@sie-js/vkp 1.0.3 → 1.0.5

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@sie-js/vkp",
-	"version": "1.0.3",
+	"version": "1.0.5",
 	"description": "Parser and utils for the .VKP files format which is used in the V-Klay.",
 	"main": "src/index.js",
 	"type": "module",
@@ -13,7 +13,8 @@
 	"license": "MIT",
 	"dependencies": {
 		"iconv-lite": "^0.6.3",
-		"moo": "^0.5.2"
+		"moo": "^0.5.2",
+		"rtf-parser": "^1.3.3"
 	},
 	"devDependencies": {
 		"glob": "^10.3.12",

package/src/index.js

@@ -1,6 +1,7 @@
 import iconv from 'iconv-lite';
 import { vkpRawParser } from './parser.js';
 import { VkpParseError } from './VkpParseError.js';
+import RTFParser from 'rtf-parser';
 
 const DEFAULT_PRAGMAS = {
 	warn_no_old_on_apply:		true,
@@ -116,8 +117,49 @@ function vkpDetectContent(text) {
 	return "UNKNOWN";
 }
 
+// CP1251 -> UTF-8 + CRLF -> LF (with RTF support)
+async function vkpNormalizeWithRTF(text) {
+	if (!Buffer.isBuffer(text))
+		throw new Error(`Patch text is not Buffer!`);
+
+	if (text.indexOf('{\\rtf1') >= 0) {
+		// Strip RTF images
+		while (true) {
+			let pictureIndex = text.indexOf('{\\pict');
+			if (pictureIndex >= 0) {
+				let pictureEndIndex = text.indexOf('}', pictureIndex);
+				if (pictureIndex >= 0) {
+					text = Buffer.concat([ text.slice(0, pictureIndex), text.slice(pictureEndIndex + 1) ]);
+					continue;
+				}
+			}
+			break;
+		}
+
+		text = text.toString('utf-8').replace(/{\\pict.*?\}/gsi, ''); // remove pictures
+		let parsed = await new Promise((resolve, reject) => {
+			RTFParser.string(text, (err, doc) => {
+				if (err) {
+					reject(err);
+				} else {
+					resolve(doc);
+				}
+			});
+		});
+
+		let lines = [];
+		for (let p of parsed.content) {
+			lines.push(p.content.map((s) => s.value).join(''));
+		}
+		return lines.join('\n');
+	}
+	return iconv.decode(text, 'windows-1251').replace(/(\r\n|\n|\r)/g, "\n");
+}
+
 // CP1251 -> UTF-8 + CRLF -> LF
 function vkpNormalize(text) {
+	if (!Buffer.isBuffer(text))
+		throw new Error(`Patch text is not Buffer!`);
 	return iconv.decode(text, 'windows-1251').replace(/(\r\n|\n|\r)/g, "\n");
 }
 
@@ -126,4 +168,4 @@ function vkpCanonicalize(text) {
 	return iconv.encode(text.replace(/(\r\n|\n|\r)/g, "\r\n"), 'windows-1251');
 }
 
-export { vkpParse, vkpRawParser, vkpNormalize, vkpCanonicalize, vkpDetectContent };
+export { vkpParse, vkpRawParser, vkpNormalize, vkpNormalizeWithRTF, vkpCanonicalize, vkpDetectContent };