| 1 |
11
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Converts a CSV dataset to XML using a mappings spreadsheet
|
| 3 |
|
|
|
| 4 |
|
|
import csv
|
| 5 |
42
|
aaronmk
|
import os.path
|
| 6 |
11
|
aaronmk
|
import re
|
| 7 |
|
|
import sys
|
| 8 |
|
|
from copy import deepcopy
|
| 9 |
|
|
from xml.dom.minidom import getDOMImplementation
|
| 10 |
|
|
|
| 11 |
43
|
aaronmk
|
sys.path.append(os.path.dirname(__file__)+"/lib")
|
| 12 |
21
|
aaronmk
|
import xpath
|
| 13 |
11
|
aaronmk
|
|
| 14 |
|
|
def main():
|
| 15 |
|
|
prog_name = sys.argv.pop(0)
|
| 16 |
|
|
try:
|
| 17 |
|
|
dest = sys.argv.pop(0)
|
| 18 |
|
|
mappings_path = sys.argv.pop(0)
|
| 19 |
|
|
except IndexError: raise Exception('Usage: '+prog_name
|
| 20 |
18
|
aaronmk
|
+' dest_mappings_column mappings_path <dataset >output')
|
| 21 |
11
|
aaronmk
|
|
| 22 |
|
|
# Get mappings
|
| 23 |
|
|
mappings = {}
|
| 24 |
22
|
aaronmk
|
has_types = False # whether outer elements are type containiners
|
| 25 |
11
|
aaronmk
|
stream = open(mappings_path, 'rb')
|
| 26 |
|
|
reader = csv.reader(stream, delimiter=',', quotechar='"')
|
| 27 |
|
|
fieldnames = reader.next()
|
| 28 |
|
|
src = fieldnames[0]
|
| 29 |
|
|
dest_idx = fieldnames.index(dest)
|
| 30 |
|
|
for row in reader:
|
| 31 |
|
|
name = row[0]
|
| 32 |
|
|
path = row[dest_idx]
|
| 33 |
18
|
aaronmk
|
if name != '' and path != '':
|
| 34 |
31
|
aaronmk
|
if path.startswith('/*s/'): has_types = True # *s used for type elem
|
| 35 |
11
|
aaronmk
|
path = path.replace('<name>', name)
|
| 36 |
21
|
aaronmk
|
mappings[name] = xpath.XpathParser(path).parse()
|
| 37 |
11
|
aaronmk
|
stream.close()
|
| 38 |
|
|
|
| 39 |
|
|
# Process dataset
|
| 40 |
|
|
doc = getDOMImplementation().createDocument(None, dest, None)
|
| 41 |
|
|
stream = sys.stdin
|
| 42 |
|
|
reader = csv.reader(stream, delimiter=',', quotechar='"')
|
| 43 |
|
|
fieldnames = reader.next()
|
| 44 |
|
|
row_idx = 0
|
| 45 |
|
|
for row in reader:
|
| 46 |
18
|
aaronmk
|
row_id = str(row_idx)
|
| 47 |
11
|
aaronmk
|
for idx, name in enumerate(fieldnames):
|
| 48 |
|
|
value = row[idx]
|
| 49 |
18
|
aaronmk
|
if value != '' and name in mappings:
|
| 50 |
11
|
aaronmk
|
path = deepcopy(mappings[name]) # don't modify main value!
|
| 51 |
22
|
aaronmk
|
xpath.set_id(path, row_id, has_types)
|
| 52 |
|
|
xpath.set_value(path, value)
|
| 53 |
21
|
aaronmk
|
xpath.get(doc, path, True)
|
| 54 |
11
|
aaronmk
|
row_idx += 1
|
| 55 |
|
|
doc.writexml(sys.stdout, addindent=' ', newl='\n')
|
| 56 |
|
|
|
| 57 |
|
|
main()
|
