1 00:00:00,060 --> 00:00:05,070 In this tutorial, I'm going to be showing you the different ways you can scrape data from a PDF documents 2 00:00:05,220 --> 00:00:07,130 using optical character recognition. 3 00:00:07,950 --> 00:00:15,700 So the most common way is to use the kit OCR text activity over here. 4 00:00:16,020 --> 00:00:22,900 So under UI automation, OCR screen scraping, we have to get OCR text activity. 5 00:00:23,220 --> 00:00:27,690 So this is what you want to use when you want to scrape a certain piece of text from a PDF. 6 00:00:28,410 --> 00:00:33,390 If you want to scrape all the text from the entire PDF, you'd have to use something else. 7 00:00:33,750 --> 00:00:35,580 So you'd have to install this package. 8 00:00:35,580 --> 00:00:41,220 So you would have to come to manage packages over here and you can come to all packages and you can 9 00:00:41,220 --> 00:00:44,640 search for PDF over here. 10 00:00:44,640 --> 00:00:49,200 You'll see a UI poster PDF DOT activities and you can install this. 11 00:00:50,190 --> 00:00:53,910 Okay, then click save to make sure that the installation completes. 12 00:00:54,660 --> 00:00:56,280 It'll download the dependencies. 13 00:00:58,430 --> 00:00:58,790 Great. 14 00:00:58,820 --> 00:01:04,730 So now the PDF activities should have been added to your activities panel on the left, you see, if 15 00:01:04,730 --> 00:01:12,360 I search for PDF, you'll see here there is now a read PDF text and a read PDF with OCR. 16 00:01:12,620 --> 00:01:14,100 So this is the one we're looking for. 17 00:01:14,330 --> 00:01:20,300 What this does, it read all the characters from a specified PDF and it'll output everything that it 18 00:01:20,300 --> 00:01:21,320 reads as text. 19 00:01:21,620 --> 00:01:26,300 So let's try and run it here just so I can show you how it works so we can drag this in. 20 00:01:26,960 --> 00:01:33,740 And you can see here it asks you for a file name and then it asks for an OCR engine. 21 00:01:34,130 --> 00:01:37,940 So the filename would simply be the PDF. 22 00:01:37,940 --> 00:01:40,640 So you can just type in PDF. 23 00:01:41,870 --> 00:01:42,410 All right. 24 00:01:42,710 --> 00:01:47,140 You can see it's probably requires a string over here. 25 00:01:47,150 --> 00:01:47,990 Let's just come here. 26 00:01:48,020 --> 00:01:49,520 Yes, it requires a string. 27 00:01:49,940 --> 00:01:52,330 So you can say Dotto String can see there. 28 00:01:52,340 --> 00:01:53,030 It's a string. 29 00:01:53,040 --> 00:01:54,020 That's how I knew that. 30 00:01:54,020 --> 00:01:56,010 Click OK, click away. 31 00:01:56,780 --> 00:02:00,820 So there's still an area that's probably because we haven't added our OCR engine yet. 32 00:02:00,830 --> 00:02:02,390 Let's just see what it says. 33 00:02:02,690 --> 00:02:06,050 Yes, it says no OCR engine assigned so you can close this. 34 00:02:06,050 --> 00:02:14,830 We can type in OCR and there should be a section with OCR engines here and a UI automation OCR engine. 35 00:02:15,320 --> 00:02:21,050 These are the different OCR engines which you could use, which will actually do the converting from 36 00:02:21,050 --> 00:02:22,070 image into text. 37 00:02:22,310 --> 00:02:23,870 So here they are, four listed ones. 38 00:02:23,870 --> 00:02:28,340 There's Ebbe, there's Google, there's Microsoft and there's Tesseract. 39 00:02:28,520 --> 00:02:29,360 What tends to work? 40 00:02:29,360 --> 00:02:35,930 Well, when you use it, when you're scraping components of a PDF would be the Google OCR. 41 00:02:36,140 --> 00:02:42,530 But when you're scraping an entire PDF, usually from my experience, I find the Microsoft OCR works 42 00:02:42,530 --> 00:02:42,910 better. 43 00:02:43,520 --> 00:02:43,910 All right. 44 00:02:43,910 --> 00:02:50,450 So you can experiment with all the different ones depending on your PDF, because sometimes one might 45 00:02:50,450 --> 00:02:51,680 work better over another. 46 00:02:52,580 --> 00:02:56,360 So we can use Microsoft OCR over here and we'll drag that into. 47 00:02:56,690 --> 00:03:01,460 What we can then do is you can click on our read PDF with OCR and over here you can see there's a text 48 00:03:01,460 --> 00:03:05,120 output so we can create a variable which will then store that text. 49 00:03:05,450 --> 00:03:07,250 So I'll just say control. 50 00:03:07,250 --> 00:03:09,530 Okay, to create a variable and I'll call it 51 00:03:12,050 --> 00:03:13,010 OCR. 52 00:03:14,340 --> 00:03:14,970 Output's. 53 00:03:18,860 --> 00:03:23,700 Just like that, OK, we're not going to be keeping this, this is just for demonstration purposes of 54 00:03:23,720 --> 00:03:25,700 they use a right line activity. 55 00:03:27,030 --> 00:03:29,050 To write everything that read the output panel. 56 00:03:29,400 --> 00:03:34,730 So over here, we want to write our OCR output and that is all. 57 00:03:35,310 --> 00:03:37,800 So I'm just going to clear our output panel. 58 00:03:37,830 --> 00:03:41,780 I'll make sure there's no PDF open and we can run this and see what happens. 59 00:03:44,980 --> 00:03:49,720 So now it's opened all three and it should have scraped all the data and here you can see. 60 00:03:50,790 --> 00:03:51,450 There you go. 61 00:03:51,870 --> 00:03:56,970 You'll notice that it spits out the titles first, such as product name, trademark middle name, that 62 00:03:56,970 --> 00:04:00,840 is all of these here on the left hand side, product name, trademark middle name, et cetera. 63 00:04:01,170 --> 00:04:07,950 And on the right hand side, it's, for example, GRV Sports is listed just below that. 64 00:04:07,950 --> 00:04:12,570 It should be there, GRV, Sports, etc. and then it has all the values just below that. 65 00:04:12,960 --> 00:04:19,170 So it doesn't spit it out in a nice formatted order, unfortunately, but it does scrape all the data 66 00:04:19,170 --> 00:04:20,100 from the PDF. 67 00:04:20,610 --> 00:04:26,460 An application I could think of for this would be something like where you want to search all the text 68 00:04:26,460 --> 00:04:31,680 inside a scanned PDF document and you want to search for a certain term. 69 00:04:32,040 --> 00:04:39,290 If that term exists, then you can then, for example, save that PDF into a certain folder. 70 00:04:39,600 --> 00:04:46,060 And if that term doesn't exist, then it saves that PDF into a different folder or something like that. 71 00:04:46,260 --> 00:04:51,990 So the way you could go about doing that is you could use something like a if statement she could use. 72 00:04:51,990 --> 00:04:58,830 And if you can search for the text, which it outputs it, in our case, it was OCR output. 73 00:04:59,160 --> 00:05:09,430 So you can select OCR output, you can say DOT contains and then you can search for a certain term. 74 00:05:09,690 --> 00:05:19,020 For example, let's say you're looking for all the invoices which have the product GRV eSport in it. 75 00:05:19,410 --> 00:05:28,350 Then what you do is you add simply right here, g r v dash e dash sports. 76 00:05:29,040 --> 00:05:34,950 Then from there you can assign certain sequences to do whatever you want. 77 00:05:35,490 --> 00:05:40,180 If it does contain that, do this, if it doesn't contain that, do that, etc.. 78 00:05:40,710 --> 00:05:41,130 All right. 79 00:05:41,140 --> 00:05:47,010 So that is just a quick summary of how to use the read PDF with OCR activity. 80 00:05:47,460 --> 00:05:54,390 In the next video, I'll be showing you how to scrape a certain elements from a PDF using OCR.