Project

General

Profile

1 13 aaronmk
# XML-database conversion
2
3 2418 aaronmk
import copy
4 16 aaronmk
import re
5 13 aaronmk
from xml.dom import Node
6
7 3718 aaronmk
import dicts
8 446 aaronmk
import exc
9 2436 aaronmk
import Parser
10 46 aaronmk
import sql
11 3077 aaronmk
import sql_io
12 2268 aaronmk
import sql_gen
13 84 aaronmk
import strings
14 133 aaronmk
import util
15 46 aaronmk
import xml_dom
16 2113 aaronmk
import xml_func
17 2436 aaronmk
import xpath
18 13 aaronmk
19 139 aaronmk
def name_of(node): return re.sub(r'^.*\.', r'', node.tagName)
20 16 aaronmk
21 13 aaronmk
ptr_suffix = '_id'
22
23 138 aaronmk
def is_ptr(node_name): return node_name.lower().endswith(ptr_suffix)
24 13 aaronmk
25 172 aaronmk
def ptr_type_guess(node_name):
26 13 aaronmk
    assert is_ptr(node_name)
27
    return node_name[:-len(ptr_suffix)]
28
29
def ptr_target(node):
30
    assert is_ptr(name_of(node))
31 2113 aaronmk
    return xml_dom.value_node(node)
32 13 aaronmk
33
def find_by_name(node, name):
34 46 aaronmk
    for parent in xml_dom.NodeParentIter(node):
35 13 aaronmk
        if name_of(parent) == name: return parent
36
        else:
37 46 aaronmk
            for child in xml_dom.NodeElemIter(parent):
38 16 aaronmk
                child_name = name_of(child)
39 13 aaronmk
                if is_ptr(child_name):
40 172 aaronmk
                    target = ptr_target(child)
41
                    if target.tagName == name: return target
42 13 aaronmk
                elif child_name == name: return child
43
    return None
44
45 2039 aaronmk
class ColRef:
46
    '''A reference to a table column'''
47
    def __init__(self, name, idx):
48
        self.name = name
49
        self.idx = idx
50
51
    def __str__(self): return self.name
52
53 4242 aaronmk
input_col_prefix = xml_func.var_name_prefix
54 4241 aaronmk
55 6402 aaronmk
put_special_funcs = set(['_setDefault', '_simplifyPath'])
56 2549 aaronmk
57 5725 aaronmk
no_parent_ids_loc = object() # tells put() there is no parent_ids_loc
58
59
def put(db, node, row_ins_ct_ref=None, on_error=exc.reraise, col_defaults={},
60
    in_table=None, parent_ids_loc=no_parent_ids_loc, next=None):
61 3676 aaronmk
    '''
62
    @param node To use an entire XML document, pass root.firstChild.
63
    '''
64
    if node == None: return None # when no rows, root.firstChild == None
65
66 3685 aaronmk
    def put_(node):
67
        return put(db, node, row_ins_ct_ref, on_error, col_defaults, in_table,
68
            parent_ids_loc, next)
69 3621 aaronmk
70 4491 aaronmk
    def augment_error(e): exc.add_msg(e, 'node:\n'+strings.ustr(node))
71 3646 aaronmk
    def on_error_(e):
72 3677 aaronmk
        augment_error(e)
73 3646 aaronmk
        on_error(e)
74
75 3719 aaronmk
    def wrap_e(e):
76
        augment_error(e)
77
        raise xml_func.SyntaxError(e)
78 3718 aaronmk
79 2551 aaronmk
    is_func = xml_func.is_func(node)
80 2005 aaronmk
    out_table = name_of(node)
81 2432 aaronmk
82
    # Divide children into fields and children with fkeys to parent
83 3718 aaronmk
    row = dicts.OnceOnlyDict()
84 2005 aaronmk
    children = []
85 3718 aaronmk
    try:
86
        for child in xml_dom.NodeElemIter(node):
87
            child_name = name_of(child)
88
            if xml_dom.is_empty(child): row[child_name] = None
89
            elif xml_dom.is_text(child):
90
                row[child_name] = strings.to_unicode(xml_dom.value(child))
91
            else:
92
                child_value = xml_dom.value_node(child)
93 5508 aaronmk
                if ((is_func or is_ptr(child_name)
94
                    or xml_func.is_func(child_value))
95
                    and not xml_func.is_func(child)):
96 3718 aaronmk
                    row[child_name] = child_value
97
                else: children.append(child)
98
    except dicts.KeyExistsError, e: wrap_e(e)
99 2434 aaronmk
100
    # Special handling for structural XML functions
101 6402 aaronmk
    if out_table == '_setDefault':
102 2436 aaronmk
        # Parse args
103 6406 aaronmk
        try: path = row.pop('path')
104 6402 aaronmk
        except KeyError, e: wrap_e(e)
105
106 6406 aaronmk
        col_defaults.update([(k, put_(v)) for k, v in row.iteritems()])
107 6402 aaronmk
        return put_(path)
108
    elif out_table == '_simplifyPath':
109
        # Parse args
110
        try:
111 3650 aaronmk
            next = row['next'] # modifies outer next var used by put_()
112 2434 aaronmk
            path = row['path']
113 2436 aaronmk
        except KeyError, e: wrap_e(e)
114
        try: next = xpath.parse(next)
115
        except Parser.SyntaxError, e: wrap_e(e)
116
        try: next = next[0].name
117
        except IndexError, e: wrap_e(e)
118 2434 aaronmk
119 3650 aaronmk
        return put_(path)
120 2434 aaronmk
121 3687 aaronmk
    is_literals = in_table == None
122 3624 aaronmk
    in_tables = []
123
    no_empty = set()
124 3687 aaronmk
    if not is_literals:
125 3624 aaronmk
        in_tables.append(in_table)
126
        no_empty.add(in_table)
127 2966 aaronmk
128 5388 aaronmk
    def pkey_name(table): return sql.pkey_name(db, table, True)
129 3107 aaronmk
130 2177 aaronmk
    # Add fkey to parent
131 5725 aaronmk
    if parent_ids_loc is not no_parent_ids_loc:
132 3601 aaronmk
        if sql_gen.is_table_col(parent_ids_loc):
133
            no_empty.add(parent_ids_loc.table)
134 2177 aaronmk
        parent_ptr = node.getAttribute('fkey')
135 5388 aaronmk
        if parent_ptr == '': parent_ptr = pkey_name(name_of(node.parentNode))
136 2177 aaronmk
        row[parent_ptr] = parent_ids_loc
137
138 5071 aaronmk
    # Parse input columns
139 3718 aaronmk
    row = row.inner # now allow keys to be overwritten
140 2060 aaronmk
    for out_col, value in row.iteritems():
141 5071 aaronmk
        if (not is_literals and util.is_str(value)
142
            and value.startswith(input_col_prefix)): # value is input column
143
            row[out_col] = sql_gen.Col(strings.remove_prefix(input_col_prefix,
144
                value), in_table)
145
146 5075 aaronmk
    # Optimizations for structural XML functions
147
    if out_table == '_alt': # return first arg if non-NULL
148
        args = row.items()
149
        args.sort()
150
        out_col, value = min(args) # first arg
151
        if xml_dom.is_node(value): row[out_col] = value = put_(value)
152
        if not sql_gen.is_nullable(db, value): return value
153
154 5071 aaronmk
    # Process values
155 5725 aaronmk
    parent_ids_loc = no_parent_ids_loc # applies to this section
156 5071 aaronmk
    for out_col, value in row.iteritems():
157 2432 aaronmk
        # Handle forward pointers
158 3650 aaronmk
        if xml_dom.is_node(value): row[out_col] = value = put_(value)
159 2432 aaronmk
160
        # Translate values
161 5071 aaronmk
        if isinstance(value, sql_gen.Col): # value is table column
162 3687 aaronmk
            assert sql_gen.is_table_col(value)
163 5071 aaronmk
            if value.table is not in_table: in_tables.append(value.table)
164 3428 aaronmk
        else: # value is literal value
165 2323 aaronmk
            row[out_col] = sql_gen.NamedCol(out_col, value)
166 2060 aaronmk
167
    # Insert node
168 3657 aaronmk
    try: pkeys_loc = sql_io.put_table(db, out_table, in_tables, row,
169
        row_ins_ct_ref, next, col_defaults, on_error_)
170
    except Exception, e:
171 3677 aaronmk
        augment_error(e)
172
        raise
173 5382 aaronmk
    if sql_gen.is_table_col(pkeys_loc): no_empty.add(pkeys_loc.table)
174 2060 aaronmk
175 2975 aaronmk
    sql.empty_temp(db, set(in_tables) - no_empty)
176 2966 aaronmk
177 2177 aaronmk
    # Insert children with fkeys to parent
178 2506 aaronmk
    parent_ids_loc = pkeys_loc # applies to this section
179 5725 aaronmk
    for child in children: put_(child)
180 2177 aaronmk
181 2133 aaronmk
    return pkeys_loc
182 3651 aaronmk
183
def get(db, node, limit=None, start=None):
184 5388 aaronmk
    def pkey_name(table): return sql.pkey_name(db, table)
185 3651 aaronmk
186
    node = node.firstChild
187
    table = name_of(node)
188 5388 aaronmk
    pkey_ = pkey_name(table)
189 3651 aaronmk
190
    fields = []
191
    conds = {}
192
    for child in xml_dom.NodeElemIter(node):
193
        child_name = name_of(child)
194
        if xml_dom.is_empty(child): fields.append(child_name)
195
        elif xml_dom.is_text(child): conds[child_name] = xml_dom.value(child)
196
        else: raise Exception('Joins not supported yet')
197
    id_ = xml_dom.get_id(node)
198 5388 aaronmk
    if id_ != None: conds[pkey_name(table)] = id_ # replace any existing value
199 3651 aaronmk
    if fields == []: fields.append(pkey_)
200
201
    return sql.select(db, table, fields, conds, limit, start)
202
203
# Controls when and how put_table() will partition the input table
204 5706 aaronmk
partition_size = 1000000 # rows; must be >= NCBI.nodes size
205 3651 aaronmk
206
def put_table(db, node, in_table, in_row_ct_ref=None, row_ins_ct_ref=None,
207 5007 aaronmk
    limit=None, start=0, on_error=exc.reraise, col_defaults={},
208
    partition_size=partition_size):
209 3651 aaronmk
    '''
210
    @param node The XML tree that transforms the input to the output. Similar to
211
        put()'s node param, but with the input column name prefixed by
212
        input_col_prefix in place of the column value.
213
    @return sql_gen.Col Where the pkeys (from INSERT RETURNING) are made
214
        available
215
    '''
216
    in_table = sql_gen.as_Table(in_table)
217 4437 aaronmk
    sql_io.mk_errors_table(db, in_table)
218 3651 aaronmk
    in_table.set_srcs([in_table], overwrite=False)
219 4491 aaronmk
    db.src = strings.ustr(in_table)
220 3651 aaronmk
221
    db.autoanalyze = True # but don't do this in row-based import
222
    db.autoexplain = True # but don't do this in row-based import
223
224
    # Subset and partition in_table
225
    # OK to do even if table already the right size because it takes <1 sec.
226
    full_in_table = in_table
227 5087 aaronmk
    pkeys_loc = None # used if loop is never executed
228 3651 aaronmk
    total = 0
229
    while limit == None or total < limit:
230
        # Adjust partition size if last partition
231
        this_limit = partition_size
232
        if limit != None: this_limit = min(this_limit, limit - total)
233
234
        # Row # is interally 0-based, but 1-based to the user
235
        db.log_debug('********** Partition: rows '+str(start+1)+'-'
236
            +str(start+this_limit)+' **********', level=1.2)
237
238
        # Subset in_table
239 5022 aaronmk
        in_table = sql_gen.Table(strings.ustr(full_in_table),
240
            srcs=full_in_table.srcs, is_temp=True) # prepend schema to name
241 5023 aaronmk
        sql.copy_table_struct(db, full_in_table, in_table)
242 5385 aaronmk
        try: sql.add_row_num(db, in_table, 'row_num')
243 5024 aaronmk
        except sql.DatabaseErrors: pass # already has pkey
244 5023 aaronmk
        cur = sql.insert_select(db, in_table, None, sql.mk_select(db,
245 5523 aaronmk
            full_in_table, limit=this_limit, start=start))
246 3651 aaronmk
247
        this_ct = cur.rowcount
248
        total += this_ct
249
        start += this_ct # advance start to fetch next set
250
        if this_ct == 0: break # in_table size is multiple of partition_size
251
252 5018 aaronmk
        # Import data
253
        pkeys_loc = put(db, node, row_ins_ct_ref, on_error, col_defaults,
254
            in_table)
255 3651 aaronmk
        if in_row_ct_ref != None: in_row_ct_ref[0] += this_ct
256
257
        sql.empty_temp(db, in_table)
258
259
        if this_ct < partition_size: break # partial partition = last
260
261
        # Work around PostgreSQL's temp table disk space leak
262
        db.reconnect()
263
264
    return pkeys_loc