1 |
13
|
aaronmk
|
# 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 is_field(node_name): return node_name.find('.') >= 0
|
24 |
|
|
|
25 |
|
|
def field_name(node_name):
|
26 |
|
|
if is_field(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 |
|
|
def _main(node):
|
44 |
|
|
for child in xml_util.NodeElemIter(node):
|
45 |
|
|
if not xml_util.is_text(child): _obj(child) # not XML metadata
|
46 |
|
|
|
47 |
|
|
def _obj(node):
|
48 |
|
|
table = name_of(node)
|
49 |
|
|
db_util.check_name(table)
|
50 |
|
|
pkey = pkey_name(table)
|
51 |
|
|
row = {}
|
52 |
|
|
children = []
|
53 |
|
|
iter_ = xml_util.NodeElemIter(node)
|
54 |
|
|
|
55 |
|
|
# Skip any pkey
|
56 |
|
|
child_name = name_of(iter_.curr()) # first child
|
57 |
|
|
if is_db_export(child_name) and is_ptr(child_name)\
|
58 |
|
|
or field_name(child_name) == pkey: iter_.next()
|
59 |
|
|
|
60 |
|
|
# Divide children into fields and children with fkeys to parent
|
61 |
|
|
for child in iter_:
|
62 |
|
|
if is_field(name_of(child)): row.update([_field(child)])
|
63 |
|
|
else: children.append(child)
|
64 |
|
|
|
65 |
|
|
# Add fkey to parent
|
66 |
|
|
parent_id = xml_util.get_id(node.parentNode)
|
67 |
|
|
if parent_id != '': row[pkey_name(name_of(node.parentNode))] = parent_id
|
68 |
|
|
|
69 |
|
|
# Insert node
|
70 |
|
|
for try_num in range(2):
|
71 |
|
|
try:
|
72 |
|
|
xml_util.set_id(node, db_util.insert_or_get(db, table, row,
|
73 |
|
|
pkey, row_ct_ref))
|
74 |
|
|
break
|
75 |
|
|
except db_util.NullValueException, ex:
|
76 |
|
|
if try_num > 0: raise # exception still raised after retry
|
77 |
14
|
aaronmk
|
# Search for required column in ancestors and their children
|
78 |
13
|
aaronmk
|
target = find_by_name(node, ptr_type(ex.col))
|
79 |
|
|
if target == None: raise
|
80 |
|
|
row[ex.col] = xml_util.get_id(target)
|
81 |
|
|
|
82 |
|
|
# Insert children with fkeys to parent
|
83 |
|
|
for child in children: _obj(child)
|
84 |
|
|
|
85 |
|
|
def _field(node):
|
86 |
|
|
if xml_util.is_text(node): value = xml_util.value(node)
|
87 |
|
|
else:
|
88 |
|
|
child = xml_util.first_elem(node)
|
89 |
|
|
_obj(child)
|
90 |
|
|
value = xml_util.get_id(child)
|
91 |
|
|
return (field_name(name_of(node)), value)
|
92 |
|
|
|
93 |
|
|
row_ct_ref = [0]
|
94 |
|
|
_main(node)
|
95 |
|
|
return row_ct_ref[0]
|