| 1 |
13
|
aaronmk
|
# XML-database conversion
|
| 2 |
|
|
|
| 3 |
16
|
aaronmk
|
import re
|
| 4 |
294
|
aaronmk
|
import traceback
|
| 5 |
13
|
aaronmk
|
from xml.dom import Node
|
| 6 |
|
|
|
| 7 |
46
|
aaronmk
|
import sql
|
| 8 |
84
|
aaronmk
|
import strings
|
| 9 |
133
|
aaronmk
|
import util
|
| 10 |
46
|
aaronmk
|
import xml_dom
|
| 11 |
13
|
aaronmk
|
|
| 12 |
139
|
aaronmk
|
def name_of(node): return re.sub(r'^.*\.', r'', node.tagName)
|
| 13 |
16
|
aaronmk
|
|
| 14 |
13
|
aaronmk
|
ptr_suffix = '_id'
|
| 15 |
|
|
|
| 16 |
138
|
aaronmk
|
def is_ptr(node_name): return node_name.lower().endswith(ptr_suffix)
|
| 17 |
13
|
aaronmk
|
|
| 18 |
172
|
aaronmk
|
def ptr_type_guess(node_name):
|
| 19 |
13
|
aaronmk
|
assert is_ptr(node_name)
|
| 20 |
|
|
return node_name[:-len(ptr_suffix)]
|
| 21 |
|
|
|
| 22 |
|
|
def ptr_target(node):
|
| 23 |
|
|
assert is_ptr(name_of(node))
|
| 24 |
46
|
aaronmk
|
return xml_dom.first_elem(node)
|
| 25 |
13
|
aaronmk
|
|
| 26 |
|
|
def find_by_name(node, name):
|
| 27 |
46
|
aaronmk
|
for parent in xml_dom.NodeParentIter(node):
|
| 28 |
13
|
aaronmk
|
if name_of(parent) == name: return parent
|
| 29 |
|
|
else:
|
| 30 |
46
|
aaronmk
|
for child in xml_dom.NodeElemIter(parent):
|
| 31 |
16
|
aaronmk
|
child_name = name_of(child)
|
| 32 |
13
|
aaronmk
|
if is_ptr(child_name):
|
| 33 |
172
|
aaronmk
|
target = ptr_target(child)
|
| 34 |
|
|
if target.tagName == name: return target
|
| 35 |
13
|
aaronmk
|
elif child_name == name: return child
|
| 36 |
|
|
return None
|
| 37 |
|
|
|
| 38 |
146
|
aaronmk
|
def get(db, node, pkeys=None, limit=None):
|
| 39 |
126
|
aaronmk
|
if pkeys == None: pkeys = {}
|
| 40 |
|
|
def pkey(table): return sql.pkey(db, pkeys, table)
|
| 41 |
|
|
|
| 42 |
141
|
aaronmk
|
node = node.firstChild
|
| 43 |
135
|
aaronmk
|
table = name_of(node)
|
| 44 |
|
|
pkey_ = pkey(table)
|
| 45 |
|
|
|
| 46 |
|
|
fields = []
|
| 47 |
|
|
conds = {}
|
| 48 |
133
|
aaronmk
|
for child in xml_dom.NodeElemIter(node):
|
| 49 |
135
|
aaronmk
|
child_name = name_of(child)
|
| 50 |
|
|
if xml_dom.is_empty(child): fields.append(child_name)
|
| 51 |
|
|
elif xml_dom.is_text(child): conds[child_name] = xml_dom.value(child)
|
| 52 |
|
|
else: raise Exception('Joins not supported yet')
|
| 53 |
|
|
id_ = xml_dom.get_id(node)
|
| 54 |
|
|
if id_ != '': conds[pkey(table)] = id_ # replace any existing pkey value
|
| 55 |
|
|
if fields == []: fields.append(pkey_)
|
| 56 |
133
|
aaronmk
|
|
| 57 |
146
|
aaronmk
|
return sql.select(db, table, fields, conds, limit)
|
| 58 |
126
|
aaronmk
|
|
| 59 |
48
|
aaronmk
|
def put(db, node, store_ids=False, row_ct_ref=None, pkeys=None, parent_id=None):
|
| 60 |
138
|
aaronmk
|
'''store_ids enables searching the tree for missing fields'''
|
| 61 |
40
|
aaronmk
|
if pkeys == None: pkeys = {}
|
| 62 |
126
|
aaronmk
|
def pkey(table): return sql.pkey(db, pkeys, table)
|
| 63 |
15
|
aaronmk
|
|
| 64 |
48
|
aaronmk
|
table = name_of(node)
|
| 65 |
|
|
pkey_ = pkey(table)
|
| 66 |
|
|
row = {}
|
| 67 |
|
|
children = []
|
| 68 |
13
|
aaronmk
|
|
| 69 |
48
|
aaronmk
|
# Divide children into fields and children with fkeys to parent
|
| 70 |
|
|
for child in xml_dom.NodeElemIter(node):
|
| 71 |
|
|
child_name = name_of(child)
|
| 72 |
84
|
aaronmk
|
if xml_dom.is_text(child):
|
| 73 |
|
|
row[child_name] = strings.to_unicode(xml_dom.value(child))
|
| 74 |
48
|
aaronmk
|
elif is_ptr(child_name): row[child_name] = put(db, ptr_target(child),
|
| 75 |
|
|
store_ids, row_ct_ref, pkeys)
|
| 76 |
|
|
else: children.append(child)
|
| 77 |
|
|
try: del row[pkey_]
|
| 78 |
|
|
except KeyError: pass
|
| 79 |
|
|
|
| 80 |
|
|
# Add fkey to parent
|
| 81 |
59
|
aaronmk
|
if parent_id != None:
|
| 82 |
|
|
parent_ptr = node.getAttribute('fkey')
|
| 83 |
|
|
if parent_ptr == '': parent_ptr = pkey(name_of(node.parentNode))
|
| 84 |
|
|
row[parent_ptr] = parent_id
|
| 85 |
48
|
aaronmk
|
|
| 86 |
|
|
# Insert node
|
| 87 |
70
|
aaronmk
|
for try_num in xrange(2):
|
| 88 |
48
|
aaronmk
|
try:
|
| 89 |
|
|
id_ = sql.get(db, table, row, pkey_, True, row_ct_ref)
|
| 90 |
|
|
if store_ids: xml_dom.set_id(node, id_)
|
| 91 |
|
|
break
|
| 92 |
300
|
aaronmk
|
except sql.NullValueException, e:
|
| 93 |
48
|
aaronmk
|
if try_num > 0: raise # exception still raised after retry
|
| 94 |
300
|
aaronmk
|
if store_ids and is_ptr(e.col):
|
| 95 |
73
|
aaronmk
|
# Search for required column in ancestors and their children
|
| 96 |
300
|
aaronmk
|
target = find_by_name(node, ptr_type_guess(e.col))
|
| 97 |
73
|
aaronmk
|
if target == None: raise
|
| 98 |
300
|
aaronmk
|
row[e.col] = xml_dom.get_id(target)
|
| 99 |
73
|
aaronmk
|
else: raise
|
| 100 |
48
|
aaronmk
|
|
| 101 |
|
|
# Insert children with fkeys to parent
|
| 102 |
|
|
for child in children: put(db, child, store_ids, row_ct_ref, pkeys, id_)
|
| 103 |
|
|
|
| 104 |
|
|
return id_
|
| 105 |
40
|
aaronmk
|
|
| 106 |
295
|
aaronmk
|
def xml2db(db, node, commit=False, row_ct_ref=None):
|
| 107 |
48
|
aaronmk
|
iter_ = xml_dom.NodeElemIter(node)
|
| 108 |
133
|
aaronmk
|
util.skip(iter_, xml_dom.is_text) # skip metadata
|
| 109 |
294
|
aaronmk
|
for child in iter_:
|
| 110 |
295
|
aaronmk
|
try:
|
| 111 |
|
|
put(db, child, False, row_ct_ref)
|
| 112 |
|
|
if commit: db.commit()
|
| 113 |
|
|
except Exception:
|
| 114 |
|
|
db.rollback()
|
| 115 |
|
|
traceback.print_exc()
|
