1
|
# XML-database conversion
|
2
|
|
3
|
from xml.dom import Node
|
4
|
|
5
|
import db_util
|
6
|
import xml_util
|
7
|
from xml_util import name_of
|
8
|
|
9
|
ptr_suffix = '_id'
|
10
|
|
11
|
def is_ptr(node_name): return node_name.endswith(ptr_suffix)
|
12
|
|
13
|
def ptr_type(node_name):
|
14
|
assert is_ptr(node_name)
|
15
|
return node_name[:-len(ptr_suffix)]
|
16
|
|
17
|
def ptr_target(node):
|
18
|
assert is_ptr(name_of(node))
|
19
|
return xml_util.first_elem(node)
|
20
|
|
21
|
def pkey_name(node_name): return node_name+ptr_suffix
|
22
|
|
23
|
def isfield(node_name): return node_name.find('.') >= 0
|
24
|
|
25
|
def field_name(node_name):
|
26
|
if isfield(node_name): return node_name.partition('.')[2]
|
27
|
else: return node_name
|
28
|
|
29
|
def is_db_export(node_name): return node_name.find('.') >= 1 # has prefix
|
30
|
|
31
|
def find_by_name(node, name):
|
32
|
for parent in xml_util.NodeParentIter(node):
|
33
|
if name_of(parent) == name: return parent
|
34
|
else:
|
35
|
for child in xml_util.NodeElemIter(parent):
|
36
|
child_name = field_name(name_of(child))
|
37
|
if is_ptr(child_name):
|
38
|
if ptr_type(child_name) == name: return ptr_target(child)
|
39
|
elif child_name == name: return child
|
40
|
return None
|
41
|
|
42
|
def xml2db(node, db):
|
43
|
pkeys = {}
|
44
|
def pkey(table):
|
45
|
if table not in pkeys: pkeys[table] = db_util.pkey(db, table)
|
46
|
return pkeys[table]
|
47
|
|
48
|
def main(node):
|
49
|
for child in xml_util.NodeElemIter(node):
|
50
|
if not xml_util.is_text(child): obj(child) # not XML metadata
|
51
|
|
52
|
def obj(node):
|
53
|
table = name_of(node)
|
54
|
db_util.check_name(table)
|
55
|
pkey_ = pkey(table)
|
56
|
row = {}
|
57
|
children = []
|
58
|
iter_ = xml_util.NodeElemIter(node)
|
59
|
|
60
|
# Skip any pkey
|
61
|
child_name = name_of(iter_.curr()) # first child
|
62
|
if is_db_export(child_name) and is_ptr(child_name)\
|
63
|
or field_name(child_name) == pkey: iter_.next()
|
64
|
|
65
|
# Divide children into fields and children with fkeys to parent
|
66
|
for child in iter_:
|
67
|
if isfield(name_of(child)): row.update([field(child)])
|
68
|
else: children.append(child)
|
69
|
|
70
|
# Add fkey to parent
|
71
|
parent_id = xml_util.get_id(node.parentNode)
|
72
|
if parent_id != '': row[pkey(name_of(node.parentNode))] = parent_id
|
73
|
|
74
|
# Insert node
|
75
|
for try_num in range(2):
|
76
|
try:
|
77
|
xml_util.set_id(node, db_util.insert_or_get(db, table, row,
|
78
|
pkey_, row_ct_ref))
|
79
|
break
|
80
|
except db_util.NullValueException, ex:
|
81
|
if try_num > 0: raise # exception still raised after retry
|
82
|
# Search for required column in ancestors and their children
|
83
|
target = find_by_name(node, ptr_type(ex.col))
|
84
|
if target == None: raise
|
85
|
row[ex.col] = xml_util.get_id(target)
|
86
|
|
87
|
# Insert children with fkeys to parent
|
88
|
for child in children: obj(child)
|
89
|
|
90
|
def field(node):
|
91
|
if xml_util.is_text(node): value = xml_util.value(node)
|
92
|
else:
|
93
|
child = xml_util.first_elem(node)
|
94
|
obj(child)
|
95
|
value = xml_util.get_id(child)
|
96
|
return (field_name(name_of(node)), value)
|
97
|
|
98
|
row_ct_ref = [0]
|
99
|
main(node)
|
100
|
return row_ct_ref[0]
|