add xmp metadata parsing for videos (#4882)

Author: taylorli2000

fixtures/empty_metadata.jpg

@@ -48,6 +48,11 @@ module.exports = function(config) {
         files: [
             'scripts/tests.ts',
             'scripts/**/*.html',
+            { 
+                pattern: 'fixtures/**/*',
+                watched: false,
+                included: false,
+            }
         ],
 
         ngHtml2JsPreprocessor: {

fixtures/empty_metadata.mov

@@ -29,7 +29,6 @@
   "main": "scripts/index.js",
   "types": "scripts/core/superdesk-api.d.ts",
   "dependencies": {
-    "@metadata/exif": "github:superdesk/exif#431066d",
     "@popperjs/core": "2.10.2",
     "@sourcefabric/common": "0.0.69",
     "@sourcefabric/date-fns-tz": "^1.2.1",
@@ -44,6 +43,7 @@
     "@types/react-dom": "16.8.0",
     "@types/react-redux": "7.1.9",
     "@typescript-eslint/parser": "5.57.0",
+    "@uswriting/exiftool": "1.0.5",
     "angular": "1.6.9",
     "angular-contenteditable": "0.3.9",
     "angular-dynamic-locale": "0.1.32",

fixtures/metadata.jpg

@@ -1,3 +1,6 @@
+import {IPTCMetadata, XMPMetadata} from 'superdesk-api';
+import {getInvertObject} from 'utils/object';
+
 export enum ITEM_STATE {
     /**
      * Item created in user workspace.
@@ -114,3 +117,53 @@ export const CANCELED_STATES = KILLED_STATES.concat([ITEM_STATE.SPIKED]);
  * KILLED | RECALLED | UNPUBLISHED | SPIKED | SCHEDULED
  */
 export const READONLY_STATES = CANCELED_STATES.concat([ITEM_STATE.SCHEDULED]);
+
+export const IPTC_XMP_TAGS = {
+    'IPTC:Destination': 'XMP:Destination',
+    'IPTC:ServiceIdentifier': 'XMP:ServiceIdentifier',
+    'IPTC:ProductID': 'XMP:ProductID',
+    'IPTC:DateSent': 'XMP:DateSent',
+    'IPTC:TimeSent': 'XMP:TimeSent',
+    'IPTC:ObjectName': 'XMP:Title',
+    'IPTC:EditStatus': 'XMP:EditStatus',
+    'IPTC:Urgency': 'XMP:Urgency',
+    'IPTC:SubjectReference': 'XMP:SubjectCode',
+    'IPTC:Category': 'XMP:Category',
+    'IPTC:SupplementalCategories': 'XMP:SupplementalCategories',
+    'IPTC:Keywords': 'XMP:Subject',
+    'IPTC:ContentLocationCode': 'XMP:LocationCode',
+    'IPTC:ContentLocationName': 'XMP:LocationName',
+    'IPTC:ReleaseDate': 'XMP:ReleaseDate',
+    'IPTC:ReleaseTime': 'XMP:ReleaseTime',
+    'IPTC:ExpirationDate': 'XMP:ExpirationDate',
+    'IPTC:ExpirationTime': 'XMP:ExpirationTime',
+    'IPTC:SpecialInstructions': 'XMP:Instructions',
+    'IPTC:DateCreated': 'XMP:DateCreated',
+    'IPTC:TimeCreated': 'XMP:DateCreated',
+    'IPTC:By-line': 'XMP:Creator',
+    'IPTC:By-lineTitle': 'XMP:AuthorsPosition',
+    'IPTC:City': 'XMP:City',
+    'IPTC:Sub-location': 'XMP:Location',
+    'IPTC:Province-State': 'XMP:State',
+    'IPTC:Country-PrimaryLocationCode': 'XMP:CountryCode',
+    'IPTC:Country-PrimaryLocationName': 'XMP:Country',
+    'IPTC:OriginalTransmissionReference': 'XMP:TransmissionReference',
+    'IPTC:Headline': 'XMP:Headline',
+    'IPTC:Credit': 'XMP:Credit',
+    'IPTC:Source': 'XMP:Source',
+    'IPTC:CopyrightNotice': 'XMP:Rights',
+    'IPTC:Contact': 'XMP:CreatorContactInfo',
+    'IPTC:Caption-Abstract': 'XMP:Description',
+    'IPTC:Writer-Editor': 'XMP:CaptionWriter',
+    'IPTC:LanguageIdentifier': 'XMP:Language',
+} as const satisfies Record<`IPTC:${keyof IPTCMetadata}`, `XMP:${XMPMetadata}`>;
+export const XMP_IPTC_TAGS = getInvertObject(IPTC_XMP_TAGS);
+
+export const EXIFTOOL_ARGS = {
+    COMPOSITE: ['-use MWG', '-mwg:all'],
+    IPTC: '-iptc:all',
+    JSON: '-j',
+    showDuplicates: '-a',
+    showGroupNames: '-G',
+    XMP: '-xmp:all',
+} as const;

fixtures/metadata.mov

@@ -2,39 +2,22 @@ import _ from 'lodash';
 import {getDataUrl} from 'core/upload/image-preview-directive';
 import {gettext} from 'core/utils';
 import {isEmpty, pickBy} from 'lodash';
-import {handleBinaryFile} from '@metadata/exif';
 import {extensions} from 'appConfig';
 import {IPTCMetadata, IUser, IArticle} from 'superdesk-api';
 import {appConfig} from 'appConfig';
 import {fileUploadErrorModal} from './file-upload-error-modal';
 import {showModal} from '@superdesk/common';
 import {sdApi} from 'api';
+import {getMetadata} from 'apps/archive/parse-metadata';
 
 const isNotEmptyString = (value: any) => value != null && value !== '';
 
 /* eslint-disable complexity */
 
-function getExifData(file: File): Promise<IPTCMetadata> {
-    return new Promise((resolve, reject) => {
-        const reader = new FileReader();
-
-        reader.onloadend = () => {
-            try {
-                const exif: { iptcdata: IPTCMetadata } = handleBinaryFile(reader.result);
-
-                resolve(exif.iptcdata);
-            } catch (error) {
-                console.error(error);
-                reject(error);
-            }
-        };
-
-        reader.onerror = reject;
-        reader.readAsArrayBuffer(file);
-    });
-}
-
-function mapIPTCExtensions(metadata: IPTCMetadata, user: IUser, parent?: IArticle): Promise<Partial<IArticle>> {
+function mapIPTCExtensions(
+    metadata: Partial<IPTCMetadata>,
+    user: IUser, parent?: IArticle,
+): Promise<Partial<IArticle>> {
     const meta: Partial<IPTCMetadata> = Object.assign({
         'By-line': user.byline,
     }, pickBy(metadata, isNotEmptyString));
@@ -361,10 +344,10 @@ export function UploadController(
                 ? Promise.resolve()
                 : Promise.all(acceptedFiles.map(
                     ({file, getThumbnail}) =>
-                        getExifData(file)
+                        getMetadata(file)
                             .then(
                                 (fileMeta) => mapIPTCExtensions(fileMeta, $scope.currentUser, $scope.parent),
-                                () => ({}), // proceed with upload on exif parsing error
+                                (): Partial<IArticle> => ({}), // proceed with upload on exif parsing error
                             )
                             .then((meta) => {
                                 const item = initFile(file, meta, getPseudoId());

karma.conf.js

@@ -0,0 +1,80 @@
+import {parseMetadata} from '@uswriting/exiftool/cjs';
+import {IContentProfileType} from 'apps/workspace/content/controllers/ContentProfilesController';
+import {IPTCMetadata} from 'superdesk-api';
+import {getObjectEntries} from 'utils/object';
+import {EXIFTOOL_ARGS, XMP_IPTC_TAGS} from './constants';
+
+type RawMetadata = {
+  data: Array<Record<string, unknown>>;
+  contentType: IContentProfileType.picture | IContentProfileType.video;
+};
+
+type ExiftoolOptions = Parameters<typeof parseMetadata>[1];
+
+const getMetadata = (f: File, options?: ExiftoolOptions) =>
+    (f.type.startsWith('video/')
+        ? getVideoMetadata(f, options)
+        : getPictureMetadata(f, options)
+    ).then(processMetadata, (): Partial<IPTCMetadata> => ({}));
+
+const getVideoMetadata = (
+    f: File,
+    options?: ExiftoolOptions,
+): Promise<RawMetadata> =>
+    parseMetadata<any>(f, {
+        args: [EXIFTOOL_ARGS.showGroupNames, EXIFTOOL_ARGS.JSON, EXIFTOOL_ARGS.XMP],
+        transform: (d) => JSON.parse(d),
+        ...options,
+    }).then((r) => ({...r, contentType: IContentProfileType.video}));
+
+const getPictureMetadata = (
+    f: File,
+    options?: ExiftoolOptions,
+): Promise<RawMetadata> =>
+    parseMetadata<any>(f, {
+        args: [
+            EXIFTOOL_ARGS.showGroupNames,
+            EXIFTOOL_ARGS.JSON,
+            EXIFTOOL_ARGS.XMP,
+            EXIFTOOL_ARGS.showDuplicates,
+            EXIFTOOL_ARGS.IPTC,
+            ...EXIFTOOL_ARGS.COMPOSITE,
+        ],
+        transform: (d) => JSON.parse(d),
+        ...options,
+    }).then((r) => ({...r, contentType: IContentProfileType.picture}));
+
+const processMetadata = (metadata: RawMetadata): Partial<IPTCMetadata> => {
+    const data = metadata.data?.[0] ?? {};
+
+    if (metadata.contentType === IContentProfileType.video)
+        return stripGroupNames(mapXMPtoIPTC(data));
+    return stripGroupNames(data);
+};
+
+const mapXMPtoIPTC = (metadata: RawMetadata['data'][number]) =>
+    getObjectEntries(metadata).reduce<RawMetadata['data'][number]>(
+        (acc, [k, v]) => {
+            acc[XMP_IPTC_TAGS[k] ?? k] = v;
+            return acc;
+        },
+        {},
+    );
+
+const stripGroupNames = (metadata: RawMetadata['data'][number]) =>
+    (({IPTC, XMP, Composite}) => ({...IPTC, ...XMP, ...Composite}))(
+        getObjectEntries(metadata).reduce<
+      Record<string, Partial<Record<string, unknown>>>
+    >(
+        (a, [k, v]) => {
+            const [group, tag] = k.split(':');
+
+            if (!group || !tag) return a;
+            a[group][tag] = v;
+            return a;
+        },
+        {IPTC: {}, XMP: {}, Composite: {}},
+    ),
+    );
+
+export {getMetadata, getPictureMetadata, getVideoMetadata};