1 |
13
|
aaronmk
|
# XML-database conversion
|
2 |
|
|
|
3 |
16
|
aaronmk
|
import re
|
4 |
13
|
aaronmk
|
from xml.dom import Node
|
5 |
|
|
|
6 |
|
|
import db_util
|
7 |
|
|
import xml_util
|
8 |
|
|
|
9 |
16
|
aaronmk
|
def name_of(node): return re.sub(r'^.*\.', r'', xml_util.name_of(node))
|
10 |
|
|
|
11 |
13
|
aaronmk
|
ptr_suffix = '_id'
|
12 |
|
|
|
13 |
|
|
def is_ptr(node_name): return node_name.endswith(ptr_suffix)
|
14 |
|
|
|
15 |
|
|
def ptr_type(node_name):
|
16 |
|
|
assert is_ptr(node_name)
|
17 |
|
|
return node_name[:-len(ptr_suffix)]
|
18 |
|
|
|
19 |
|
|
def ptr_target(node):
|
20 |
|
|
assert is_ptr(name_of(node))
|
21 |
|
|
return xml_util.first_elem(node)
|
22 |
|
|
|
23 |
|
|
def find_by_name(node, name):
|
24 |
|
|
for parent in xml_util.NodeParentIter(node):
|
25 |
|
|
if name_of(parent) == name: return parent
|
26 |
|
|
else:
|
27 |
|
|
for child in xml_util.NodeElemIter(parent):
|
28 |
16
|
aaronmk
|
child_name = name_of(child)
|
29 |
13
|
aaronmk
|
if is_ptr(child_name):
|
30 |
|
|
if ptr_type(child_name) == name: return ptr_target(child)
|
31 |
|
|
elif child_name == name: return child
|
32 |
|
|
return None
|
33 |
|
|
|
34 |
|
|
def xml2db(node, db):
|
35 |
15
|
aaronmk
|
pkeys = {}
|
36 |
|
|
def pkey(table):
|
37 |
|
|
if table not in pkeys: pkeys[table] = db_util.pkey(db, table)
|
38 |
|
|
return pkeys[table]
|
39 |
|
|
|
40 |
|
|
def main(node):
|
41 |
13
|
aaronmk
|
for child in xml_util.NodeElemIter(node):
|
42 |
15
|
aaronmk
|
if not xml_util.is_text(child): obj(child) # not XML metadata
|
43 |
13
|
aaronmk
|
|
44 |
15
|
aaronmk
|
def obj(node):
|
45 |
13
|
aaronmk
|
table = name_of(node)
|
46 |
15
|
aaronmk
|
pkey_ = pkey(table)
|
47 |
13
|
aaronmk
|
row = {}
|
48 |
|
|
children = []
|
49 |
|
|
|
50 |
|
|
# Divide children into fields and children with fkeys to parent
|
51 |
16
|
aaronmk
|
for child in xml_util.NodeElemIter(node):
|
52 |
|
|
child_name = name_of(child)
|
53 |
|
|
if xml_util.is_text(child): row[child_name] = xml_util.value(child)
|
54 |
|
|
elif is_ptr(child_name):
|
55 |
|
|
child = ptr_target(child)
|
56 |
|
|
obj(child)
|
57 |
|
|
row[child_name] = xml_util.get_id(child)
|
58 |
13
|
aaronmk
|
else: children.append(child)
|
59 |
16
|
aaronmk
|
try: del row[pkey_]
|
60 |
|
|
except KeyError: pass
|
61 |
13
|
aaronmk
|
|
62 |
|
|
# Add fkey to parent
|
63 |
|
|
parent_id = xml_util.get_id(node.parentNode)
|
64 |
15
|
aaronmk
|
if parent_id != '': row[pkey(name_of(node.parentNode))] = parent_id
|
65 |
13
|
aaronmk
|
|
66 |
|
|
# Insert node
|
67 |
|
|
for try_num in range(2):
|
68 |
|
|
try:
|
69 |
|
|
xml_util.set_id(node, db_util.insert_or_get(db, table, row,
|
70 |
15
|
aaronmk
|
pkey_, row_ct_ref))
|
71 |
13
|
aaronmk
|
break
|
72 |
|
|
except db_util.NullValueException, ex:
|
73 |
|
|
if try_num > 0: raise # exception still raised after retry
|
74 |
14
|
aaronmk
|
# Search for required column in ancestors and their children
|
75 |
13
|
aaronmk
|
target = find_by_name(node, ptr_type(ex.col))
|
76 |
|
|
if target == None: raise
|
77 |
|
|
row[ex.col] = xml_util.get_id(target)
|
78 |
|
|
|
79 |
|
|
# Insert children with fkeys to parent
|
80 |
15
|
aaronmk
|
for child in children: obj(child)
|
81 |
13
|
aaronmk
|
|
82 |
|
|
row_ct_ref = [0]
|
83 |
15
|
aaronmk
|
main(node)
|
84 |
13
|
aaronmk
|
return row_ct_ref[0]
|