In [1]:

    

!pip install Pillow


    

    




Requirement already satisfied: Pillow in c:\programdata\anaconda3\lib\site-packages (6.2.1)

In [2]:

    

!pip install pdf2image


    

    




Requirement already satisfied: pdf2image in c:\programdata\anaconda3\lib\site-packages (1.10.0)
Requirement already satisfied: pillow in c:\programdata\anaconda3\lib\site-packages (from pdf2image) (6.2.1)

In [3]:

    

!pip install pytesseract


    

    




Requirement already satisfied: pytesseract in c:\programdata\anaconda3\lib\site-packages (0.3.0)
Requirement already satisfied: Pillow in c:\programdata\anaconda3\lib\site-packages (from pytesseract) (6.2.1)

In [4]:

    

!pip install opencv-python


    

    




Requirement already satisfied: opencv-python in c:\programdata\anaconda3\lib\site-packages (4.1.2.30)
Requirement already satisfied: numpy>=1.14.5 in c:\programdata\anaconda3\lib\site-packages (from opencv-python) (1.17.4)

In [5]:

    

from PIL import Image 
import sys 
from pdf2image import convert_from_path 
import os


    

In [6]:

    

# Path of the pdf 
PDF_file = "Lista inscrisi Admitere Licenta sept 11.09.2015.pdf"


    

In [7]:

    

if not os.path.exists('pdf/'+PDF_file+'/'):
    os.makedirs('pdf/'+PDF_file+'/')


    

In [40]:

    

# Store all the pages of the PDF in a variable 
#a masodik szam a felbontas, ezt 300-600 kozott probalgasd
pages = convert_from_path(PDF_file, 300) 
  
# Counter to store images of each page of PDF to image 
image_counter = 1
  
# Iterate through all the pages stored above 
for page in pages: 
  
    # Declaring filename for each page of PDF as JPG 
    # For each page, filename will be: 
    # PDF page 1 -> page_1.jpg 
    # PDF page 2 -> page_2.jpg 
    # PDF page 3 -> page_3.jpg 
    # .... 
    # PDF page n -> page_n.jpg 
    filename = 'pdf/'+PDF_file+"/page_"+str(image_counter)+".jpg"
    print(image_counter,'oldal kész..')
      
    # Save the image of the page in system 
    page.save(filename, 'JPEG') 
  
    # Increment the counter to update filename 
    image_counter = image_counter + 1


    

    




1 oldal kész..
2 oldal kész..
3 oldal kész..
4 oldal kész..
5 oldal kész..

In [9]:

    

import pytesseract


    

In [47]:

    

# Variable to get count of total number of pages 
filelimit = image_counter-1
  
# Creating a text file to write the output 
outfile = 'pdf/'+PDF_file+"/text.txt"
  
# Open the file in append mode so that  
# All contents of all images are added to the same file 
f = open(outfile, "a") 
  
# Iterate from 1 to total number of pages 
for i in range(1, filelimit + 1): 
  
    # Set filename to recognize text from 
    # Again, these files will be: 
    # page_1.jpg 
    # page_2.jpg 
    # .... 
    # page_n.jpg 
    filename = 'pdf/'+PDF_file+"/page_"+str(i)+".jpg"
          
    # Recognize the text as string in image using pytesserct 
    text = str(((pytesseract.image_to_string(Image.open(filename))))) 
    print(i,'oldal kész..')
  
    # The recognized text is stored in variable text 
    # Any string processing may be applied on text 
    # Here, basic formatting has been done: 
    # In many PDFs, at line ending, if a word can't 
    # be written fully, a 'hyphen' is added. 
    # The rest of the word is written in the next line 
    # Eg: This is a sample text this word here GeeksF- 
    # orGeeks is half on first line, remaining on next. 
    # To remove this, we replace every '-\n' to ''. 
    text = text.replace('-\n', '')     
  
    # Finally, write the processed text to the file. 
    f.write(text) 

# Close the file after writing all the text. 
f.close()


    

    




1 oldal kész..
2 oldal kész..
3 oldal kész..
4 oldal kész..
5 oldal kész..

In [48]:

    

import pandas as pd


    

In [325]:

    

pages=open(outfile,'r').read()


    

In [326]:

    

lines=[i for i in pages.split('\n') if i]


    

In [327]:

    

good_lines=[line for line in lines if line[0].isdigit()]


    

In [328]:

    

good_lines=[line.replace('_',' ').replace('. ',' ').replace('-',' ').replace('—',' ')\
     .replace('~',' ').replace('=',' ').replace('  ',' ').replace('  ',' ')\
     .replace('»',' ')for line in good_lines]


    

In [329]:

    

good_lines[:10]


    

    
Out[329]:





['1 4696 ACHIM O.A ANDREEA MAGDALENA 9,56 9,50 9,70 19 18 16 14 13 11 15 20 12',
 '3 4441 ALBERT J CSILLA EDIT 6,76 5,70 7,80 48 ,',
 '2 208 ALBERT Vv IULAN 6,10 5,20 600 51.',
 '4 4500 ALEXAN A MARIOARA 7,82 7,00 855 41',
 '5 4565 AMBRUS T TIBOR 8,36 9,10 940 46.',
 '6 4424 ANDREI C IOAN CLAUDIU 7,66 7,00 6,80 19.',
 '7 210 ASANOV A REFAT 7,20 7,20 7,20 36 37',
 '8 4913 BACIU I IOAN ALEXANDRU 7,11 6,60 7,55 59 61',
 '9 4509 BALANESCU D.M RAZVAN SAUL 7,75 9,85 545 12 13',
 '10 4962 BALAS I MARIA ELENA 7,73 7,85 7,60 70.']

In [330]:

    

for i,l in enumerate(good_lines):
    print(str(i)+':::'+l)


    

    




0:::1 4696 ACHIM O.A ANDREEA MAGDALENA 9,56 9,50 9,70 19 18 16 14 13 11 15 20 12
1:::3 4441 ALBERT J CSILLA EDIT 6,76 5,70 7,80 48 ,
2:::2 208 ALBERT Vv IULAN 6,10 5,20 600 51.
3:::4 4500 ALEXAN A MARIOARA 7,82 7,00 855 41
4:::5 4565 AMBRUS T TIBOR 8,36 9,10 940 46.
5:::6 4424 ANDREI C IOAN CLAUDIU 7,66 7,00 6,80 19.
6:::7 210 ASANOV A REFAT 7,20 7,20 7,20 36 37
7:::8 4913 BACIU I IOAN ALEXANDRU 7,11 6,60 7,55 59 61
8:::9 4509 BALANESCU D.M RAZVAN SAUL 7,75 9,85 545 12 13
9:::10 4962 BALAS I MARIA ELENA 7,73 7,85 7,60 70.
10:::11 44114 BALINT c ROBERT ISTVAN 7,93 6,80 9,00 36 37,16 18
11:::12 4571 BALOG | MARIAN BOGDAN 6,80 5,25 6,10 46.
12:::13 4526 BANDICI .M RAZVAN ADRIAN 8,93 5,80 815 46.
13:::15 4902 BARABAS L SZENDE 7,01 7,05 7,60 64 59 67
14:::14 4572 BARABAS O BENIAMIN OVIDIU 8,10 650 8,60 11 16 14 15
15:::16 4508 BARATI Vv NINEL EMIL 6,95 6,30 7,50 47.46 41
16:::17 4695 BARDAN Vv SIMINA 9,30 8,70 9,20 37 18 27 36 16 19
17:::18 4669 BARSAN S MARIA ANCA 7,65 6,30 8,30 41 44 46
18:::19 4651 BEJINARIU Vv GIANINA ANAMARIA 6,18 5,20 5,50 11 16 04 14 13 15
19:::20 4462 BENCZE L NOEMI 9,18 9,25 8,90 47 46
20:::21 4570 BERCHI t CRISTINA MARIA 7,53 6,10 7,60 46.
21:::22 222 BERKI J ANDREEA ALEXANDRA 7,40 .7,90 9,05 53
22:::23 4602 BEUCA D VASILINA 8,16 7,00 7,50 16 13 12
23:::24 4611 BIDIAN I CARMEN ANDREEA 8,23 8,60 850 41
24:::25 4961 BILICI Vv CRISTIAN 7,38 5,50 825 70.
25:::26 213 BIZGU A DIANA 8,25 7,00 9,00 36.
26:::27 4557 BLAI D.| | DAN FLORIN 6,66 5,45 5,65 46 41 47 43
27:::28 4676 BLAJAN G ANDREEA ROXANA 7,80 6,20 895 O07 16 11 18 19 04 14 15
28:::29 4671 BOEREAN PL ALISA PAULA 6,00 6,80 5,10 12.15 13 18 16 20 19 14 11
29:::30 4909 BOGDAN ROBERT "6,20 5,35 8,30 67.
30:::31 4490 BOICIUC | VASILE IONEL 6,13 6,70 620 18 19 16
31:::32 4918 BOKOR F A ATTILA LASZLO 8,80 0,00 0,00 67 
32:::33 4635 BORODI | SIMONA LENUTA 8,63 8,80 8,20 41
33:::34 4657 BORS A ALPAR 7,13 6,10 6,50 33 32 30 31 19 18 16 15
34:::35 4655 BORZA A ALEXANDRU FLORIAN 6,38 5,30 655 37 27 36 18 16
35:::36 4656 BRATU I LAURIAN IONEL 7,66 7,00 685 27 36
36:::37 4678 BRUDAN D TEODORA MADALINA 6,50 6,30 6,00 47.41 44 46 42 45 43
37:::38 4977 BUCATEA s MELISA STELIANA 6,20 0,00 0,00 69 
38:::39 4499 BUDULAU 1, LOREDANA IOANA 7,80 7,50 8,00 41
39:::40 4521 BULZ 1 IULIA ELENA 8,80 7,80 960 16 11 07
40:::41 4665 BUMBU 7 IONUT ADRIAN 7,16 6,70 7,00 16 11 18 12 19
41:::42 4603 BURA Vv LAVINIA 8,75 9,70 860 11
42:::43 4639 BUTUC GH ANE MARY 7,56 8,32 5,75 18 46 12 19
43:::44 4654 CAIENAR A DAN SEBASTIAN 8,60 8,75 860 04 16 18 19 11
44:::45 218 CALARAS S VASILE 6,50 5,00 7,00 15 18 12
45:::47 206 CALIN M MITICA 9,76 9,76 9,76 36.
46:::48 4561 CAPUSAN Vv MARIUS 6,00 5,70 635 46 47
47:::49 4551 CATINAS Ih RADU OTNIEL 8,40 9,20 615 45 46
48:::50 4489 CAZAC N SILVIU CONSTANTIN 6,03 600 5,20 18 19 15.
49:::51 4685 CAZACU GH CONSTANTIN ADRIAN 7,06 6,70 6,70 12 20
50:::46 4434 CALBOREAN P.S ADRIAN SERGIU 9,15 8,70 8,50 46.
51:::52 4403 CEBUC E.l ANA MARIA 8,06 8,25 940 34 36 16 12
52:::53 4427 CENAN S RADU CRISTIAN 8,25 9,60 6,70 O7 11 16 20 18
53:::54 4621 CHIHAIA ( IULIANA 7,95 6,90 9,00 16.
54:::55 4431 CHIOREAN Cc ELISA LAVINIA 648 6,95 6,40 19 18 16 11 14
55:::56 4694 CHIUHAN M DAIANA CRISTINA 6,83 7,00 5,40 16 44 19 47,Nr Medie Gen NotaPr Nota
56:::57 4566 CHMARA I VICTOR ALEXANDRU 8,41 8,15 850 04 14 16 19 11 15
57:::58 4554 CIDOIU ANDREEA H ANDREEA 9,19 8,50 9,20 41 46 43.
58:::59 4501 CIOANCA F HILDA ALETTA 7,63 5,70 9,00 44 46
59:::60 4649 CIOC DZ DANIEL ALEXANDRU 686 8,05 6,10 18 19 12 16 11 14 46 47 42 44 41
60:::61 4422 CIOCULESCU R EDUARD MIHAI 6,76 5,85 7,15 18 19 16 11°20
61:::62 4573 CIOROBITCA C CONSTANTIN ANDREI 6,55 5,70 6,50 37 36 35
62:::63 4659 CIOT P DRAGOS BOGDAN 6,63 7,70 5,00 19 18
63:::64 4698 CIPCIGAN I CRISTINA OANA 10,00 9,45 8,75 07 
64:::66 4479 CIUCA S PAULA ANDREEA 6,96 6,80 6,25 45 47 46 42 41 43
65:::65 4436 CIRLAN G DIANA ANDREIA 7,76 7,45 880 15 18 13 12 19
66:::67 4433 CLAPA SG AUN 7,26 5,20 845 37 36 19 12 18 16
67:::68 4684 CONSTANTIN V ELENA IULIA 8,10 8,00 7,50 41 44
68:::69 4429 CORCHIS N RADU NICOLAE 8,43 10,00 6,30 46.
69:::70 4476 CORDIS N NICOLETA MARIA , 8,92 7,30 955 41
70:::71 4511 COSOVAN G.N INGRID 7,70 6,50 7,45 46.
71:::72 4689 COSTE I RAUL ADRIAN 9,38 9,05 9,10 13 
72:::73 4443 COSTIN V ADRIAN TIBERIU 9,43 9,15 955 36 37,
73:::74 4494 COTROAZA Cc AUREL COSMIN 6,58 5,20 7,55 13 
74:::75 4976 COVACI G MARIA MONICA 9,27 9,60 7,30 69.
75:::76 4912 CRACIUN P DANIEL LEONARD 7,78 5,35 6,50 59 61 62.
76:::77 4495 CRISAN DM ADINA MIHAELA 9,81 9,20 990 46 41
77:::78 4646 CRISAN | ANCA MARIA 8,13 8,40 830 11 16 15 04 14 18 19 12 13 20
78:::79 4667 CURAC Vv ILEANA 6,43 7,00 5,80 46 41 47 44
79:::80 4419 CURTEA M EMANUELA IULIA 9,41 9,50 9,00 44 41
80:::81 4505 DAN D LORENA ANTONIA 7,35 6,40 7,00 46 41 47
81:::82 4412 DANCI M | ROXANA CLAUDIA 7,43 8,05 6,05 16 15 18 19 12 20 11 14
82:::83 4683 DANCI V.l IOANA TIMEEA 9,44 9,45 8,60 43 45 46 42 47 41
83:::84 4525 DARABONT I NATALIA 7,28 6,95 815 18 19 16
84:::85 4524 DASCALESCU P | ANCA NICOLETA 7,20 8,10 5,40 11, 16 15 18 19
85:::86 4408 DEAC N DENISA NICOLETA 8,11 8,00 7,150 44.
86:::87 4640 DEMIAN A.M ALEXANDRU IOSIF 7,05 6,00 6,15 14.16 15 12 13 18 14 19 20
87:::88 4901 DEZSI Zz HENRIETTA 6,57 7,10 8,05 65 64 66 60 62.
88:::89 4423 DINCA | DAYANA 9,20 9,70 9,70 46 
89:::90 4666 DOHOTAR G GHEORGHE 6,38 5,00 6,85 16 11,18 15 12 19 41 44
90:::91 201 DONOS P CONSTANTIN 6,50 5,00 7,00 16 141,
91:::92 4914 DRAGU Cc DORIN NICOLAE 8,14 10,00 615 59 62
92:::93 4439 DRON T ALINA 7,43 6,10 8,50 44 42
93:::94 221 DUCA M TUDOR 7,50 7,00 8,00 15 16
94:::95 4904 DUMITRU s FLORIN STELIAN 7,04 7,60 6,80 59 61 62,
95:::96 44480 ENACHE E.V IOANA MAGDALENA 9,10 8,35 925 46 43 42
96:::97 4461 ENACHE MM LILICA 7,40 6,50 7,50 46 47 41
97:::98 4630 ERINCZ A ROLAND 8,21 6,50 7,00 46 
98:::99 4507 FALAUS LV CALIN VASILE 6,50 6,65 6,20 42 46 41 43 47 44
99:::100 4675 FARAGAU A PAUL BOGDAN 7,01 6,20 5,75 47 44
100:::101 4482 FARCAS I SINZEANA MELISA 7,18 7,65 7,80 16 11
101:::102 4498 FILIP L. L MARIA MAGDALENA 7,78 8,15 6,95 43 15
102:::103 4601 FLOREA D.C MARCELA SILVIA 7,80 8,10 7,40 .12 18 19 15 13 16
103:::104 4618 FLORESCU Hl | MIHAELA HRISA 9,15 9,25 8,70 46 
104:::105 4606 FOGAS E. H SERGIU IONEL 9,93 8,70 915 11
105:::106 4688 GALL Vs LEVENTE 7,23 5,65 810 37, 27 36 35 32 30 31
106:::107 4454 GAZDAC c CATALIN 9,10 0,00 9,10 19 18 11
107:::108 4563 GERGELY E IMRE SZABOLCS 8,06 8,25 835 30 31 32 33
108:::109 4664 GILOAN T ANGELO VLAD 9,04 9,35 7,05 15 
109:::110 4686 GOLLER A MIRCEA ADRIAN 8,65 8,40 855 34 15 12
110:::112 4632 GORON I, ANTON IOAN 9,32 8,50 880 41 46 47
111:::113 4497 GREABU N.D OANA MARIA 8,86 8,60 935 141 16 15 18 19
112:::114 4612 GREC Vv RAMONA ROXANA 9,01 9,90 820 46 45 43 41 44 42
113:::115 4472 GRUAC | NARCISA MARIA 9,30 8,85 9,05 0O7 16
114:::116 4496 GROZAV R ALINA 9,24 8,80 915 45.
115:::117 202 GURDUZ DUMITRU 6,60 6,75 7,05 20 16
116:::118 4631 GUSAVAN D DUMITRU BOGDAN 8,80 9,25 8,60 45 46 41
117:::119 4444 GYURKA zZ HUNOR 7,83 5,45 7,30 48 46 47 45
118:::120 4426 HABA V. AL DIANA 7,76 6,00 880 12.16 20 19 11 18 15 14 13
119:::121 4445 HAGAU N OANA ANDREEA 6,98 6,25 6,20 .45 41
120:::122 217 HAMPEA N RARES MARIAN 6,43 5,70 5,00 56.
121:::123 4663 HARAP D VERONICA ' 7,74 5,15 640 41.
122:::124 4401 HAZOTA A ALEXANDRU CRISTIAN 6,93 9,15 6,35 18 46
123:::125 4969 HAZOTA A ALEXANDRU CRISTIAN 6,93 9,15 6,35 69.
124:::126 4405 HEGYI I LAURA 7,52 8,05 7,00 32 33 30 48
125:::127 4447 HUIDES LE LAURA SIMONA 6,43 7,00 510 11 16 18 19
126:::128 220 HULPE V VASILE DARIUS 5,12 5,12 5,12 18 11 13 15,ve Nr Leg Nume iF Prenume Menle Gen voce e ih nna, Optiuni
127:::129 4620 HUZMEZAN I IONUT 6,36 6,40 5,70 44 41 42
128:::130 4523 IANUS E MADALIN CONSTANTIN 7,01 7,30 5,20 46 
129:::131 4474 IAZ F ALEXANDRA 6,82 5,00 7,55 48,
130:::132 4407 |IAZU H.G ADRIANA CRISTINA 9,33 9,20 9,30 38, 19 18 15 12
131:::133 4430 ILIES LF ALINA 931 9,35 9,00 1 16 18 19 14 12 15 13 20
132:::134 4415 ILLES V NATALIA BIANCA 645 5,60 6,30 27.36 37 35 15 19 18
133:::135 215 IONAS I ROMAN 8,00 8,00 7,00 16 36
134:::136 4515 100 A ANDREI RARES 868 8,00 870 19 18 47 46
135:::137 4682 IUGA S MARIA NICOLETA 805 825 6,90 07.11 16 04 14
136:::138 4648 JURGIU T TITUS TIBERIU TREBONIU CORIC 7,91 9,35 7,10 18 19 15 20 11 16 14
137:::139 4905 KATO A FERENC 6,00 6,30 6,90 64 66
138:::140 4910 KOPACZ B LEVENTE BENEDEK 8,00 8,50 8,00 67 
139:::141 4627 KORAZSAN T TIBERIA TEODORA 8,47 8,40 7,00 34.
140:::142 4558 KOVACS S SZABOLCS ISTVAN 6,386 7,80 600 46 41 47 44
141:::143 4911 LAKATOS L LASZLO 9,02 9,05 8,25 59 61, 62,
142:::144 219 LAPTIACRU N NICOLAE 6,50 6,00 6,00 16.15 14
143:::145 4450 LAZAR I LAVINIA DANIELA 6,56 6,50 6,60 16 19 18 14 21 12 13 15
144:::146 4564 LAZAR Vv ANDREEA DANIELA 6,73 7,70 5,15 16.11 19 18
145:::147 4413 LAZAR M CELIA 8,13 8,00 8,40 11 
146:::148 4456 LIBER P MIHAELA 7 AL 6,65 7,00 11 16 15 13 12 14 18 19
147:::149 205 LISTIUGHIN ‘GH TATIANA 7,50 7,50 7,50 35 11
148:::150 4504 LORINCZ RO SZABOLCS CSABA 7,22 5,75 6,95 45 44 46
149:::151 4463 LUBANSKI V DENISPIOTR 7,83 830 7,90 15.12 0414
150:::452 4438 LUNGU G OVIDIU 780 7,45 7,15 14 15 12 16
151:::153 4506 MAGDA Vv EMANUEL 8,64 7,55 6,10 46 18
152:::154 4492 MAGHIAR Vv ADRIAN 7,34 5,20 7,00 42 
153:::155 209 MALNASI AL IZABELLA ERZSEBET 7,20 7,20 7,20 19,
154:::157 4556 MARIAN A. L ADELA 7,01 5,40 7,30 44 41
155:::156 4487 MARIAN 1A CATALIN 6,13 5,00 7,65 46 41
156:::158 4418 MARIN 1 DAVID ROBERT 6,08 5,75 5,50 47.
157:::159 4626 MARINESCU E. M SABINA GABRIELA 6,78 5,00 7,45 19 18 12,
158:::162 4464 MAXIAN A ROBERT ALEXANDRU 7,26 835 5,20 18 16 19 37 3615
159:::163 4518 MAXIM F.A FLAVIUS 7,36 510 7,80 16 11 13
160:::160 4510 MARMUREANU N ALEXANDRA CRISTINA 6,23 7,60 5,50 41.
161:::164 4613 MELEN P ANDREI SEBASTIAN 651 600 6,00 07 19 1112 13 14 15 20 18
162:::166 4973 MESAROS I IOAN 6,23 5,00 6,00 69 
163:::165 4660 MESAROS Vv ADRIAN MESAROS 6,50 5,00 8,70 45 46 44
164:::167 4644 MEZEI A KRISTINA ANAMARIA 7,98 8,10 7,90 16 44 15 43 11 41
165:::168 4406 MICLEAN CIAPA Cl BIANCA 6,46 5,95 6,00 46 47 43 42 44 41 45
166:::169 4624 MICULAICIUC S SONIA 711 600 660 18 11 07 16 15 12
167:::170 4966 MIHALI P ALINA 8,22 8,00 7,30 70 
168:::171 4513 MIKOLA VA GRETA MARTA 817 600 9,50 31 33 32 16 13 11
169:::172 4691 MITRANICA | GABRIEL 861 7,80 835 46 47.
170:::173 4559 MODOG M OCTAVIANA RALUCA 8,90 8,30 8,90 45 46
171:::174 4619 MOLDOVAN Vv BIANCA DENISA 820 945 7,15 16 11 15
172:::175 4617 MORA T FLAVIA ANDREEA 845 980 835 04 07 11 15 16 13 19 18 12 20
173:::176 4552 MORARIU Vv OANA MARIA 7,68 8,80 8,75 43 47 44
174:::177 207 MUNTEAN Ss ION 7,00 6,00 8,00 16, 15 19
175:::179 4641 MURESAN G GEORGIANA COSMINA 8,98 9,75 8,35 44 
176:::180 4679 MURESAN | ANAMARIA 610 5,10 630 411 19 18 12 13 15,
177:::178 4465 MURESAN M ANDRADA 7,11 7,50 6,00 41 
178:::181 4457 MUSAT S$ ROBERT MIHAI 685 7,85 5,10 11 16 12
179:::182 4970 MUSTATA Cc ALEXANDRA IOANA 6,18 5,00 7,75 69.
180:::183 4628 NEAMT A ANA MARIA 8,06 7,80 8,20 41 46
181:::185 4636 NEAMTU Vv RAZVAN IOAN 640 7,90 5,90 15.1219 1811
182:::184 4469 NEAMTU Vv VASILE MIHAI 6,73   5,00 6,70 41 46
183:::186 4967 NEGREA P PAUL VASILE 9,15 9,05 860 70.
184:::187 4574 NEGREA Vv MARIA SONIA 8,23 7,90 9,00 44 
185:::188 4637 NISTOR M IULIAN MARIUS 7,66 7,20 8,00 20 16 11 15 07 18 19 12 13
186:::189 4622 OANRA T CRINA ELENA 8,30 7,25 7,85 47 46 41 44 43
187:::190 4485 OARA I ADRIAN CATALIN 6,36 6,50 7,30 45 11 41 18 46 19 47
188:::191 4687 OLTEAN .F tULIU MATEI 9,28 8,90 8,40 45"
189:::192 4916 OLTEAN Vv ANA MARIA 7,06 5,00 7,85 59 61 62.
190:::193 4634 ONTI M CRINA CAROLINA 848 645 9,25 O07 04 16 1118
191:::194 4915 OPREA E CRISTIAN 6,53 5,10 6,00 59 62.
192:::195 212 PALAMARCIUC 0 CRISTINA 7,25 6,00 7,00 12.19 18
193:::196 4442 PANES A ELENA 7,28 6,40 7,45 19 18 11,
194:::197 4700 PANESCU Vv ADINA 8,41 8,00 9,65 04 12 14 16 15 ‘
195:::198 4981 PASARE DAN 6,13 500 7,90 70 199 4514 PASCA G FLORINA DENISA MIHAELA 8,28 8,85 8,10 43 i
196:::200 4625 PAUL Vv CLAUDIA MARIOARA 6,73 5,95 6,10 16 19 11, S$ : "ED .
197:::4 Bp Sp
198:::201 4421 PAUSAN V ALEXANDRU VICTOR : 6,03 5,80 6,55 20 16 19 12 15 18 14 11 13
199:::202 4608 PERDE B ANDREEA DENISA 7,23 8,30 595 11 19 04 15 16 18 12 20
200:::203 4491 PETA O.A LORENA 7,28 6,75 8,50 37 18 19
201:::204 4460 PETEAN P PETRINA 7,14 7,85 5,00 18
202:::205 4975 PETREUS V VASILE JUSTIN 6,71 7,00 5,00 69 
203:::206 4493 PETRISOR D CRISTINA 7,96 6,60 8,30 18.
204:::207 4553 PETROESC 1S ALEXANDRA DANIELA 6,90 5,00 7,20 43 46 42 47.
205:::208 214 PIVEN] E EVGHENII 5,00 5,00 5,00 19.
206:::209 4633 PLESCAN A PAUL ANDREI 6,80 6,00 6,00 11, 41 19 18
207:::211 4673 PODAR N STEFAN 6,83 9,10 5,05 46 47 41 42
208:::212 4409 PODEA T IONELA ANDRADA 6,56 7,00 7,70 13 18 19 12
209:::213 211 POGRIBNIC M ELENA 7,75 8,00 7,00 16 11 15 19
210:::219 4680 POP A DAN 7,68 6,30 685 11 15 16 12
211:::217 4642 POP D RAMONA 9,38 .10,00 835 41 44
212:::214 4428 POP El EDUARD , 6,33 5,85 8,00 35 37 36
213:::220 4690 POP G DIANARALUCA 9,12 9,15 6,80 18 15 14 12
214:::215 4484 POP P,V IONELA MARIA 6,83 6,40 7,50 11 16 18 19 41 44 46 47
215:::218 4647 POP Vv LORENA LUMINITA 7,91 7,30 9,10 41, 44
216:::216 4610 POP V.R REBEKA ANNA 7,28 8,20 7,60 33 32, 30 31
217:::221 4615 PORFIRE N MADALINA NICOLETA 8,36 8,40 7,00 16 11 19 12 18
218:::222 4402 POSTELNICU S BOGDAN 8,21 10,00 7,30 07 35, 11, 15 27 36.
219:::223 4483 PREDA Ss MIHAI 9,36 9,60 850 07 16 15 18 19 11
220:::224 4417 PRESCURA N ADELA 7,78 8,40 7,95 41.
221:::225 4435 PRUTEANU A ANTONIU ALEXANDRU 6,20 5,00 6,20 15 14 19 18 13 12 11 16 20
222:::226 4502 PUICAR T DENISA MARIANA 7,15 6,00 7,55 16 11 18 19 12 44 41 46 47 42
223:::227 4478 RADU | ZAHARIE OVIDIU 8,36 9,00 7,37 44 47 46
224:::228 4906 RAPO Vv ATTILA 8,22 7,90 6,85 59 61 62.
225:::229 4414 RATIU Vv ALEXANDRU NICOLAE 6,66 5,40 6,45 18 19
226:::230 4437 REDENSTEIN Cc ELENA OANA 8,86 9,90 7,45 15 16 19 18
227:::231 4971 REDNIC P IULIANA MARIA 6,56 5,00 6,50 70 
228:::232 4681 ROBERT M ROBERT , 8,00 8,65 7,745 48 
229:::233 4662 ROMAN DACIAN LUCIAN 9,44 8,85 8,00 45 41
230:::239 4965 RUS CV DANIEL 6,91 5,90 7,40 70.
231:::235 4470 RUS | ANA MARIA 6,11 7,55 5,40 41 46 43 44
232:::234 4452 RUS I IOANA NATALIA 9,53 10,00 8,70 16 18 19 12 15 20 13
233:::238 4604 RUS M ANDREI 6,43 5,40 7,90 18 16 11 19 12
234:::237 4568 RUS 0 DARIUS OVIDIU 6,15 5,70 5,80 19 18 15 12 13 20 16 11
235:::236 4481 RUS “R OVIDIU ANDREI 7,45 7,40 7,85 46 43 42
236:::240 4569 RUSEN L ALEXANDRA ELENA 8,33 7,05 9,55 19 18 12
237:::241 4522 RUSU D ALEXANDRA COSMINA 9,17 8,70 9,40 43 44 42
238:::242 4432 SACOTA N NICOLETA 6,31 5,00 645 11 12 15 19 18 16 13 14 20 41 42 43 47 46 44
239:::243 4609 SATMAR Vv OVIDIU LUCIAN 6,18 5,60 5,30 44 43
240:::244 4512 SAVU S.P. CARMEN VOICHITA 9,90 9,85 955 04 14
241:::245 4980 SCOPET T FEDOR 6,66 6,00 7,00 70 
242:::246 4446 SCRIPCA T TATIANA FLORENTINA 6,60 6,00 6,00 41 44 42 43
243:::247 4629 SCROBOTA I EMANUELA MADALINA 7,80 7,40 845 41 47 44 46
244:::248 4623 SCURTU M.G RAZVAN ANDREI 7,48 7,40 7,20 19 12, 13,
245:::249 4692 SESERMAN R ADELA 9,34 9,15 10,00 41 
246:::251 4661 SILAGHI D PATRIK ADRIAN 6,23 5,00 7,50 11, 16 19 18
247:::250 4614 SILAGHI 1 ADRIAN FLAVIU IONUT 7,05 7,10 680 07 20 35 11 12 37 18 19 36 16
248:::252 4477 SIMON A ROBERT 6,18 6,85 5,80 32 33 30 18 19 06
249:::254 4616 SIMON G ALEXANDRU IOAN 6,50 6,20 5,55 14.16 14 19 18 12 13 2015
250:::253 4607 SIMON M MIHAELA FLORENTINA 8,36 8,05 9,30 16 19 18 15
251:::255 4488 SIPOS I ADELA LAURA 8,06 7,40 850 18 16 19 12 14 11 15
252:::256 4519 SMADICI M ALEXANDRA 6,50 5,00 6,20 11 16 18
253:::258 4520 SOPON I, ELENA ALEXANDRA 9,75 9,65 9,50 41 44.
254:::259 4468 SORA M ALEXANDRU CATALIN 7,75 8,65 5,65 18, 19 12
255:::260 216 SOS LS ROXANA STELIANA 6,40 6,40 6,40 11, 16 20
256:::261 4674 STANCIU D | DAN ANDREI 6,38 6,90 6,90 16 19 18
257:::262 4453 STETCO G.G, {!USTIN RENE IONUT 6,46 5,00 6,80 43 46 42 41,
258:::263 4404 STICLARU M MARCELA ALEXANDRA 7,30 5,30 7,80 16.
259:::265 4672 STOIAN Vv GHEORGHE LADISLAU 7,68 8,30 6,50 44 41 46 47.
260:::266 4555 STRUGAR G ANCA IOANA 8,55 9,30 7,25 44 47, 46
261:::267 4645 SZALASI A EDWARD DACIAN 6,15 5,10 6,30 19 16 11 18 15 12 13 14 20
262:::270 4670 SZEKELY G IULIA MARIA 8,33 8,30 7,60 12, 19 18
263:::269 4475 SZEKELY G EVA ROZALIA 6,86 5,45 8,70 41 48 44
264:::268 4471 SZEKELY K TIMEA ERZSEBET 8,12 6,65 9,00 12 18 11 19
265:::271 4517 SZEP A ATTILA 7,28 8,05 6,45 33 32:30 15 14
266:::272 4416 SZOCS AS PETER ROBERT 6,10 5,25 5,95 48
267:::257 4473 SOMCUTEAN L ROZALIA LOREDANA 7,50 7,40 6,30 45 42, 43 44 46 47.
268:::264 4467 STIR R RUDOLF EMANUEL 6,38 680 7,35 19Nr Medie Gen NotaPr Nota
269:::273 4560 TAKACS T BERNADETT 7,53 7,05 845 41 44
270:::274 4652 TANTAU I ILEANA LETITIA 8,15 9,00 8,25 46 43
271:::275 4653 TAUT l RALUCA IOANA 7,11 7,30 7,30 16 11.15 18 12
272:::276 4974 TIFRAC | DANIELA 6,02 5,00 5,00 69.
273:::277 4964 TIMIS G GEORGIANA ECATERINA 8,33 7,40 8,90 69.
274:::278 4963 TIPLEA P MARIA 7,30 8,35 5,85 69 
275:::279 4486 TISE P IOANA CAMELIA 6,28 5,75 5,00 44.
276:::280 4459 TODOR A ALEXANDRU , 6,21 5,40 6,20 44
277:::281 4668 TODOR I VALTER 6,17 6,50 6,10 31 32 33,
278:::282 4466 TOMA A ANDREI 6,35 8,70 5,00 42 47,
279:::283 4978 TOMOIAGA M ALINA MARIA 9,14 8,40 8,10 70,
280:::284 4979 TOMOIAGA V MARIA 6,16 6,00 5,10 70.
281:::285 4516 TOPAN D BOGDAN MIHAI 6,88 5,05 7,40 18 16 19 15
282:::286 4643 TOROK l NORBERT ZSOLT 8,23 8,10 7,95 47 46 43.
283:::287 4903 TUZES E IBOLYA 6,58 5,00 8,12 59 67,
284:::288 4693 UJVARI N GABOR OTTO 6,65 6,35 8,90 32 33 30
285:::289 4458 UNGUREANU Vv RAZVAN NICOLAE 6,98 5,40 850 46.
286:::290 4448 VADAN | ANAMARIA MERY 7,50 5,40 5,40 43 42
287:::291 4605 VAIDA Zz HELGA 6,32 5,75 5,10 32, 33 30
288:::292 4917 VAJDA A EVA 7,71 8,15 9,30 64 66
289:::293 4420 VALEAN L STEFAN ILIE 7,21 7,20 5,90 47 46.
290:::294 4697 VANCEA 1 IOAN CLAUDIU 6,10 5,80 5,80 41 43
291:::295 4638 VASIAC D LIVIU DANIEL LUIGI 6,73 6,80 7,70 14 16 15 18 20 12 19
292:::296 4968 VASIES | IONELA CRISTINA 6,48 6,40 5,70 70 
293:::297 4658 VASILICA 1 RALUCA CLAUDIA 8,40 9,60 6,40 11 16
294:::298 4449 VASS D DENES MADALIN 9,11 8,10 6,85 O07.
295:::299 4451 VASTAG R RAUL STEFAN 6,23 5,50 6,70 16 14 15 20 18 11,13 12 19
296:::300 8 ©4455 VIGA I IONEL DANIEL 6,26 6,60 5,50 44 47 46 16 19 18 13
297:::301 4650 VILD E ERNEST 7,10 6,20 6,00 04 18 19 16 15 11 14 13 12 20
298:::302 4907 VIRAG I JOZSEF 7,84 8,80 7,50 67 
299:::303 4567 VITISAN G.B ADELINAEMILIA 9,40 7,90 9,50 44 41
300:::304 4503 VLASA M MIHAI 6,45 5,00 5,00 46 47
301:::305 204 ZACHARIS K GEORGIOS 5,00 5,00 5,00 19 15 18 37 16 11
302:::306 203 ZADOROINII © A IULIANA 6,50 6,00 6,00 15, 16 19
303:::307 4677 ZAGREAN D ANCUJA DIANA 6,36 5,00 5,30 41.
304:::308 4425 ZEGREAN D ANA MARIA 7,51 6,60 8,00 11
305:::309 7908 ZSIGA M KAROLA , 7,32 6,95 8,40 64 59
306:::310 4410 ZSURZSAN G. ZS, VIVIEN MERCEDESZ 8,02 7,20 8,65 27 36 37 18 19 16
307:::311 4562 ZUGRAV | IOANA / 6,58 5,00 6,75 15 12

In [331]:

    

good_lines[126]=good_lines[126].split(',ve Nr Leg')[0]


    

In [332]:

    

good_lines[146]=good_lines[146].replace('7 AL','7,41')


    

In [333]:

    

helper=good_lines[195].split('199 4514')
good_lines[195]=helper[0]
good_lines.append('199 4514'+helper[1])


    

In [334]:

    

clean_lines=[]
for line in good_lines:
    values=line.split(' ')
    
    try:
        for k in [2,3,4]:
            values[k]=values[k].replace('1','I').replace('l','I')\
                .replace('|','I').replace('7','I').replace('0','O')

        counter=0
        name_start=100
        name_end=0
        while counter<len(values)-1:
            counter+=1
            try:
                if not values[counter][0].isdigit():
                    name_start=min(name_start,counter)
                    if values[counter+1][0].isdigit():
                        name_end=counter+1
                        counter=len(values)
            except:
                print('HIBA: '+line)
    
        clean_values=values[:name_start]+[' '.join(values[name_start:name_end])]+values[name_end:]
        clean_values=[szo for szo in clean_values if szo]
        clean_lines.append(clean_values)

    except:
        print('HIBA: '+line)


    

    




HIBA: 4 Bp Sp

In [341]:

    

df=pd.DataFrame(clean_lines)
df=df[range(6)]
df.columns=['ID','Leg','Nume','Admitere','Bac 2','Lb. Mat.']
df.head()


    

    
Out[341]:







  

    

      

      
ID
      
Leg
      
Nume
      
Admitere
      
Bac 2
      
Lb. Mat.
    
  
  

    

      
0
      
1
      
4696
      
ACHIM O.A ANDREEA MAGDALENA
      
9.56
      
9.5
      
9.70
    
    

      
1
      
3
      
4441
      
ALBERT J CSILLA EDIT
      
6.76
      
5.7
      
7.80
    
    

      
2
      
2
      
208
      
ALBERT Vv IULAN
      
6.10
      
5.2
      
6.00
    
    

      
3
      
4
      
4500
      
ALEXAN A MARIOARA
      
7.82
      
7.0
      
8.55
    
    

      
4
      
5
      
4565
      
AMBRUS T TIBOR
      
8.36
      
9.1
      
9.40
    
  

In [350]:

    

df.to_excel('pdf/'+PDF_file+'/data.xlsx')


    

In [340]:

    

for line in clean_lines:
    try:
        for i in [3,4,5]:
            jegy=line[i]
            jegy=jegy.replace('.',',')
            jegy=jegy.replace('I','1').replace('l','1')\
                .replace('|','1').replace('A','4').replace('O','0')

            if jegy[0]==',':jegy=jegy[1:]
            if jegy[-1]==',':jegy=jegy[:-1]
            if jegy!='10,00':
                if jegy!='1000':
                    if len(jegy)<4:
                        jegy=jegy[0]+','+jegy[1:]
                    jegy=float(jegy.replace(',','.'))
                else: jegy=10
            else: jegy=10
            line[i]=jegy
    except:
        print(line)


    

In [342]:

    

df=pd.DataFrame(clean_lines)
df=df[range(6)]
df.columns=['ID','Leg','Nume','Admitere','Bac 2','Lb. Mat.']
df.head()


    

    
Out[342]:







  

    

      

      
ID
      
Leg
      
Nume
      
Admitere
      
Bac 2
      
Lb. Mat.
    
  
  

    

      
0
      
1
      
4696
      
ACHIM O.A ANDREEA MAGDALENA
      
9.56
      
9.5
      
9.70
    
    

      
1
      
3
      
4441
      
ALBERT J CSILLA EDIT
      
6.76
      
5.7
      
7.80
    
    

      
2
      
2
      
208
      
ALBERT Vv IULAN
      
6.10
      
5.2
      
6.00
    
    

      
3
      
4
      
4500
      
ALEXAN A MARIOARA
      
7.82
      
7.0
      
8.55
    
    

      
4
      
5
      
4565
      
AMBRUS T TIBOR
      
8.36
      
9.1
      
9.40
    
  

In [349]:

    

df['Mean']=df.mean(axis=1)
df


    

    
Out[349]:







  

    

      

      
ID
      
Leg
      
Nume
      
Admitere
      
Bac 2
      
Lb. Mat.
      
Mean
    
  
  

    

      
0
      
1
      
4696
      
ACHIM O.A ANDREEA MAGDALENA
      
9.56
      
9.50
      
9.70
      
9.586667
    
    

      
1
      
3
      
4441
      
ALBERT J CSILLA EDIT
      
6.76
      
5.70
      
7.80
      
6.753333
    
    

      
2
      
2
      
208
      
ALBERT Vv IULAN
      
6.10
      
5.20
      
6.00
      
5.766667
    
    

      
3
      
4
      
4500
      
ALEXAN A MARIOARA
      
7.82
      
7.00
      
8.55
      
7.790000
    
    

      
4
      
5
      
4565
      
AMBRUS T TIBOR
      
8.36
      
9.10
      
9.40
      
8.953333
    
    

      
...
      
...
      
...
      
...
      
...
      
...
      
...
      
...
    
    

      
303
      
308
      
4425
      
ZEGREAN D ANA MARIA
      
7.51
      
6.60
      
8.00
      
7.370000
    
    

      
304
      
309
      
7908
      
ZSIGA M KAROLA ,
      
7.32
      
6.95
      
8.40
      
7.556667
    
    

      
305
      
310
      
4410
      
ZSURZSAN G. ZS, VIVIEN MERCEDESZ
      
8.02
      
7.20
      
8.65
      
7.956667
    
    

      
306
      
311
      
4562
      
ZUGRAV I IOANA /
      
6.58
      
5.00
      
6.75
      
6.110000
    
    

      
307
      
199
      
4514
      
PASCA G FLORINA DENISA MIHAELA
      
8.28
      
8.85
      
8.10
      
8.410000
    
  

308 rows × 7 columns