Project

General

Profile

1
# XML-database conversion
2

    
3
import re
4
import traceback
5
from xml.dom import Node
6

    
7
import exc
8
import sql
9
import strings
10
import util
11
import xml_dom
12

    
13
def name_of(node): return re.sub(r'^.*\.', r'', node.tagName)
14

    
15
ptr_suffix = '_id'
16

    
17
def is_ptr(node_name): return node_name.lower().endswith(ptr_suffix)
18

    
19
def ptr_type_guess(node_name):
20
    assert is_ptr(node_name)
21
    return node_name[:-len(ptr_suffix)]
22

    
23
def ptr_target(node):
24
    assert is_ptr(name_of(node))
25
    return xml_dom.first_elem(node)
26

    
27
def find_by_name(node, name):
28
    for parent in xml_dom.NodeParentIter(node):
29
        if name_of(parent) == name: return parent
30
        else:
31
            for child in xml_dom.NodeElemIter(parent):
32
                child_name = name_of(child)
33
                if is_ptr(child_name):
34
                    target = ptr_target(child)
35
                    if target.tagName == name: return target
36
                elif child_name == name: return child
37
    return None
38

    
39
def get(db, node, pkeys=None, limit=None):
40
    if pkeys == None: pkeys = {}
41
    def pkey(table): return sql.pkey(db, pkeys, table)
42
    
43
    node = node.firstChild
44
    table = name_of(node)
45
    pkey_ = pkey(table)
46
    
47
    fields = []
48
    conds = {}
49
    for child in xml_dom.NodeElemIter(node):
50
        child_name = name_of(child)
51
        if xml_dom.is_empty(child): fields.append(child_name)
52
        elif xml_dom.is_text(child): conds[child_name] = xml_dom.value(child)
53
        else: raise Exception('Joins not supported yet')
54
    id_ = xml_dom.get_id(node)
55
    if id_ != '': conds[pkey(table)] = id_ # replace any existing pkey value
56
    if fields == []: fields.append(pkey_)
57
    
58
    return sql.select(db, table, fields, conds, limit)
59

    
60
def put(db, node, store_ids=False, row_ct_ref=None, pkeys=None, parent_id=None):
61
    '''store_ids enables searching the tree for missing fields'''
62
    if pkeys == None: pkeys = {}
63
    def pkey(table): return sql.pkey(db, pkeys, table)
64
    
65
    def on_error(e):
66
        exc.add_msg(e, 'node:\n'+str(node))
67
        raise
68
    
69
    table = name_of(node)
70
    try: pkey_ = pkey(table)
71
    except sql.DatabaseErrors, e: on_error(e)
72
    row = {}
73
    children = []
74
    
75
    # Divide children into fields and children with fkeys to parent
76
    for child in xml_dom.NodeElemIter(node):
77
        child_name = name_of(child)
78
        if xml_dom.is_text(child):
79
            row[child_name] = strings.to_unicode(xml_dom.value(child))
80
        elif not xml_dom.has_elems(child): pass # ignore empty field
81
        elif is_ptr(child_name):
82
            row[child_name] = put(db, ptr_target(child), store_ids,
83
                row_ct_ref, pkeys)
84
        else: children.append(child)
85
    try: del row[pkey_]
86
    except KeyError: pass
87
    
88
    # Add fkey to parent
89
    if parent_id != None:
90
        parent_ptr = node.getAttribute('fkey')
91
        if parent_ptr == '': parent_ptr = pkey(name_of(node.parentNode))
92
        row[parent_ptr] = parent_id
93
    
94
    # Insert node
95
    try:
96
        for try_num in xrange(2):
97
            try:
98
                id_ = sql.get(db, table, row, pkey_, True, row_ct_ref)
99
                if store_ids: xml_dom.set_id(node, id_)
100
                break
101
            except sql.NullValueException, e:
102
                if try_num > 0: raise # exception still raised after retry
103
                if store_ids and is_ptr(e.col):
104
                    # Search for required column in ancestors and their children
105
                    target = find_by_name(node, ptr_type_guess(e.col))
106
                    if target == None: raise
107
                    row[e.col] = xml_dom.get_id(target)
108
                else: raise
109
    except sql.DatabaseErrors, e: on_error(e)
110
    
111
    # Insert children with fkeys to parent
112
    for child in children: put(db, child, store_ids, row_ct_ref, pkeys, id_)
113
    
114
    return id_
(2-2/11)