Pypdf extracts code from one PDF, but not from another?

Question

I am trying to make a primitive crawler for my own pdf files. For that, I use Pypdf to extract the Data (Customer, Product, Amount, etc.) and use that data.

Now, I have the code, its pretty easy, but it doesn't seem to be able to extract anything out of my PDFs while I tried it on some random PDF from google and it works. I tried with multiple of my documents, pdfs, don't work, random pdf off the internet works.

I use Spyder.

Below is the code I am using:

import PyPDF2 as p2

PDFfile=open("pdf_barrierefrei.pdf","rb") # Random PFD off the Internet
pdfread = p2.PdfFileReader(PDFfile)

x = pdfread.getPage(0)
print(x.extractText())
PDFfile.close

PDFfile=open("2.pdf","rb")
pdfread = p2.PdfFileReader(PDFfile) # My PDF

y = pdfread.getPage(0)
print(y.extractText())
PDFfile.close

My Expected output is a string of mashed-up data that is part of my PDF, like customer, My name, date, etc which I then sort and clear out.

The actual result is either:

" [b'\n', b'endobj\n', b'11 0 obj\n', b'<<\n', b'/Length 1011\n', b'>>\n', b'stream\n', b'/CIDInit /ProcSet findresource begin 12 dict begin begincmap /CIDSystemInfo << /Registry (Adobe) /Ordering (UCS) /Supplement 0 >> def /CMapName /Adobe-Identity-UCS def /CMapType 2 def 1 begincodespacerange <0000> endcodespacerange 49 beginbfchar <0003> <0020> <0004> <0041> <0011> <0042> <0012> <0043> <001C> <0045> <0026> <0046> <0027> <0047> <002C> <0048> <002F> <0049> <003E> <004C> <0045> <004E> <004B> <004F> <0057> <0050> <005A> <0052> <005E> <0053> <0064> <0054> <0068> <0055> <0073> <0056> <0102> <0061> <010F> <0062> <0110> <0063> <011A> <0064> <011E> <0065> <0128> <0066> <0150> <0067> <015A> <0068> <015D> <0069> <016C> <006B> <0175> <006D> <0176> <006E> <017D> <006F> <018C> <0072> <0190> <0073> <019A> <0074> <01B5> <0075> <0355> <002C> <0357> <003A> <0358> <002E> <036C> <002F> <0372> <002D> <03A6> <20AC> <03EC> <0030> <03ED> <0031> <03EE> <0032> <03F0> <0034> <03F1> <0035> <03F2> <0036> <03F3> <0037> <03F5> <0039> endbfchar endcmap CMapName currentdict /CMap defineresource pop end end \n', b'endstream\n', b'endobj\n', b'10 0 obj\n', b'[ 3 3 226 4 4 605 17 17 560 18 18 529 28 28 487 38 38 458 39 39 637 44 44 630 47 47 266 62 62 422 69 69 658 75 75 676 87 87 532 90 90 562 94 94 472 100 100 495 104 104 652 115 115 591 258 258 493 271 271 536 272 272 418 282 282 536 286 286 503 296 296 316 336 336 474 346 346 536 349 349 245 364 364 479 373 373 813 374 374 536 381 381 537 396 396 355 400 400 398 410 410 346 437 437 536 853 853 257 855 855 275 856 856 267 876 876 429 882 882 306 934 934 506 1004 1004 506 1005 1005 506 1006 1006 506 1008 1008 506 1009 1009 506 1010 1010 506 1011 1011 506 1013 1013 506 ]\n', b'endobj\n', b'7 0 obj\n', b'[ -813 -268 813 952 ]\n', b'endobj\n', b'8 0 obj\n', b'813\n', b'endobj\n', b'19 0 obj\n', b'<<\n', b'/Length 1207\n', b'>>\n', b'stream\n', b'/CIDInit /ProcSet findresource begin 12 dict begin begincmap /CIDSystemInfo << /Registry (Adobe) /Ordering (UCS) /Supplement 0 >> def /CMapName /Adobe-Identity-UCS def /CMapType 2 def 1 begincodespacerange <0000> endcodespacerange 63 beginbfchar <0003> <0020> <0004> <0041> <0011> <0042> <0012> <0043> <0018> <0044> <0026> <0046> <0027> <0047> <002C> <0048> <002F> <0049> <003C> <004B> <003E> <004C> <0044> <004D> <0045> <004E> <004B> <004F> <0057> <0050> <005A> <0052> <005E> <0053> <0064> <0054> <0068> <0055> <0073> <0056> <0074> <0057> <007F> <005A> <0102> <0061> <010F> <0062> <0110> <0063> <011A> <0064> <011E> <0065> <0128> <0066> <0150> <0067> <015A> <0068> <015D> <0069> <016C> <006B> <016F> <006C> <0175> <006D> <0176> <006E> <017D> <006F> <0189> <0070> <018C> <0072> <0190> <0073> <0198> <00DF> <019A> <0074> <01B5> <0075> <01C0> <0076> <01C1> <0077> <01CC> <007A> <0355> <002C> <0358> <002E> <036C> <002F> <0372> <002D> <039B> <0040> <03A6> <20AC> <03EC> <0030> <03ED> <0031> <03EE> <0032> <03EF> <0033> <03F0> <0034> <03F1> <0035> <03F2> <0036> <03F3> <0037> <03F4> <0038> <03F5> <0039> <0439> <0025> <043D> <002B> endbfchar endcmap CMapName currentdict /CMap defineresource pop end end \n', b'endstream\n', b'endobj\n', b'18 0 obj\n', b'[ 3 3 226 4 4 578 17 17 543 18 18 533 24 24 615 38 38 459 39 39 630 44 44 623 47 47 251 60 60 519 62 62 420 68 68 854 69 69 645 75 75 662 87 87 516 90 90 542 94 94 459 100 100 487 104 104 641 115 115 567 116 116 889 127 127 468 258 258 479 271 271 525 272 272 422 282 282 525 286 286 497 296 296 305 336 336 470 346 346 525 349 349 229 364 364 454 367 367 229 373 373 798 374 374 525 381 381 527 393 393 525 396 396 348 400 400 391 408 408 527 410 410 334 437 437 525 448 448 451 449 449 714 460 460 395 853 853 249 856 856 252 876 876 386 882 882 306 923 923 894 934 934 506 1004 1004 506 1005 1005 506 1006 1006 506 1007 1007 506 1008 1008 506 1009 1009 506 1010 1010 506 1011 1011 506 1012 1012 506 1013 1013 506 1081 1081 714 1085 1085 498 ]\n', b'endobj\n', b'15 0 obj\n', b'[ -894 -268 894 952 ]\n', b'endobj\n', b'16 0 obj\n', b'894\n', b'endobj\n', b'2 0 obj\n', b'<<\n', b'/Count 1\n', b'/Kids [ 3 0 R ]\n', b'/Type /Pages\n', b'>>\n', b'endobj\n', b'1 0 obj\n', b'<<\n', b'/Pages 2 0 R\n', b'/Type /Catalog\n', b'>>\n', b'endobj\n', b'23 0 obj\n', b'<<\n', b'/Author (Schicki)\n', b"/CreationDate (D:20190913102353+02'00')\n", b"/ModDate (D:20190913102353+02'00')\n", b'/Producer (Microsoft: Print To PDF)\n', b'/Title (Produktliste Schickmaier Excel.xlsx)\n', b'>>\n', b'endobj\n', b'xref\n', b'0 24\r\n', b'0000000000 65535 f\r\n', b'0000585703 00000 n\r\n', b'0000585644 00000 n\r\n', b'0000581675 00000 n\r\n', b'0000000009 00000 n\r\n', b'0000148586 00000 n\r\n', b'0000148612 00000 n\r\n', b'0000583508 00000 n\r\n', b'0000583545 00000 n\r\n', b'0000148635 00000 n\r\n', b'0000582918 00000 n\r\n', b'0000581854 00000 n\r\n', b'0000355352 00000 n\r\n', b'0000355825 00000 n\r\n', b'0000355852 00000 n\r\n', b'0000585586 00000 n\r\n', b'0000585624 00000 n\r\n', b'0000355876 00000 n\r\n', b'0000584824 00000 n\r\n', b'0000583564 00000 n\r\n', b'0000577725 00000 n\r\n', b'0000578203 00000 n\r\n', b'0000581590 00000 n\r\n', b'0000585752 00000 n\r\n', b'trailer\n', b'<<\n', b'/Info 23 0 R\n', b'/Root 1 0 R\n', b'/Size 24\n', b'>>\n', b'startxref\n', b'585949\n', b'%%EOF\n']"

or half a page of nothing, just "In [40]:" and then literally 20 sentences of nothing, no error or anything

The output got from the random Pdf:

"Aktion Mensch e.V. Fachartikel —PDF-Dokume....

16.10.03 Autor: Roland Heuwinkel 17. Oktober 2003 Seite 1 von 24 "

It is just some random scribbly, don't care about the content, just the fact that it gets me some text.

I am really new to coding, and I have no idea where to start, as both of them are PDFs. I am just wondering if it is because I used "Print to PDF" to create them?

please give us sample pdf and will take a look. Looking at your output, looks like the pdf is an image rather than readable text. — jose_bacoy, Nov 08 '19 at 14:40
how do i do that, it doesnt let me upload a wetransfer link, is there a good site to share that is supported on here? — Schicki, Nov 08 '19 at 15:01
I am using jupyther notebook and I can see text from pdf_barrierefrei.pdf. But from 2.pdf, I only see blank because it is an image table. I would convert it to text using adobe or tika — jose_bacoy, Nov 08 '19 at 16:24
hmm yes, but i would need the Program to continuously crawl the PDFs and add data if a new PDF is added, is there a way to do it in Python? Also, i will look into Print to PDF if i have something calibrated wrong there, because i make the invoice in Excel and then convert it to PDF with "Print to PDF" so that has to be the problem there.... If it would make proper PDFs then i would not have this problem — Schicki, Nov 08 '19 at 17:04
I am also using PyPDF2 in my work but to ensure that ALL pdf is readable, I used adobe pro reader. You may try it for 30days. (Note: I am not endorsing any product, this is based on my work experience only). Goodluck! — jose_bacoy, Nov 08 '19 at 17:48
I tried numerous times now to use that data, to either clean it up in the buffer or write it to txt or csv and then clean it up, but nothing works, it would already help a lot if i at least could write it to a txt and then go from there, which is not nice at all, but i am new and i have limited possibilities :/ best would be to write it to a csv in cleaned up form,, add all the other invoices and then use the data - which i am planning to do, but programming is hard xD maybe you got a tip, i will keep searching for a way. I already went and worked on the GUI, but this data issue hurts :P — Schicki, Nov 12 '19 at 14:06

score 0 · Answer 1 · edited Feb 05 '23 at 22:39

For pdf_barrierefrei.pdf, looks like the text result is converted into unicode format. By the way, I'm using jupyter notebook and not spyder, so I'm getting correct result.

For 2.pdf, the pdf has an image logo and table. Install tika (pip install tika) then use the code below to read the non-images into text.

from tika import parser
raw = parser.from_file('2.pdf')
print(raw['content'])

Result:

Produktliste Schickmaier Excel.xlsx


LIEFERSCHEIN

Kunde Customer
Adresse Adress

Adress Data
Data

K/DB-Nr. 211 Contact

Preis/N M Gesamtpreis

Bio Erdbeer-Chilischokolade 3,05 €       20 61,09 €                                                   
Bio Beuscherl 5,23 €       6 31,36 €                                                   
Bio ChiliconCarne 5,98 €       15 89,77 €                                                   
Bio Geschnetzeltes 5,23 €       15 78,41 €                                                   

Versand Brutto
Versand Netto -  €                                                       

Warenwert netto 10% 260,64 €                                                
Umsatzsteuer 10% 26,06 €                                                   

RECHNUNGSBETRAG  BRUTTO 286,70 €                                                
Seite 1/1

2019/

Data

Pypdf extracts code from one PDF, but not from another?

1 Answers1

Linked