5.0.0 • Published 25 days ago

@myrmidon/cadmus-img-annotator v5.0.0

Weekly downloads
-
License
-
Repository
github
Last release
25 days ago

CadmusImgAnnotator

This library was generated with Angular CLI version 17.3.0.

Since its version 3, this library has been refactored to use Annotorious headless mode. In this mode, the Annotorious library is used only for drawing, while disposing of its default popup UI. This allows for much more customization, even if at a price of higher code complexity.

Overview

architecture

The architecture of the image annotation subsystem is represented by these libraries:

  • cadmus-img-annotator: core components using Recogito Annotorious.
  • cadmus-img-gallery: gallery-specific components.
  • cadmus-img-gallery-iiif: IIIF-related services for galleries.

Note that since version 3, the GalleryImgAnnotator component found in cadmus-img-gallery is obsolete, and is replaced by custom implementations based on annotations list editors. In future versions, this component will be removed.

At the core there is the annotator component, an Angular directive (=template-less component) wrapping the Annotorious library used in headless mode.

Headless mode disallows the usage of Annotorious standard UI for editing W3C-based annotations, leaving to the consumer code the task of providing its own UI.

This component is used as an img element attribute, and wires Annotorious to the image, while providing a number of events reporting user interaction to the outer world. Among these events, there is also one which provides an Annotorious core instance (depicted as the gray circled A in the above schema), to be directly used for interacting with this subsystem.

The essential task of this component is providing an interface between the underlying Annotorious library, used to draw on top of images, and the rest of the system.

The Cadmus bricks annotation subsystem couples the W3C compliant annotations produced by Annotorious with custom-schema metadata for each of them.

As customary in Cadmus, each Cadmus-based project can provide its own schema for such metadata, and also its own editing UI. In the above scheme, yellow components are those which can be implemented for each specific project, and represent the customizable portion of the annotations subsystem.

At the core of this system there is the annotations list, which collects annotations merging data from two sources:

  • the Annotorious drawing system, which provides shapes in W3C compliant annotations;
  • the Cadmus annotation metadata editor, which provides project-specific metadata entered via a custom UI. This custom UI (annotation editor in the above scheme) is directly popped up on top of the image whenever the user draws a shape on it, via an annotation dialog wrapper.

The result of combining these data is a list annotation, including:

  • metadata about the target image;
  • the original W3C annotation produced by Annotorious, and optionally modified by the annotation editor;
  • the additional metadata (payload) produced by the annotation editor.

So, the annotator, annotations list, and annotation editor work all together to provide a unified UX where users draw on top of an image and enter metadata in a custom UI.

The annotations list editor base provides the foundation for a component orchestrating all these components. This is just a code base, whereas the UI is fully provided by a project-specific implementation (the yellow annotations list editor in the above scheme). Usually, the UI consists of a table listing the various annotations created by users.

Finally, at the top level of this components hierarchy we find the gallery image annotator, which just assembles all the pieces together. Its task is providing the UI which includes:

  • the image being annotated with Annotorious attached;
  • the annotations list;
  • the gallery images list. This is a generic component whose task is fetching either a virtual page of images or a single one, with their metadata, while optionally filtering them. The gallery does not impose a specific technology as the source of these images: it just relies on an abstraction, the gallery service, which has different implementations according to the technology used. In the default scenario this is IIIF.

Thus, the gallery image annotator provides a way of selecting an image from a set, annotating it using a custom UI, and outputting the results as a set of list annotations. This is usually consumed by a project-specific part editor, which adapts the backend annotations to the UI annotations and vice-versa, thus providing persistent storage in the context of Cadmus architecture.

Requirements

This library uses Annotorius. The consumer app must install it as follows:

(1) npm install @recogito/annotorious;

(2) if you want selectors other than the default rect/polygon, you must also install the selector pack plugin: npm i @recogito/annotorious-selector-pack.

(3) in angular.json, add CSS to styles:

"styles": [
  "node_modules/@recogito/annotorious/dist/annotorious.min.css",
  "src/styles.css"
],

(4) in your src root folder, add a types.d.ts file with this declaration:

declare module "@recogito/annotorious";

You can now import and use like this:

import { Component } from '@angular/core';
// @ts-ignore
import { Annotorious } from '@recogito/annotorious';
// @ts-ignore
import SelectorPack from '@recogito/annotorious-selector-pack';

@Component({
  selector: 'app-root',
  templateUrl: './app.component.html',
  styleUrls: ['./app.component.css'],
})
export class AppComponent {
  ngAfterViewInit(): void {
    // the DOM must be initialized before creating instance
    const cfg = this.config || {};
    cfg.image = this._elementRef.nativeElement;
    this._ann = new Annotorious(cfg);

    // plugin
    if (this.additionalTools?.length) {
      SelectorPack(this._ann, {
        tools: this.additionalTools,
      });
    }
  }
}

Components

This library contains the following components:

  • ImgAnnotatorDirective is the lowest level component, in charge of communicating with an Annotorious annotator. To use it, just add cadmusImgAnnotator as an attribute to your img element.

    • ➡️ input:
      • config: AnnotoriousConfig | undefined
      • disableEditor: boolean | undefined
      • tool: string
      • annotations: any[]
      • selectedAnnotation: any | undefined
      • additionalTools: string[] | undefined
    • ⬅️ output:
      • annotatorInit: any
      • cancelSelected: Annotation
      • changeSelectionTarget: any
      • clickAnnotation: AnnotationEvent
      • createAnnotation: AnnotationEvent
      • createSelection: Annotation
      • deleteAnnotation: AnnotationEvent
      • mouseEnterAnnotation: AnnotationEvent
      • mouseLeaveAnnotation: AnnotationEvent
      • selectAnnotation: Annotation
      • updateAnnotation: AnnotationEvent

  • ImgAnnotatorToolbar is a dumb component used to provide a toolbar for selecting the drawing tool when using ImgAnnotatorDirective.

  • ➡️ input:
    • tools: list of tools, each with its Annotorious ID, icon, and tip. Usually you won't set this but just use the default.
  • ⬅️ output:
    • toolChange: emitted when the control is initialized, and whenever the tool is changed.

Toolbar consumers just handle the toolChange event to set a tool property; the annotator directive's tool property is bound to it.

  • ImgAnnotationList<T> (where T is the annotation payload type) is a list of image annotations. This list empowers an image annotations list component by maintaining a list of annotation/payload pairs for each annotation, where the payload type is defined by T. This list requires an instance of Annotorious annotator, and the type of the editor component to use for editing each annotation. It will then use the annotator to keep in synch with Annotorious, and a dialog wrapper to edit each annotation via the provided editor. Consumers should thus provide an annotation editor and a corresponding dialog component wrapper, which wires the annotation to the editor.

  • ImgAnnotationListComponent<T> (where T is the annotation payload type) is a base class for image annotations list components. Derive your component from this, wiring its input annotator and editorComponent properties. Once both these are set, the list is initialized and ready to be used.

    • ➡️ input:
      • annotator: any
      • editorComponent: any
      • annotationToString: (object: ListAnnotation<any>) => string | null
    • ⬅️ output:
      • listInit: ImgAnnotationList<T>: emitted when the list is initialized, which happens as soon as both annotator and editorComponent are set.

This component takes care of creating the inner list core (of type ImgAnnotationList<T>), lazily instantiated by this base class as soon as all its dependencies are satisfied (via bound input properties).

Usage

Annotated Image

The typical usage of these components is represented by a customized component which has:

  • an annotator directive, properly configured via bound properties, and emitting a number of events for initialization (which provides the annotator instance) and user interaction.
  • a customized list component (derived from ImgAnnotationListComponent<T>) which provides the list of annotations with their payloads synched with Annotorious, and allows editing each in its own editor wrapped in a popup dialog. So, this component requires:
    • a component to edit the annotation and its payload;
    • a dialog component wrapping this editor component so that it can popup as a dialog.

As the annotator directive emits events and the list consumes most of them, this component orchestrates their interaction by handling the annotator events via calls to the inner list core of the list component.

Also, this component must provide the annotation editor type, which is bound to the list component.

You can find an example of this in this app's image gallery page (img-gallery-pg.component).

Annotated Image with Gallery

A more complex scenario involves the usage of an images gallery, which allows users to pick images from a list, whatever its source (typically, but not limited to, IIIF).

To build such a gallery, you can follow the steps outlined here.

1. Modeling Annotation

First, define the model of your annotation. This should be done both at the backend and at the frontend level. The essential model from Annotorious is here represented by Annotation:

  • id: a GUID prefixed by #
  • value:
    • @context (http://www.w3.org/ns/anno.jsonld)
    • type (Annotation)
    • body: an array including objects having:
      • type (TextualBody)
      • value (the text entered by user)
      • purpose (commenting or tagging)
    • target:
      • source: image URI
      • selector
        • type (FragmentSelector)
        • conformsTo (http://www.w3.org/TR/media-frags/)
        • value (e.g. xywh=pixel:42,32,127,200)

Here, annotations are included in lists, and wrap custom payload data; their type is ListAnnotation<T>. This type pairs an Annotorious Annotation with a custom payload:

  • id: the ID as derived from Annotorious.
  • image: the image (GalleryImage) being annotated.
  • value (Annotation): the Annotorious annotation.
  • payload: this is up to you.

2. Creating Gallery Components

A gallery implies an image, picked from a list; and a set of annotations on it. A gallery image is any object implementing the GalleryImage, which provides the barely minimum properties of each image: a string ID, a URI, a title, and a short description.

A gallery component is designed to get a gallery image and optionally a set of list annotations to edit; and interactively emit as output a gallery annotated image (GalleryAnnotatedImage<T>), which pairs a GalleryImage with its list annotations.

This component requires:

(2A) an editor component to edit list annotations. These components mostly edit the payload portion of the annotation, and of course vary according to the payload used. This is a dumb component which gets an annotation and emits its change event.

A template for its body can be:

private _annotation?: ListAnnotation<any>;

@Input()
public get annotation(): ListAnnotation<any> | undefined {
  return this._annotation;
}
public set annotation(value: ListAnnotation<any> | undefined) {
  if (this._annotation === value) {
    return;
  }
  this._annotation = value;
  this.updateForm(this._annotation);
}

@Output()
public cancel: EventEmitter<any>;

@Output()
public annotationChange: EventEmitter<ListAnnotation<any>>;

private getAnnotation(): ListAnnotation<any> {
  // TODO get annotation from form
}

public close(): void {
  this.cancel.emit();
}

public save(): void {
  if (this.form.invalid) {
    return;
  }
  this._annotation = this.getAnnotation();
  this.annotationChange.emit(this.annotation);
}

with its template:

<fieldset>
  <legend>annotation</legend>
  <form [formGroup]="form" (submit)="save()">
    TODO your form controls here...
    <div class="center-content">
      <button type="button" mat-icon-button (click)="close()" color="warn">
        <mat-icon>cancel</mat-icon>
      </button>
      <button
        type="submit"
        mat-icon-button
        color="primary"
        [disabled]="form.pristine || form.invalid"
      >
        <mat-icon>check_circle</mat-icon>
      </button>
    </div>
  </form>
</fieldset>

and styles:

fieldset {
  border: 1px solid silver;
  border-radius: 4px;
  padding: 6px;
}
legend {
  color: silver;
}
.center-content {
  display: flex;
  justify-content: center;
}

You can find an example of this component in this demo app at edit-annotation-component. This is a simple editor which has no payload at all and just deals with a single text annotation added to the annotation body. In real world, you would have your own payload or at least more annotations (comments and/or tags) like in the default Annotorious UI.

(2B) a dialog wrapper for this editor. This is used to wire the list annotation received by the editor to the bindings in your editor, providing a frame to be used as a popup dialog whose content is determined by the editor.

For instance, here is a wrapper using an any-type payload for the demo editor (which has no payload at all):

import { Component, Inject } from '@angular/core';
import { MAT_DIALOG_DATA, MatDialogRef } from '@angular/material/dialog';

import {
  Annotation,
  ListAnnotation,
} from 'projects/myrmidon/cadmus-img-annotator/src/public-api';

/**
 * A dialog wrapping an annotation editor. This just wires the received
 * data with the editor.
 */
@Component({
  selector: 'app-edit-annotation-dialog',
  templateUrl: './edit-annotation-dialog.component.html',
  styleUrls: ['./edit-annotation-dialog.component.css'],
})
export class EditAnnotationDialogComponent {
  constructor(
    public dialogRef: MatDialogRef<EditAnnotationDialogComponent>,
    @Inject(MAT_DIALOG_DATA) public data: ListAnnotation<any>
  ) {}

  onCloseClick(): void {
    this.dialogRef.close();
  }

  onSaveClick(annotation: ListAnnotation<any>): void {
    this.dialogRef.close(annotation);
  }
}

and its template:

<h1 mat-dialog-title>Edit Annotation</h1>
<div mat-dialog-content>
  <!-- this is your own list annotation editor -->
  <app-edit-annotation
    [annotation]="data"
    (cancel)="onCloseClick()"
    (annotationChange)="onSaveClick($event)"
  ></app-edit-annotation>
</div>

(2C) a customized version of an annotations list component, derived from ImgAnnotationListComponent<T>. Typically all what you have to do is implement a few methods for user actions, and use the type argument corresponding to your payload (in this sample we just have any because it is related to the app's sample with no payload).

import { Component } from '@angular/core';
import { ImgAnnotationListComponent } from 'projects/myrmidon/cadmus-img-annotator/src/public-api';

@Component({
  selector: 'app-my-img-annotation-list',
  templateUrl: './my-img-annotation-list.component.html',
  styleUrls: ['./my-img-annotation-list.component.css'],
})
export class MyImgAnnotationListComponent extends ImgAnnotationListComponent<any> {
  public selectAnnotation(annotation: any): void {
    this.list?.selectAnnotation(annotation);
  }

  public removeAnnotation(index: number): void {
    this.list?.removeAnnotation(index);
  }

  public editAnnotation(annotation: any): void {
    this.list?.editAnnotation(annotation);
  }
}

Its template should represent the list of annotations, so here you are free to implement the design you prefer. For instance, here is a table:

<div *ngIf="list">
  <table *ngIf="list!.annotations$ | async as annotations">
    <thead>
      <tr>
        <th></th>
        <th>annotation</th>
        <th class="noif-lt-md">ID</th>
      </tr>
    </thead>
    <tbody>
      <tr
        *ngFor="let a of annotations; let i = index"
        [class.selected]="a === (list!.selectedAnnotation$ | async)"
      >
        <td class="fit-width">
          <button
            type="button"
            mat-icon-button
            (click)="selectAnnotation(i)"
            matTooltip="Select annotation"
          >
            <mat-icon>check_circle</mat-icon>
          </button>
          <button
            type="button"
            mat-icon-button
            color="warn"
            (click)="removeAnnotation(i)"
            matTooltip="Remove annotation"
          >
            <mat-icon>delete</mat-icon>
          </button>
          <button
            type="button"
            mat-icon-button
            color="primary"
            (click)="editAnnotation(i)"
            matTooltip="Edit annotation"
          >
            <mat-icon>edit</mat-icon>
          </button>
        </td>
        <td>{{ a | objectToString : annotationToString }}</td>
        <td class="muted noif-lt-md">{{ a.id }}</td>
      </tr>
    </tbody>
  </table>
</div>

with its styles:

table {
  width: 100%;
  border-collapse: collapse;
}
.muted {
  color: silver;
}
tr:nth-child(odd) {
  background-color: #fafafa;
}
th {
  font-weight: normal;
  color: silver;
  text-align: left;
}
tr.selected {
  border: 1px solid orange;
}
td.fit-width {
  width: 1px;
  white-space: nowrap;
}
@media only screen and (max-width: 959px) {
  .noif-lt-md {
    display: none;
  }
}

3. Creating Gallery

Finally, you must create a gallery component to orchestrate all these components:

  • an image decorated with the annotator directive and linked to an annotator tools bar for selecting the drawing tool. The directive is bound to the selected image, and its events are handled by the parent component.

  • a list of annotations, represented by your annotations list component (see above, 2C). This is bound to the annotator instance, got from the annotator directive; and to the annotation editor type, provided by a property in the parent component; its event provides the list instance which manages data.

  • a gallery list, which provides the list of images to pick from; the parent component handles the image pick event.

Here is a sample code:

@Component({
  selector: 'app-my-gallery-image-annotator',
  templateUrl: './my-gallery-image-annotator.component.html',
  styleUrls: ['./my-gallery-image-annotator.component.css'],
})
export class MyGalleryImageAnnotatorComponent implements OnInit, OnDestroy {
  private _sub?: Subscription;
  private _image?: GalleryImage;
  private _list?: ImgAnnotationList<MyAnnotationPayload>;

  public entries: ThesaurusEntry[];
  public annotator?: any;
  public editorComponent = EditAnnotationDialogComponent;
  public tool: string = 'rect';
  public tabIndex: number = 0;
 
  /**
   * The gallery image to annotate.
   */
  @Input()
  public get image(): GalleryImage | undefined | null {
    return this._image;
  }
  public set image(value: GalleryImage | undefined | null) {
    if (this._image === value) {
      return;
    }
    this._image = value || undefined;
    // reset annotations if image URI changed
    if (this._image?.uri !== value?.uri) {
      this._list?.clearAnnotations();
    }
    // switch to image tab
    setTimeout(() => {
      this.tabIndex = value ? 0 : 1;
    });
  }

  /**
   * The annotations being edited.
   */
  @Input()
  public get annotations(): ListAnnotation<MyAnnotationPayload>[] {
    return this._list?.getAnnotations() || [];
  }
  public set annotations(value: ListAnnotation<MyAnnotationPayload>[]) {
    this._list?.setAnnotations(value);
  }

  /**
   * Emitted whenever annotations change.
   */
  @Output()
  public annotationsChange: EventEmitter<
    GalleryAnnotatedImage<MyAnnotationPayload>
  >;

  constructor(
    public dialog: MatDialog,
    @Inject(MAT_DIALOG_DEFAULT_OPTIONS) public dlgConfig: MatDialogConfig,
    @Inject(IMAGE_GALLERY_SERVICE_KEY)
    private _galleryService: GalleryService,
    private _options: GalleryOptionsService,
    formBuilder: FormBuilder
  ) {
    this.annotationsChange = new EventEmitter<
      GalleryAnnotatedImage<MyAnnotationPayload>
    >();

    // mock filter entries
    this.entries = [
      {
        id: 'title',
        value: 'title',
      },
      {
        id: 'dsc',
        value: 'description',
      },
    ];
  }

  public ngOnInit(): void {
    if (!this._image) {
      this.tabIndex = 1;
    }
  }

  public ngOnDestroy(): void {
    this._sub?.unsubscribe();
  }

  public onToolChange(tool: string): void {
    this.tool = tool;
  }

  public onAnnotatorInit(annotator: any) {
    setTimeout(() => {
      this.annotator = annotator;
    });
  }

  public onListInit(list: ImgAnnotationList<MyAnnotationPayload>) {
    this._list = list;

    // emit annotations whenever they change
    this._sub?.unsubscribe();
    this._sub = this._list.annotations$.subscribe((annotations) => {
      if (this._image) {
        this.annotationsChange.emit({
          image: this._image,
          annotations: annotations,
        });
      }
    });
  }

  public setAnnotations(): void {
    if (this.form.invalid) {
      return;
    }
    const annotations = JSON.parse(this.json.value || '[]');
    this._list?.setAnnotations(annotations);
  }

  public onCreateSelection(annotation: Annotation) {
    this._list?.onCreateSelection(annotation);
  }

  public onSelectAnnotation(annotation: Annotation) {
    this._list?.onSelectAnnotation(annotation);
  }

  public onCancelSelected(annotation: Annotation) {
    this._list?.onCancelSelected(annotation);
  }

  public editAnnotation(index: number): void {
    this._list?.editAnnotation(index);
  }

  public selectAnnotation(index: number): void {
    this._list?.selectAnnotation(index);
  }

  public removeAnnotation(index: number): void {
    this._list?.removeAnnotation(index);
  }

  public onCreateAnnotation(event: AnnotationEvent) {
    this._list?.onCreateAnnotation(event);
  }

  public onUpdateAnnotation(event: AnnotationEvent) {
    this._list?.onUpdateAnnotation(event);
  }

  public onDeleteAnnotation(event: AnnotationEvent) {
    this._list?.onDeleteAnnotation(event);
  }

  public onImagePick(image: GalleryImage): void {
    this._galleryService
      .getImage(image.id, this._options.get())
      .pipe(take(1))
      .subscribe((image) => {
        this.image = image!;
      });
    this.tabIndex = 1;
  }
}

and its template:

<mat-tab-group [(selectedIndex)]="tabIndex">
  <mat-tab label="Annotator">
    <div id="container">
      <div id="image" *ngIf="image">
        <div>
          <cadmus-img-annotator-toolbar
            (toolChange)="onToolChange($event)"
          ></cadmus-img-annotator-toolbar>
        </div>
        <div>
          <img
            alt="image"
            cadmusImgAnnotator
            (createAnnotation)="onCreateAnnotation($event)"
            (updateAnnotation)="onUpdateAnnotation($event)"
            (deleteAnnotation)="onDeleteAnnotation($event)"
            (createSelection)="onCreateSelection($event)"
            (selectAnnotation)="onSelectAnnotation($event)"
            (cancelSelected)="onCancelSelected($event)"
            (annotatorInit)="onAnnotatorInit($event)"
            [disableEditor]="true"
            [tool]="tool"
            [additionalTools]="['circle', 'ellipse', 'freehand']"
            [src]="image!.uri"
          />
        </div>
      </div>
      <div id="list">
        <app-my-img-annotation-list
          [image]="image!"
          [annotator]="annotator"
          [editorComponent]="editorComponent"
          (listInit)="onListInit($event)"
        ></app-my-img-annotation-list>
      </div>
    </div>
  </mat-tab>
  <mat-tab label="Gallery">
    <cadmus-gallery-list
      [entries]="entries"
      (imagePick)="onImagePick($event)"
    ></cadmus-gallery-list>
  </mat-tab>
</mat-tab-group>

with styles:

div#container {
  display: grid;
  grid-template-rows: 1fr;
  grid-template-columns: auto minmax(0, 1fr);
  grid-template-areas: "image list";
  gap: 8px;
}
div#image {
  grid-area: image;
}
div#list {
  grid-area: list;
}
@media only screen and (max-width: 959px) {
  div#container {
    grid-template-columns: 1fr;
    grid-template-areas:
      "image"
      "list";
  }
}
Cadmusbricks
tslib
@infinitebrahmanuniverse/nolb-_myr@everything-registry/sub-chunk-630
5.0.0

25 days ago

3.0.4

10 months ago

3.0.3

10 months ago

3.0.2

10 months ago

3.0.1

10 months ago

3.0.0

10 months ago

4.0.0

6 months ago

2.0.0

12 months ago

1.1.1

1 year ago

1.1.0

1 year ago

1.1.2

1 year ago

1.0.1

1 year ago

1.0.0

1 year ago

0.0.2

1 year ago