, compact as compact2, isEmpty as isEmpty5 } from "lodash-es";
+function jsonToCsv(json, options) {
+  const jsonArray = compact2(castArray5(json));
+  if (isEmpty5(jsonArray)) {
+    return "";
+  }
+  const parser = new Parser(options);
+  return parser.parse(jsonArray);
+}
+
 // src/functions/jsToXml.ts
 import { js2xml } from "xml-js";
 function jsToXml_default(json, options = { compact: true, spaces: 4 }) {
@@ -2849,14 +2960,14 @@ function mostSimilar(value, possibleValues, threshold = 0.8) {
 var mostSimilar_default = mostSimilar;
 
 // src/functions/parseDocument.ts
-import pRetry2, { AbortError as AbortError2 } from "p-retry";
+import pRetry3, { AbortError as AbortError3 } from "p-retry";
 async function parseDocument(file, fileType) {
   const documentParserApiUrl = this.environment.lookup("documentParserApiUrl");
   const documentParserApiKey = this.environment.lookup("documentParserApiKey");
   if (!documentParserApiKey) {
     throw new Error("API key not found in environment");
   }
-  return await pRetry2(async () => {
+  return await pRetry3(async () => {
     const response = await fetch(`${documentParserApiUrl}/parse`, {
       method: "POST",
       headers: {
@@ -2874,7 +2985,7 @@ async function parseDocument(file, fileType) {
       if (response.status >= 500) {
         throw new Error("Server error");
       }
-      throw new AbortError2(await response.text());
+      throw new AbortError3(await response.text());
     }
     const data = await response.json();
     return data.content;
@@ -3084,6 +3195,7 @@ function registerJsonataExtensions(expression) {
   expression.registerFunction("difference", function(arr1, arr2) {
     return difference(arr1, arr2);
   });
+  expression.registerFunction("convertMdToPdf", convertMdToPdf);
   expression.registerFunction("toNumber", toNumber_default);
   expression.registerFunction("jsonParse", jsonParse_default);
   expression.registerFunction("getMimeType", getMimeType_default);
@@ -3100,10 +3212,10 @@ function registerJsonataExtensions(expression) {
   expression.registerFunction("jsToXml", jsToXml_default);
   expression.registerFunction("generateEmbeddingsCohere", generateEmbeddingsCohere_default);
   expression.registerFunction("groupBy", function(array, key) {
-    return groupBy3(castArray5(array), key);
+    return groupBy3(castArray6(array), key);
   });
   expression.registerFunction("keyBy", function(array, key) {
-    return keyBy(castArray5(array), key);
+    return keyBy(castArray6(array), key);
   });
   expression.registerFunction("pick", function(obj, keys) {
     return pick(obj, keys);
@@ -3112,28 +3224,28 @@ function registerJsonataExtensions(expression) {
     return omit2(obj, keys);
   });
   expression.registerFunction("compact", function(arr) {
-    return compact2(castArray5(arr));
+    return compact3(castArray6(arr));
   });
   expression.registerFunction("join", function(arr, separator) {
-    return join2(castArray5(arr), separator);
+    return join3(castArray6(arr), separator);
   });
   expression.registerFunction("orderBy", function(arr, attr, order) {
-    return orderBy(castArray5(arr), attr, order);
+    return orderBy(castArray6(arr), attr, order);
   });
   expression.registerFunction("find", function(arr, attr) {
-    return find2(castArray5(arr), attr);
+    return find2(castArray6(arr), attr);
   });
   expression.registerFunction("lofilter", function(arr, attr) {
-    return filter2(castArray5(arr), attr);
+    return filter2(castArray6(arr), attr);
   });
   expression.registerFunction("values", function(obj) {
     return values(obj);
   });
   expression.registerFunction("chunk", function(arr, size) {
-    return chunk2(castArray5(arr), size);
+    return chunk2(castArray6(arr), size);
   });
   expression.registerFunction("wrap", function(value, wrapper, endWrapper) {
-    return join2([wrapper, value, endWrapper || wrapper], "");
+    return join3([wrapper, value, endWrapper || wrapper], "");
   });
   expression.registerFunction("parseDocument", parseDocument_default);
   expression.registerFunction("recursiveCharacterTextSplitter", recursiveCharacterTextSplitter);
@@ -3143,14 +3255,15 @@ function registerJsonataExtensions(expression) {
   expression.registerFunction("parseQuery", parseQuery);
   expression.registerFunction("stringifyQuery", stringifyQuery);
   expression.registerFunction("flatten", function(arr) {
-    return flatten2(castArray5(arr));
+    return flatten2(castArray6(arr));
   });
   expression.registerFunction("flattenDeep", function(arr) {
-    return flattenDeep3(castArray5(arr));
+    return flattenDeep3(castArray6(arr));
   });
   expression.registerFunction("flattenDepth", function(arr, depth) {
-    return flattenDepth(castArray5(arr), depth);
+    return flattenDepth(castArray6(arr), depth);
   });
+  expression.registerFunction("jsonToCsv", jsonToCsv);
   return expression;
 }
 
@@ -3162,4 +3275,4 @@ export {
   trutoJsonata as default
 };
 
-//# debugId=EADA1C7ED032523764756E2164756E21
+//# debugId=555CAE470CE91F1064756E2164756E21