workaround for issue with pikepdf

Feature: merge tool UI

src-ui/package-lock.json

@@ -8361,6 +8361,21 @@
         "tslib": "^2.0.0"
       }
     },
+    "ngx-drag-drop": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/ngx-drag-drop/-/ngx-drag-drop-2.0.0.tgz",
+      "integrity": "sha512-t+4/eiC8zaXKqU1ruNfFEfGs1GpMNwpffD0baopvZFKjQHCb5rhNqFilJ54wO4T0OwGp4/RnsVhlcxe1mX6UJg==",
+      "requires": {
+        "tslib": "^1.9.0"
+      },
+      "dependencies": {
+        "tslib": {
+          "version": "1.14.1",
+          "resolved": "https://registry.npmjs.org/tslib/-/tslib-1.14.1.tgz",
+          "integrity": "sha512-Xni35NKzjgMrwevysHTCArtLDpPvye8zV/0E4EyYn43P7/7qvQwPh9BGkHewbMulVntbigmcT7rdX3BNo9wRJg=="
+        }
+      }
+    },
     "ngx-file-drop": {
       "version": "10.0.0",
       "resolved": "https://registry.npmjs.org/ngx-file-drop/-/ngx-file-drop-10.0.0.tgz",

src-ui/package.json

@@ -28,6 +28,7 @@
     "ng2-pdf-viewer": "^6.3.2",
     "ngx-color": "^6.2.0",
     "ngx-cookie-service": "^10.1.1",
+    "ngx-drag-drop": "^2.0.0",
     "ngx-file-drop": "^10.0.0",
     "ngx-infinite-scroll": "^9.1.0",
     "rxjs": "~6.6.0",

src-ui/src/app/app-routing.module.ts

@@ -11,12 +11,14 @@ import { SettingsComponent } from './components/manage/settings/settings.compone
 import { TagListComponent } from './components/manage/tag-list/tag-list.component';
 import { NotFoundComponent } from './components/not-found/not-found.component';
 import {DocumentAsnComponent} from "./components/document-asn/document-asn.component";
+import { SplitMergeComponent } from './components/split-merge/split-merge.component';
 
 const routes: Routes = [
   {path: '', redirectTo: 'dashboard', pathMatch: 'full'},
   {path: '', component: AppFrameComponent, children: [
     {path: 'dashboard', component: DashboardComponent },
     {path: 'documents', component: DocumentListComponent },
+    {path: 'split_merge', component: SplitMergeComponent },
     {path: 'view/:id', component: DocumentListComponent },
     {path: 'documents/:id', component: DocumentDetailComponent },
       {path: 'asn/:id', component: DocumentAsnComponent },

src-ui/src/app/app.module.ts

@@ -33,6 +33,7 @@ import { DocumentCardSmallComponent } from './components/document-list/document-
 import { BulkEditorComponent } from './components/document-list/bulk-editor/bulk-editor.component';
 import { NgxFileDropModule } from 'ngx-file-drop';
 import { TextComponent } from './components/common/input/text/text.component';
+import { InputDebounceComponent } from './components/common/input/debounce/input-debounce.component';
 import { SelectComponent } from './components/common/input/select/select.component';
 import { CheckComponent } from './components/common/input/check/check.component';
 import { SaveViewConfigDialogComponent } from './components/document-list/save-view-config-dialog/save-view-config-dialog.component';
@@ -64,6 +65,10 @@ import { ApiVersionInterceptor } from './interceptors/api-version.interceptor';
 import { ColorSliderModule } from 'ngx-color/slider';
 import { ColorComponent } from './components/common/input/color/color.component';
 import { DocumentAsnComponent } from './components/document-asn/document-asn.component';
+import { SplitMergeComponent } from './components/split-merge/split-merge.component';
+import { DndModule } from 'ngx-drag-drop';
+import { DocumentChooserComponent } from './components/common/document-chooser/document-chooser.component';
+import { PageChooserComponent } from './components/common/page-chooser/page-chooser.component';
 
 import localeFr from '@angular/common/locales/fr';
 import localeNl from '@angular/common/locales/nl';
@@ -119,6 +124,7 @@ registerLocaleData(localeSv)
     DocumentCardSmallComponent,
     BulkEditorComponent,
     TextComponent,
+    InputDebounceComponent,
     SelectComponent,
     CheckComponent,
     SaveViewConfigDialogComponent,
@@ -140,7 +146,10 @@ registerLocaleData(localeSv)
     CustomDatePipe,
     DateComponent,
     ColorComponent,
-    DocumentAsnComponent
+    DocumentAsnComponent,
+    SplitMergeComponent,
+    DocumentChooserComponent,
+    PageChooserComponent
   ],
   imports: [
     BrowserModule,
@@ -153,7 +162,8 @@ registerLocaleData(localeSv)
     InfiniteScrollModule,
     PdfViewerModule,
     NgSelectModule,
-    ColorSliderModule
+    ColorSliderModule,
+    DndModule
   ],
   providers: [
     DatePipe,

src-ui/src/app/components/app-frame/app-frame.component.html

@@ -68,6 +68,13 @@
               </svg>&nbsp;<ng-container i18n>Documents</ng-container>
             </a>
           </li>
+          <li class="nav-item" *ngIf="splitMergeToolVisible">
+            <a class="nav-link" routerLink="split_merge" routerLinkActive="active" (click)="closeMenu()">
+              <svg class="sidebaricon" fill="currentColor">
+                <use xlink:href="assets/bootstrap-icons.svg#scissors"/>
+              </svg>&nbsp;<ng-container i18n>Split & merge</ng-container>
+            </a>
+          </li>
         </ul>
 
         <h6 class="sidebar-heading px-3 mt-4 mb-1 text-muted" *ngIf='savedViewService.sidebarViews.length > 0'>

src-ui/src/app/components/app-frame/app-frame.component.ts

@@ -12,6 +12,7 @@ import { DocumentDetailComponent } from '../document-detail/document-detail.comp
 import { Meta } from '@angular/platform-browser';
 import { DocumentListViewService } from 'src/app/services/document-list-view.service';
 import { FILTER_FULLTEXT_QUERY } from 'src/app/data/filter-rule-type';
+import { SplitMergeService } from 'src/app/services/split-merge.service';
 
 @Component({
   selector: 'app-app-frame',
@@ -27,7 +28,8 @@ export class AppFrameComponent implements OnInit {
     private searchService: SearchService,
     public savedViewService: SavedViewService,
     private list: DocumentListViewService,
-    private meta: Meta
+    private meta: Meta,
+    private splitMergeService: SplitMergeService
     ) {
 
   }
@@ -106,6 +108,10 @@ export class AppFrameComponent implements OnInit {
     }
   }
 
+  get splitMergeToolVisible() {
+    return this.splitMergeService.hasDocuments()
+  }
+
   ngOnInit() {
   }
 

src-ui/src/app/components/common/document-chooser/document-chooser.component.html

@@ -0,0 +1,15 @@
+<div class="modal-header">
+  <h4 class="modal-title" id="modal-basic-title" i18n>Choose Documents</h4>
+  <button type="button" class="close" aria-label="Close" (click)="cancelClicked()">
+    <span aria-hidden="true">&times;</span>
+  </button>
+</div>
+<div class="modal-body">
+  <div class="m-n2 row row-cols-paperless-cards">
+    <app-document-card-small [simpleCard]="true" [selected]="list.isSelected(d)" (toggleSelected)="toggleSelected(d, $event)" [document]="d" *ngFor="let d of list.documents"></app-document-card-small>
+  </div>
+</div>
+<div class="modal-footer">
+  <button type="button" class="btn btn-outline-danger" (click)="cancelClicked()" i18n>Cancel</button>
+  <button type="button" class="btn btn-outline-primary" [disabled]="list.selected.size == 0" (click)="confirmClicked.emit()" i18n>Add Selected</button>
+</div>

src-ui/src/app/components/common/document-chooser/document-chooser.component.scss

@@ -0,0 +1,28 @@
+@import "/src/theme";
+
+.modal-body {
+  max-height: 70vh;
+  overflow-y: scroll;
+}
+
+$paperless-card-breakpoints: (
+  0: 2, // xs
+  768px: 3, //md
+  992px: 4, //lg
+  1200px: 5, //xl
+  1400px: 6, // xxl
+  1600px: 7,
+  1800px: 8,
+  2000px: 9
+);
+
+.row-cols-paperless-cards {
+  @each $width, $n_cols in $paperless-card-breakpoints {
+    @media(min-width: $width) {
+      > * {
+        flex: 0 0 auto;
+        width: 100% / $n_cols;
+      }
+    }
+  }
+}

src-ui/src/app/components/common/document-chooser/document-chooser.component.spec.ts

@@ -0,0 +1,25 @@
+import { ComponentFixture, TestBed } from '@angular/core/testing';
+
+import { DocumentChooserComponent } from './document-chooser.component';
+
+describe('DocumentChooserComponent', () => {
+  let component: DocumentChooserComponent;
+  let fixture: ComponentFixture<DocumentChooserComponent>;
+
+  beforeEach(async () => {
+    await TestBed.configureTestingModule({
+      declarations: [ DocumentChooserComponent ]
+    })
+    .compileComponents();
+  });
+
+  beforeEach(() => {
+    fixture = TestBed.createComponent(DocumentChooserComponent);
+    component = fixture.componentInstance;
+    fixture.detectChanges();
+  });
+
+  it('should create', () => {
+    expect(component).toBeTruthy();
+  });
+});

src-ui/src/app/components/common/document-chooser/document-chooser.component.ts

@@ -0,0 +1,36 @@
+import { Component, Input, Output, EventEmitter, OnInit } from '@angular/core';
+import { NgbActiveModal } from '@ng-bootstrap/ng-bootstrap';
+import { PaperlessDocument } from 'src/app/data/paperless-document';
+import { DocumentListViewService } from 'src/app/services/document-list-view.service';
+
+@Component({
+  selector: 'app-document-chooser',
+  templateUrl: './document-chooser.component.html',
+  styleUrls: ['./document-chooser.component.scss']
+})
+export class DocumentChooserComponent implements OnInit {
+
+  constructor(
+    public activeModal: NgbActiveModal,
+    public list: DocumentListViewService
+  ) { }
+
+  @Output()
+  public confirmClicked = new EventEmitter()
+
+  ngOnInit(): void {
+    this.list.selectNone()
+    this.list.activateSavedView(null)
+    this.list.reload()
+  }
+
+  cancelClicked() {
+    this.list.selectNone()
+    this.activeModal.close()
+  }
+
+  toggleSelected(d: PaperlessDocument, event: MouseEvent): void {
+    if (!event.shiftKey) this.list.toggleSelected(d)
+    else this.list.selectRangeTo(d)
+  }
+}

src-ui/src/app/components/common/input/debounce/input-debounce.component.html

@@ -0,0 +1,2 @@
+<input type="text" [class]="cssClasses" [placeholder]="placeholder" [(ngModel)]="inputValue" (keyup)="keyUp($event)" />
+<span *ngIf="allowClear && inputValue?.length > 0" class="clear-wrapper" title="Clear all" (click)="clear()"><span aria-hidden="true" class="clear">×</span></span>

src-ui/src/app/components/common/input/debounce/input-debounce.component.scss

@@ -0,0 +1,23 @@
+.clear-wrapper {
+  -moz-user-select: none;
+  -ms-user-select: none;
+  -webkit-user-select: none;
+  cursor: pointer;
+  position: absolute;
+  top: 5px;
+  right: 5px;
+  user-select: none;
+  width: 17px;
+  color: #999;
+
+  .clear {
+    display: inline-block;
+    font-size: 18px;
+    line-height: 1;
+    pointer-events: none;
+  }
+
+  &:hover .clear {
+    color: #D0021B;
+  }
+}

src-ui/src/app/components/common/input/debounce/input-debounce.component.spec.ts

@@ -0,0 +1,25 @@
+import { ComponentFixture, TestBed } from '@angular/core/testing';
+
+import { InputDebounceComponent } from './text.component';
+
+describe('InputDebounceComponent', () => {
+  let component: InputDebounceComponent;
+  let fixture: ComponentFixture<InputDebounceComponent>;
+
+  beforeEach(async () => {
+    await TestBed.configureTestingModule({
+      declarations: [ InputDebounceComponent ]
+    })
+    .compileComponents();
+  });
+
+  beforeEach(() => {
+    fixture = TestBed.createComponent(InputDebounceComponent);
+    component = fixture.componentInstance;
+    fixture.detectChanges();
+  });
+
+  it('should create', () => {
+    expect(component).toBeTruthy();
+  });
+});

src-ui/src/app/components/common/input/debounce/input-debounce.component.ts

@@ -0,0 +1,66 @@
+import { Component, Input, Output, ElementRef, EventEmitter } from '@angular/core';
+import { fromEvent } from 'rxjs';
+import { map, debounceTime, distinctUntilChanged } from 'rxjs/operators';
+
+@Component({
+  selector: 'app-input-debounce',
+  templateUrl: './input-debounce.component.html',
+  styleUrls: ['./input-debounce.component.scss']
+})
+export class InputDebounceComponent {
+
+  @Input()
+  placeholder: string
+
+  @Input()
+  delay: number = 500
+
+  @Input()
+  classes: string
+
+  @Input()
+  allowClear: boolean = true
+
+  @Input()
+  set pattern(regexStr: string) {
+    this._pattern = new RegExp(regexStr)
+  }
+
+  _pattern: RegExp
+
+  @Output()
+  value: EventEmitter<string> = new EventEmitter<string>()
+
+  @Input()
+  inputValue: string;
+
+  constructor(private elementRef: ElementRef) {
+    fromEvent(elementRef.nativeElement, 'keyup').pipe(
+      map(() => this.inputValue),
+      debounceTime(this.delay),
+      distinctUntilChanged()
+    ).subscribe(input =>
+      this.value.emit(input)
+    );
+  }
+
+  keyUp(event: any) {
+    if (!this._pattern) return
+    if (event.key.length == 1 && !this._pattern.test(event.key)) { // e.g. dont do for backspace
+      // invalid character, prevent input
+      event.preventDefault()
+      event.stopImmediatePropagation()
+      this.inputValue = this.inputValue.slice(this.inputValue.length, 1)
+    }
+  }
+
+  get cssClasses(): string {
+    return 'form-control ' + this.classes
+  }
+
+  clear() {
+    this.inputValue = ''
+    this.value.emit(this.inputValue)
+  }
+
+}

src-ui/src/app/components/common/page-chooser/page-chooser.component.html

@@ -0,0 +1,13 @@
+<div class="modal-header">
+  <h4 class="modal-title" id="modal-basic-title">{{title}}: {{document.title}}</h4>
+  <button type="button" class="close" aria-label="Close" (click)="cancelClicked()">
+    <span aria-hidden="true">&times;</span>
+  </button>
+</div>
+<div class="modal-body">
+  <pdf-viewer class="page-chooser-viewer" [class.splitting]="splitting" [src]="previewUrl" [original-size]="false" [zoom]=".25" [autoresize]="true" [show-borders]="true" [show-all]="true" [render-text-mode]="2" (page-rendered)="afterPageRendered($event)" (after-load-complete)="pdfLoaded($event)"></pdf-viewer>
+</div>
+<div class="modal-footer">
+  <button type="button" class="btn btn-outline-danger" (click)="cancelClicked()" i18n>Cancel</button>
+  <button type="button" class="btn btn-outline-primary" (click)="confirmClicked()" [disabled]="pages.length == 0">{{title}}</button>
+</div>

src-ui/src/app/components/common/page-chooser/page-chooser.component.scss

@@ -0,0 +1,89 @@
+@import "src/theme";
+
+.modal-body {
+  max-height: 70vh;
+  overflow-y: scroll;
+}
+
+::ng-deep .page-chooser-viewer {
+  min-height: 400px;
+  width: 100%;
+
+  .pdfViewer {
+    display: flex !important;
+    flex-wrap: wrap;
+    justify-content: space-between;
+    align-items: flex-start;
+
+    .page:not(.disabled) {
+      margin: 2px 3% 2rem !important;
+      cursor: pointer;
+
+      .textLayer {
+        pointer-events: none;
+      }
+
+      &:hover {
+        .textLayer {
+          background-color: $primary !important;
+          opacity: 0.5;
+        }
+      }
+    }
+  }
+
+  &:not(.splitting) {
+    .pdfViewer {
+      .page.selected {
+        &::after {
+          position: absolute;
+          top: 0.5rem;
+          left: 0.5rem;
+          display: block;
+          width: 1rem;
+          height: 1rem;
+          border-radius: 0.25rem;
+          content: "";
+          background: $primary no-repeat 50%/50% 50%;
+          background-image: url("data:image/svg+xml,%3csvg xmlns='http://www.w3.org/2000/svg' width='8' height='8' viewBox='0 0 8 8'%3e%3cpath fill='%23fff' d='M6.564.75l-3.59 3.612-1.538-1.55L0 4.26l2.974 2.99L8 2.193z'/%3e%3c/svg%3e");
+        }
+
+        .textLayer {
+          background-color: $primary !important;
+          opacity: 0.5;
+        }
+      }
+    }
+  }
+
+  &.splitting {
+    .pdfViewer {
+      .page:not(:last-child).selected::after,
+      .page:not(:last-child):not(.disabled):hover::after {
+        background-color: $primary;
+        width: 2rem;
+        height: 100%;
+        content: "Split Here";
+        color: #abb2bf; // dark mode text color
+        position: absolute;
+        top: 0;
+        right: -2rem;
+        writing-mode: vertical-lr;
+        padding: 0.5rem;
+        text-align: center;
+        text-transform: uppercase;
+        letter-spacing: 0.1rem;
+        font-size: 80%;
+        font-weight: bold;
+      }
+
+      .page.disabled {
+        .canvasWrapper,
+        .textLayer {
+          opacity: 0.25;
+          pointer-events: none;
+        }
+      }
+    }
+  }
+}

src-ui/src/app/components/common/page-chooser/page-chooser.component.spec.ts

@@ -0,0 +1,25 @@
+import { ComponentFixture, TestBed } from '@angular/core/testing';
+
+import { PageChooserComponent } from './page-chooser.component';
+
+describe('PageChooserComponent', () => {
+  let component: PageChooserComponent;
+  let fixture: ComponentFixture<PageChooserComponent>;
+
+  beforeEach(async () => {
+    await TestBed.configureTestingModule({
+      declarations: [ PageChooserComponent ]
+    })
+    .compileComponents();
+  });
+
+  beforeEach(() => {
+    fixture = TestBed.createComponent(PageChooserComponent);
+    component = fixture.componentInstance;
+    fixture.detectChanges();
+  });
+
+  it('should create', () => {
+    expect(component).toBeTruthy();
+  });
+});

src-ui/src/app/components/common/page-chooser/page-chooser.component.ts

@@ -0,0 +1,90 @@
+import { Component, Input, Output, EventEmitter, OnInit } from '@angular/core';
+import { NgbActiveModal } from '@ng-bootstrap/ng-bootstrap';
+import { PaperlessDocument, PaperlessDocumentPart } from 'src/app/data/paperless-document';
+import { DocumentService } from 'src/app/services/rest/document.service';
+import { PDFDocumentProxy } from 'ng2-pdf-viewer';
+
+@Component({
+  selector: 'app-page-chooser',
+  templateUrl: './page-chooser.component.html',
+  styleUrls: ['./page-chooser.component.scss']
+})
+export class PageChooserComponent implements OnInit {
+
+  constructor(
+    public activeModal: NgbActiveModal,
+    private documentService: DocumentService
+  ) { }
+
+  get previewUrl(): string {
+    return this.documentService.getPreviewUrl(this.document.id)
+  }
+
+  @Input()
+  document: PaperlessDocument
+
+  @Input()
+  splitting: boolean = false
+
+  public pages: number[]
+
+  public enabledPages: number[]
+
+  private nPages: number
+
+  @Output()
+  public confirmPages = new EventEmitter<number[]>()
+
+  get title(): string {
+    return this.splitting ? 'Select Split' : 'Select Pages'
+  }
+
+  ngOnInit(): void {
+    this.pages = (this.document as PaperlessDocumentPart).pages
+    if (this.splitting && this.pages?.length) this.enabledPages = this.pages.concat([])
+    if (!this.pages) this.pages = []
+  }
+
+  cancelClicked() {
+    this.activeModal.close()
+  }
+
+  confirmClicked() {
+    if (this.splitting) {
+      // we need pages as the full range since we wont know length of document later
+      const startPage = this.pages.pop()
+      for (let page = startPage + 1; page <= this.nPages; page++) {
+        if (!this.enabledPages || (this.enabledPages && this.enabledPages.indexOf(page) !== -1)) this.pages.push(page)
+      }
+    }
+    this.confirmPages.emit(this.pages.sort((a, b) => { return a - b }))
+  }
+
+  pdfLoaded(pdf: PDFDocumentProxy) {
+    this.nPages = pdf.numPages
+  }
+
+  public afterPageRendered(pageRenderedEvent: any) {
+    const pageView = pageRenderedEvent.source /* as PDFPageView */
+    const div = pageView.div as HTMLDivElement
+    if (this.splitting && this.enabledPages?.length) {
+      div.classList.toggle('disabled', this.enabledPages.indexOf(parseInt(div.dataset.pageNumber)) == -1)
+    } else {
+      div.classList.toggle('selected', this.pages.indexOf(parseInt(div.dataset.pageNumber)) !== -1)
+    }
+    div.onclick = () => {
+      let pageNum: number = parseInt(div.dataset.pageNumber)
+      if (this.splitting) { // only select 1 page
+        if (pageNum == this.nPages || (this.enabledPages?.length && pageNum == Math.max(...this.enabledPages))) return
+        this.pages = [pageNum]
+        div.parentNode.childNodes.forEach(pageEl => {
+          (pageEl as Element).classList.toggle('selected', false)
+        })
+        div.classList.toggle('selected', true)
+      } else {
+        this.pages.indexOf(pageNum) !== -1 ? this.pages.splice(this.pages.indexOf(pageNum), 1) : this.pages.push(pageNum)
+        div.classList.toggle('selected', this.pages.indexOf(pageNum) !== -1)
+      }
+    }
+  }
+}

src-ui/src/app/components/dashboard/widgets/upload-file-widget/upload-file-widget.component.scss

@@ -4,11 +4,6 @@ form {
   position: relative;
 }
 
-.alert-heading {
-  font-size: 80%;
-  font-weight: bold;
-}
-
 .alerts-hidden {
   .btn {
     line-height: 1;

src-ui/src/app/components/document-detail/document-detail.component.html

@@ -38,6 +38,12 @@
         </svg>&nbsp;<span class="d-none d-lg-inline" i18n>More like this</span>
     </button>
 
+    <button type="button" class="btn btn-sm btn-outline-primary mr-2" (click)="splitMerge()">
+        <svg class="buttonicon" fill="currentColor">
+            <use xlink:href="assets/bootstrap-icons.svg#scissors" />
+        </svg>&nbsp;<span class="d-none d-lg-inline" i18n>Add to split/merge</span>
+    </button>
+
     <button type="button" class="btn btn-sm btn-outline-primary" (click)="close()">
         <svg class="buttonicon" fill="currentColor">
             <use xlink:href="assets/bootstrap-icons.svg#x" />

src-ui/src/app/components/document-detail/document-detail.component.ts

@@ -21,6 +21,7 @@ import { TextComponent } from '../common/input/text/text.component';
 import { SettingsService, SETTINGS_KEYS } from 'src/app/services/settings.service';
 import { PaperlessDocumentSuggestions } from 'src/app/data/paperless-document-suggestions';
 import { FILTER_FULLTEXT_MORELIKE } from 'src/app/data/filter-rule-type';
+import { SplitMergeService } from 'src/app/services/split-merge.service';
 
 @Component({
   selector: 'app-document-detail',
@@ -76,7 +77,8 @@ export class DocumentDetailComponent implements OnInit {
     private documentListViewService: DocumentListViewService,
     private documentTitlePipe: DocumentTitlePipe,
     private toastService: ToastService,
-    private settings: SettingsService) { }
+    private settings: SettingsService,
+    private splitMergeService: SplitMergeService) { }
 
   get useNativePdfViewer(): boolean {
     return this.settings.get(SETTINGS_KEYS.USE_NATIVE_PDF_VIEWER)
@@ -225,6 +227,11 @@ export class DocumentDetailComponent implements OnInit {
     this.documentListViewService.quickFilter([{rule_type: FILTER_FULLTEXT_MORELIKE, value: this.documentId.toString()}])
   }
 
+  splitMerge() {
+    this.splitMergeService.addDocument(this.document)
+    this.router.navigate(["split_merge"])
+  }
+
   hasNext() {
     return this.documentListViewService.hasNext(this.documentId)
   }

src-ui/src/app/components/document-list/bulk-editor/bulk-editor.component.html

@@ -56,6 +56,12 @@
     </div>
   </div>
   <div class="col-auto ml-auto mb-2 mb-xl-0 d-flex">
+    <button type="button" class="btn btn-sm btn-outline-primary mr-2" (click)="addToSplitMerge()">
+      <svg class="buttonicon" fill="currentColor">
+        <use xlink:href="assets/bootstrap-icons.svg#scissors" />
+      </svg>&nbsp;<span class="d-none d-lg-inline" i18n>Add to split/merge</span>
+    </button>
+
     <div class="btn-group btn-group-sm mr-2">
       <button type="button" class="btn btn-outline-primary btn-sm" (click)="downloadSelected()">
         <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">

src-ui/src/app/components/document-list/bulk-editor/bulk-editor.component.ts

@@ -1,4 +1,4 @@
-import { Component } from '@angular/core';
+import { Component, EventEmitter, Output } from '@angular/core';
 import { PaperlessTag } from 'src/app/data/paperless-tag';
 import { PaperlessCorrespondent } from 'src/app/data/paperless-correspondent';
 import { PaperlessDocumentType } from 'src/app/data/paperless-document-type';
@@ -16,6 +16,8 @@ import { MatchingModel } from 'src/app/data/matching-model';
 import { SettingsService, SETTINGS_KEYS } from 'src/app/services/settings.service';
 import { ToastService } from 'src/app/services/toast.service';
 import { saveAs } from 'file-saver';
+import { SplitMergeService } from 'src/app/services/split-merge.service';
+import { Router } from '@angular/router';
 
 @Component({
   selector: 'app-bulk-editor',
@@ -41,7 +43,9 @@ export class BulkEditorComponent {
     private modalService: NgbModal,
     private openDocumentService: OpenDocumentsService,
     private settings: SettingsService,
-    private toastService: ToastService
+    private toastService: ToastService,
+    private splitMergeService: SplitMergeService,
+    private router: Router
   ) { }
 
   applyOnClose: boolean = this.settings.get(SETTINGS_KEYS.BULK_EDIT_APPLY_ON_CLOSE)
@@ -214,4 +218,9 @@ export class BulkEditorComponent {
       saveAs(result, 'documents.zip');
     })
   }
+
+  addToSplitMerge() {
+    this.splitMergeService.addDocuments(this.list.selectedDocuments)
+    this.router.navigate(["split_merge"])
+  }
 }

src-ui/src/app/components/document-list/document-card-small/document-card-small.component.html

@@ -10,7 +10,7 @@
         </div>
       </div>
 
-      <div style="top: 0; right: 0; font-size: large" class="text-right position-absolute mr-1">
+      <div *ngIf="!simpleCard" style="top: 0; right: 0; font-size: large" class="text-right position-absolute mr-1">
         <div *ngFor="let t of getTagsLimited$() | async">
           <app-tag [tag]="t" (click)="clickTag.emit(t.id);$event.stopPropagation()" [clickable]="true" linkTitle="Filter by tag" i18n-linkTitle></app-tag>
         </div>
@@ -22,13 +22,13 @@
 
     <div class="card-body p-2">
       <p class="card-text">
-        <ng-container *ngIf="document.correspondent">
+        <ng-container *ngIf="document.correspondent && !simpleCard">
           <a [routerLink]="" title="Filter by correspondent" i18n-title (click)="clickCorrespondent.emit(document.correspondent);$event.stopPropagation()" class="font-weight-bold">{{(document.correspondent$ | async)?.name}}</a>:
         </ng-container>
         {{document.title | documentTitle}}
       </p>
     </div>
-    <div class="card-footer pt-0 pb-2 px-2">
+    <div *ngIf="!simpleCard" class="card-footer pt-0 pb-2 px-2">
       <div class="list-group list-group-flush border-0 pt-1 pb-2 card-info">
         <button *ngIf="document.document_type" type="button" class="list-group-item list-group-item-action bg-transparent pl-0 p-1 border-0" title="Filter by document type"
          (click)="clickDocumentType.emit(document.document_type);$event.stopPropagation()">

src-ui/src/app/components/document-list/document-card-small/document-card-small.component.ts

@@ -17,12 +17,15 @@ export class DocumentCardSmallComponent implements OnInit {
   @Input()
   selected = false
 
-  @Output()
-  toggleSelected = new EventEmitter()
-
   @Input()
   document: PaperlessDocument
 
+  @Input()
+  simpleCard: boolean = false
+  
+  @Output()
+  toggleSelected = new EventEmitter()
+
   @Output()
   clickTag = new EventEmitter<number>()
 
@@ -93,6 +96,6 @@ export class DocumentCardSmallComponent implements OnInit {
   }
 
   mouseLeaveCard() {
-    this.popover.close()
+    this.popover?.close()
   }
 }

src-ui/src/app/components/split-merge/split-merge.component.html

@@ -0,0 +1,166 @@
+<app-page-header i18n-title title="Split & Merge Editor">
+  <div class="btn-group mt-2 mt-sm-0">
+    <button class="btn btn-sm btn-outline-primary" i18n-title title="Add Documents" (click)="chooseDocuments()">
+      <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+        <use xlink:href="assets/bootstrap-icons.svg#plus-circle" />
+      </svg>&nbsp;<ng-container i18n>Add Documents</ng-container>
+    </button>
+  </div>
+  <div class="btn-group ml-2 mt-2 mt-sm-0">
+    <button type="button" class="btn btn-sm btn-outline-primary"
+      [popoverTitle]="optionsTitle" [autoClose]="'outside'" [ngbPopover]="optionsPopoverBody" popoverClass="shadow" #optionsPopover="ngbPopover">
+      <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+        <use xlink:href="assets/bootstrap-icons.svg#gear" />
+      </svg>&nbsp;<ng-container i18n>Options</ng-container>
+    </button>
+    <ng-template #optionsTitle>
+      <ng-container i18n>Options</ng-container>
+      <button type="button" class="close" aria-label="Close" (click)="optionsPopover.close()">
+        <span aria-hidden="true">&times;</span>
+      </button>
+    </ng-template>
+    <ng-template #optionsPopoverBody>
+      <div class="form-group row text-dark">
+        <label for="source_options" class="col-sm-3 col-form-label pt-0" i18n>Delete source documents</label>
+        <div class="col-sm-9">
+          <div class="custom-control custom-switch">
+            <input type="checkbox" class="custom-control-input" id="source_options-delete" [(ngModel)]="delete_source_documents">
+            <label class="custom-control-label" for="source_options-delete" [class.text-danger]="delete_source_documents"><ng-container i18n>Delete</ng-container><br/><small [class.text-muted]="!delete_source_documents" i18n>Source documents will be immediately deleted.</small></label>
+          </div>
+        </div>
+      </div>
+      <div class="form-group mt-4 row text-dark">
+        <label for="metadata_options" class="col-sm-3 col-form-label pt-0" i18n>Output document metadata</label>
+        <div class="col-sm-9">
+          <div class="custom-control custom-radio">
+            <input type="radio" id="metadata_setting-copy" name="metadata_setting" class="custom-control-input" [value]="splitMergeMetadata.COPY_FIRST" [(ngModel)]="metadata_setting">
+            <label class="custom-control-label" for="metadata_setting-copy"><ng-container i18n>First document</ng-container><br/><small class="text-muted" i18n>Metadata will be copied from first document.</small></label>
+          </div>
+          <div class="custom-control custom-radio">
+            <input type="radio" id="metadata_setting-redo" name="metadata_setting" class="custom-control-input" [value]="splitMergeMetadata.REDO" [(ngModel)]="metadata_setting">
+            <label class="custom-control-label" for="metadata_setting-redo"><ng-container i18n>Redo</ng-container><br/><small class="text-muted" i18n>Metadata detection will be re-run on new document.</small></label>
+          </div>
+        </div>
+      </div>
+    </ng-template>
+  </div>
+
+  <div class="btn-group ml-2 mr-2 mr-sm-5 mt-2 mt-sm-0">
+    <button class="btn btn-sm btn-outline-primary" i18n-title title="Save" (click)="save()" [disabled]="splitMergeService.getDocuments().length < 1">
+      <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+        <use xlink:href="assets/bootstrap-icons.svg#check-circle" />
+      </svg>&nbsp;<ng-container i18n>Save</ng-container>
+    </button>
+  </div>
+
+  <div class="btn-group ml-sm-5 mt-2 mt-sm-0">
+    <button class="btn btn-sm btn-outline-danger" i18n-title title="Cancel" (click)="cancel()">
+      <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+        <use xlink:href="assets/bootstrap-icons.svg#slash-circle" />
+      </svg>&nbsp;<ng-container i18n>Cancel</ng-container>
+    </button>
+  </div>
+</app-page-header>
+
+<div class="row">
+  <div class="col-md-6 col-lg-5">
+    <div class="h-100" dndDropzone (dndDrop)="onDrop($event, documents)">
+      <div class="spacer bg-primary" dndPlaceholderRef>
+      </div>
+      <div *ngFor="let d of documents; let i = index" class="card w-100 mb-3" [class.separator-card]="d.is_separator" [dndDraggable]="d" dndEffectAllowed="move" (dndMoved)="onDragged(d, documents)">
+        <ng-container *ngIf="d.is_separator">
+          <div class="separator"></div>
+          <h6 class="text-muted small font-weight-bold d-inline-block py-2 px-3 mb-0 mx-auto bg-body" i18n>Document End</h6>
+          <button class="btn btn-sm btn-danger" title="Remove" (click)="removeDocument(d, i)">
+            <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+              <use xlink:href="assets/bootstrap-icons.svg#trash" />
+            </svg>
+          </button>
+        </ng-container>
+        <div *ngIf="!d.is_separator" class="row no-gutters">
+          <div class="col-md-3 bg-light d-flex handle" dndHandle>
+            <svg width="2em" height="100%" viewBox="0 0 16 16" fill="currentColor" class="mx-auto">
+              <use xlink:href="assets/bootstrap-icons.svg#grip-horizontal" />
+            </svg>
+            <div class="doc-thumb" [style.background-image]="'url(' + getThumbUrl(d.id) + ')'"></div>
+          </div>
+          <div class="col-md-9">
+            <div class="card-body p-3">
+              <h6 class="card-title mb-2">{{d.title}}</h6>
+              <div class="card-actions">
+                <div class="btn-toolbar btn-toolbar-sm justify-content-between w-100">
+                  <button class="btn btn-sm btn-outline-danger flex-fill" i18n-title title="Remove" (click)="removeDocument(d, i)">
+                    <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+                      <use xlink:href="assets/bootstrap-icons.svg#trash" />
+                    </svg><span class="d-none d-xl-inline">&nbsp;<ng-container i18n>Remove</ng-container></span>
+                  </button>
+                  <button class="btn btn-sm btn-outline-primary ml-3 flex-fill" i18n-title title="Split" (click)="chooseSplit(d, i)">
+                    <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+                      <use xlink:href="assets/bootstrap-icons.svg#scissors" />
+                    </svg><span class="d-none d-xl-inline">&nbsp;<ng-container i18n>Split</ng-container></span>
+                  </button>
+                </div>
+                <div class="input-group input-group-sm w-100 mt-3">
+                  <div class="input-group-prepend">
+                    <div class="input-group-text"><ng-container i18n>Pages</ng-container>:</div>
+                  </div>
+                  <app-input-debounce delay="500" classes="form-control-sm" placeholder="All" ngbTooltip="e.g. 1-3,5" pattern="[0-9\,\-]" [inputValue]="formatPages(d.pages)" (value)="pagesFieldChange($event, d, i)"></app-input-debounce>
+                  <div class="input-group-append">
+                    <button class="btn btn-outline-primary" type="button" i18n-title title="Select Pages" (click)="choosePages(d, i)">
+                      <svg width="1em" height="1em" viewBox="0 0 16 16" fill="currentColor">
+                        <use xlink:href="assets/bootstrap-icons.svg#pencil" />
+                      </svg><span class="d-none d-xl-inline">&nbsp;<ng-container i18n>Select Pages</ng-container></span>
+                    </button>
+                  </div>
+                </div>
+              </div>
+            </div>
+          </div>
+        </div>
+      </div>
+    </div>
+  </div>
+  <div class="col-md-6 col-lg-7">
+    <ngb-alert *ngIf="error" type="danger" class="mt-2 mb-0" [dismissible]="false">
+      <h6 class="alert-heading" i18n>An error occurred during operation</h6>
+      <p class="mb-0 pb-1" i18n>Please check document settings and parts.</p>
+      <p class="mb-0 pb-1 small"><ng-container i18n>Message</ng-container>: <span class="font-italic">{{error}}</span></p>
+    </ngb-alert>
+    <ng-container *ngIf="!error && previewUrls.length > 1">
+      <ul ngbNav #previewNav="ngbNav" class="nav-tabs">
+        <li *ngFor="let previewUrl of previewUrls; let j = index" [ngbNavItem]="j">
+          <a class="bg-light" ngbNavLink><ng-container i18n>Document</ng-container> {{j + 1}}</a>
+          <ng-template ngbNavContent>
+            <div class="preview-container px-2 tabbed text-center bg-light">
+              <ng-container [ngTemplateOutlet]="previewHeader" [ngTemplateOutletContext]="{ $implicit: j }"></ng-container>
+              <pdf-viewer [src]="previewUrl" [original-size]="false" [show-borders]="true" [show-all]="true" [render-text-mode]="2" [(page)]="previewCurrentPages[j]" (after-load-complete)="pdfPreviewLoaded($event, j)" (page-rendered)="pdfPageRendered($event, j)"></pdf-viewer>
+            </div>
+          </ng-template>
+        </li>
+      </ul>
+      <div [ngbNavOutlet]="previewNav"></div>
+    </ng-container>
+    <div *ngIf="!error && previewUrls.length <= 1" class="preview-container px-2 text-center bg-light">
+      <ng-container [ngTemplateOutlet]="previewHeader" [ngTemplateOutletContext]="{ $implicit: 0 }"></ng-container>
+      <pdf-viewer *ngIf="previewUrls.length > 0" [src]="previewUrls[0]" [original-size]="false" [show-borders]="true" [show-all]="true" [render-text-mode]="2" [(page)]="previewCurrentPages[0]" (after-load-complete)="pdfPreviewLoaded($event, 0)" (page-rendered)="pdfPageRendered($event, 0)"></pdf-viewer>
+    </div>
+  </div>
+</div>
+
+<ng-template #previewHeader let-index>
+  <div class="d-flex mt-2 p-2 w-100 justify-space-between">
+    <h6 class="text-muted mt-1" [class.mx-auto]="previewNumPages[index] == undefined" [class.mr-auto]="previewNumPages[index] !== undefined" i18n>Preview</h6>
+    <div *ngIf="previewNumPages[index] !== undefined" class="input-group input-group-sm w-auto">
+      <div class="input-group-prepend">
+        <div class="input-group-text" i18n>Page</div>
+      </div>
+      <input class="form-control flex-grow-0 w-auto" type="number" min="1" [max]="previewNumPages[index]" [(ngModel)]="previewCurrentPages[index]" />
+      <div class="input-group-append">
+        <div class="input-group-text" i18n>of {{previewNumPages[index]}}</div>
+      </div>
+    </div>
+  </div>
+  <div *ngIf="loading" class="spinner-border" role="status">
+    <span class="sr-only" i18n>Loading...</span>
+  </div>
+</ng-template>

src-ui/src/app/components/split-merge/split-merge.component.scss

@@ -0,0 +1,116 @@
+@import "/src/theme";
+
+.card {
+  .handle {
+    cursor: move;
+  }
+
+  .doc-thumb {
+    background-position: top center;
+    background-size: cover;
+    background-repeat: no-repeat;
+    min-height: 5rem;
+    height: 100%;
+    width: 70%;
+  }
+
+  &.separator-card {
+    border: none;
+    background: transparent;
+  }
+}
+
+.spacer {
+  min-height: 0.25rem;
+  width: 100%;
+  margin: 1rem;
+}
+
+.preview-container {
+  height: calc(100vh - 170px);
+  overflow: hidden;
+
+  &.tabbed {
+    height: calc(100vh - 200px);
+  }
+
+  .spinner-border {
+    margin-top: 40%;
+  }
+}
+
+.split-document-preview ::ng-deep .ng2-pdf-viewer-container {
+  width: 1px !important;
+  height: 1px !important;
+  overflow: hidden !important;
+}
+
+::ng-deep .document-selector {
+  width: 80vw;
+  min-width: 50vw;
+}
+
+button span {
+  margin-left: 0.15rem;
+}
+
+app-input-debounce {
+  flex: 1 1;
+  position: relative;
+
+  ::ng-deep .form-control {
+    border-radius: 0;
+  }
+}
+
+.nav-link.bg-light {
+  border-bottom-color: #f8f9fa;
+
+  &:not(.active) {
+    background-color: #fff !important;
+    border-bottom-color: #dee2e6;
+  }
+}
+
+.separator-card {
+  .separator {
+    border: 0;
+    height: 20px;
+    opacity: 0.5;
+    background-color: transparent !important;
+    background-image: url("data:image/svg+xml;utf8, <svg viewBox='0 0 200 110' xmlns='http://www.w3.org/2000/svg'><path d='M -15 110 L100 10 L215 110' fill='none' stroke='%23000000' stroke-width='1' vector-effect='non-scaling-stroke'/></svg>");
+    background-position: bottom left;
+    background-size: 2% auto;
+    background-repeat: repeat-x;
+  }
+
+  h6 {
+    text-transform: uppercase;
+    text-align: center;
+    letter-spacing: 0.1em;
+    margin-top: -20px !important;
+    z-index: 10;
+  }
+
+  button {
+    transition: all 0.3s ease;
+    opacity: 0;
+    pointer-events: none;
+    position: absolute;
+    top: 0;
+    right: 0;
+  }
+
+  &:hover button {
+    opacity: 1;
+    pointer-events: all;
+  }
+}
+
+::ng-deep .popover {
+  min-width: 30rem;
+
+  .close {
+    line-height: 0.7;
+  }
+}

src-ui/src/app/components/split-merge/split-merge.component.spec.ts

@@ -0,0 +1,25 @@
+import { ComponentFixture, TestBed } from '@angular/core/testing';
+
+import { SplitMergeComponent } from './split-merge.component';
+
+describe('SplitMergeComponent', () => {
+  let component: SplitMergeComponent;
+  let fixture: ComponentFixture<SplitMergeComponent>;
+
+  beforeEach(async () => {
+    await TestBed.configureTestingModule({
+      declarations: [ SplitMergeComponent ]
+    })
+    .compileComponents();
+  });
+
+  beforeEach(() => {
+    fixture = TestBed.createComponent(SplitMergeComponent);
+    component = fixture.componentInstance;
+    fixture.detectChanges();
+  });
+
+  it('should create', () => {
+    expect(component).toBeTruthy();
+  });
+});

src-ui/src/app/components/split-merge/split-merge.component.ts

@@ -0,0 +1,212 @@
+import { Component, OnInit, OnDestroy } from '@angular/core';
+import { Router } from '@angular/router';
+import { SplitMergeMetadata } from 'src/app/data/split-merge-request';
+import { SplitMergeService } from 'src/app/services/split-merge.service';
+import { DocumentService } from 'src/app/services/rest/document.service';
+import { DocumentListViewService } from 'src/app/services/document-list-view.service';
+import { PaperlessDocument, PaperlessDocumentPart } from 'src/app/data/paperless-document';
+import { DndDropEvent } from 'ngx-drag-drop';
+import { Subject, Subscription } from 'rxjs';
+import { debounceTime } from 'rxjs/operators';
+import { NgbModal } from '@ng-bootstrap/ng-bootstrap';
+import { DocumentChooserComponent } from 'src/app/components/common/document-chooser/document-chooser.component'
+import { PageChooserComponent } from 'src/app/components/common/page-chooser/page-chooser.component'
+import { PDFDocumentProxy } from 'ng2-pdf-viewer';
+
+@Component({
+  selector: 'app-split-merge',
+  templateUrl: './split-merge.component.html',
+  styleUrls: ['./split-merge.component.scss']
+})
+export class SplitMergeComponent implements OnInit, OnDestroy {
+
+  public loading: boolean = false
+
+  public previewUrls: string[] = []
+  public previewNumPages: number[] = []
+  public previewCurrentPages: number[] = []
+
+  public delete_source_documents: boolean = false
+
+  public metadata_setting: SplitMergeMetadata = SplitMergeMetadata.COPY_FIRST
+
+  private previewDebounce$ = new Subject()
+
+  private previewSub: Subscription
+
+  public error: string
+
+  constructor(
+    public splitMergeService: SplitMergeService,
+    private documentService: DocumentService,
+    private list: DocumentListViewService,
+    private router: Router,
+    private modalService: NgbModal
+  ) { }
+
+  ngOnInit() {
+    this.previewSub = this.previewDebounce$.pipe(
+      debounceTime(400)
+    ).subscribe(() => {
+      if (this.splitMergeService.hasDocuments()) {
+        this.save(true)
+      } else {
+        this.previewUrls = []
+        this.error = undefined
+      }
+    })
+    this.previewDebounce$.next()
+  }
+
+  ngOnDestroy() {
+    this.previewDebounce$.complete()
+  }
+
+  get documents(): PaperlessDocument[] {
+    return this.splitMergeService.getDocuments()
+  }
+
+  getThumbUrl(documentId: number) {
+    return this.documentService.getThumbUrl(documentId)
+  }
+
+  get splitMergeMetadata() {
+    return SplitMergeMetadata
+  }
+
+  chooseDocuments() {
+    let modal = this.modalService.open(DocumentChooserComponent, { backdrop: 'static', size: 'xl' })
+    modal.componentInstance.confirmClicked.subscribe(() => {
+      modal.componentInstance.buttonsEnabled = false
+      this.splitMergeService.addDocuments(this.list.selectedDocuments)
+      this.list.selectNone()
+      modal.close()
+      this.previewDebounce$.next()
+    })
+  }
+
+  onDragged(document: PaperlessDocument, documents: PaperlessDocument[]) {
+    const index = this.documents.indexOf(document)
+    this.documents.splice(index, 1)
+  }
+
+  onDrop(event: DndDropEvent, documents: PaperlessDocument[]) {
+    let index = event.index
+    if (typeof index === "undefined") {
+      index = this.documents.length
+    }
+    this.documents.splice(index, 0, event.data)
+    this.previewDebounce$.next()
+  }
+
+  cancel() {
+    this.splitMergeService.clear()
+    this.router.navigate([""])
+  }
+
+  save(preview: boolean = false) {
+    this.loading = true
+    this.previewUrls = []
+    this.splitMergeService.executeSplitMerge(preview, this.delete_source_documents, this.metadata_setting).subscribe(
+      result => {
+        this.loading = false
+        this.error = undefined
+        if (preview) {
+          this.previewUrls = result.map(r => this.splitMergeService.getPreviewUrl(r))
+        } else {
+          this.splitMergeService.clear()
+          this.router.navigate([""])
+        }
+      },
+      error => {
+        this.error = error
+        this.loading = false
+      }
+    )
+  }
+
+  removeDocument(d: PaperlessDocument, index: number) {
+    this.splitMergeService.removeDocument(d, index)
+    this.previewNumPages[index] = this.previewCurrentPages[index] = undefined
+    this.previewDebounce$.next()
+  }
+
+  choosePages(d: PaperlessDocument, index: number) {
+    let modal = this.modalService.open(PageChooserComponent, { backdrop: 'static', size: 'lg' })
+    modal.componentInstance.document = d
+    modal.componentInstance.confirmPages.subscribe((pages) => {
+      this.splitMergeService.setDocumentPages(d, index, pages)
+      modal.componentInstance.buttonsEnabled = false
+      modal.close()
+      this.previewDebounce$.next()
+    })
+  }
+
+  chooseSplit(d: PaperlessDocument, index: number) {
+    let modal = this.modalService.open(PageChooserComponent, { backdrop: 'static', size: 'lg' })
+    modal.componentInstance.document = d
+    const enabledPages = (d as PaperlessDocumentPart).pages
+    modal.componentInstance.splitting = true
+    modal.componentInstance.confirmPages.subscribe((pages) => {
+      this.splitMergeService.splitDocument(d, index, pages, enabledPages)
+      modal.componentInstance.buttonsEnabled = false
+      modal.close()
+      this.previewDebounce$.next()
+    })
+  }
+
+  pagesFieldChange(pageStr: string, d: PaperlessDocument, index: number) {
+    let pages = pageStr.split(',').map(p => {
+      if (p.indexOf('-') !== -1) {
+        const minmax = p.split('-')
+        let range = []
+        for (let i = parseInt(minmax[0]); i <= parseInt(minmax[1]); i++) {
+          range.push(i)
+        }
+        return range
+      } else if (p.length > 0) {
+        return parseInt(p)
+      } else {
+        return null
+      }
+    })
+    pages = [].concat.apply([], pages) // e.g. flat()
+    pages = pages.filter(page => page !== null)
+    this.splitMergeService.setDocumentPages(d, index, pages as number[])
+    this.previewDebounce$.next()
+  }
+
+  formatPages(pages: number[]): string {
+    let pageStrings = []
+    let rangeStart, rangeEnd
+
+    pages?.forEach(page => {
+      if (rangeStart == undefined) {
+        rangeStart = page
+      } else if (page - rangeStart == 1) {
+        rangeEnd = page
+      } else if (rangeEnd !== undefined && page - rangeEnd == 1) {
+        rangeEnd = page
+      } else {
+        pageStrings.push(rangeEnd !== undefined ? rangeStart + '-' + rangeEnd : rangeStart)
+        rangeStart = page
+        rangeEnd = undefined
+      }
+    })
+
+    if (rangeEnd !== undefined) {
+      pageStrings.push(rangeStart + '-' + rangeEnd)
+    } else if (rangeStart !== undefined) {
+      pageStrings.push(rangeStart)
+    }
+    return pageStrings.join(',')
+  }
+
+  pdfPreviewLoaded(pdf: PDFDocumentProxy, index: number) {
+    this.previewNumPages[index] = pdf.numPages
+  }
+
+  pdfPageRendered(event: any, index) { // CustomEvent is es6
+    if (event.pageNumber > 0 && this.previewCurrentPages[index] == undefined) this.previewCurrentPages[index] = 1
+  }
+}

src-ui/src/app/data/paperless-document.ts

@@ -52,3 +52,11 @@ export interface PaperlessDocument extends ObjectWithId {
     __search_hit__?: SearchHit
 
 }
+
+export interface PaperlessDocumentPart extends PaperlessDocument {
+  pages: number[]
+}
+
+export interface PaperlessDocumentSeparator extends PaperlessDocument {
+  is_separator: boolean
+}

src-ui/src/app/data/split-merge-request.ts

@@ -0,0 +1,29 @@
+export enum SplitMergeMetadata {
+
+  REDO = "redo",
+
+  COPY_FIRST = "copy_first"
+
+}
+
+export interface SplitMergePlanDocumentPart {
+
+  document: number,
+
+  pages?: string
+
+}
+
+export type SplitMergePlan = SplitMergePlanDocumentPart[][]
+
+export interface SplitMergeRequest {
+
+  split_merge_plan: SplitMergePlan
+
+  delete_source: boolean
+
+  metadata: SplitMergeMetadata
+
+  preview: boolean
+
+}

src-ui/src/app/services/document-list-view.service.ts

@@ -189,6 +189,10 @@ export class DocumentListViewService {
     return this.activeListViewState.selected
   }
 
+  get selectedDocuments(): PaperlessDocument[] {
+    return this.activeListViewState.documents.filter(d => this.activeListViewState.selected.has(d.id))
+  }
+
   setSort(field: string, reverse: boolean) {
     this.activeListViewState.sortField = field
     this.activeListViewState.sortReverse = reverse

src-ui/src/app/services/split-merge.service.spec.ts

@@ -0,0 +1,16 @@
+import { TestBed } from '@angular/core/testing';
+
+import { SplitMergeService } from './split-merge.service';
+
+describe('SplitMergeService', () => {
+  let service: SplitMergeService;
+
+  beforeEach(() => {
+    TestBed.configureTestingModule({});
+    service = TestBed.inject(SplitMergeService);
+  });
+
+  it('should be created', () => {
+    expect(service).toBeTruthy();
+  });
+});

src-ui/src/app/services/split-merge.service.ts

@@ -0,0 +1,103 @@
+import { HttpClient, HttpErrorResponse } from '@angular/common/http';
+import { Injectable } from '@angular/core';
+import { Observable, throwError } from 'rxjs';
+import { catchError } from 'rxjs/operators';
+import { environment } from 'src/environments/environment';
+import { PaperlessDocument, PaperlessDocumentPart, PaperlessDocumentSeparator } from '../data/paperless-document';
+import { SplitMergeMetadata, SplitMergeRequest } from '../data/split-merge-request';
+
+@Injectable({
+  providedIn: 'root'
+})
+export class SplitMergeService {
+
+  // this also needs to incorporate pages, if we want to support that.
+  private documents: PaperlessDocument[] = []
+
+  constructor(private http: HttpClient) { }
+
+  addDocument(document: PaperlessDocument, atIndex?: number) {
+    if (atIndex !== undefined) this.documents.splice(atIndex, 0, document)
+    else this.documents.push(document)
+  }
+
+  addDocuments(documents: PaperlessDocument[]) {
+    this.documents = this.documents.concat(documents)
+  }
+
+  removeDocument(document: PaperlessDocument, atIndex?: number) {
+    if (!atIndex) atIndex = this.documents.indexOf(document)
+    this.documents.splice(atIndex, 1)
+    this.sanitizeStartEndSeparators()
+  }
+
+  getDocuments() {
+    return this.documents
+  }
+
+  hasDocuments(): boolean {
+    return this.documents.length > 0
+  }
+
+  clear() {
+    this.documents = []
+  }
+
+  setDocumentPages(d: PaperlessDocument, index: number, pages: number[]) {
+    (this.documents[index] as PaperlessDocumentPart).pages = pages.length > 0 ? pages : null
+  }
+
+  splitDocument(d: PaperlessDocument, index: number, secondPages: number[], enabledPages?: number[]) {
+    const firstPages = []
+    for (let page = 1; page < secondPages[0]; page++) {
+      if (!enabledPages || (enabledPages?.length && enabledPages.indexOf(page) !== -1)) firstPages.push(page)
+    }
+    (this.documents[index] as PaperlessDocumentPart).pages = firstPages
+    this.documents.splice(index + 1, 0, { is_separator: true } as PaperlessDocumentSeparator)
+    let d2 = { ...d } as PaperlessDocumentPart
+    d2.pages = secondPages
+    this.documents.splice(index + 2, 0, d2)
+  }
+
+  sanitizeStartEndSeparators() {
+    if (this.documents.length > 0 && (this.documents[this.documents.length - 1] as PaperlessDocumentSeparator).is_separator) {
+      this.documents.pop()
+    } else if (this.documents.length > 0 && (this.documents[0] as PaperlessDocumentSeparator).is_separator) {
+      this.documents.shift()
+    }
+  }
+
+  executeSplitMerge(preview: boolean, delete_source: boolean, metadata: SplitMergeMetadata): Observable<string[]> {
+    this.sanitizeStartEndSeparators()
+    let currentDocument = []
+    let split_merge_plan = [currentDocument]
+    this.documents.forEach(d => {
+      if ((d as PaperlessDocumentSeparator).is_separator) {
+        currentDocument = []
+        split_merge_plan.push(currentDocument)
+      } else {
+        currentDocument.push({document: d.id, pages: (d as PaperlessDocumentPart).pages?.join(',')})
+      }
+    })
+
+    let request: SplitMergeRequest = {
+      delete_source: delete_source,
+      preview: preview,
+      metadata: metadata,
+      split_merge_plan: split_merge_plan
+    }
+    return this.http.post<string[]>(`${environment.apiBaseUrl}split_merge/`, request).pipe(
+      catchError(this.handleError)
+    )
+  }
+
+  handleError(error: HttpErrorResponse) {
+    console.log('API split_merge result error:', error)
+    return throwError(error.message)
+  }
+
+  getPreviewUrl(previewKey: string) {
+    return `${environment.apiBaseUrl}split_merge/${previewKey}/`
+  }
+
+}

src-ui/src/theme.scss

@@ -9,3 +9,8 @@ $theme-colors: (
 .bg-body {
   background-color: #fff;
 }
+
+.alert-heading {
+  font-size: 80%;
+  font-weight: bold;
+}

src-ui/src/theme_dark.scss

@@ -94,6 +94,23 @@ $border-color-dark-mode: #47494f;
         color: $text-color-dark-mode !important;
       }
     }
+
+    .nav-link.bg-light {
+      border-bottom-color: $border-color-dark-mode !important;
+
+      &:not(.active) {
+        background-color: $bg-dark-mode !important;
+
+        &:hover {
+          background-color: $bg-dark-mode !important;
+        }
+      }
+
+      &.active {
+        background-color: $bg-light-dark-mode !important;
+        border-bottom-color: $bg-light-dark-mode !important;
+      }
+    }
   }
 
   .dropdown-menu {
@@ -424,6 +441,10 @@ $border-color-dark-mode: #47494f;
   .ngb-dp-month {
     background-color: $bg-light-dark-mode;
   }
+
+  app-split-merge .separator {
+    filter: invert(1);
+  }
 }
 
 body.color-scheme-dark {

src/documents/consumer.py

@@ -74,6 +74,7 @@ class Consumer(LoggingMixin):
         self.path = None
         self.filename = None
         self.override_title = None
+        self.override_date = None
         self.override_correspondent_id = None
         self.override_tag_ids = None
         self.override_document_type_id = None
@@ -167,6 +168,7 @@ class Consumer(LoggingMixin):
                          path,
                          override_filename=None,
                          override_title=None,
+                         override_date=None,
                          override_correspondent_id=None,
                          override_document_type_id=None,
                          override_tag_ids=None,
@@ -178,6 +180,7 @@ class Consumer(LoggingMixin):
         self.path = path
         self.filename = override_filename or os.path.basename(path)
         self.override_title = override_title
+        self.override_date = override_date
         self.override_correspondent_id = override_correspondent_id
         self.override_document_type_id = override_document_type_id
         self.override_tag_ids = override_tag_ids
@@ -399,6 +402,9 @@ class Consumer(LoggingMixin):
             for tag_id in self.override_tag_ids:
                 document.tags.add(Tag.objects.get(pk=tag_id))
 
+        if self.override_date:
+            document.created = self.override_date
+
     def _write(self, storage_type, source, target):
         with open(source, "rb") as read_file:
             with open(target, "wb") as write_file:

src/documents/merge.py

@@ -0,0 +1,201 @@
+import logging
+import os
+import re
+import tempfile
+
+from django_q.tasks import async_task
+
+from documents import tasks
+from documents.consumer import ConsumerError
+from documents.models import Document
+
+from pikepdf import Pdf
+
+
+logger = logging.getLogger("paperless.merge")
+
+
+class MergeError(Exception):
+    pass
+
+
+class PdfCache:
+
+    def __init__(self):
+        self.cache = dict()
+
+    def open_from_document(self, document: Document):
+        if document.pk in self.cache:
+            return self.cache[document.pk]
+
+        if document.mime_type == 'application/pdf':
+            filename = document.source_path
+        elif document.has_archive_version:
+            filename = document.archive_path
+        else:
+            raise MergeError(f"Document {document.pk} does not have a PDF.")
+
+        if not os.path.exists(filename):
+            raise MergeError(f"{filename} does not exist.")
+
+        pdf = Pdf.open(filename)
+        self.cache[document.pk] = pdf
+
+        return pdf
+
+    def close_all(self):
+        for pk in self.cache:
+            self.cache[pk].close()
+
+        self.cache.clear()
+
+
+def parse_page_list(page_list: str):
+
+    if not page_list:
+        return []
+
+    result = []
+
+    re_simple = re.compile(r"^\d+$")
+    re_range = re.compile(r"^(\d+)-(\d+)$")
+
+    for page_list_part in page_list.split(","):
+        match_simple = re_simple.match(page_list_part)
+        match_range = re_range.match(page_list_part)
+        if match_simple:
+            result.append(int(page_list_part))
+        elif match_range:
+            first = int(match_range.group(1))
+            last = int(match_range.group(2))
+            if first <= last:
+                result.extend(range(first, last+1))
+            else:
+                result.extend(reversed(range(last, first+1)))
+        else:
+            raise MergeError(f"Invalid page range: {page_list}")
+
+    return result
+
+
+def consume_many_files(kwargs_list, delete_document_ids=None):
+    new_document_ids = []
+
+    try:
+        for kwargs in kwargs_list:
+            new_document_ids.append(tasks.consume_file(**kwargs))
+
+    except ConsumerError:
+        # in case something goes wrong, delete all previously created documents
+        for document_id in new_document_ids:
+            Document.objects.get(id=document_id).delete()
+        raise
+    else:
+        # If everything goes well, optionally delete source documents
+        if delete_document_ids:
+            for document_id in delete_document_ids:
+                Document.objects.get(id=document_id).delete()
+
+
+def copy_pdf_metadata(source: Pdf, target: Pdf):
+    with source.open_metadata() as source_meta:
+        with target.open_metadata() as target_meta:
+            for k in source_meta:
+                try:
+                    target_meta[k] = source_meta[k]
+                except TypeError:
+                    # TODO: https://github.com/pikepdf/pikepdf/issues/188
+                    logger.warning(f"Could not copy metadata {k} while "
+                                   f"merging documents", exc_info=True)
+
+
+def copy_document_metadata(document: Document, consume_task):
+    if document.correspondent:
+        consume_task["override_correspondent_id"] = document.correspondent.id
+    if document.document_type:
+        consume_task["override_document_type_id"] = document.document_type.id
+    if document.tags.count() > 0:
+        consume_task["override_tag_ids"] = [
+            tag.id for tag in document.tags.all()
+        ]
+
+    consume_task['override_date'] = document.created
+
+
+def execute_split_merge_plan(plan,
+                             tempdir: str,
+                             metadata: str = "redo",
+                             delete_source: bool = False,
+                             preview: bool = True):
+
+    consume_tasks = []
+    cache = PdfCache()
+    source_documents = set()
+
+    try:
+        for target_doc_spec in plan:
+            # create a new document from documents in target_doc_spec
+
+            target_pdf: Pdf = None
+            try:
+                target_pdf = Pdf.new()
+                target_pdf_filename = tempfile.NamedTemporaryFile(
+                    prefix="merge_", suffix="_pdf", dir=tempdir).name
+                version = target_pdf.pdf_version
+                consume_task = {"path": target_pdf_filename}
+
+                for (i, source_doc_spec) in enumerate(target_doc_spec):
+                    source_document_id = source_doc_spec['document']
+                    source_documents.add(source_document_id)
+
+                    if 'pages' in source_doc_spec:
+                        pages = parse_page_list(source_doc_spec['pages'])
+                    else:
+                        pages = None
+
+                    try:
+                        source_document: Document = Document.objects.get(
+                            id=source_document_id)
+                    except Document.DoesNotExist:
+                        raise MergeError(
+                            f"Document {source_document_id} does not exist.")
+
+                    source_pdf: Pdf = cache.open_from_document(source_document)
+                    version = max(version, source_pdf.pdf_version)
+
+                    if i == 0:
+                        # first source document for this target
+                        consume_task["override_title"] = source_document.title
+                        copy_pdf_metadata(source_pdf, target_pdf)
+                        if metadata == "copy_first":
+                            copy_document_metadata(
+                                source_document, consume_task)
+
+                    if pages is not None:
+                        for page in pages:
+                            if page > len(source_pdf.pages) or page < 1:
+                                raise MergeError(
+                                    f"Page {page} is out of range.")
+                            target_pdf.pages.append(source_pdf.pages[page - 1])
+                    else:
+                        target_pdf.pages.extend(source_pdf.pages)
+
+                target_pdf.remove_unreferenced_resources()
+                target_pdf.save(target_pdf_filename, min_version=version)
+                target_pdf.close()
+
+                consume_tasks.append(consume_task)
+            finally:
+                if target_pdf is not None:
+                    target_pdf.close()
+    finally:
+        cache.close_all()
+
+    if not preview:
+        async_task(
+            "documents.merge.consume_many_files",
+            kwargs_list=consume_tasks,
+            delete_document_ids=list(source_documents) if delete_source else None  # NOQA: E501
+        )
+
+    return [t["path"] for t in consume_tasks]

src/documents/serialisers.py

@@ -494,3 +494,94 @@ class BulkDownloadSerializer(DocumentListSerializer):
             "bzip2": zipfile.ZIP_BZIP2,
             "lzma": zipfile.ZIP_LZMA
         }[compression]
+
+
+class DocumentSplitMergePlanSerializer(serializers.Serializer):
+    """
+    A split/merge plan describes how a set of documents should be transformed
+    into a new set of documents.
+
+    split_merge_plan is a list. The split and merge tool will create a new
+    document for each element in that list. Each item in split_merge_plan
+    is by itself a list that lists document IDs and pages to take from this
+    document.
+
+    Example:
+
+        split_merge_plan = [
+            [
+                {
+                    "document": 1,
+                    "pages": "1-3"
+                },
+                {
+                    "document": 2,
+                },
+                {
+                    "document": 1,
+                    "pages": "4"
+                }
+            ],
+            [
+                {
+                    "document": 1,
+                    "pages": "6,5"
+                }
+            ]
+        ]
+
+    This will split document 1 into two documents. The first document contains
+    pages 1 to 4, the second document contains pages 5 and 6 in reverse.
+    Additionally, all pages from document 2 are inserted into the first
+    document between pages 3 and 4.
+
+    If delete_source is true, the split and merge tool will delete the source
+    files in the plan after all resulting documents have been added
+    successfully to paperless.
+
+    metadata selects how to assign new metadata to the created documents.
+    With 'redo', paperless will run the matching algorithms on the resulting
+    document again. With 'copy_first', each document created by the merge plan
+    will keep the metadata of the first document used to create that new
+    document.
+
+    If preview is true, the split and merge tool will execute the split and
+    merge plan without modifying any data in paperless. The previews are
+    temporarily made available for download. If preview is true,
+    delete_source and metadata is ignored.
+
+    """
+
+    split_merge_plan = serializers.ListField(
+        required=True,
+        min_length=1,
+        child=serializers.ListField(
+            required=True,
+            min_length=1,
+            child=serializers.DictField()
+        )
+    )
+
+    delete_source = serializers.BooleanField(default=False)
+
+    metadata = serializers.ChoiceField(
+        choices=("redo", "copy_first"),
+        default="copy_first"
+    )
+
+    preview = serializers.BooleanField(default=True)
+
+    def validate_split_merge_plan(self, split_merge_plan):
+        for target_doc in split_merge_plan:
+            for source_doc in target_doc:
+                if "document" not in source_doc:
+                    raise serializers.ValidationError(
+                        '"document" is required'
+                    )
+                try:
+                    Document.objects.get(id=source_doc["document"])
+                except Document.DoesNotExist:
+                    raise serializers.ValidationError(
+                        f"document {source_doc['document']} does not exist"
+                    )
+        return split_merge_plan

src/documents/tasks.py

@@ -66,6 +66,7 @@ def train_classifier():
 def consume_file(path,
                  override_filename=None,
                  override_title=None,
+                 override_date=None,
                  override_correspondent_id=None,
                  override_document_type_id=None,
                  override_tag_ids=None,
@@ -75,6 +76,7 @@ def consume_file(path,
         path,
         override_filename=override_filename,
         override_title=override_title,
+        override_date=override_date,
         override_correspondent_id=override_correspondent_id,
         override_document_type_id=override_document_type_id,
         override_tag_ids=override_tag_ids,
@@ -82,9 +84,7 @@ def consume_file(path,
     )
 
     if document:
-        return "Success. New document id {} created".format(
-            document.pk
-        )
+        return document.pk
     else:
         raise ConsumerError("Unknown error: Returned document was null, but "
                             "no error message was given.")

src/documents/tests/test_api.py

@@ -11,6 +11,7 @@ import pytest
 from django.conf import settings
 from django.contrib.auth.models import User
 from django.test import override_settings
+from pikepdf import Pdf
 from rest_framework.test import APITestCase
 from whoosh.writing import AsyncWriter
 
@@ -1372,6 +1373,75 @@ class TestBulkDownload(DirectoriesMixin, APITestCase):
             "compression": "lzma"
         }), content_type='application/json')
 
+
+class TestSplitMerge(DirectoriesMixin, APITestCase):
+
+    def setUp(self):
+        super(TestSplitMerge, self).setUp()
+
+        user = User.objects.create_superuser(username="temp_admin")
+        self.client.force_login(user=user)
+
+        self.doc1 = Document.objects.create(title="document A", filename="docA.pdf", mime_type="application/pdf", checksum="A")
+        self.doc2 = Document.objects.create(title="document B", filename="docB.jpg", mime_type="image/jpeg", checksum="B", archive_filename="docB.pdf", archive_checksum="D")
+        self.doc3 = Document.objects.create(title="document C", filename="docC.jpg", mime_type="image/jpeg", checksum="C")
+
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.pdf"), self.doc1.source_path)
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.jpg"), self.doc2.source_path)
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.pdf"), self.doc2.archive_path)
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.jpg"), self.doc3.source_path)
+
+    @mock.patch("documents.merge.async_task")
+    def test_merge(self, m):
+        result = self.client.post("/api/split_merge/", json.dumps({
+            "split_merge_plan": [
+                [{"document": self.doc1.pk}, {"document": self.doc2.pk}]
+            ],
+            "preview": False
+        }), content_type="application/json")
+        self.assertEqual(result.status_code, 200)
+        m.assert_called_once()
+        args, kwargs = m.call_args
+        self.assertEqual(kwargs["delete_document_ids"], None)
+
+        self.assertEqual(len(result.data), 1)
+        preview_key = result.data[0]
+
+        preview_response = self.client.get(f"/api/split_merge/{preview_key}/")
+        self.assertEqual(200, preview_response.status_code)
+
+        pdf: Pdf = Pdf.open(io.BytesIO(preview_response.content))
+
+        self.assertEqual(2, len(pdf.pages))
+
+        pdf.close()
+
+    @mock.patch("documents.merge.async_task")
+    def test_merge_delete_source(self, m):
+        result = self.client.post("/api/split_merge/", json.dumps({
+            "split_merge_plan": [
+                [{"document": self.doc1.pk}, {"document": self.doc2.pk}]
+            ],
+            "preview": False,
+            "delete_source": True
+        }), content_type="application/json")
+        self.assertEqual(result.status_code, 200)
+        m.assert_called_once()
+        args, kwargs = m.call_args
+        self.assertCountEqual(kwargs["delete_document_ids"], [self.doc1.pk, self.doc2.pk])
+
+    @mock.patch("documents.merge.async_task")
+    def test_merge_invalid_doc(self, m):
+        result = self.client.post("/api/split_merge/", json.dumps({
+            "split_merge_plan": [
+                [{"document": self.doc1.pk}, {"document": self.doc3.pk}]
+            ],
+            "preview": False
+        }), content_type="application/json")
+        self.assertEqual(result.status_code, 400)
+        m.assert_not_called()
+
+
 class TestApiAuth(APITestCase):
 
     def test_auth_required(self):
@@ -1396,6 +1466,7 @@ class TestApiAuth(APITestCase):
         self.assertEqual(self.client.get("/api/documents/bulk_edit/").status_code, 401)
         self.assertEqual(self.client.get("/api/documents/bulk_download/").status_code, 401)
         self.assertEqual(self.client.get("/api/documents/selection_data/").status_code, 401)
+        self.assertEqual(self.client.get("/api/split_merge/").status_code, 401)
 
     def test_api_version_no_auth(self):
 

src/documents/tests/test_merge.py

@@ -0,0 +1,313 @@
+import datetime
+import hashlib
+import os
+import shutil
+import tempfile
+import uuid
+from contextlib import contextmanager
+from unittest import mock
+from unittest.mock import MagicMock
+
+from django.conf import settings
+from django.test import TestCase, override_settings
+from django.utils import timezone
+
+from documents.consumer import ConsumerError
+from documents.merge import PdfCache, MergeError, execute_split_merge_plan, consume_many_files, parse_page_list
+from documents.models import Document, Correspondent, Tag, DocumentType
+from documents.tests.utils import DirectoriesMixin
+
+
+class TestPdfCache(DirectoriesMixin, TestCase):
+
+    def setUp(self) -> None:
+        super(TestPdfCache, self).setUp()
+        self.cache = PdfCache()
+
+    def test_open_no_pdf(self):
+        doc = Document.objects.create(mime_type="image/jpeg")
+        self.assertRaisesMessage(MergeError, "does not have a PDF.", self.cache.open_from_document, doc)
+
+    def test_open_original_pdf(self):
+        doc = Document.objects.create(mime_type="application/pdf", filename="test.pdf")
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.pdf"), doc.source_path)
+        self.cache.open_from_document(doc)
+        self.assertIn(doc.pk, self.cache.cache)
+
+    def test_document_does_not_exist(self):
+        doc = Document.objects.create(mime_type="application/pdf", filename="test.pdf")
+        self.assertRaisesMessage(MergeError, "does not exist.", self.cache.open_from_document, doc)
+
+    def test_open_archive_pdf(self):
+        doc = Document.objects.create(mime_type="image/jpef", archive_checksum="B", archive_filename="test.pdf")
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.pdf"), doc.archive_path)
+        self.cache.open_from_document(doc)
+        self.assertIn(doc.pk, self.cache.cache)
+
+    @mock.patch("documents.merge.Pdf.open")
+    def test_cached(self, pdf_open: MagicMock):
+        pdf_open.return_value = "the_pdf"
+        doc = Document.objects.create(mime_type="application/pdf", filename="test.pdf")
+        shutil.copy(os.path.join(os.path.dirname(__file__), "samples", "simple.pdf"), doc.source_path)
+        self.assertEqual(self.cache.open_from_document(doc), "the_pdf")
+        pdf_open.assert_called_once()
+        self.assertIn(doc.pk, self.cache.cache)
+
+        pdf_open.reset_mock()
+        self.assertEqual(self.cache.open_from_document(doc), "the_pdf")
+        pdf_open.assert_not_called()
+
+    def test_close(self):
+        self.test_open_original_pdf()
+
+        self.assertEqual(len(self.cache.cache), 1)
+
+        self.cache.close_all()
+
+        self.assertEqual(len(self.cache.cache), 0)
+
+
+class TestParsePageList(TestCase):
+
+    def test_ranges(self):
+        self.assertListEqual(parse_page_list(""), [])
+        self.assertListEqual(parse_page_list(None), [])
+        self.assertListEqual(parse_page_list("1,2,3"), [1, 2, 3])
+        self.assertListEqual(parse_page_list("1,3,2"), [1, 3, 2])
+        self.assertListEqual(parse_page_list("1-3,5"), [1, 2, 3, 5])
+        self.assertListEqual(parse_page_list("1-1,1,2-2,2,6-6,6"), [1, 1, 2, 2, 6, 6])
+        self.assertListEqual(parse_page_list("5-3,1"), [5, 4, 3, 1])
+
+    def test_error(self):
+        self.assertRaises(MergeError, parse_page_list, "as")
+        self.assertRaises(MergeError, parse_page_list, "1--2")
+        self.assertRaises(MergeError, parse_page_list, "1,2,3,")
+        self.assertRaises(MergeError, parse_page_list, "1-8,1-a")
+        self.assertRaises(MergeError, parse_page_list, "8,1-,8")
+        self.assertRaises(MergeError, parse_page_list, "1,,2")
+
+
+class FakePdf:
+
+    def __init__(self):
+        self.pages = list()
+
+    def remove_unreferenced_resources(self):
+        pass
+
+    def close(self):
+        pass
+
+    def save(self, filename, min_version=None):
+        with open(filename, "w") as f:
+            f.write(" ".join(self.pages))
+
+    @property
+    def pdf_version(self):
+        return 0
+
+    @contextmanager
+    def open_metadata(self):
+        yield dict()
+
+
+def fake_open(self, document):
+    pdf = FakePdf()
+    pdf.pages = document.content.split(" ")
+    return pdf
+
+
+@mock.patch("documents.merge.Pdf.new", lambda: FakePdf())
+@mock.patch("documents.merge.PdfCache.open_from_document", fake_open)
+class TestExecuteSplitMergePlan(DirectoriesMixin, TestCase):
+
+    def setUp(self) -> None:
+        super(TestExecuteSplitMergePlan, self).setUp()
+        self.tempdir = os.path.join(settings.SCRATCH_DIR, "split_merge")
+        os.makedirs(self.tempdir, exist_ok=True)
+
+    def _verify_results(self, files, expected_content):
+        self.assertEqual(len(files), len(expected_content))
+        for (file, content) in zip(files, expected_content):
+            with open(file) as f:
+                self.assertEqual(f.read(), content)
+
+    def test_delete_page(self):
+        doc1 = Document.objects.create(content="A B C")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc1.pk, "pages": "1,3"}]
+            ],
+            tempdir=self.tempdir,
+        )
+        self._verify_results(result, ["A C"])
+
+    def test_split(self):
+        doc1 = Document.objects.create(content="A B C")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc1.pk, "pages": "1,2"}],
+                [{"document": doc1.pk, "pages": "3"}],
+            ],
+            tempdir=self.tempdir,
+        )
+        self._verify_results(result, ["A B", "C"])
+
+    def test_merge(self):
+        doc1 = Document.objects.create(content="A B C", checksum="A")
+        doc2 = Document.objects.create(content="D E", checksum="B")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc2.pk}, {"document": doc1.pk}],
+            ],
+            tempdir=self.tempdir,
+        )
+        self._verify_results(result, ["D E A B C"])
+
+    def test_partial_merge_reorder(self):
+        doc1 = Document.objects.create(content="A B C", checksum="A")
+        doc2 = Document.objects.create(content="D E", checksum="B")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc1.pk, "pages": "3,1"}, {"document": doc2.pk, "pages": "2"}],
+            ],
+            tempdir=self.tempdir,
+        )
+        self._verify_results(result, ["C A E"])
+
+    def test_document_does_not_exist(self):
+        self.assertRaisesMessage(
+            MergeError,
+            "Document 3456 does not exist.",
+            execute_split_merge_plan,
+            plan=[[{"document": 3456}]],
+            tempdir=self.tempdir,
+        )
+
+    def test_page_out_of_range(self):
+        doc1 = Document.objects.create(content="A B C")
+        self.assertRaisesMessage(
+            MergeError,
+            "Page 4 is out of range.",
+            execute_split_merge_plan,
+            plan=[[{"document": doc1.pk, "pages": "4"}]],
+            tempdir=self.tempdir,
+        )
+
+    @mock.patch("documents.merge.async_task")
+    def test_no_preview(self, async_task: MagicMock):
+        doc1 = Document.objects.create(content="A B C", checksum="A")
+        doc2 = Document.objects.create(content="D E", checksum="B")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc2.pk}, {"document": doc1.pk}],
+            ],
+            tempdir=self.tempdir,
+            preview=False,
+            delete_source=False
+        )
+        async_task.assert_called_once()
+        args, kwargs = async_task.call_args
+        self.assertListEqual(kwargs['kwargs_list'], [{"path": result[0], "override_title": ""}])
+        self.assertIsNone(kwargs['delete_document_ids'])
+
+    @mock.patch("documents.merge.async_task")
+    def test_no_preview_delete_source(self, async_task: MagicMock):
+        doc1 = Document.objects.create(content="A B C", checksum="A")
+        doc2 = Document.objects.create(content="D E", checksum="B")
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc2.pk}, {"document": doc1.pk}],
+            ],
+            tempdir=self.tempdir,
+            preview=False,
+            delete_source=True
+        )
+        async_task.assert_called_once()
+        args, kwargs = async_task.call_args
+        self.assertCountEqual(kwargs['delete_document_ids'], [doc1.pk, doc2.pk])
+
+    @mock.patch("documents.merge.async_task")
+    @override_settings(TIME_ZONE="UTC")
+    def test_metadata_copy_first(self, async_task: MagicMock):
+        c1 = Correspondent.objects.create(name="c1")
+        c2 = Correspondent.objects.create(name="c2")
+        t1 = Tag.objects.create(name="t1")
+        t2 = Tag.objects.create(name="t2")
+        dt1 = DocumentType.objects.create(name="dt1")
+        dt2 = DocumentType.objects.create(name="dt2")
+        doc1 = Document.objects.create(content="A B C", checksum="A", title="titleA", correspondent=c1, document_type=dt1, created=timezone.datetime(2020, 12, 1, 0, 0, 0))
+        doc1.tags.add(t1)
+        doc2 = Document.objects.create(content="D E", checksum="B", title="titleB", correspondent=c2, document_type=dt2, created=timezone.datetime(2020, 2, 20, 0, 0, 0))
+        doc2.tags.add(t2)
+
+        result = execute_split_merge_plan(
+            plan=[
+                [{"document": doc2.pk}, {"document": doc1.pk}],
+            ],
+            tempdir=self.tempdir,
+            preview=False,
+            metadata="copy_first",
+        )
+        async_task.assert_called_once()
+        args, kwargs = async_task.call_args
+        self.assertEqual(1, len(kwargs['kwargs_list']))
+
+        kwargs = kwargs['kwargs_list'][0]
+        self.assertEqual(kwargs["override_title"], "titleB")
+        self.assertEqual(kwargs["override_correspondent_id"], c2.pk)
+        self.assertEqual(kwargs["override_tag_ids"], [t2.pk])
+        self.assertEqual(kwargs["override_document_type_id"], dt2.pk)
+        self.assertEqual(kwargs["override_date"], timezone.make_aware(datetime.datetime(2020, 2, 20, 0, 0, 0)))
+
+
+def fake_consume_file(fail, *args, **kwargs):
+    if fail:
+        raise ConsumerError()
+    return Document.objects.create(checksum=str(Document.objects.count())).pk
+
+
+class TestConsumeManyFiles(DirectoriesMixin, TestCase):
+
+    @mock.patch("documents.merge.tasks.consume_file")
+    def test_success(self, consume_file: MagicMock):
+        consume_file.side_effect = fake_consume_file
+        consume_many_files([{"fail": False}, {"fail": False}])
+        self.assertEqual(consume_file.call_count, 2)
+        self.assertEqual(Document.objects.count(), 2)
+
+    @mock.patch("documents.merge.tasks.consume_file")
+    def test_fail(self, consume_file: MagicMock):
+        consume_file.side_effect = fake_consume_file
+        self.assertRaises(
+            ConsumerError,
+            consume_many_files,
+            [{"fail": False}, {"fail": True}, {"fail": False}]
+        )
+        self.assertEqual(consume_file.call_count, 2)
+        self.assertEqual(Document.objects.count(), 0)
+
+    @mock.patch("documents.merge.tasks.consume_file")
+    def test_success_delete_source(self, consume_file: MagicMock):
+        doc_id = Document.objects.create().id
+        self.assertTrue(Document.objects.filter(id=doc_id).exists())
+        consume_file.side_effect = fake_consume_file
+        consume_many_files([{"fail": False}, {"fail": False}], delete_document_ids=[doc_id])
+        self.assertEqual(consume_file.call_count, 2)
+        self.assertEqual(Document.objects.count(), 2)
+        self.assertFalse(Document.objects.filter(id=doc_id).exists())
+
+    @mock.patch("documents.merge.tasks.consume_file")
+    def test_fail_no_delete_source(self, consume_file: MagicMock):
+        doc_id = Document.objects.create().id
+        self.assertTrue(Document.objects.filter(id=doc_id).exists())
+        consume_file.side_effect = fake_consume_file
+        self.assertRaises(
+            ConsumerError,
+            consume_many_files,
+            [{"fail": False}, {"fail": True}],
+            delete_document_ids=[doc_id]
+        )
+        self.assertEqual(consume_file.call_count, 2)
+        self.assertEqual(Document.objects.count(), 1)
+        self.assertTrue(Document.objects.filter(id=doc_id).exists())

src/documents/views.py

@@ -6,6 +6,7 @@ import zipfile
 from datetime import datetime
 from time import mktime
 
+import pathvalidate
 from django.conf import settings
 from django.db.models import Count, Max, Case, When, IntegerField
 from django.db.models.functions import Lower
@@ -57,9 +58,10 @@ from .serialisers import (
     DocumentTypeSerializer,
     PostDocumentSerializer,
     SavedViewSerializer,
-    BulkEditSerializer,
     DocumentListSerializer,
-    BulkDownloadSerializer
+    BulkDownloadSerializer,
+    BulkEditSerializer,
+    DocumentSplitMergePlanSerializer
 )
 
 
@@ -624,3 +626,53 @@ class BulkDownloadView(GenericAPIView):
                 "attachment", "documents.zip")
 
             return response
+
+
+class DocumentSplitMergeViewSet(GenericViewSet):
+
+    permission_classes = (IsAuthenticated,)
+    serializer_class = DocumentSplitMergePlanSerializer
+
+    def __init__(self, **kwargs):
+        super(DocumentSplitMergeViewSet, self).__init__(**kwargs)
+        self.tempdir = os.path.join(settings.SCRATCH_DIR, "split-merge")
+        os.makedirs(self.tempdir, exist_ok=True)
+
+    def get_queryset(self):
+        return os.listdir(self.tempdir)
+
+    def retrieve(self, request, pk, *args, **kwargs):
+        filename = os.path.join(
+            self.tempdir, pathvalidate.sanitize_filename(pk))
+        if not os.path.isfile(filename):
+            raise Http404()
+
+        with open(filename, "rb") as f:
+            return HttpResponse(f, content_type="application/pdf")
+
+    def list(self, request, *args, **kwargs):
+        return Response(self.get_queryset())
+
+    def create(self, request, *args, **kwargs):
+        serializer = self.get_serializer(data=request.data)
+        serializer.is_valid(raise_exception=True)
+
+        from .merge import execute_split_merge_plan, MergeError
+
+        split_merge_plan = serializer.validated_data.get("split_merge_plan")
+        preview = serializer.validated_data.get("preview")
+        delete_source = serializer.validated_data.get("delete_source")
+        metadata = serializer.validated_data.get("metadata")
+        try:
+            pdf_files = execute_split_merge_plan(
+                plan=split_merge_plan,
+                preview=preview,
+                delete_source=delete_source,
+                metadata=metadata,
+                tempdir=self.tempdir
+            )
+        except MergeError:
+            logger.exception("Error during merge")
+            return HttpResponseBadRequest()
+
+        return Response([os.path.basename(file) for file in pdf_files])

src/paperless/urls.py

@@ -23,7 +23,8 @@ from documents.views import (
     SavedViewViewSet,
     BulkEditView,
     SelectionDataView,
-    BulkDownloadView
+    BulkDownloadView,
+    DocumentSplitMergeViewSet
 )
 from paperless.views import FaviconView
 
@@ -35,6 +36,9 @@ api_router.register(r"logs", LogViewSet, basename="logs")
 api_router.register(r"tags", TagViewSet)
 api_router.register(r"saved_views", SavedViewViewSet)
 
+api_router.register(
+    r"split_merge", DocumentSplitMergeViewSet, basename="split_merge")
+
 
 urlpatterns = [
     re_path(r"^api/", include([