add more utilities

src/app.module.ts

@@ -5,7 +5,7 @@ import { PdfController } from './pdf/pdf.controller';
 import { ConfigModule } from '@nestjs/config';
 import config from './config/configuration';
 import { HttpModule } from '@nestjs/axios';
-import { FileUtil } from './utils/pdf/fileUtils';
+import { FileUtil } from './utils/pdf/fileUtil';
 
 @Module({
   imports: [

src/config/configuration.ts

@@ -1,6 +1,5 @@
 import * as path from 'path';
 
 export default () => ({
-  port: parseInt(process.env.PORT) || 3000,
   pdfOutputPath: path.resolve(__dirname, '../../test/pdf/output'),
 });

src/pdf/pdf-parser/horaire/HoraireCours.ts

@@ -1,13 +1,7 @@
 import { CourseCodeValidationPipe } from '../../pipes/course-code-validation-pipe';
 import { Group } from './Group';
 import { Period } from './Period';
-
-interface IHoraireCours {
-  code: string;
-  title: string;
-  prerequisites: string;
-  groups: { [groupNumber: string]: Group };
-}
+import { IHoraireCours } from './horaire-cours.types';
 
 export class HoraireCours implements IHoraireCours {
   private static readonly TITLE_FONT_SIZE = 10.998999999999999;

src/pdf/pdf-parser/horaire/horaire-cours.service.ts

@@ -1,11 +1,26 @@
-import PDFParser, { Output, Page, Text } from 'pdf2json';
+import { Output, Page, Text } from 'pdf2json';
 import { Injectable } from '@nestjs/common';
 import { HttpService } from '@nestjs/axios';
-import { FileUtil } from '../../../utils/pdf/fileUtils';
+import { FileUtil } from '../../../utils/pdf/fileUtil';
 import { firstValueFrom } from 'rxjs';
 import { HoraireCours } from './HoraireCours';
 import { Period } from './Period';
 import { Group } from './Group';
+import { TextExtractor } from '../../../utils/pdf/parser/textExtractorUtil';
+import { PdfParserUtil } from '../../../utils/pdf/parser/pdfParserUtil';
+
+/**
+ *   private processPdfData(
+    err: null | Error,
+    pdfData: any,
+  ): PlanificationCourse[] {
+ *
+ * Je penses que cette fonction pourrait être déplacé vers un fichier utilitaire qui serait appeler par le service,
+ * cela créerait un segregation du code en deux portions :
+ *
+ * 1. parsing du input data (le fichier util),
+ * 2. gestion et maintenance des données (dans le service)
+ */
 
 @Injectable()
 export class HoraireCoursService {
@@ -14,10 +29,7 @@ export class HoraireCoursService {
   private readonly START_PAGE_CONTENT_Y_AXIS = 14.019;
   private readonly END_PAGE_CONTENT_Y_AXIS = 59;
 
-  constructor(
-    private httpService: HttpService,
-    private fileUtil: FileUtil,
-  ) {}
+  constructor(private httpService: HttpService) {}
 
   public async parsePdfFromUrl(pdfUrl: string) {
     try {
@@ -34,24 +46,10 @@ export class HoraireCoursService {
   private async parseHoraireCoursPdf(
     pdfBuffer: Buffer,
   ): Promise<HoraireCours[]> {
-    const parser = new PDFParser();
-    return new Promise((resolve, reject) => {
-      parser.on('pdfParser_dataError', (errData: string) =>
-        console.error(errData),
-      );
-      parser.on('pdfParser_dataReady', async (pdfData) => {
-        try {
-          console.info('Parsing PDF...');
-          await this.fileUtil.writeDataToFile(pdfData, 'inputHoraire.json');
-          const courses = this.processPdfData(pdfData);
-          await this.fileUtil.writeDataToFile(courses, 'coursesHoraire.json');
-          resolve(courses);
-        } catch (error) {
-          reject(error);
-        }
-      });
-      parser.parseBuffer(pdfBuffer);
-    });
+    return PdfParserUtil.parsePdfBuffer(
+      pdfBuffer,
+      this.processPdfData.bind(this),
+    );
   }
 
   // Processes the raw PDF data to extract course information
@@ -69,12 +67,12 @@ export class HoraireCoursService {
           const {
             textContent: text,
             fontSize,
-            bold,
             xPos,
             yPos,
-          } = this.extractTextDetails(textItem);
+            bold,
+          } = TextExtractor.extractTextDetails(textItem);
 
-          if (!text || yPos > this.END_PAGE_CONTENT_Y_AXIS || bold) return; //Les cells du header (COURS, ...) sont en gras
+          if (!text || yPos > this.END_PAGE_CONTENT_Y_AXIS || bold) return;
 
           if (HoraireCours.isCourseCode(text, xPos)) {
             // Finalize the last group of the current course if necessary
@@ -143,13 +141,4 @@ export class HoraireCoursService {
       throw new Error('Error processing PDF data');
     }
   }
-
-  private extractTextDetails(textItem: Text) {
-    const textContent = decodeURIComponent(textItem.R[0].T).trim();
-    const fontSize = textItem.R[0].TS[1];
-    const bold = textItem.R[0].TS[2];
-    const xPos = textItem.x;
-    const yPos = textItem.y;
-    return { textContent, fontSize, bold, xPos, yPos };
-  }
 }

src/pdf/pdf-parser/horaire/horaire-cours.types.ts

@@ -0,0 +1,8 @@
+import { Group } from './Group';
+
+export interface IHoraireCours {
+  code: string;
+  title: string;
+  prerequisites: string;
+  groups: { [groupNumber: string]: Group };
+}

src/pdf/pdf-parser/planification/Row.ts

@@ -11,7 +11,7 @@ export class Row {
     this.endX = this.truncateToFiveDecimals(endX);
   }
 
-  private truncateToFiveDecimals(num) {
+  private truncateToFiveDecimals(num: number) {
     return Math.floor(num * 100000) / 100000;
   }
 

src/pdf/pdf-parser/planification/planification-cours.service.ts

@@ -1,11 +1,13 @@
 import PDFParser, { Output, Fill, Page, Text } from 'pdf2json';
 import { HttpService } from '@nestjs/axios';
-import { FileUtil } from '../../../utils/pdf/fileUtils';
+import { FileUtil } from '../../../utils/pdf/fileUtil';
 import { Injectable } from '@nestjs/common';
 import { firstValueFrom } from 'rxjs';
 import { CourseCodeValidationPipe } from '../../pipes/course-code-validation-pipe';
 import { Row } from './Row';
 import { PlanificationCours } from './planification-cours.types';
+import { TextExtractor } from '../../../utils/pdf/parser/textExtractorUtil';
+import { PdfParserUtil } from '../../../utils/pdf/parser/pdfParserUtil';
 
 //TODO Add title to the course (cycles superieurs ont plusieurs cours avec le meme code)
 //https://horaire.etsmtl.ca/Horairepublication/Planification-CyclesSuperieurs.pdf
@@ -36,31 +38,10 @@ export class PlanificationCoursService {
   private parsePlanificationCoursPdf(
     pdfBuffer: Buffer,
   ): Promise<PlanificationCours[]> {
-    const parser = new PDFParser(this, 1);
-
-    return new Promise((resolve, reject) => {
-      parser.on('pdfParser_dataError', (errData: string) =>
-        console.error(errData),
-      );
-      parser.on('pdfParser_dataReady', async (pdfData: Output) => {
-        try {
-          await this.fileUtil.writeDataToFile(
-            pdfData,
-            'inputPlanification.json',
-          );
-          const courses = this.processPdfData(pdfData);
-          await this.fileUtil.writeDataToFile(
-            courses,
-            'coursesPlanification.json',
-          );
-          resolve(courses);
-        } catch (error) {
-          console.error('Error parsing pdf data: ' + error);
-          reject(error);
-        }
-      });
-      parser.parseBuffer(pdfBuffer);
-    });
+    return PdfParserUtil.parsePdfBuffer(
+      pdfBuffer,
+      this.processPdfData.bind(this),
+    );
   }
 
   private processPdfData(pdfData: Output): PlanificationCours[] {
@@ -73,7 +54,8 @@ export class PlanificationCoursService {
       pdfData.Pages.forEach((page: Page) => {
         page.Texts.forEach((textItem: Text) => {
           // eslint-disable-next-line @typescript-eslint/no-unused-vars
-          const { textContent, xPos, yPos } = this.extractTextDetails(textItem); //TODO Check yPos later
+          const { textContent, xPos, yPos } =
+            TextExtractor.extractTextDetails(textItem); //TODO Check yPos later
 
           const currentColumn = Row.getColumnHeaderName(headerCells, xPos);
           // Process course code

src/pdf/pdf.controller.ts

@@ -10,7 +10,7 @@ import { HoraireCoursService } from './pdf-parser/horaire/horaire-cours.service'
 import { PlanificationCoursService } from './pdf-parser/planification/planification-cours.service';
 import { isValidUrl } from '../utils/url/urlUtils';
 import { PlanificationCours } from './pdf-parser/planification/planification-cours.types';
-import { HoraireCours } from './pdf-parser/horaire/horaire-cours.types';
+import { IHoraireCours } from './pdf-parser/horaire/horaire-cours.types';
 
 @Controller('pdf')
 export class PdfController {
@@ -22,7 +22,7 @@ export class PdfController {
   @Get('parseHoraireCoursPdf')
   async parseHoraireCoursPdf(
     @Query('pdfUrl') pdfUrl: string,
-  ): Promise<HoraireCours[]> {
+  ): Promise<IHoraireCours[]> {
     try {
       console.log('Controller file', pdfUrl);
       if (!pdfUrl || !isValidUrl(pdfUrl)) {

src/utils/pdf/fileUtils.ts → src/utils/pdf/fileUtil.ts

@@ -7,7 +7,7 @@ import { ConfigService } from '@nestjs/config';
 export class FileUtil {
   constructor(private configService: ConfigService) {}
 
-  writeDataToFile(data: any, fileName: string): Promise<string | null> {
+  public writeDataToFile(data: any, fileName: string): Promise<string | null> {
     const pdfOutputPath =
       this.configService.get<string>('pdfOutputPath') ||
       path.join(__dirname, fileName);
@@ -34,8 +34,10 @@ export class FileUtil {
             console.error('Error encountered while writing file: ', err);
             reject(err);
           } else {
-            console.log(`File "${fileName}" successfully written`);
-            resolve(data);
+            console.log(
+              `File "${fileName}" successfully written to "${filePath}"`,
+            );
+            resolve(filePath);
           }
         },
       );

src/utils/pdf/parser/pdfParserUtil.ts

@@ -0,0 +1,24 @@
+import PDFParser, { Output } from 'pdf2json';
+import { TextExtractor } from './textExtractorUtil';
+
+export class PdfParserUtil {
+  public static async parsePdfBuffer(
+    pdfBuffer: Buffer,
+    processData: (pdfData: Output) => any,
+  ): Promise<any> {
+    const parser = new PDFParser();
+
+    return new Promise((resolve, reject) => {
+      parser.on('pdfParser_dataError', (errData) => console.error(errData));
+      parser.on('pdfParser_dataReady', async (pdfData) => {
+        try {
+          const result = await processData(pdfData);
+          resolve(result);
+        } catch (error) {
+          reject(error);
+        }
+      });
+      parser.parseBuffer(pdfBuffer);
+    });
+  }
+}

src/utils/pdf/parser/textExtractorUtil.ts

@@ -0,0 +1,19 @@
+import { Text } from 'pdf2json';
+
+export class TextExtractor {
+  public static extractTextDetails(textItem: Text): {
+    fontSize: number;
+    textContent: string;
+    xPos: number;
+    yPos: number;
+    bold: boolean;
+  } {
+    const textContent: string = decodeURIComponent(textItem.R[0].T).trim();
+    const fontSize: number = textItem.R[0].TS[1];
+    const xPos: number = textItem.x;
+    const yPos: number = textItem.y;
+    const bold: boolean = textItem.R[0].TS[2] === 1;
+
+    return { textContent, fontSize, xPos, yPos, bold };
+  }
+}