Python：将pdf转csv（多行列）

Question

我的 CSV 是：

,Élément,État général,Observations
0,ENTRÉE,Etat d'usage,
1,PORTES,Etat d'usage,Chaînette cassé
Serrure du bas en mauvais état le système est
cassé au niveau de la chaînette
2,ENTRÉE / PORTESENTRÉE / PORTES,,
3,Type de porte,,Porte blindée
4,Poignée,,Bon état
5,Couleur,,Bois

但我想要这个：

,Élément,État général,Observations
0,ENTRÉE,Etat d'usage,
1,PORTES,Etat d'usage,Chaînette cassé; Serrure du bas en mauvais état le système ...
2,ENTRÉE / PORTESENTRÉE / PORTES,,
3,Type de porte,,Porte blindée
4,Poignée,,Bon état
5,Couleur,,Bois

我的代码只是将每个页面的一个或多个 pdf 转换为 csv，如下所示：

import os
import io
import shutil
import tabula
import time

start_time = time.time()
path = './'

i=0
j=0

for( directory, subdirectories, file ) in os.walk(path):
    for f in file:
        if f.endswith('.pdf'):
                df = tabula.read_pdf(str(directory) + "/" + str(f), pages='all')
                i=0
                j+=1
                for curr_df in df:
                    i+=1
                    curr_df.to_csv('./' + str(directory) + '-' + str(i) + '.csv')

print("--- convert %d .PDF to %d .CSV in %s seconds ---" % (j, i, time.time() - start_time))

我的问题也是因为我无法逐案处理。我需要能够以相同的方式处理所有 csv

Answer 1

您可以打开 csv，阅读各行，并将不以空开头 (header) 或以数字开头的字符串添加到上一行。然后将这些行写入一个新的 csv 文件：

with open('filename.csv') as f:
    text = [line.rstrip() for line in f.readlines()] #remove newline character with rstrip()
    lines = []
    for i in text:
        try:
            if i[0] ==',' or i[0].isnumeric():
                lines.append(i)
            else:
                lines[-1] = lines[-1] + "; " + i
        except:
            continue
            
with open('new_file.csv', mode='wt', encoding='utf-8') as newfile:
    newfile.write('\n'.join(lines)) # reinsert newline characters with '\n'.join()

要处理目录中的所有文件，我们可以将其放入一个函数并将目录中的所有文件提供给该函数：

import os as os
import glob as glob

def process_csv(filepath):
    with open(filepath) as f:
        text = [line.rstrip() for line in f.readlines()] #remove newline character with rstrip()
        lines = []
        for i in text:
            try:
                if i[0] ==',' or i[0].isnumeric():
                    lines.append(i)
                else:
                    lines[-1] = lines[-1] + "; " + i
            except:
                continue

    with open(os.path.basename(filepath) + '_fixed.csv', mode='wt', encoding='utf-8') as newfile:
        newfile.write('\n'.join(lines)) # reinsert newline characters with '\n'.join()
        print('fixed: ' + os.path.basename(filepath) + '_fixed.csv')

files = glob.glob('./*.csv') #use glob to create a list of filepath of csv files in a directory

for file in files: # loop through the list and feed each file to the function process_csv
    process_csv(file)

Answer 2

对于@SergeBallesta，这是我所拥有的：

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 8 entries, 0 to 7
Data columns (total 4 columns):
 #   Column        Non-Null Count  Dtype 
---  ------        --------------  ----- 
 0   Unnamed: 0    8 non-null      object
 1   Élément       6 non-null      object
 2   État général  2 non-null      object
 3   Observations  4 non-null      object
dtypes: object(4)
memory usage: 384.0+ bytes
None
                                      Unnamed: 0                         Élément  État général     Observations
0                                              0                          ENTRÉE  Etat d'usage              NaN
1                                              1                          PORTES  Etat d'usage  Chaînette cassé
2  Serrure du bas en mauvais état le système est                             NaN           NaN              NaN
3                cassé au niveau de la chaînette                             NaN           NaN              NaN
4                                              2  ENTRÉE / PORTESENTRÉE / PORTES           NaN              NaN
5                                              3                   Type de porte           NaN    Porte blindée
6                                              4                         Poignée           NaN         Bon état
7                                              5                         Couleur           NaN             Bois

和

import pandas as pd 

df = pd.read_csv('../CSV/Entire/PDF-1.csv') 
print(df.info()) 
print(df)

Answer 3

对于@RJAdriaansen，这是我得到的错误：

fixed: PDF-8.csv_fixed.csv
fixed: PDF-5.csv_fixed.csv
fixed: PDF-7.csv_fixed.csv
fixed: PDF-6.csv_fixed.csv
fixed: PDF-2.csv_fixed.csv
fixed: PDF-10.csv_fixed.csv
fixed: PDF-3.csv_fixed.csv
fixed: PDF-4.csv_fixed.csv
Traceback (most recent call last):
  File "corrCSV_v2.py", line 24, in <module>
    process_csv(file)
  File "corrCSV_v2.py", line 12, in process_csv
    if i[0] ==',' or i[0].isnumeric():
IndexError: string index out of range

错误来自这个.csv

,Élément,État général,Observations
0,CUISINE,Etat d'usage,
1,CUISINECUISINE 15CUISINE 18

CUISINE 19,,

我认为这是由于空行

Python：将pdf转csv（多行列）

Python: convert pdf to csv (multi-line column)

python

csv

pdf

multiline

tabula