| 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 |
39
|
aaronmk
|
def obj(node, parent_id=None):
|
| 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 |
39
|
aaronmk
|
elif is_ptr(child_name): row[child_name] = obj(ptr_target(child))
|
| 55 |
13
|
aaronmk
|
else: children.append(child)
|
| 56 |
16
|
aaronmk
|
try: del row[pkey_]
|
| 57 |
|
|
except KeyError: pass
|
| 58 |
13
|
aaronmk
|
|
| 59 |
|
|
# Add fkey to parent
|
| 60 |
39
|
aaronmk
|
if parent_id != None: row[pkey(name_of(node.parentNode))] = parent_id
|
| 61 |
13
|
aaronmk
|
|
| 62 |
|
|
# Insert node
|
| 63 |
|
|
for try_num in range(2):
|
| 64 |
|
|
try:
|
| 65 |
39
|
aaronmk
|
id_ = db_util.insert_or_get(db, table, row, pkey_, row_ct_ref)
|
| 66 |
|
|
xml_util.set_id(node, id_)
|
| 67 |
13
|
aaronmk
|
break
|
| 68 |
|
|
except db_util.NullValueException, ex:
|
| 69 |
|
|
if try_num > 0: raise # exception still raised after retry
|
| 70 |
14
|
aaronmk
|
# Search for required column in ancestors and their children
|
| 71 |
13
|
aaronmk
|
target = find_by_name(node, ptr_type(ex.col))
|
| 72 |
|
|
if target == None: raise
|
| 73 |
|
|
row[ex.col] = xml_util.get_id(target)
|
| 74 |
|
|
|
| 75 |
|
|
# Insert children with fkeys to parent
|
| 76 |
39
|
aaronmk
|
for child in children: obj(child, id_)
|
| 77 |
|
|
|
| 78 |
|
|
return id_
|
| 79 |
13
|
aaronmk
|
|
| 80 |
|
|
row_ct_ref = [0]
|
| 81 |
15
|
aaronmk
|
main(node)
|
| 82 |
13
|
aaronmk
|
return row_ct_ref[0]
|
