KevinMidboe/immich
1
0
Fork 0
mirror of https://github.com/KevinMidboe/immich.git synced 2025-10-29 17:40:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(server): use exiftool decoded values and unify metadata extraction (#2908)

Browse Source 
* refactor(server): metadata extraction

* chore: upgrade exiftool

* chore: remove log

* fix: other rotation cases

---------

Co-authored-by: Jason Rasmussen <jrasm91@gmail.com>
This commit is contained in:
Thomas
2023-09-18 14:06:03 +01:00
committed by GitHub
parent b6c6a7e403
commit 49ef86173f
8 changed files with 440 additions and 403 deletions
Download Patch File Download Diff File Expand all files Collapse all files

349
server/src/microservices/processors/metadata-extraction.processor.ts
View File

@@ -18,27 +18,12 @@ import {
import { AssetEntity, AssetType, ExifEntity } from '@app/infra/entities';
import { Inject, Logger } from '@nestjs/common';
import { ConfigService } from '@nestjs/config';
import tz_lookup from '@photostructure/tz-lookup';
import { exiftool, Tags } from 'exiftool-vendored';
import ffmpeg, { FfprobeData } from 'fluent-ffmpeg';
import { DefaultReadTaskOptions, ExifDateTime, exiftool, ReadTaskOptions, Tags } from 'exiftool-vendored';
import { firstDateTime } from 'exiftool-vendored/dist/FirstDateTime';
import * as geotz from 'geo-tz';
import { Duration } from 'luxon';
import fs from 'node:fs/promises';
import path from 'node:path';
import sharp from 'sharp';
import { promisify } from 'util';
import { parseLatitude, parseLongitude } from '../utils/exif/coordinates';
import { exifTimeZone, exifToDate } from '../utils/exif/date-time';
import { parseISO } from '../utils/exif/iso';
import { toNumberOrNull } from '../utils/numbers';
const ffprobe = promisify<string, FfprobeData>(ffmpeg.ffprobe);
interface MotionPhotosData {
isMotionPhoto: string | number | null;
isMicroVideo: string | number | null;
videoOffset: string | number | null;
directory: DirectoryEntry[] | null;
}
interface DirectoryItem {
Length?: number;
@@ -56,8 +41,12 @@ interface ImmichTags extends Tags {
MotionPhoto?: number;
MotionPhotoVersion?: number;
MotionPhotoPresentationTimestampUs?: number;
MediaGroupUUID?: string;
}
const exifDate = (dt: ExifDateTime | string | undefined) => (dt instanceof ExifDateTime ? dt?.toDate() : null);
const validate = <T>(value: T): T | null => (typeof value === 'string' ? null : value ?? null);
export class MetadataExtractionProcessor {
private logger = new Logger(MetadataExtractionProcessor.name);
private reverseGeocodingEnabled: boolean;
@@ -153,249 +142,48 @@ export class MetadataExtractionProcessor {
return false;
}
if (asset.type === AssetType.VIDEO) {
return this.handleVideoMetadataExtraction(asset);
} else {
return this.handlePhotoMetadataExtraction(asset);
}
}
const [exifData, tags] = await this.exifData(asset);
private async handlePhotoMetadataExtraction(asset: AssetEntity) {
const mediaExifData = await exiftool.read<ImmichTags>(asset.originalPath).catch((error: any) => {
this.logger.warn(
`The exifData parsing failed due to ${error} for asset ${asset.id} at ${asset.originalPath}`,
error?.stack,
);
return null;
});
await this.applyMotionPhotos(asset, tags);
await this.applyReverseGeocoding(asset, exifData);
const sidecarExifData = asset.sidecarPath
? await exiftool.read<ImmichTags>(asset.sidecarPath).catch((error: any) => {
this.logger.warn(
`The exifData parsing failed due to ${error} for asset ${asset.id} at ${asset.originalPath}`,
error?.stack,
);
return null;
})
: {};
const getExifProperty = <T extends keyof ImmichTags>(
...properties: T[]
): NonNullable<ImmichTags[T]> | string | null => {
for (const property of properties) {
const value = sidecarExifData?.[property] ?? mediaExifData?.[property];
if (value !== null && value !== undefined) {
// Can also be string when the value cannot be parsed
return value;
}
}
return null;
};
const timeZone = exifTimeZone(getExifProperty('DateTimeOriginal', 'CreateDate') ?? asset.fileCreatedAt);
const fileCreatedAt = exifToDate(getExifProperty('DateTimeOriginal', 'CreateDate') ?? asset.fileCreatedAt);
const fileModifiedAt = exifToDate(getExifProperty('ModifyDate') ?? asset.fileModifiedAt);
const fileStats = await fs.stat(asset.originalPath);
const fileSizeInBytes = fileStats.size;
const newExif = new ExifEntity();
newExif.assetId = asset.id;
newExif.fileSizeInByte = fileSizeInBytes;
newExif.make = getExifProperty('Make');
newExif.model = getExifProperty('Model');
newExif.exifImageHeight = toNumberOrNull(getExifProperty('ExifImageHeight', 'ImageHeight'));
newExif.exifImageWidth = toNumberOrNull(getExifProperty('ExifImageWidth', 'ImageWidth'));
newExif.exposureTime = getExifProperty('ExposureTime');
newExif.orientation = getExifProperty('Orientation')?.toString() ?? null;
newExif.dateTimeOriginal = fileCreatedAt;
newExif.modifyDate = fileModifiedAt;
newExif.timeZone = timeZone;
newExif.lensModel = getExifProperty('LensModel');
newExif.fNumber = toNumberOrNull(getExifProperty('FNumber'));
newExif.focalLength = toNumberOrNull(getExifProperty('FocalLength'));
// Handle array values by converting to string
const iso = getExifProperty('ISO')?.toString();
newExif.iso = iso ? parseISO(iso) : null;
const latitude = getExifProperty('GPSLatitude');
const longitude = getExifProperty('GPSLongitude');
const lat = parseLatitude(latitude);
const lon = parseLongitude(longitude);
if (lat === 0 && lon === 0) {
this.logger.warn(`Latitude & Longitude were on Null Island (${lat},${lon}), not assigning coordinates`);
} else {
newExif.latitude = lat;
newExif.longitude = lon;
}
const projectionType = getExifProperty('ProjectionType');
if (projectionType) {
newExif.projectionType = String(projectionType).toUpperCase();
}
newExif.livePhotoCID = getExifProperty('MediaGroupUUID');
const rawDirectory = getExifProperty('Directory');
await this.applyMotionPhotos(asset, {
isMotionPhoto: getExifProperty('MotionPhoto'),
isMicroVideo: getExifProperty('MicroVideo'),
videoOffset: getExifProperty('MicroVideoOffset'),
directory: Array.isArray(rawDirectory) ? (rawDirectory as DirectoryEntry[]) : null,
});
await this.applyReverseGeocoding(asset, newExif);
/**
* IF the EXIF doesn't contain the width and height of the image,
* We will use Sharpjs to get the information.
*/
if (!newExif.exifImageHeight || !newExif.exifImageWidth || !newExif.orientation) {
const metadata = await sharp(asset.originalPath).metadata();
if (newExif.exifImageHeight === null) {
newExif.exifImageHeight = metadata.height || null;
}
if (newExif.exifImageWidth === null) {
newExif.exifImageWidth = metadata.width || null;
}
if (newExif.orientation === null) {
newExif.orientation = metadata.orientation !== undefined ? `${metadata.orientation}` : null;
}
}
await this.assetRepository.upsertExif(newExif);
await this.assetRepository.upsertExif(exifData);
await this.assetRepository.save({
id: asset.id,
fileCreatedAt: fileCreatedAt || undefined,
updatedAt: new Date(),
duration: tags.Duration ? Duration.fromObject({ seconds: tags.Duration }).toFormat('hh:mm:ss.SSS') : null,
fileCreatedAt: exifData.dateTimeOriginal ?? undefined,
});
return true;
}
private async handleVideoMetadataExtraction(asset: AssetEntity) {
const data = await ffprobe(asset.originalPath);
const durationString = this.extractDuration(data.format.duration || asset.duration);
let fileCreatedAt = asset.fileCreatedAt;
const videoTags = data.format.tags;
if (videoTags) {
if (videoTags['com.apple.quicktime.creationdate']) {
fileCreatedAt = new Date(videoTags['com.apple.quicktime.creationdate']);
} else if (videoTags['creation_time']) {
fileCreatedAt = new Date(videoTags['creation_time']);
}
}
const exifData = await exiftool.read<ImmichTags>(asset.sidecarPath || asset.originalPath).catch((error: any) => {
this.logger.warn(
`The exifData parsing failed due to ${error} for asset ${asset.id} at ${asset.originalPath}`,
error?.stack,
);
return null;
});
const newExif = new ExifEntity();
newExif.assetId = asset.id;
newExif.fileSizeInByte = data.format.size || null;
newExif.dateTimeOriginal = fileCreatedAt ? new Date(fileCreatedAt) : null;
newExif.modifyDate = null;
newExif.timeZone = null;
newExif.latitude = null;
newExif.longitude = null;
newExif.city = null;
newExif.state = null;
newExif.country = null;
newExif.fps = null;
newExif.livePhotoCID = exifData?.ContentIdentifier || null;
if (videoTags && videoTags['location']) {
const location = videoTags['location'] as string;
const locationRegex = /([+-][0-9]+\.[0-9]+)([+-][0-9]+\.[0-9]+)\/$/;
const match = location.match(locationRegex);
if (match?.length === 3) {
newExif.latitude = parseLatitude(match[1]);
newExif.longitude = parseLongitude(match[2]);
}
} else if (videoTags && videoTags['com.apple.quicktime.location.ISO6709']) {
const location = videoTags['com.apple.quicktime.location.ISO6709'] as string;
const locationRegex = /([+-][0-9]+\.[0-9]+)([+-][0-9]+\.[0-9]+)([+-][0-9]+\.[0-9]+)\/$/;
const match = location.match(locationRegex);
if (match?.length === 4) {
newExif.latitude = parseLatitude(match[1]);
newExif.longitude = parseLongitude(match[2]);
}
}
if (newExif.longitude && newExif.latitude) {
try {
newExif.timeZone = tz_lookup(newExif.latitude, newExif.longitude);
} catch (error: any) {
this.logger.warn(`Error while calculating timezone from gps coordinates: ${error}`, error?.stack);
}
}
await this.applyReverseGeocoding(asset, newExif);
for (const stream of data.streams) {
if (stream.codec_type === 'video') {
newExif.exifImageWidth = stream.width || null;
newExif.exifImageHeight = stream.height || null;
if (typeof stream.rotation === 'string') {
newExif.orientation = stream.rotation;
} else if (typeof stream.rotation === 'number') {
newExif.orientation = `${stream.rotation}`;
} else {
newExif.orientation = null;
}
if (stream.r_frame_rate) {
const fpsParts = stream.r_frame_rate.split('/');
if (fpsParts.length === 2) {
newExif.fps = Math.round(parseInt(fpsParts[0]) / parseInt(fpsParts[1]));
}
}
}
}
await this.assetRepository.upsertExif(newExif);
await this.assetRepository.save({ id: asset.id, duration: durationString, fileCreatedAt });
return true;
}
private async applyReverseGeocoding(asset: AssetEntity, newExif: ExifEntity) {
const { latitude, longitude } = newExif;
if (this.reverseGeocodingEnabled && longitude && latitude) {
try {
const { country, state, city } = await this.geocodingRepository.reverseGeocode({ latitude, longitude });
newExif.country = country;
newExif.state = state;
newExif.city = city;
} catch (error: any) {
this.logger.warn(
`Unable to run reverse geocoding due to ${error} for asset ${asset.id} at ${asset.originalPath}`,
error?.stack,
);
}
}
}
private async applyMotionPhotos(asset: AssetEntity, data: MotionPhotosData) {
if (asset.livePhotoVideoId) {
private async applyReverseGeocoding(asset: AssetEntity, exifData: ExifEntity) {
const { latitude, longitude } = exifData;
if (!this.reverseGeocodingEnabled || !longitude || !latitude) {
return;
}
const { isMotionPhoto, isMicroVideo, directory, videoOffset } = data;
try {
const { city, state, country } = await this.geocodingRepository.reverseGeocode({ latitude, longitude });
Object.assign(exifData, { city, state, country });
} catch (error: Error | any) {
this.logger.warn(
`Unable to run reverse geocoding due to ${error} for asset ${asset.id} at ${asset.originalPath}`,
error?.stack,
);
}
}
private async applyMotionPhotos(asset: AssetEntity, tags: ImmichTags) {
if (asset.type !== AssetType.IMAGE || asset.livePhotoVideoId) {
return;
}
const rawDirectory = tags.Directory;
const isMotionPhoto = tags.MotionPhoto;
const isMicroVideo = tags.MicroVideo;
const videoOffset = tags.MicroVideoOffset;
const directory = Array.isArray(rawDirectory) ? (rawDirectory as DirectoryEntry[]) : null;
let length = 0;
let padding = 0;
@@ -464,12 +252,63 @@ export class MetadataExtractionProcessor {
}
}
private extractDuration(duration: number | string | null) {
const videoDurationInSecond = Number(duration);
if (!videoDurationInSecond) {
return null;
}
private async exifData(asset: AssetEntity): Promise<[ExifEntity, ImmichTags]> {
const readTaskOptions: ReadTaskOptions = {
...DefaultReadTaskOptions,
return Duration.fromObject({ seconds: videoDurationInSecond }).toFormat('hh:mm:ss.SSS');
defaultVideosToUTC: true,
backfillTimezones: true,
inferTimezoneFromDatestamps: true,
useMWG: true,
numericTags: DefaultReadTaskOptions.numericTags.concat(['FocalLength']),
geoTz: (lat: number, lon: number): string => geotz.find(lat, lon)[0],
};
const mediaTags = await exiftool
.read<ImmichTags>(asset.originalPath, undefined, readTaskOptions)
.catch((error: any) => {
this.logger.warn(`error reading exif data (${asset.id} at ${asset.originalPath}): ${error}`, error?.stack);
return null;
});
const sidecarTags = asset.sidecarPath
? await exiftool.read<ImmichTags>(asset.sidecarPath, undefined, readTaskOptions).catch((error: any) => {
this.logger.warn(`error reading exif data (${asset.id} at ${asset.sidecarPath}): ${error}`, error?.stack);
return null;
})
: null;
const stats = await fs.stat(asset.originalPath);
const tags = { ...mediaTags, ...sidecarTags };
this.logger.verbose('Exif Tags', tags);
return [
<ExifEntity>{
// altitude: tags.GPSAltitude ?? null,
assetId: asset.id,
dateTimeOriginal: exifDate(firstDateTime(tags)) ?? asset.fileCreatedAt,
exifImageHeight: validate(tags.ImageHeight),
exifImageWidth: validate(tags.ImageWidth),
exposureTime: tags.ExposureTime ?? null,
fileSizeInByte: stats.size,
fNumber: validate(tags.FNumber),
focalLength: validate(tags.FocalLength),
fps: validate(tags.VideoFrameRate),
iso: validate(tags.ISO),
latitude: validate(tags.GPSLatitude),
lensModel: tags.LensModel ?? null,
livePhotoCID: (asset.type === AssetType.VIDEO ? tags.ContentIdentifier : tags.MediaGroupUUID) ?? null,
longitude: validate(tags.GPSLongitude),
make: tags.Make ?? null,
model: tags.Model ?? null,
modifyDate: exifDate(tags.ModifyDate) ?? asset.fileModifiedAt,
orientation: validate(tags.Orientation)?.toString() ?? null,
projectionType: tags.ProjectionType ? String(tags.ProjectionType).toUpperCase() : null,
timeZone: tags.tz,
},
tags,
];
}
}

36
server/src/microservices/utils/exif/date-time.spec.ts
View File

@@ -1,36 +0,0 @@
import { describe, expect, it } from '@jest/globals';
import { ExifDateTime } from 'exiftool-vendored';
import { exifTimeZone, exifToDate } from './date-time';
describe('converts exif date to JS date', () => {
it('returns null for invalid inputs', () => {
expect(exifToDate(undefined)).toBeNull();
expect(exifToDate('invalid')).toBeNull();
expect(exifToDate(new Date('invalid'))).toBeNull();
expect(exifToDate(ExifDateTime.fromEXIF('invalid'))).toBeNull();
});
it('returns a valid date object for valid inputs', () => {
const date = new Date('2023');
expect(exifToDate(date)).toBeInstanceOf(Date);
expect(exifToDate(date)?.toISOString()).toBe('2023-01-01T00:00:00.000Z');
expect(exifToDate('2023')).toBeInstanceOf(Date);
const exifDateTime = ExifDateTime.fromISO('2023-01-01T00:00:00.000Z');
expect(exifToDate(exifDateTime)).toBeInstanceOf(Date);
expect(exifToDate(exifDateTime)?.toISOString()).toBe('2023-01-01T00:00:00.000Z');
});
});
describe('extracts the timezone from a date', () => {
it('returns null for invalid inputs', () => {
expect(exifTimeZone(undefined)).toBeNull();
expect(exifTimeZone('')).toBeNull();
expect(exifTimeZone(new Date('2023'))).toBeNull();
expect(exifTimeZone(ExifDateTime.fromEXIF('invalid'))).toBeNull();
});
it('returns the timezone for valid inputs', () => {
expect(exifTimeZone(ExifDateTime.fromEXIF('2020:12:29 14:24:45.700-05:00'))).toBe('UTC-5');
});
});

24
server/src/microservices/utils/exif/date-time.ts
View File

@@ -1,24 +0,0 @@
import { ExifDateTime } from 'exiftool-vendored';
import { isDecimalNumber } from '../numbers';
export function exifToDate(exifDate: string | Date | ExifDateTime | undefined): Date | null {
if (!exifDate) {
return null;
}
const date = exifDate instanceof ExifDateTime ? exifDate.toDate() : new Date(exifDate);
if (!isDecimalNumber(date.valueOf())) {
return null;
}
return date;
}
export function exifTimeZone(exifDate: string | Date | ExifDateTime | undefined): string | null {
const isExifDate = exifDate instanceof ExifDateTime;
if (!isExifDate) {
return null;
}
return exifDate.zone ?? null;
}

24
server/src/microservices/utils/exif/iso.spec.ts
View File

@@ -1,24 +0,0 @@
import { describe, expect, it } from '@jest/globals';
import { parseISO } from './iso';
describe('parsing ISO values', () => {
it('returns null for invalid values', () => {
expect(parseISO('')).toBeNull();
expect(parseISO(',,,')).toBeNull();
expect(parseISO('invalid')).toBeNull();
expect(parseISO('-5')).toBeNull();
expect(parseISO('99999999999999')).toBeNull();
});
it('returns the ISO number for valid inputs', () => {
expect(parseISO('0.0')).toBe(0);
expect(parseISO('32000.9')).toBe(32000);
});
it('returns the first valid ISO number in a comma separated list', () => {
expect(parseISO('400, 200, 100')).toBe(400);
expect(parseISO('-1600,800')).toBe(800);
expect(parseISO('-1, a., 1200')).toBe(1200);
expect(parseISO('NaN,50,100')).toBe(50);
});
});

14
server/src/microservices/utils/exif/iso.ts
View File

@@ -1,14 +0,0 @@
import { isNumberInRange } from '../numbers';
export function parseISO(input: string): number | null {
const values = input.split(',');
for (const value of values) {
const iso = Number.parseInt(value, 10);
if (isNumberInRange(iso, 0, 2 ** 32)) {
return iso;
}
}
return null;
}