| 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.path
|
| 7 |
|
|
import sys
|
| 8 |
|
|
import xml.dom.minidom
|
| 9 |
|
|
|
| 10 |
|
|
sys.path.append(os.path.dirname(__file__)+"/lib")
|
| 11 |
|
|
|
| 12 |
64
|
aaronmk
|
import opts
|
| 13 |
133
|
aaronmk
|
from Parser import SyntaxException
|
| 14 |
131
|
aaronmk
|
import sql
|
| 15 |
133
|
aaronmk
|
import xml_dom
|
| 16 |
86
|
aaronmk
|
import xml_func
|
| 17 |
53
|
aaronmk
|
|
| 18 |
84
|
aaronmk
|
def metadata_value(name):
|
| 19 |
|
|
if name.startswith(':'): return name[1:]
|
| 20 |
|
|
else: return None
|
| 21 |
|
|
|
| 22 |
53
|
aaronmk
|
def main():
|
| 23 |
131
|
aaronmk
|
env_names = []
|
| 24 |
|
|
def usage_err():
|
| 25 |
|
|
raise SystemExit('Usage: '+opts.env_usage(env_names, True)
|
| 26 |
|
|
+' [commit=1] '+sys.argv[0]+' [map_path] [<input] [>output]')
|
| 27 |
146
|
aaronmk
|
limit = opts.get_env_var('n', None, env_names)
|
| 28 |
|
|
if limit != None: limit = int(limit)
|
| 29 |
135
|
aaronmk
|
commit = opts.env_flag('commit')
|
| 30 |
131
|
aaronmk
|
|
| 31 |
53
|
aaronmk
|
# Get db config from env vars
|
| 32 |
131
|
aaronmk
|
db_config_names = ['engine', 'host', 'user', 'password', 'database']
|
| 33 |
53
|
aaronmk
|
def get_db_config(prefix):
|
| 34 |
64
|
aaronmk
|
return opts.get_env_vars(db_config_names, prefix, env_names)
|
| 35 |
67
|
aaronmk
|
in_db_config = get_db_config('in')
|
| 36 |
|
|
out_db_config = get_db_config('out')
|
| 37 |
131
|
aaronmk
|
in_is_db = 'engine' in in_db_config
|
| 38 |
|
|
out_is_db = 'engine' in out_db_config
|
| 39 |
53
|
aaronmk
|
|
| 40 |
|
|
# Parse args
|
| 41 |
73
|
aaronmk
|
map_path = None
|
| 42 |
67
|
aaronmk
|
try: _prog_name, map_path = sys.argv
|
| 43 |
53
|
aaronmk
|
except ValueError:
|
| 44 |
135
|
aaronmk
|
if in_is_db: usage_err()
|
| 45 |
53
|
aaronmk
|
|
| 46 |
57
|
aaronmk
|
# Load map header
|
| 47 |
130
|
aaronmk
|
in_is_xpaths = True
|
| 48 |
73
|
aaronmk
|
if map_path != None:
|
| 49 |
56
|
aaronmk
|
import copy
|
| 50 |
|
|
import csv
|
| 51 |
|
|
|
| 52 |
53
|
aaronmk
|
import xpath
|
| 53 |
|
|
|
| 54 |
133
|
aaronmk
|
metadata = []
|
| 55 |
73
|
aaronmk
|
mappings = []
|
| 56 |
|
|
stream = open(map_path, 'rb')
|
| 57 |
|
|
reader = csv.reader(stream)
|
| 58 |
161
|
aaronmk
|
in_label, out_label = reader.next()[:2]
|
| 59 |
61
|
aaronmk
|
def split_col_name(name):
|
| 60 |
72
|
aaronmk
|
name, sep, root = name.partition(':')
|
| 61 |
|
|
return name, sep != '', root
|
| 62 |
161
|
aaronmk
|
in_label, in_is_xpaths, in_root = split_col_name(in_label)
|
| 63 |
|
|
out_label, out_is_xpaths, out_root = split_col_name(out_label)
|
| 64 |
133
|
aaronmk
|
assert out_is_xpaths # CSV output not supported yet
|
| 65 |
161
|
aaronmk
|
has_types = out_root.startswith('/*s/') # outer elements are types
|
| 66 |
73
|
aaronmk
|
for row in reader:
|
| 67 |
|
|
in_, out = row[:2]
|
| 68 |
|
|
if out != '':
|
| 69 |
133
|
aaronmk
|
value = metadata_value(in_)
|
| 70 |
161
|
aaronmk
|
if out_is_xpaths: out = xpath.parse(out_root+out)
|
| 71 |
|
|
if value != None: metadata.append((value, out))
|
| 72 |
133
|
aaronmk
|
else: mappings.append((in_, out))
|
| 73 |
73
|
aaronmk
|
stream.close()
|
| 74 |
130
|
aaronmk
|
in_is_xml = in_is_xpaths and not in_is_db
|
| 75 |
56
|
aaronmk
|
|
| 76 |
57
|
aaronmk
|
# Input datasource to XML tree, mapping if needed
|
| 77 |
161
|
aaronmk
|
if in_is_xml:
|
| 78 |
|
|
doc0 = xml.dom.minidom.parse(sys.stdin)
|
| 79 |
73
|
aaronmk
|
if map_path != None:
|
| 80 |
161
|
aaronmk
|
doc1 = xml_dom.create_doc(out_label)
|
| 81 |
141
|
aaronmk
|
root = doc1.documentElement
|
| 82 |
126
|
aaronmk
|
if in_is_db:
|
| 83 |
130
|
aaronmk
|
assert in_is_xpaths
|
| 84 |
126
|
aaronmk
|
|
| 85 |
117
|
aaronmk
|
import db_xml
|
| 86 |
|
|
|
| 87 |
161
|
aaronmk
|
in_root_xml = xpath.path2xml(in_root)
|
| 88 |
|
|
mappings = [(xpath.path2xml(in_root+in_), out) for in_, out in mappings]
|
| 89 |
126
|
aaronmk
|
|
| 90 |
131
|
aaronmk
|
in_db = sql.connect(in_db_config)
|
| 91 |
133
|
aaronmk
|
in_pkeys = {}
|
| 92 |
162
|
aaronmk
|
for row in sql.rows(db_xml.get(in_db, in_root_xml, in_pkeys,
|
| 93 |
|
|
limit)):
|
| 94 |
135
|
aaronmk
|
row_id, = row
|
| 95 |
|
|
row_id = str(row_id)
|
| 96 |
|
|
|
| 97 |
133
|
aaronmk
|
def put_col(path, value):
|
| 98 |
141
|
aaronmk
|
xpath.put_obj(root, path, row_id, has_types, value)
|
| 99 |
133
|
aaronmk
|
for value, out in metadata: put_col(out, value)
|
| 100 |
|
|
for in_, out in mappings:
|
| 101 |
141
|
aaronmk
|
in_ = in_.cloneNode(True) # don't modify orig value!
|
| 102 |
161
|
aaronmk
|
xml_dom.set_id(xpath.get(in_, in_root), row_id)
|
| 103 |
140
|
aaronmk
|
value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys))
|
| 104 |
|
|
if value != None: put_col(out, str(value))
|
| 105 |
117
|
aaronmk
|
in_db.close()
|
| 106 |
161
|
aaronmk
|
elif in_is_xml:
|
| 107 |
|
|
row = xpath.get(doc0.documentElement, in_root)
|
| 108 |
|
|
for row_idx, row in enumerate(xml_dom.NodeElemIter(row.parentNode)):
|
| 109 |
|
|
if not (limit == None or row_idx < limit): break
|
| 110 |
|
|
row_id = str(row_idx)
|
| 111 |
|
|
|
| 112 |
|
|
def put_col(path, value):
|
| 113 |
|
|
xpath.put_obj(root, path, row_id, has_types, value)
|
| 114 |
|
|
for value, out in metadata: put_col(out, value)
|
| 115 |
|
|
for in_, out in mappings:
|
| 116 |
|
|
node = xpath.get(row, in_)
|
| 117 |
|
|
if node != None: put_col(out, xml_dom.value(node))
|
| 118 |
56
|
aaronmk
|
else: # input is CSV
|
| 119 |
133
|
aaronmk
|
map_ = dict(mappings)
|
| 120 |
59
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
| 121 |
84
|
aaronmk
|
cols = reader.next()
|
| 122 |
162
|
aaronmk
|
col_idxs = dict([(value, idx) for idx, value in enumerate(cols)])
|
| 123 |
|
|
mappings = [(col_idxs.get(in_, None), out) for in_, out in mappings]
|
| 124 |
|
|
mappings = filter(lambda item: item[0] != None, mappings)
|
| 125 |
|
|
|
| 126 |
83
|
aaronmk
|
for row_idx, row in enumerate(reader):
|
| 127 |
146
|
aaronmk
|
if not (limit == None or row_idx < limit): break
|
| 128 |
56
|
aaronmk
|
row_id = str(row_idx)
|
| 129 |
135
|
aaronmk
|
|
| 130 |
84
|
aaronmk
|
def put_col(path, value):
|
| 131 |
141
|
aaronmk
|
xpath.put_obj(root, path, row_id, has_types, value)
|
| 132 |
84
|
aaronmk
|
for value, out in metadata: put_col(out, value)
|
| 133 |
162
|
aaronmk
|
for in_, out in mappings:
|
| 134 |
|
|
if row[in_] != '': put_col(out, row[in_])
|
| 135 |
142
|
aaronmk
|
xml_func.process(root)
|
| 136 |
116
|
aaronmk
|
else: doc1 = doc0
|
| 137 |
53
|
aaronmk
|
|
| 138 |
|
|
# Output XML tree
|
| 139 |
130
|
aaronmk
|
if out_is_db:
|
| 140 |
53
|
aaronmk
|
from psycopg2.extensions import ISOLATION_LEVEL_SERIALIZABLE
|
| 141 |
|
|
import db_xml
|
| 142 |
|
|
|
| 143 |
133
|
aaronmk
|
out_db = sql.connect(out_db_config)
|
| 144 |
|
|
out_db.set_isolation_level(ISOLATION_LEVEL_SERIALIZABLE)
|
| 145 |
53
|
aaronmk
|
try:
|
| 146 |
|
|
row_ct_ref = [0]
|
| 147 |
133
|
aaronmk
|
db_xml.xml2db(out_db, doc1.documentElement, row_ct_ref)
|
| 148 |
53
|
aaronmk
|
print 'Inserted '+str(row_ct_ref[0])+' rows'
|
| 149 |
133
|
aaronmk
|
if commit: out_db.commit()
|
| 150 |
53
|
aaronmk
|
finally:
|
| 151 |
133
|
aaronmk
|
out_db.rollback()
|
| 152 |
|
|
out_db.close()
|
| 153 |
135
|
aaronmk
|
else: xml_dom.writexml(sys.stdout, doc1) # output is XML
|
| 154 |
53
|
aaronmk
|
|
| 155 |
133
|
aaronmk
|
try: main()
|
| 156 |
|
|
except SyntaxException, ex: raise SystemExit(str(ex))
|
