Merge pull request #325 from axa-group/feature/page-orientation

Feature/page orientation

demo/vue-viewer/src/components/DocumentPreview/Page.vue

@@ -3,7 +3,18 @@
     <svg
       class="Page"
       :id="'SVG_' + page.pageNumber"
-      :style="{ zoom: zoom * zoomToFitPage }"
+      :style="{
+        zoom: zoom * zoomToFitPage,
+        transform:
+          'translateX(' +
+          page.rotation.translation.x +
+          'px) translateY(' +
+          page.rotation.translation.y +
+          'px) rotate(' +
+          page.rotation.degrees +
+          'deg)',
+        transformOrigin: page.rotation.origin.x + 'px ' + page.rotation.origin.y + 'px',
+      }"
       :width="page.box.w"
       :height="page.box.h"
     >
@@ -41,19 +52,7 @@
           style="stroke: #aeaeae"
         />
       </svg>
-      <g
-        :style="{
-          transform:
-            'translateX(' +
-            page.rotation.translation.x +
-            'px) translateY(' +
-            page.rotation.translation.y +
-            'px) rotate(' +
-            page.rotation.degrees +
-            'deg)',
-          transformOrigin: page.rotation.origin.x + 'px ' + page.rotation.origin.y + 'px',
-        }"
-      >
+      <g>
         <imageData
           v-for="element in images"
           :key="element.id"
@@ -267,16 +266,14 @@ export default {
 <style lang="scss">
 .PageContainer {
   margin: 0 auto;
+  display: flex;
+  align-items: center;
 }
 .Page {
   background-color: white;
   border: 1px solid rgb(204, 204, 231);
   margin: 0 auto;
   position: relative;
-  top: 50%;
-  -webkit-transform: translateY(-50%);
-  -ms-transform: translateY(-50%);
-  transform: translateY(-50%);
 }
 
 .Page text {

demo/vue-viewer/src/components/Thumbnails.vue

@@ -137,6 +137,7 @@ export default {
   top: 38px;
   left: 0;
   right: 0;
+  zoom: 0.9;
 }
 
 .Thumb:last-of-type {

demo/vue-viewer/src/components/Thumbnails/Thumbnail.vue

@@ -54,7 +54,6 @@ export default {
   box-sizing: border-box;
 }
 .Thumb img {
-  max-width: 80%;
   max-height: 100%;
   box-shadow: 0 0 2px 1px #ebebf1;
   margin: 10px;

server/src/input/pdfminer/PdfminerExtractor.ts

@@ -13,7 +13,7 @@
  * See the License for the specific language governing permissions and
  * limitations under the License.
  */
-
+import * as limit from 'limit-async';
 import { Document } from '../../types/DocumentRepresentation';
 import * as CommandExecuter from '../../utils/CommandExecuter';
 import logger from '../../utils/Logger';
@@ -46,7 +46,7 @@ export class PdfminerExtractor extends Extractor {
             const totalSeconds = (Date.now() - startTime) / 1000;
             logger.info(
               `Total PdfMiner ${
-                totalPages != null ? '(' + totalPages.toString() + ')' : ''
+              totalPages != null ? '(' + totalPages.toString() + ')' : ''
               } time: ${totalSeconds} sec - ${totalSeconds / 60} min`,
             );
             return doc;
@@ -78,8 +78,9 @@ export class PdfminerExtractor extends Extractor {
     const extractPages = this.pagesToExtract(pageIndex, maxPages, totalPages);
     return pdfminer
       .extractPages(inputFile, totalPages != null ? extractPages : null)
-      .then((xmlOutputFile: string) => pdfminer.xmlParser(xmlOutputFile))
-      .then((json: any) => pdfminer.jsParser(json))
+      .then(pdfminer.xmlParser)
+      .then(pdfminer.jsParser)
+      .then(this.detectAndFixPageRotation(inputFile))
       .then((doc: Document) => {
         document.pages = document.pages.concat(doc.pages);
         document.pages.forEach((page, index) => (page.pageNumber = index + 1));
@@ -91,6 +92,51 @@ export class PdfminerExtractor extends Extractor {
       });
   }
 
+  private async rotatePages(doc: Document, pages: number[], rotation: number): Promise<Document> {
+    logger.info(`pages ${pages.join(', ')} will be reprocessed with a rotation angle of ${rotation} degrees`);
+
+    const fixedDoc: Document = await pdfminer
+      .extractPages(doc.inputFile, pages.join(','), rotation)
+      .then(pdfminer.xmlParser)
+      .then(pdfminer.jsParser);
+    pages.forEach(pageNumber => {
+      doc.pages[pageNumber - 1] = fixedDoc.pages[pages.indexOf(pageNumber)];
+      doc.pages[pageNumber - 1].pageRotation = {
+        fileName: doc.inputFile,
+        degrees: -rotation,
+        translation: { x: 0, y: 0 },
+        origin: {
+          x: doc.pages[pageNumber - 1].width / 2,
+          y: doc.pages[pageNumber - 1].height / 2,
+        },
+      };
+
+    });
+    return doc;
+  }
+
+  private detectAndFixPageRotation(inputFile: string): (doc: Document) => Promise<Document> {
+    const limiter = limit(1);
+    return async (doc: Document): Promise<Document> => {
+      doc.inputFile = inputFile;
+      const startTime: number = Date.now();
+      const pageRotations = doc.pages.map(p => p.getMainRotationAngle()).reduce(this.groupByRotation, {});
+      const promises = Object.keys(pageRotations)
+        .filter(r => r !== '0')
+        .map(rotation => limiter(this.rotatePages)(doc, pageRotations[rotation], rotation));
+
+      await Promise.all(promises);
+      logger.info(`Page rotation detection and correction finished in ${(Date.now() - startTime) / 1000}s`);
+      return doc;
+    };
+  }
+
+  private groupByRotation(acc, value, index): any {
+    acc[value] = acc[value] || [];
+    acc[value].push(index + 1);
+    return acc;
+  }
+
   private pagesToExtract(pageIndex: number, maxPages: number, totalPages: number) {
     const fromPage = (pageIndex - 1) * maxPages;
     return [...Array(maxPages).keys()]

server/src/input/pdfminer/pdfminer.ts

@@ -41,10 +41,10 @@ import logger from '../../utils/Logger';
  * @returns The promise of a valid document (in the format DocumentRepresentation).
  */
 
-export function extractPages(pdfInputFile: string, pages: string): Promise<string> {
+export function extractPages(pdfInputFile: string, pages: string, rotationDegrees: number = 0): Promise<string> {
   return new Promise<string>((resolveXml, rejectXml) => {
     const startTime: number = Date.now();
-    CommandExecuter.pdfMinerExtract(pdfInputFile, pages)
+    CommandExecuter.pdfMinerExtract(pdfInputFile, pages, rotationDegrees)
       .then(xmlOutputPath => {
         logger.info(`PdfMiner xml: ${(Date.now() - startTime) / 1000}s`);
         try {

server/src/types/DocumentRepresentation/Page.ts

@@ -1,5 +1,5 @@
 /**
- * Copyright 2019 AXA Group Operations S.A.
+ * Copyright 2020 AXA Group Operations S.A.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -18,10 +18,12 @@ import { RotationCorrection } from '../../input/OcrExtractor';
 import { findMostCommonFont, isInBox } from '../../utils';
 import logger from '../../utils/Logger';
 import { BoundingBox } from './BoundingBox';
+import { Character } from './Character';
 import { Element } from './Element';
 import { Font } from './Font';
 import { Paragraph } from './Paragraph';
 import { Text } from './Text';
+import { Word } from './Word';
 
 export type directionType = 'horizontal' | 'vertical';
 /**
@@ -71,6 +73,29 @@ export class Page {
     this.computePageOccupancy();
   }
 
+  public getMainRotationAngle(): number {
+    const rotations = this.getElementsOfType<Word>(Word, true)
+      .map((word) => {
+        if (Array.isArray(word.content) && word.content.length > 1) {
+          const { left: x1, bottom: y1 } = word.content[0] as Character;
+          const { left: x2, bottom: y2 } = word.content[word.content.length - 1] as Character;
+          const arcTan = Math.round(Math.atan((y1 - y2) / (x1 - x2)) * 180 / Math.PI);
+          return arcTan === 0 ? (x1 < x2 ? 0 : 180) : arcTan;
+        }
+        return 0;
+      });
+
+    const elementsPerRotation = rotations.reduce((acc, value) => {
+      acc[value] = acc[value] || 0;
+      acc[value] += 1;
+      return acc;
+    }, {});
+
+    const highestValue: number = Math.max(...(Object.values(elementsPerRotation) as number[]));
+    const mainRotation = Object.keys(elementsPerRotation).find(k => elementsPerRotation[k] === highestValue);
+    return parseInt(mainRotation, 10);
+  }
+
   /**
    * Computes all horizontal and vertical page occupancies
    */

server/src/utils/CommandExecuter.ts

@@ -180,9 +180,16 @@ export async function detectTables(
   });
 }
 
-export async function pdfMinerExtract(filePath: string, pages: string): Promise<string> {
+export async function pdfMinerExtract(filePath: string, pages: string, rotationDegrees: number = 0): Promise<string> {
   const xmlOutputFile: string = getTemporaryFile('.xml');
-  let pdf2txtArguments: string[] = ['-c', 'utf-8', '-t', 'xml', '-o', xmlOutputFile, filePath];
+  let pdf2txtArguments: string[] = [
+    '--detect-vertical',
+    '-R', rotationDegrees.toString(),
+    '-c', 'utf-8',
+    '-t', 'xml',
+    '-o', xmlOutputFile,
+    filePath,
+  ];
 
   if (pages != null) {
     pdf2txtArguments = ['-p', pages].concat(pdf2txtArguments);