@truto/truto-jsonata 1.0.11 → 1.0.13

package/dist/module.js

@@ -1,5 +1,5 @@
 import $hgUW1$jsonata from "jsonata";
-import {difference as $hgUW1$difference, groupBy as $hgUW1$groupBy, castArray as $hgUW1$castArray, keyBy as $hgUW1$keyBy, pick as $hgUW1$pick, omit as $hgUW1$omit, compact as $hgUW1$compact, join as $hgUW1$join, orderBy as $hgUW1$orderBy, find as $hgUW1$find, filter as $hgUW1$filter, values as $hgUW1$values, chunk as $hgUW1$chunk, reduce as $hgUW1$reduce, each as $hgUW1$each, get as $hgUW1$get, isEmpty as $hgUW1$isEmpty, concat as $hgUW1$concat, isEqual as $hgUW1$isEqual, flatMap as $hgUW1$flatMap, map as $hgUW1$map, set as $hgUW1$set, isString as $hgUW1$isString, flattenDeep as $hgUW1$flattenDeep, sortBy as $hgUW1$sortBy, repeat as $hgUW1$repeat, flatten as $hgUW1$flatten, reject as $hgUW1$reject, isNil as $hgUW1$isNil, isFunction as $hgUW1$isFunction, toNumber as $hgUW1$toNumber} from "lodash-es";
+import {difference as $hgUW1$difference, groupBy as $hgUW1$groupBy, castArray as $hgUW1$castArray, keyBy as $hgUW1$keyBy, pick as $hgUW1$pick, omit as $hgUW1$omit, compact as $hgUW1$compact, join as $hgUW1$join, orderBy as $hgUW1$orderBy, find as $hgUW1$find, filter as $hgUW1$filter, values as $hgUW1$values, chunk as $hgUW1$chunk, reduce as $hgUW1$reduce, each as $hgUW1$each, get as $hgUW1$get, isEmpty as $hgUW1$isEmpty, concat as $hgUW1$concat, isEqual as $hgUW1$isEqual, flatMap as $hgUW1$flatMap, map as $hgUW1$map, set as $hgUW1$set, isString as $hgUW1$isString, flattenDeep as $hgUW1$flattenDeep, sortBy as $hgUW1$sortBy, repeat as $hgUW1$repeat, flatten as $hgUW1$flatten, reject as $hgUW1$reject, isNil as $hgUW1$isNil, isFunction as $hgUW1$isFunction, toNumber as $hgUW1$toNumber, includes as $hgUW1$includes} from "lodash-es";
 import {DateTime as $hgUW1$DateTime} from "luxon";
 import {Lexer as $hgUW1$Lexer, marked as $hgUW1$marked} from "marked";
 import {diceCoefficient as $hgUW1$diceCoefficient} from "dice-coefficient";
@@ -13,10 +13,8 @@ import $hgUW1$pmap from "p-map";
 import $hgUW1$pretry from "p-retry";
 import {Buffer as $hgUW1$Buffer} from "buffer";
 import {parseOfficeAsync as $hgUW1$parseOfficeAsync} from "officeparser";
-import {fileTypeFromBuffer as $hgUW1$fileTypeFromBuffer} from "file-type";
 import {resolvePDFJS as $hgUW1$resolvePDFJS} from "pdfjs-serverless";
 import {RecursiveCharacterTextSplitter as $hgUW1$RecursiveCharacterTextSplitter} from "@langchain/textsplitters";
-import {ReadableStream as $hgUW1$ReadableStream} from "node:stream/web";
 
 function $parcel$interopDefault(a) {
   return a && a.__esModule ? a.default : a;
@@ -1402,6 +1400,19 @@ var $bf2fdaa075f6a5e5$export$2e2bcd8739ae039 = $bf2fdaa075f6a5e5$var$generateEmb
 
 
 
+var $65fc862361d7fe06$require$Buffer = $hgUW1$Buffer;
+async function $65fc862361d7fe06$var$bufferToString(value, encoding) {
+    if (value instanceof ReadableStream) {
+        const chunks = [];
+        for await (const chunk of value)chunks.push($65fc862361d7fe06$require$Buffer.from(chunk));
+        const buffer = $65fc862361d7fe06$require$Buffer.concat(chunks);
+        return buffer.toString(encoding);
+    }
+    return value.toString(encoding);
+}
+var $65fc862361d7fe06$export$2e2bcd8739ae039 = $65fc862361d7fe06$var$bufferToString;
+
+
 
 var $f29f00a7536a3143$require$Buffer = $hgUW1$Buffer;
 async function $f29f00a7536a3143$var$parsePdf(buffer) {
@@ -1427,16 +1438,23 @@ async function $f29f00a7536a3143$var$parsePdf(buffer) {
     // Return the results as JSON
     return (0, $hgUW1$join)(output, "\n");
 }
-async function $f29f00a7536a3143$var$parseDocument(file) {
-    if (file) {
+async function $f29f00a7536a3143$var$parseDocument(file, fileType) {
+    let buffer;
+    if (file instanceof ReadableStream) {
         const chunks = [];
         for await (const chunk of file)chunks.push($f29f00a7536a3143$require$Buffer.from(chunk));
-        const buffer = $f29f00a7536a3143$require$Buffer.concat(chunks);
-        const fileExt = (await (0, $hgUW1$fileTypeFromBuffer)(buffer))?.ext;
-        if (fileExt === "pdf") return await $f29f00a7536a3143$var$parsePdf(buffer);
-        return await (0, $hgUW1$parseOfficeAsync)(buffer);
+        buffer = $f29f00a7536a3143$require$Buffer.concat(chunks);
     }
-    return await (0, $hgUW1$parseOfficeAsync)(file);
+    if ((0, $hgUW1$includes)([
+        "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+        "application/vnd.openxmlformats-officedocument.presentationml.presentation",
+        "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        "application/vnd.oasis.opendocument.text",
+        "application/vnd.oasis.opendocument.presentation",
+        "application/vnd.oasis.opendocument.spreadsheet"
+    ], fileType)) return await (0, $hgUW1$parseOfficeAsync)(buffer);
+    if (fileType === "application/pdf") return await $f29f00a7536a3143$var$parsePdf(buffer);
+    return (0, $65fc862361d7fe06$export$2e2bcd8739ae039)(buffer, "utf-8");
 }
 var $f29f00a7536a3143$export$2e2bcd8739ae039 = $f29f00a7536a3143$var$parseDocument;
 
@@ -1452,28 +1470,32 @@ function $a12871c9a32cbf3b$export$e600492876ee595b(text, options = {
 
 
 
-
-
 var $6b4f9d8bcf8ab446$require$Buffer = $hgUW1$Buffer;
-async function $6b4f9d8bcf8ab446$var$getDataUri(file) {
-    if (file instanceof (0, $hgUW1$ReadableStream)) {
+async function $6b4f9d8bcf8ab446$var$getDataUri(file, mimeType) {
+    if (!mimeType) throw new Error("Mime type is required");
+    if (file instanceof ReadableStream) {
         const chunks = [];
         for await (const chunk of file)chunks.push($6b4f9d8bcf8ab446$require$Buffer.from(chunk));
         const buffer = $6b4f9d8bcf8ab446$require$Buffer.concat(chunks);
         const base64Image = buffer.toString("base64");
-        const mimeType = (await (0, $hgUW1$fileTypeFromBuffer)(buffer))?.mime;
         // Construct the data URI for a PNG image
         return `data:${mimeType};base64,${base64Image}`;
     }
     const arrayBuffer = file instanceof Blob ? await file.arrayBuffer() : file;
     const buffer = $6b4f9d8bcf8ab446$require$Buffer.from(arrayBuffer);
     const base64Image = buffer.toString("base64");
-    const mimeType = (await (0, $hgUW1$fileTypeFromBuffer)(buffer))?.mime;
     return `data:${mimeType};base64,${base64Image}`;
 }
 var $6b4f9d8bcf8ab446$export$2e2bcd8739ae039 = $6b4f9d8bcf8ab446$var$getDataUri;
 
 
+async function $5e08a3608e719b52$var$teeStream(stream) {
+    return stream.tee();
+}
+var $5e08a3608e719b52$export$2e2bcd8739ae039 = $5e08a3608e719b52$var$teeStream;
+
+
+
 function $3e1c03caddb74419$export$2e2bcd8739ae039(expression) {
     expression.registerFunction("dtFromIso", (0, $9e00534fc92334a4$export$2e2bcd8739ae039));
     expression.registerFunction("base64decode", (0, $52da679a67e0b066$export$2e2bcd8739ae039));
@@ -1554,6 +1576,8 @@ function $3e1c03caddb74419$export$2e2bcd8739ae039(expression) {
     expression.registerFunction("parseDocument", (0, $f29f00a7536a3143$export$2e2bcd8739ae039));
     expression.registerFunction("recursiveCharacterTextSplitter", (0, $a12871c9a32cbf3b$export$e600492876ee595b));
     expression.registerFunction("getDataUri", (0, $6b4f9d8bcf8ab446$export$2e2bcd8739ae039));
+    expression.registerFunction("teeStream", (0, $5e08a3608e719b52$export$2e2bcd8739ae039));
+    expression.registerFunction("bufferToString", (0, $65fc862361d7fe06$export$2e2bcd8739ae039));
     return expression;
 }