| 1 |
355
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
1447
|
aaronmk
|
# Concatenates spreadsheets, removing any duplicated headers
|
| 3 |
|
|
# Usage: self [sheet...] >out_sheet
|
| 4 |
3071
|
aaronmk
|
#
|
| 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.
|
| 8 |
355
|
aaronmk
|
|
| 9 |
1447
|
aaronmk
|
import os.path
|
| 10 |
355
|
aaronmk
|
import sys
|
| 11 |
|
|
|
| 12 |
1447
|
aaronmk
|
sys.path.append(os.path.dirname(__file__)+"/../lib")
|
| 13 |
|
|
|
| 14 |
|
|
import csvs
|
| 15 |
3071
|
aaronmk
|
import strings
|
| 16 |
1447
|
aaronmk
|
import util
|
| 17 |
|
|
|
| 18 |
355
|
aaronmk
|
def main():
|
| 19 |
1447
|
aaronmk
|
paths = sys.argv[1:]
|
| 20 |
355
|
aaronmk
|
|
| 21 |
3071
|
aaronmk
|
header_written = False
|
| 22 |
1447
|
aaronmk
|
first_path = None
|
| 23 |
|
|
first_info = None
|
| 24 |
3071
|
aaronmk
|
|
| 25 |
1447
|
aaronmk
|
for path in paths:
|
| 26 |
1493
|
aaronmk
|
if path == '-': stream = sys.stdin
|
| 27 |
|
|
else: stream = open(path, 'rb')
|
| 28 |
1447
|
aaronmk
|
|
| 29 |
|
|
# Get dialect and process first line
|
| 30 |
|
|
info = csvs.stream_info(stream)
|
| 31 |
1448
|
aaronmk
|
try:
|
| 32 |
1661
|
aaronmk
|
if info.dialect == None: continue # dialect of None = empty stream
|
| 33 |
|
|
|
| 34 |
3071
|
aaronmk
|
overwrite_ref = [False] # whether to overwrite the existing header
|
| 35 |
|
|
info.header_line = strings.remove_prefix('!', info.header_line,
|
| 36 |
|
|
overwrite_ref)
|
| 37 |
|
|
|
| 38 |
1448
|
aaronmk
|
def write_header(): sys.stdout.write(info.header_line)
|
| 39 |
3071
|
aaronmk
|
|
| 40 |
1448
|
aaronmk
|
if first_info == None:
|
| 41 |
|
|
first_path = path
|
| 42 |
3071
|
aaronmk
|
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
|
| 47 |
1448
|
aaronmk
|
else:
|
| 48 |
|
|
if not util.classes_eq(info.dialect, first_info.dialect):
|
| 49 |
|
|
raise SystemExit('Spreadsheet error: "'+path
|
| 50 |
|
|
+'" dialect doesn\'t match "'+first_path+'" dialect')
|
| 51 |
|
|
if info.header_line != first_info.header_line: write_header()
|
| 52 |
|
|
# not a duplicated header
|
| 53 |
|
|
|
| 54 |
|
|
# Copy remaining lines
|
| 55 |
|
|
while True:
|
| 56 |
|
|
line = stream.readline()
|
| 57 |
|
|
if line == '': break
|
| 58 |
|
|
sys.stdout.write(line)
|
| 59 |
|
|
except IOError, e: # abort if output stream closed
|
| 60 |
|
|
if str(e) != '[Errno 32] Broken pipe': raise # other IOErrors
|
| 61 |
1661
|
aaronmk
|
finally: stream.close() # still run if continue is called
|
| 62 |
355
|
aaronmk
|
|
| 63 |
|
|
main()
|
