Support non-identical file names between .wav and .eaf, and recognise media offsets

elpis/gui/.eslintrc

@@ -5,6 +5,9 @@
     "eslint:recommended",
     "plugin:react/recommended",
   ],
+  env: {
+    es6: true,
+  },
   settings: {
     react: {
       version: "detect",

elpis/gui/src/components/Dataset/FileUpload.js

@@ -1,16 +1,64 @@
 import React, {Component} from "react";
 import classNames from "classnames";
 import Dropzone from "react-dropzone";
-import {Button} from "semantic-ui-react";
+import {Button, Message, MessageHeader} from "semantic-ui-react";
 import {fromEvent} from "file-selector";
 import {withTranslation} from "react-i18next";
 import {datasetFiles} from "redux/actions/datasetActions";
 import {connect} from "react-redux";
 
 class FileUpload extends Component {
-    onDrop = (acceptedFiles) => {
+    constructor(props) {
+        super(props);
+        this.state = {missingFiles: []};
+    }
+
+    parseElan = async (file) => {
+        return new Promise((resolve, reject) => {
+            const reader = new window.FileReader();
+
+            reader.readAsText(file);
+            reader.onload = () => {
+                const parser = new window.DOMParser();
+                const eafDoc = parser.parseFromString(reader.result, "application/xml");
+                const wavUrl = eafDoc
+                    .getElementsByTagName("ANNOTATION_DOCUMENT")[0]
+                    .getElementsByTagName("HEADER")[0]
+                    .getElementsByTagName("MEDIA_DESCRIPTOR")[0]
+                    .getAttribute("RELATIVE_MEDIA_URL").split("./")[1];
+
+                resolve(wavUrl);
+            };
+            reader.onerror = () => {
+                reject(reader.error);
+            };
+        });
+    }
+
+    onDrop = async (acceptedFiles) => {
+        // TODO: Behaviour when dropping multiple times is still undefined
+        // (do we empty the dataset?)
+        this.setState({missingFiles: []});
         console.log("files dropped:", acceptedFiles);
 
+        const eafFiles = acceptedFiles
+            .filter(file => file.name.split(".").pop() === "eaf");
+        const wavFileNames = acceptedFiles
+            .filter(file => file.name.split(".").pop() === "wav")
+            .map(file => file.name);
+
+        // for each is not supported with await for some reason...
+        for (let i = 0; i < eafFiles.length; i++) {
+            const parsedWavFile = await this.parseElan(eafFiles[i]);
+            const identicalWavFile = eafFiles[i].name.split(".")[0].concat(".wav");
+
+            if (!wavFileNames.includes(parsedWavFile) && !wavFileNames.includes(identicalWavFile)) {
+                this.setState(prevState => ({
+                    missingFiles: [...prevState.missingFiles, [identicalWavFile, parsedWavFile]],
+                }));
+            }
+        }
+
         var formData = new FormData();
 
         acceptedFiles.forEach(file => {
@@ -25,6 +73,24 @@ class FileUpload extends Component {
 
         return (
             <div className="FileUpload">
+                {this.state.missingFiles.length > 0 && 
+                    <Message negative>
+                        <MessageHeader>
+                            {t("dataset.files.missingAudioFiles")}
+                        </MessageHeader>
+                        <p>
+                            {t("dataset.files.missingAudioFilesDescription")}
+                        </p>
+                        <ul>
+                            {this.state.missingFiles.map(wavFile => {
+                        return (
+                            <li key={wavFile[0]}>
+                                {wavFile[0]} / {wavFile[1]}
+                            </li>
+                        );
+                    })}
+                        </ul>
+                    </Message>}
                 <Dropzone
                     disabled={interactionDisabled}
                     className="dropzone"

elpis/gui/src/translations/en-GB/common.json

@@ -79,7 +79,9 @@
             "tierNameDescription": "or use a tier name.",
             "puncDescription": "Punctuation will be removed from the transcriptions. Add punctuation marks here that you want to be replaced by spaces when removed.",
             "uploadButton": "Upload",
-            "saveButton": "Save Settings"
+            "saveButton": "Save Settings",
+            "missingAudioFiles": "Missing audio files",
+            "missingAudioFilesDescription": "The following audio files are missing:"
         },
         "fileUpload": {
             "audioLabel": "Audio files:",

elpis/gui/src/translations/fr/common.json

@@ -70,7 +70,9 @@
             "tierNameDescription": "ou utilisez un nom d’échelon.",
             "puncDescription": "La ponctuation sera supprimée des transcriptions. Ajoutez ici les signes de ponctuation que vous désirez remplacer par des espaces.",
             "uploadButton": "Téléverser",
-            "saveButton": "Enregistrer les paramètres"
+            "saveButton": "Enregistrer les paramètres",
+            "missingAudioFiles": "Fichiers audio manquants",
+            "missingAudioFilesDescription": "Les fichiers audio suivants sont manquants:"
         },
         "fileUpload": {
             "audioLabel": "Fichiers audio :",

elpis/transformer/elan.py

@@ -216,11 +216,20 @@ def import_eaf_file(eaf_paths: List[str],
             end = annotation[1]
             annotation = annotation[2]
 
+            time_origin = input_eaf.get_linked_files()[0].get("TIME_ORIGIN")
+            offset = int(time_origin) if time_origin is not None else 0
+
+            # If matching file, use that. Otherwise, fall back to the relative media url.
+            if os.path.isfile(f"{file_name}.wav"):
+                audio_file_name = f"{file_name}.wav"
+            else:
+                audio_file_name = input_eaf.get_linked_files()[0].get("RELATIVE_MEDIA_URL")
+
             utterance = {
-                "audio_file_name": f"{file_name}.wav",
+                "audio_file_name": audio_file_name,
                 "transcript": annotation,
-                "start_ms": start,
-                "stop_ms": end,
+                "start_ms": start + offset,
+                "stop_ms": end + offset,
                 "speaker_id": speaker_id
             }