Project

General

Profile

« Previous | Next » 

Revision 3071

Added by Aaron Marcuse-Kubitza over 12 years ago

cat_csv: Support overwriting the existing header using a separate header file

View differences:

bin/cat_csv
1 1 
#!/usr/bin/env python
2 2 
# Concatenates spreadsheets, removing any duplicated headers
3 3 
# Usage: self [sheet...] >out_sheet
4 
#
5 
# The existing header can be overwritten by adding a separate header file whose
6 
# header line is prefixed with "!". This is useful if the existing header's
7 
# column names are too long, not descriptive enough, etc.
4 8 

  		




  5
  9
  
    
import os.path


  		




  6
  10
  
    
import sys


  		




  ......




  8
  12
  
    
sys.path.append(os.path.dirname(__file__)+"/../lib")


  		




  9
  13
  
    

  		




  10
  14
  
    
import csvs


  		




  
  15
  
    
import strings


  		




  11
  16
  
    
import util


  		




  12
  17
  
    

  		




  13
  18
  
    
def main():


  		




  14
  19
  
    
    paths = sys.argv[1:]


  		




  15
  20
  
    
    


  		




  
  21
  
    
    header_written = False


  		




  16
  22
  
    
    first_path = None


  		




  17
  23
  
    
    first_info = None


  		




  
  24
  
    
    


  		




  18
  25
  
    
    for path in paths:


  		




  19
  26
  
    
        if path == '-': stream = sys.stdin


  		




  20
  27
  
    
        else: stream = open(path, 'rb')


  		




  ......




  24
  31
  
    
        try:


  		




  25
  32
  
    
            if info.dialect == None: continue # dialect of None = empty stream


  		




  26
  33
  
    
            


  		




  
  34
  
    
            overwrite_ref = [False] # whether to overwrite the existing header


  		




  
  35
  
    
            info.header_line = strings.remove_prefix('!', info.header_line,


  		




  
  36
  
    
                overwrite_ref)


  		




  
  37
  
    
            


  		




  27
  38
  
    
            def write_header(): sys.stdout.write(info.header_line)


  		




  
  39
  
    
            


  		




  28
  40
  
    
            if first_info == None:


  		




  29
  41
  
    
                first_path = path


  		




  30
  
  
    
                first_info = info


  		




  31
  
  
    
                write_header()


  		




  
  42
  
    
                if not overwrite_ref[0]: first_info = info


  		




  
  43
  
    
                    # otherwise, next header becomes the compare-to header


  		




  
  44
  
    
                if not header_written:


  		




  
  45
  
    
                    write_header()


  		




  
  46
  
    
                    header_written = True


  		




  32
  47
  
    
            else:


  		




  33
  48
  
    
                if not util.classes_eq(info.dialect, first_info.dialect):


  		




  34
  49
  
    
                    raise SystemExit('Spreadsheet error: "'+path


  		












Also available in:  Unified diff