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 |
294
|
aaronmk
|
import traceback
|
9 |
299
|
aaronmk
|
import xml.dom.minidom as minidom
|
10 |
53
|
aaronmk
|
|
11 |
266
|
aaronmk
|
sys.path.append(os.path.dirname(__file__)+"/../lib")
|
12 |
53
|
aaronmk
|
|
13 |
64
|
aaronmk
|
import opts
|
14 |
281
|
aaronmk
|
import Parser
|
15 |
131
|
aaronmk
|
import sql
|
16 |
133
|
aaronmk
|
import xml_dom
|
17 |
86
|
aaronmk
|
import xml_func
|
18 |
53
|
aaronmk
|
|
19 |
84
|
aaronmk
|
def metadata_value(name):
|
20 |
164
|
aaronmk
|
if type(name) == str and name.startswith(':'): return name[1:]
|
21 |
84
|
aaronmk
|
else: return None
|
22 |
|
|
|
23 |
53
|
aaronmk
|
def main():
|
24 |
131
|
aaronmk
|
env_names = []
|
25 |
|
|
def usage_err():
|
26 |
|
|
raise SystemExit('Usage: '+opts.env_usage(env_names, True)
|
27 |
|
|
+' [commit=1] '+sys.argv[0]+' [map_path] [<input] [>output]')
|
28 |
146
|
aaronmk
|
limit = opts.get_env_var('n', None, env_names)
|
29 |
|
|
if limit != None: limit = int(limit)
|
30 |
135
|
aaronmk
|
commit = opts.env_flag('commit')
|
31 |
131
|
aaronmk
|
|
32 |
53
|
aaronmk
|
# Get db config from env vars
|
33 |
131
|
aaronmk
|
db_config_names = ['engine', 'host', 'user', 'password', 'database']
|
34 |
53
|
aaronmk
|
def get_db_config(prefix):
|
35 |
64
|
aaronmk
|
return opts.get_env_vars(db_config_names, prefix, env_names)
|
36 |
67
|
aaronmk
|
in_db_config = get_db_config('in')
|
37 |
|
|
out_db_config = get_db_config('out')
|
38 |
131
|
aaronmk
|
in_is_db = 'engine' in in_db_config
|
39 |
|
|
out_is_db = 'engine' in out_db_config
|
40 |
53
|
aaronmk
|
|
41 |
|
|
# Parse args
|
42 |
73
|
aaronmk
|
map_path = None
|
43 |
67
|
aaronmk
|
try: _prog_name, map_path = sys.argv
|
44 |
53
|
aaronmk
|
except ValueError:
|
45 |
135
|
aaronmk
|
if in_is_db: usage_err()
|
46 |
53
|
aaronmk
|
|
47 |
57
|
aaronmk
|
# Load map header
|
48 |
130
|
aaronmk
|
in_is_xpaths = True
|
49 |
73
|
aaronmk
|
if map_path != None:
|
50 |
56
|
aaronmk
|
import copy
|
51 |
|
|
import csv
|
52 |
|
|
|
53 |
53
|
aaronmk
|
import xpath
|
54 |
|
|
|
55 |
133
|
aaronmk
|
metadata = []
|
56 |
73
|
aaronmk
|
mappings = []
|
57 |
|
|
stream = open(map_path, 'rb')
|
58 |
|
|
reader = csv.reader(stream)
|
59 |
161
|
aaronmk
|
in_label, out_label = reader.next()[:2]
|
60 |
61
|
aaronmk
|
def split_col_name(name):
|
61 |
72
|
aaronmk
|
name, sep, root = name.partition(':')
|
62 |
|
|
return name, sep != '', root
|
63 |
161
|
aaronmk
|
in_label, in_is_xpaths, in_root = split_col_name(in_label)
|
64 |
|
|
out_label, out_is_xpaths, out_root = split_col_name(out_label)
|
65 |
133
|
aaronmk
|
assert out_is_xpaths # CSV output not supported yet
|
66 |
161
|
aaronmk
|
has_types = out_root.startswith('/*s/') # outer elements are types
|
67 |
73
|
aaronmk
|
for row in reader:
|
68 |
|
|
in_, out = row[:2]
|
69 |
|
|
if out != '':
|
70 |
164
|
aaronmk
|
if out_is_xpaths: out = out_root+out
|
71 |
|
|
mappings.append((in_, out))
|
72 |
73
|
aaronmk
|
stream.close()
|
73 |
130
|
aaronmk
|
in_is_xml = in_is_xpaths and not in_is_db
|
74 |
56
|
aaronmk
|
|
75 |
299
|
aaronmk
|
if in_is_xml: doc0 = minidom.parse(sys.stdin)
|
76 |
294
|
aaronmk
|
|
77 |
309
|
aaronmk
|
def process_input(use_row):
|
78 |
|
|
'''Inputs datasource to XML tree, mapping if needed'''
|
79 |
161
|
aaronmk
|
doc1 = xml_dom.create_doc(out_label)
|
80 |
141
|
aaronmk
|
root = doc1.documentElement
|
81 |
297
|
aaronmk
|
|
82 |
309
|
aaronmk
|
def process_rows(get_value, rows):
|
83 |
297
|
aaronmk
|
'''Processes input values
|
84 |
|
|
@param get_value f(in_, row):str
|
85 |
|
|
'''
|
86 |
|
|
for i, row in enumerate(rows):
|
87 |
|
|
if not (limit == None or i < limit): break
|
88 |
|
|
row_id = str(i)
|
89 |
|
|
for in_, out in mappings:
|
90 |
|
|
value = metadata_value(in_)
|
91 |
|
|
if value == None: value = get_value(in_, row)
|
92 |
|
|
if value != None:
|
93 |
|
|
try: xpath.put_obj(root, out, row_id, has_types, value)
|
94 |
|
|
except Exception: traceback.print_exc()
|
95 |
|
|
|
96 |
309
|
aaronmk
|
if map_path == None: return doc0
|
97 |
|
|
elif in_is_db:
|
98 |
130
|
aaronmk
|
assert in_is_xpaths
|
99 |
126
|
aaronmk
|
|
100 |
117
|
aaronmk
|
import db_xml
|
101 |
|
|
|
102 |
161
|
aaronmk
|
in_root_xml = xpath.path2xml(in_root)
|
103 |
164
|
aaronmk
|
for i, mapping in enumerate(mappings):
|
104 |
|
|
in_, out = mapping
|
105 |
|
|
if metadata_value(in_) == None:
|
106 |
168
|
aaronmk
|
mappings[i] = (xpath.path2xml(in_root+'/'+in_), out)
|
107 |
126
|
aaronmk
|
|
108 |
131
|
aaronmk
|
in_db = sql.connect(in_db_config)
|
109 |
133
|
aaronmk
|
in_pkeys = {}
|
110 |
297
|
aaronmk
|
def get_value(in_, row):
|
111 |
167
|
aaronmk
|
pkey, = row
|
112 |
297
|
aaronmk
|
in_ = in_.cloneNode(True) # don't modify orig value!
|
113 |
|
|
xml_dom.set_id(xpath.get(in_, in_root), pkey)
|
114 |
|
|
value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys))
|
115 |
|
|
if value != None: return str(value)
|
116 |
|
|
else: return None
|
117 |
309
|
aaronmk
|
process_rows(get_value, sql.rows(db_xml.get(in_db, in_root_xml,
|
118 |
|
|
in_pkeys, limit)))
|
119 |
117
|
aaronmk
|
in_db.close()
|
120 |
161
|
aaronmk
|
elif in_is_xml:
|
121 |
297
|
aaronmk
|
def get_value(in_, row):
|
122 |
|
|
node = xpath.get(row, in_)
|
123 |
|
|
if node != None: return xml_dom.value(node)
|
124 |
|
|
else: return None
|
125 |
|
|
row0 = xpath.get(doc0.documentElement, in_root)
|
126 |
309
|
aaronmk
|
process_rows(get_value, xml_dom.NodeElemIter(row0.parentNode))
|
127 |
56
|
aaronmk
|
else: # input is CSV
|
128 |
133
|
aaronmk
|
map_ = dict(mappings)
|
129 |
59
|
aaronmk
|
reader = csv.reader(sys.stdin)
|
130 |
84
|
aaronmk
|
cols = reader.next()
|
131 |
162
|
aaronmk
|
col_idxs = dict([(value, idx) for idx, value in enumerate(cols)])
|
132 |
164
|
aaronmk
|
for i, mapping in enumerate(mappings):
|
133 |
|
|
in_, out = mapping
|
134 |
|
|
if metadata_value(in_) == None:
|
135 |
|
|
try: mappings[i] = (col_idxs[in_], out)
|
136 |
|
|
except KeyError: pass
|
137 |
162
|
aaronmk
|
|
138 |
297
|
aaronmk
|
def get_value(in_, row):
|
139 |
|
|
value = row[in_]
|
140 |
|
|
if value != '': return value
|
141 |
|
|
else: return None
|
142 |
309
|
aaronmk
|
process_rows(get_value, reader)
|
143 |
142
|
aaronmk
|
xml_func.process(root)
|
144 |
294
|
aaronmk
|
return doc1
|
145 |
53
|
aaronmk
|
|
146 |
|
|
# Output XML tree
|
147 |
130
|
aaronmk
|
if out_is_db:
|
148 |
53
|
aaronmk
|
from psycopg2.extensions import ISOLATION_LEVEL_SERIALIZABLE
|
149 |
|
|
import db_xml
|
150 |
|
|
|
151 |
133
|
aaronmk
|
out_db = sql.connect(out_db_config)
|
152 |
|
|
out_db.set_isolation_level(ISOLATION_LEVEL_SERIALIZABLE)
|
153 |
53
|
aaronmk
|
try:
|
154 |
|
|
row_ct_ref = [0]
|
155 |
295
|
aaronmk
|
def use_row(root): db_xml.xml2db(out_db, root, commit, row_ct_ref)
|
156 |
309
|
aaronmk
|
db_xml.xml2db(out_db, process_input(use_row).documentElement,
|
157 |
|
|
commit, row_ct_ref)
|
158 |
53
|
aaronmk
|
print 'Inserted '+str(row_ct_ref[0])+' rows'
|
159 |
|
|
finally:
|
160 |
133
|
aaronmk
|
out_db.rollback()
|
161 |
|
|
out_db.close()
|
162 |
299
|
aaronmk
|
else: # output is XML
|
163 |
|
|
output = sys.stdout
|
164 |
307
|
aaronmk
|
config = xml_dom.prettyxml_config
|
165 |
308
|
aaronmk
|
doc = xml_dom.create_doc(out_label)
|
166 |
|
|
doc.write_opening(output, **config)
|
167 |
|
|
def use_row(root): root.writexml(output, indent=1, **config)
|
168 |
309
|
aaronmk
|
map(use_row, process_input(use_row).firstChild)
|
169 |
308
|
aaronmk
|
doc.write_closing(output, **config)
|
170 |
53
|
aaronmk
|
|
171 |
133
|
aaronmk
|
try: main()
|
172 |
294
|
aaronmk
|
except Parser.SyntaxException, e: raise SystemExit(str(e))
|