import copy

import re

from xml.dom import Node

import dicts

import exc

import Parser

import sql

import sql_io

import sql_gen

import strings

import util

import xml_dom

import xml_func

import xpath

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

ptr_suffix = '_id'

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

def ptr_type_guess(node_name):

    assert is_ptr(node_name)

    return xml_dom.value_node(node)

    for parent in xml_dom.NodeParentIter(node):

        if name_of(parent) == name: return parent

            for child in xml_dom.NodeElemIter(parent):

                child_name = name_of(child)

                if is_ptr(child_name):

                    target = ptr_target(child)

                elif child_name == name: return child

class ColRef:

input_col_prefix = xml_func.var_name_prefix

put_special_funcs = set(['_setDefault', '_simplifyPath'])

no_parent_ids_loc = object() # tells put() there is no parent_ids_loc

def put(db, node, row_ins_ct_ref=None, on_error=exc.reraise, col_defaults=None,

    in_table=None, parent_ids_loc=no_parent_ids_loc, next=None):

    '''

    elif xml_dom.is_text_node(node): return xml_dom.value(node)

    if col_defaults == None: col_defaults = {}

    def put_(node):

        if util.is_str(node): return node

        return put(db, node, row_ins_ct_ref, on_error, col_defaults, in_table,

    def augment_error(e): exc.add_msg(e, 'node:\n'+strings.ustr(node))

    def on_error_(e):

        augment_error(e)

        on_error(e)

    def wrap_e(e):

    is_func = xml_func.is_func(node)

    out_table = name_of(node)

    row = dicts.OnceOnlyDict()

    children = []

    try:

                if ((is_func or is_ptr(child_name)

                    row[child_name] = child_value

    if out_table == '_setDefault':

        # Parse args

        try: path = row.pop('path')

        except KeyError, e: wrap_e(e)

        col_defaults = dicts.MergeDict(dicts.WrapDict(put_, row), col_defaults)

        return put_(path)

            next = row['next'] # modifies outer next var used by put_()

            path = row['path']

        except KeyError, e: wrap_e(e)

        return put_(path)

    is_literals = in_table == None

    in_tables = []

    if not is_literals:

        in_tables.append(in_table)

    def pkey_name(table): return sql.pkey_name(db, table, True)

    # Add fkey to parent

    if parent_ids_loc is not no_parent_ids_loc:

        if sql_gen.is_table_col(parent_ids_loc):

        parent_ptr = node.getAttribute('fkey')

        if parent_ptr == '': parent_ptr = pkey_name(name_of(node.parentNode))

        row[parent_ptr] = parent_ids_loc

    # Parse input columns

    row = row.inner # now allow keys to be overwritten

    for out_col, value in row.iteritems():

        if (not is_literals and util.is_str(value)

    # Optimizations for structural XML functions

    # Process values

    parent_ids_loc = no_parent_ids_loc # applies to this section

    for out_col, value in row.iteritems():

        # Handle forward pointers

        if xml_dom.is_node(value): row[out_col] = value = put_(value)

        if isinstance(value, sql_gen.Col): # value is table column

            assert sql_gen.is_table_col(value)

            if value.table is not in_table: in_tables.append(value.table)

        else: # value is literal value

            row[out_col] = sql_gen.NamedCol(out_col, value)

    try: pkeys_loc = sql_io.put_table(db, out_table, in_tables, row,

        augment_error(e)

    if sql_gen.is_table_col(pkeys_loc): no_empty.add(pkeys_loc.table)

    sql.empty_temp(db, set(in_tables) - no_empty)

    # Insert children with fkeys to parent

    parent_ids_loc = pkeys_loc # applies to this section

    for child in children: put_(child)

    return pkeys_loc

    def pkey_name(table): return sql.pkey_name(db, table)

    pkey_ = pkey_name(table)

    if id_ != None: conds[pkey_name(table)] = id_ # replace any existing value

    if fields == []: fields.append(pkey_)

partition_size = 1000000 # rows; must be >= NCBI.nodes size

    limit=None, start=0, on_error=exc.reraise, col_defaults={},

    '''

    if in_table == None:

    in_table = sql_gen.as_Table(in_table)

    sql_io.mk_errors_table(db, in_table)

    in_table.set_srcs([in_table], overwrite=False)

    db.src = strings.ustr(in_table)

    pkeys_loc = None # used if loop is never executed

    total = 0

        in_table = sql_gen.Table(strings.ustr(full_in_table),

        sql.copy_table_struct(db, full_in_table, in_table)

        try: sql.add_row_num(db, in_table, 'row_num')

        except sql.DatabaseErrors: pass # already has pkey

        cur = sql.insert_select(db, in_table, None, sql.mk_select(db,

            full_in_table, limit=this_limit, start=start))

        start += this_ct # advance start to fetch next set

        if this_ct == 0: break # in_table size is multiple of partition_size

        # Import data

        if in_row_ct_ref != None: in_row_ct_ref[0] += this_ct