| 1 |
53
|
aaronmk
|
#!/usr/bin/env python
|
| 2 |
|
|
# Maps one datasource to another, using a map spreadsheet if needed
|
| 3 |
|
|
# For outputting an XML file to a PostgreSQL database, use the general format of
|
| 4 |
|
|
# http://vegbank.org/vegdocs/xml/vegbank_example_ver1.0.2.xml
|
| 5 |
|
|
|
| 6 |
|
|
import os
|
| 7 |
|
|
import os.path
|
| 8 |
|
|
import sys
|
| 9 |
|
|
import xml.dom.minidom
|
| 10 |
|
|
|
| 11 |
|
|
sys.path.append(os.path.dirname(__file__)+"/lib")
|
| 12 |
|
|
|
| 13 |
64
|
aaronmk
|
import opts
|
| 14 |
86
|
aaronmk
|
import xml_func
|
| 15 |
53
|
aaronmk
|
|
| 16 |
84
|
aaronmk
|
def metadata_value(name):
|
| 17 |
|
|
if name.startswith(':'): return name[1:]
|
| 18 |
|
|
else: return None
|
| 19 |
|
|
|
| 20 |
53
|
aaronmk
|
def main():
|
| 21 |
|
|
# Get db config from env vars
|
| 22 |
|
|
db_config_names = ['host', 'user', 'password', 'database']
|
| 23 |
|
|
env_names = []
|
| 24 |
|
|
def get_db_config(prefix):
|
| 25 |
64
|
aaronmk
|
return opts.get_env_vars(db_config_names, prefix, env_names)
|
| 26 |
67
|
aaronmk
|
in_db_config = get_db_config('in')
|
| 27 |
|
|
out_db_config = get_db_config('out')
|
| 28 |
130
|
aaronmk
|
in_is_db = in_db_config != {}
|
| 29 |
|
|
out_is_db = out_db_config != {}
|
| 30 |
53
|
aaronmk
|
|
| 31 |
|
|
# Parse args
|
| 32 |
73
|
aaronmk
|
map_path = None
|
| 33 |
67
|
aaronmk
|
try: _prog_name, map_path = sys.argv
|
| 34 |
53
|
aaronmk
|
except ValueError:
|
| 35 |
73
|
aaronmk
|
if in_is_db or not out_is_db: raise SystemExit('Usage: '
|
| 36 |
|
|
+opts.env_usage(env_names, True)+' [commit=1] '+sys.argv[0]
|
| 37 |
|
|
+' [map_path] [<input] [>output]')
|
| 38 |
64
|
aaronmk
|
commit = opts.env_flag('commit')
|
| 39 |
53
|
aaronmk
|
|
| 40 |
57
|
aaronmk
|
# Load map header
|
| 41 |
130
|
aaronmk
|
in_is_xpaths = True
|
| 42 |
73
|
aaronmk
|
if map_path != None:
|
| 43 |
56
|
aaronmk
|
import copy
|
| 44 |
|
|
import csv
|
| 45 |
|
|
|
| 46 |
75
|
aaronmk
|
from Parser import SyntaxException
|
| 47 |
53
|
aaronmk
|
import xpath
|
| 48 |
|
|
|
| 49 |
73
|
aaronmk
|
mappings = []
|
| 50 |
|
|
stream = open(map_path, 'rb')
|
| 51 |
|
|
reader = csv.reader(stream)
|
| 52 |
|
|
src, dest = reader.next()[:2]
|
| 53 |
61
|
aaronmk
|
def split_col_name(name):
|
| 54 |
72
|
aaronmk
|
name, sep, root = name.partition(':')
|
| 55 |
|
|
return name, sep != '', root
|
| 56 |
130
|
aaronmk
|
src, in_is_xpaths, src_root = split_col_name(src)
|
| 57 |
|
|
dest, out_is_xpaths, dest_root = split_col_name(dest)
|
| 58 |
|
|
assert out_is_xpaths
|
| 59 |
67
|
aaronmk
|
has_types = dest_root.startswith('/*s/') # outer elements are types
|
| 60 |
73
|
aaronmk
|
for row in reader:
|
| 61 |
|
|
in_, out = row[:2]
|
| 62 |
|
|
if out != '':
|
| 63 |
84
|
aaronmk
|
try: mappings.append((in_, xpath.parse(dest_root+out)))
|
| 64 |
73
|
aaronmk
|
except SyntaxException, ex: raise SystemExit(str(ex))
|
| 65 |
|
|
stream.close()
|
| 66 |
130
|
aaronmk
|
in_is_xml = in_is_xpaths and not in_is_db
|
| 67 |
56
|
aaronmk
|
|
| 68 |
57
|
aaronmk
|
# Input datasource to XML tree, mapping if needed
|
| 69 |
116
|
aaronmk
|
if in_is_xml: doc0 = xml.dom.minidom.parse(sys.stdin)
|
| 70 |
73
|
aaronmk
|
if map_path != None:
|
| 71 |
116
|
aaronmk
|
doc1 = xml.dom.minidom.getDOMImplementation().createDocument(None,
|
| 72 |
56
|
aaronmk
|
dest, None)
|
| 73 |
126
|
aaronmk
|
if in_is_db:
|
| 74 |
130
|
aaronmk
|
assert in_is_xpaths
|
| 75 |
126
|
aaronmk
|
|
| 76 |
118
|
aaronmk
|
import psycopg2
|
| 77 |
117
|
aaronmk
|
import db_xml
|
| 78 |
|
|
|
| 79 |
126
|
aaronmk
|
try: src_root = xpath.parse(src_root)
|
| 80 |
|
|
except SyntaxException, ex: raise SystemExit(str(ex))
|
| 81 |
|
|
|
| 82 |
117
|
aaronmk
|
in_db = psycopg2.connect(**in_db_config)
|
| 83 |
|
|
for in_, out in mappings:
|
| 84 |
|
|
value = metadata_value(in_)
|
| 85 |
|
|
if value == None:
|
| 86 |
118
|
aaronmk
|
raise SystemExit('DB-XML mapping not supported yet')
|
| 87 |
117
|
aaronmk
|
in_db.close()
|
| 88 |
126
|
aaronmk
|
elif in_is_xml: raise SystemExit('XML-XML mapping not supported yet')
|
| 89 |
56
|
aaronmk
|
else: # input is CSV
|
| 90 |
84
|
aaronmk
|
metadata = []
|
| 91 |
|
|
map_ = {}
|
| 92 |
|
|
for in_, out in mappings:
|
| 93 |
|
|
value = metadata_value(in_)
|
| 94 |
|
|
if value != None: metadata.append((value, out))
|
| 95 |
|
|
else: map_[in_] = out
|
| 96 |
|
|
|
| 97 |
59
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
| 98 |
84
|
aaronmk
|
cols = reader.next()
|
| 99 |
83
|
aaronmk
|
for row_idx, row in enumerate(reader):
|
| 100 |
56
|
aaronmk
|
row_id = str(row_idx)
|
| 101 |
84
|
aaronmk
|
def put_col(path, value):
|
| 102 |
116
|
aaronmk
|
xpath.put_obj(doc1, path, row_id, has_types, value)
|
| 103 |
84
|
aaronmk
|
for value, out in metadata: put_col(out, value)
|
| 104 |
|
|
for i, col in enumerate(cols):
|
| 105 |
|
|
if row[i] != '' and col in map_: put_col(map_[col], row[i])
|
| 106 |
116
|
aaronmk
|
xml_func.process(doc1)
|
| 107 |
|
|
else: doc1 = doc0
|
| 108 |
53
|
aaronmk
|
|
| 109 |
|
|
# Output XML tree
|
| 110 |
130
|
aaronmk
|
if out_is_db:
|
| 111 |
53
|
aaronmk
|
import psycopg2
|
| 112 |
|
|
from psycopg2.extensions import ISOLATION_LEVEL_SERIALIZABLE
|
| 113 |
|
|
import db_xml
|
| 114 |
|
|
|
| 115 |
67
|
aaronmk
|
db = psycopg2.connect(**out_db_config)
|
| 116 |
53
|
aaronmk
|
db.set_isolation_level(ISOLATION_LEVEL_SERIALIZABLE)
|
| 117 |
|
|
try:
|
| 118 |
|
|
row_ct_ref = [0]
|
| 119 |
116
|
aaronmk
|
db_xml.xml2db(db, doc1.documentElement, row_ct_ref)
|
| 120 |
53
|
aaronmk
|
print 'Inserted '+str(row_ct_ref[0])+' rows'
|
| 121 |
|
|
if commit: db.commit()
|
| 122 |
|
|
finally:
|
| 123 |
|
|
db.rollback()
|
| 124 |
|
|
db.close()
|
| 125 |
116
|
aaronmk
|
else: doc1.writexml(sys.stdout, addindent=' ', newl='\n') # output is XML
|
| 126 |
53
|
aaronmk
|
|
| 127 |
|
|
main()
|
