fix: ignore interpolations in source data

bin/xlf-translate

@@ -3,7 +3,7 @@
 
 const yaml = require('js-yaml');
 const fs = require('fs');
-const xml = require('xml2js');
+const cheerio = require('cheerio');
 const translate = require('../translate');
 
 const args = (() => {
@@ -28,26 +28,22 @@ const args = (() => {
         action: 'storeTrue'
     });
 
+    parser.addArgument([ '-e', '--encoding' ], {
+        help: 'Specifies the XLIFF file encoding. Default is "utf-8".',
+        defaultValue: 'utf-8'
+    });
+
     return parser.parseArgs();
 })();
 
 const langData = fs.readFileSync(args.lang_file);
 const lang = yaml.safeLoad(langData);
 
-const messageData = fs.readFileSync(args.file);
-const parser = new xml.Parser();
-parser.parseString(messageData, (err, result) => {
-    if (err) {
-        console.log('Oops, error occurred');
-        throw err;
-    }
-
-    const stats = translate(result, lang, args.force);
-
-    const builder = new xml.Builder();
-    const out = builder.buildObject(result);
+const messageData = fs.readFileSync(args.file, { encoding: args.encoding });
+const messages = cheerio.load(messageData, { xmlMode: true, decodeEntities: false });
+const stats = translate(messages, lang, args.force);
+const out = messages.html();
 
-    fs.writeFileSync(args.file, out);
-    console.log(`Done. Replaced ${stats.count} units, skipped ${stats.skip}.`);
-});
+fs.writeFileSync(args.file, out);
+console.log(`Done. Replaced ${stats.count} units, skipped ${stats.skip}.`);
 

package.json

@@ -14,9 +14,9 @@
   "author": "Tsvetomir Tsonev",
   "license": "BSD-3-Clause",
   "dependencies": {
+    "cheerio": "0.22.0",
     "js-yaml": "^3.7.0",
-    "jsonpath": "^0.2.9",
-    "xml2js": "^0.4.17"
+    "jsonpath": "^0.2.9"
   },
   "repository": {
     "type": "git",

sample/messages.fr.xlf

@@ -20,6 +20,11 @@
         <note priority="1" from="description">A goodbye message for the localized component</note>
         <note priority="1" from="meaning">localized.component.goodbye</note>
       </trans-unit>
+      <trans-unit id="225c24c98a62f1f8974e1c1206f6eac06d7e5b62" datatype="html">
+        <source>Créé par <x id="INTERPOLATION"/> le <x id="INTERPOLATION_1"/></source>
+        <target/>
+        <note priority="1" from="description">Info création par qui / quand</note>
+      </trans-unit>
     </body>
   </file>
-</xliff>
+</xliff>

sample/messages.xlf

@@ -20,6 +20,11 @@
         <note priority="1" from="description">A goodbye message for the localized component</note>
         <note priority="1" from="meaning">localized.component.goodbye</note>
       </trans-unit>
+      <trans-unit id="225c24c98a62f1f8974e1c1206f6eac06d7e5b62" datatype="html">
+        <source>Créé par <x id="INTERPOLATION"/> le <x id="INTERPOLATION_1"/></source>
+        <target/>
+        <note priority="1" from="description">Info création par qui / quand</note>
+      </trans-unit>
     </body>
   </file>
 </xliff>

spec/translate.spec.js

@@ -2,7 +2,7 @@
 
 const yaml = require('js-yaml');
 const fs = require('fs');
-const xml = require('xml2js');
+const cheerio = require('cheerio');
 const translate = require('../translate');
 
 const lang = {
@@ -18,38 +18,53 @@ describe("translate", function() {
     let messages;
     let units;
 
+    const target = (index, text) => units.eq(index).find('target').text(text);
+
     beforeEach(() => {
-        messages = require('./message-data');
-        units = messages.xliff.file[0].body[0]['trans-unit'];
+        const messageData = fs.readFileSync('./sample/messages.xlf', { encoding: 'utf-8' });
+        messages = cheerio.load(messageData, { xmlMode: true, decodeEntities: false });
+        units = messages('trans-unit');
     });
 
     it("fills tagged units", function() {
         translate(messages, lang);
 
-        expect(units[1].target[0]).toBe(lang.localized.component.hello);
-        expect(units[2].target[0]).toBe(lang.localized.component.goodbye);
+        expect(target(1)).toBe(lang.localized.component.hello);
+        expect(target(2)).toBe(lang.localized.component.goodbye);
     });
 
     it("skips translated tagged units", function() {
-        units[1].target[0] = 'Foo';
+        target(1, 'Foo');
 
         translate(messages, lang);
 
-        expect(units[1].target[0]).toBe('Foo');
+        expect(target(1)).toBe('Foo');
     });
 
     it("overwrites translated tagged units with force", function() {
-        units[1].target[0] = 'Foo';
+        target(1, 'Foo');
 
         translate(messages, lang, true);
 
-        expect(units[1].target[0]).toBe(lang.localized.component.hello);
+        expect(target(1)).toBe(lang.localized.component.hello);
     });
 
     it("skips regular units", function() {
         translate(messages, lang);
 
-        expect(units[0].target[0]).toBe('');
+        expect(target(0)).toBe('');
+    });
+
+    it("skips units with interpolations", function() {
+        translate(messages, lang);
+
+        expect(target(3)).toBe('');
+    });
+
+    it("outputs interpolations", function() {
+        translate(messages, lang);
+
+        expect(units.eq(3).find('source').html()).toBe('Créé par <x id="INTERPOLATION"/> le <x id="INTERPOLATION_1"/>');
     });
 });
 

translate.js

@@ -5,28 +5,28 @@ const jp = require('jsonpath');
 /**
  * Fills in translations in XLIFF files based on 'meaning' metadata as a key.
  *
- * @param messages an XLIFF document parsed with xml2js
+ * @param doc an XLIFF document parsed with cheerio
  * @param lang a tree of keys and their corresponding translations
  * @return statistics the messages will be filled in-place
  */
-const translate = (messages, lang, force) => {
-    const units = jp.query(messages, '$..["trans-unit"]')[0];
+const translate = (doc, lang, force) => {
+    const units = doc('trans-unit').toArray();
     let stats = { count: 0, skip: 0 };
 
     units
-        .filter(hasNotes)
+        .filter(unit => doc(unit).find('note').length > 0)
         .map(unit => ({
-            target: unit.target,
-            id: meaning(unit)
+            target: doc(unit).find('target'),
+            id: doc(unit).find('note[from=meaning]').text()
         }))
-        .filter(d => isKey(d.id))
+        .filter(d => d.id && isKey(d.id))
         .forEach(d => {
             const query = '$.' + d.id;
             const match = jp.query(lang, query);
 
             if (match.length === 1) {
-                if (d.target[0] === '' || force) {
-                    d.target[0] = match[0];
+                if (d.target.text() === '' || force) {
+                    d.target.text(match[0]);
                     stats.count++;
                 } else if (!force) {
                     stats.skip++;
@@ -37,9 +37,8 @@ const translate = (messages, lang, force) => {
     return stats;
 };
 
+
 const isKey = val => !!val.match(/^[a-zA-Z0-9.]*$/);
-const hasNotes = u => !!u.note;
-const meaning = unit => unit.note.find(note => note['$'].from === 'meaning')['_'];
 
 module.exports = translate;