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 3652 aaronmk
put_special_funcs = set(['_simplifyPath'])
56 2549 aaronmk
57 3660 aaronmk
def put(db, node, row_ins_ct_ref=None, on_error=exc.reraise,
58 3649 aaronmk
    col_defaults={}, in_table=None, parent_ids_loc=None, next=None):
59 3676 aaronmk
    '''
60
    @param node To use an entire XML document, pass root.firstChild.
61
    '''
62
    if node == None: return None # when no rows, root.firstChild == None
63
64 3685 aaronmk
    def put_(node):
65
        return put(db, node, row_ins_ct_ref, on_error, col_defaults, in_table,
66
            parent_ids_loc, next)
67 3621 aaronmk
68 4491 aaronmk
    def augment_error(e): exc.add_msg(e, 'node:\n'+strings.ustr(node))
69 3646 aaronmk
    def on_error_(e):
70 3677 aaronmk
        augment_error(e)
71 3646 aaronmk
        on_error(e)
72
73 3719 aaronmk
    def wrap_e(e):
74
        augment_error(e)
75
        raise xml_func.SyntaxError(e)
76 3718 aaronmk
77 2551 aaronmk
    is_func = xml_func.is_func(node)
78 2005 aaronmk
    out_table = name_of(node)
79 2432 aaronmk
80
    # Divide children into fields and children with fkeys to parent
81 3718 aaronmk
    row = dicts.OnceOnlyDict()
82 2005 aaronmk
    children = []
83 3718 aaronmk
    try:
84
        for child in xml_dom.NodeElemIter(node):
85
            child_name = name_of(child)
86
            if xml_dom.is_empty(child): row[child_name] = None
87
            elif xml_dom.is_text(child):
88
                row[child_name] = strings.to_unicode(xml_dom.value(child))
89
            else:
90
                child_value = xml_dom.value_node(child)
91 5508 aaronmk
                if ((is_func or is_ptr(child_name)
92
                    or xml_func.is_func(child_value))
93
                    and not xml_func.is_func(child)):
94 3718 aaronmk
                    row[child_name] = child_value
95
                else: children.append(child)
96
    except dicts.KeyExistsError, e: wrap_e(e)
97 2434 aaronmk
98
    # Special handling for structural XML functions
99
    if out_table == '_simplifyPath':
100 2436 aaronmk
        # Parse args
101 2434 aaronmk
        try:
102 3650 aaronmk
            next = row['next'] # modifies outer next var used by put_()
103 2434 aaronmk
            path = row['path']
104 2436 aaronmk
        except KeyError, e: wrap_e(e)
105
        try: next = xpath.parse(next)
106
        except Parser.SyntaxError, e: wrap_e(e)
107
        try: next = next[0].name
108
        except IndexError, e: wrap_e(e)
109 2434 aaronmk
110 3650 aaronmk
        return put_(path)
111 2434 aaronmk
112 3687 aaronmk
    is_literals = in_table == None
113 3624 aaronmk
    in_tables = []
114
    no_empty = set()
115 3687 aaronmk
    if not is_literals:
116 3624 aaronmk
        in_tables.append(in_table)
117
        no_empty.add(in_table)
118 2966 aaronmk
119 5388 aaronmk
    def pkey_name(table): return sql.pkey_name(db, table, True)
120 3107 aaronmk
121 2177 aaronmk
    # Add fkey to parent
122
    if parent_ids_loc != None:
123 3601 aaronmk
        if sql_gen.is_table_col(parent_ids_loc):
124
            no_empty.add(parent_ids_loc.table)
125 2177 aaronmk
        parent_ptr = node.getAttribute('fkey')
126 5388 aaronmk
        if parent_ptr == '': parent_ptr = pkey_name(name_of(node.parentNode))
127 2177 aaronmk
        row[parent_ptr] = parent_ids_loc
128
129 5071 aaronmk
    # Parse input columns
130 3718 aaronmk
    row = row.inner # now allow keys to be overwritten
131 2060 aaronmk
    for out_col, value in row.iteritems():
132 5071 aaronmk
        if (not is_literals and util.is_str(value)
133
            and value.startswith(input_col_prefix)): # value is input column
134
            row[out_col] = sql_gen.Col(strings.remove_prefix(input_col_prefix,
135
                value), in_table)
136
137 5075 aaronmk
    # Optimizations for structural XML functions
138
    if out_table == '_alt': # return first arg if non-NULL
139
        args = row.items()
140
        args.sort()
141
        out_col, value = min(args) # first arg
142
        if xml_dom.is_node(value): row[out_col] = value = put_(value)
143
        if not sql_gen.is_nullable(db, value): return value
144
145 5071 aaronmk
    # Process values
146
    parent_ids_loc = None # applies to this section
147
    for out_col, value in row.iteritems():
148 2432 aaronmk
        # Handle forward pointers
149 3650 aaronmk
        if xml_dom.is_node(value): row[out_col] = value = put_(value)
150 2432 aaronmk
151
        # Translate values
152 5071 aaronmk
        if isinstance(value, sql_gen.Col): # value is table column
153 3687 aaronmk
            assert sql_gen.is_table_col(value)
154 5071 aaronmk
            if value.table is not in_table: in_tables.append(value.table)
155 3428 aaronmk
        else: # value is literal value
156 2323 aaronmk
            row[out_col] = sql_gen.NamedCol(out_col, value)
157 2060 aaronmk
158
    # Insert node
159 3657 aaronmk
    try: pkeys_loc = sql_io.put_table(db, out_table, in_tables, row,
160
        row_ins_ct_ref, next, col_defaults, on_error_)
161
    except Exception, e:
162 3677 aaronmk
        augment_error(e)
163
        raise
164 5382 aaronmk
    if sql_gen.is_table_col(pkeys_loc): no_empty.add(pkeys_loc.table)
165 2060 aaronmk
166 2975 aaronmk
    sql.empty_temp(db, set(in_tables) - no_empty)
167 2966 aaronmk
168 2177 aaronmk
    # Insert children with fkeys to parent
169 2506 aaronmk
    parent_ids_loc = pkeys_loc # applies to this section
170 3650 aaronmk
    for child in children: put_(child)
171 2177 aaronmk
172 2133 aaronmk
    return pkeys_loc
173 3651 aaronmk
174
def get(db, node, limit=None, start=None):
175 5388 aaronmk
    def pkey_name(table): return sql.pkey_name(db, table)
176 3651 aaronmk
177
    node = node.firstChild
178
    table = name_of(node)
179 5388 aaronmk
    pkey_ = pkey_name(table)
180 3651 aaronmk
181
    fields = []
182
    conds = {}
183
    for child in xml_dom.NodeElemIter(node):
184
        child_name = name_of(child)
185
        if xml_dom.is_empty(child): fields.append(child_name)
186
        elif xml_dom.is_text(child): conds[child_name] = xml_dom.value(child)
187
        else: raise Exception('Joins not supported yet')
188
    id_ = xml_dom.get_id(node)
189 5388 aaronmk
    if id_ != None: conds[pkey_name(table)] = id_ # replace any existing value
190 3651 aaronmk
    if fields == []: fields.append(pkey_)
191
192
    return sql.select(db, table, fields, conds, limit, start)
193
194
# Controls when and how put_table() will partition the input table
195 5514 aaronmk
partition_size = 1000000 # rows; must be >= NCBI.higher_taxa size
196 3651 aaronmk
197
def put_table(db, node, in_table, in_row_ct_ref=None, row_ins_ct_ref=None,
198 5007 aaronmk
    limit=None, start=0, on_error=exc.reraise, col_defaults={},
199
    partition_size=partition_size):
200 3651 aaronmk
    '''
201
    @param node The XML tree that transforms the input to the output. Similar to
202
        put()'s node param, but with the input column name prefixed by
203
        input_col_prefix in place of the column value.
204
    @return sql_gen.Col Where the pkeys (from INSERT RETURNING) are made
205
        available
206
    '''
207
    in_table = sql_gen.as_Table(in_table)
208 4437 aaronmk
    sql_io.mk_errors_table(db, in_table)
209 3651 aaronmk
    in_table.set_srcs([in_table], overwrite=False)
210 4491 aaronmk
    db.src = strings.ustr(in_table)
211 3651 aaronmk
212
    db.autoanalyze = True # but don't do this in row-based import
213
    db.autoexplain = True # but don't do this in row-based import
214
215
    # Subset and partition in_table
216
    # OK to do even if table already the right size because it takes <1 sec.
217
    full_in_table = in_table
218 5087 aaronmk
    pkeys_loc = None # used if loop is never executed
219 3651 aaronmk
    total = 0
220
    while limit == None or total < limit:
221
        # Adjust partition size if last partition
222
        this_limit = partition_size
223
        if limit != None: this_limit = min(this_limit, limit - total)
224
225
        # Row # is interally 0-based, but 1-based to the user
226
        db.log_debug('********** Partition: rows '+str(start+1)+'-'
227
            +str(start+this_limit)+' **********', level=1.2)
228
229
        # Subset in_table
230 5022 aaronmk
        in_table = sql_gen.Table(strings.ustr(full_in_table),
231
            srcs=full_in_table.srcs, is_temp=True) # prepend schema to name
232 5023 aaronmk
        sql.copy_table_struct(db, full_in_table, in_table)
233 5385 aaronmk
        try: sql.add_row_num(db, in_table, 'row_num')
234 5024 aaronmk
        except sql.DatabaseErrors: pass # already has pkey
235 5023 aaronmk
        cur = sql.insert_select(db, in_table, None, sql.mk_select(db,
236 5523 aaronmk
            full_in_table, limit=this_limit, start=start))
237 3651 aaronmk
238
        this_ct = cur.rowcount
239
        total += this_ct
240
        start += this_ct # advance start to fetch next set
241
        if this_ct == 0: break # in_table size is multiple of partition_size
242
243 5018 aaronmk
        # Import data
244
        pkeys_loc = put(db, node, row_ins_ct_ref, on_error, col_defaults,
245
            in_table)
246 3651 aaronmk
        if in_row_ct_ref != None: in_row_ct_ref[0] += this_ct
247
248
        sql.empty_temp(db, in_table)
249
250
        if this_ct < partition_size: break # partial partition = last
251
252
        # Work around PostgreSQL's temp table disk space leak
253
        db.reconnect()
254
255
    return pkeys_loc