import copy

import re

from xml.dom import Node

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

def get(db, node, limit=None, start=None):

    node = node.firstChild

    table = name_of(node)

    for child in xml_dom.NodeElemIter(node):

        child_name = name_of(child)

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

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

    return sql.select(db, table, fields, conds, limit, start)

def put(db, node, row_ct_ref=None, on_error=exc.raise_, pool=None,

    '''store_ids enables searching the tree for missing fields'''

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

    def put_(node, parent_id=None):

        args = (db, node, row_ct_ref, on_error, pool, store_ids, parent_id)

        exc.add_msg(e, 'node:\n'+str(node))

        on_error(e)

    table = name_of(node)

    try: pkey_ = pkey(table)

    except sql.DatabaseErrors, e: on_error_(e); return None

    row = {}

    # Divide children into fields and children with fkeys to parent

        if xml_dom.is_empty(child): row[child_name] = None

        elif xml_dom.is_text(child):

            row[child_name] = strings.to_unicode(xml_dom.value(child))

        elif is_ptr(child_name): row[child_name] = put_(ptr_target(child))

        else: children.append(child)

    if parent_id != None:

    try:

                id_ = sql_io.put(db, table, row, pkey_, row_ct_ref)

                if store_ids: xml_dom.set_id(node, id_)

                col = e.cols[0]

                if try_num > 0: raise # exception still raised after retry

                if store_ids and is_ptr(col):

                    # Search for required column in ancestors and their children

                    target = find_by_name(node, ptr_type_guess(col))

                    if target == None: raise

                    row[col] = xml_dom.get_id(target)

                else: raise

    except sql.DatabaseErrors, e: on_error_(e); return None

    for child in children: put_(child, id_)

class ColRef:

# Controls when and how put_table() will partition the input table

partition_size = 500000 # rows

input_col_prefix = '$'

put_table_special_funcs = set(['_simplifyPath'])

def put_table(db, node, in_table, in_row_ct_ref=None,

    row_ins_ct_ref=None, limit=None, start=0, on_error=exc.raise_,

    '''

    @param node The XML tree that transforms the input to the output. Similar to

        put()'s node param, but with the input column name prefixed by

    @return (table, col) Where the pkeys (from INSERT RETURNING) are made

        available

    '''

    in_table = sql_gen.as_Table(in_table)

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

    db.src = str(in_table)

    def put_table_(node, in_row_ct_ref=None):

        return put_table(db, node, in_table, in_row_ct_ref, row_ins_ct_ref,

            on_error=on_error, parent_ids_loc=parent_ids_loc, next=next)

    # Subset and/or partition in_table if needed

    in_row_ct = sql.table_row_count(db, in_table)

    if limit != None or start != 0 or in_row_ct > partition_size:

        end = in_row_ct

        if limit != None: end = min(start + limit, end)

        full_in_table = in_table

        for start_ in xrange(start, end, partition_size):

            in_table.name = str(in_table) # prepend schema

            sql.run_query_into(db, sql.mk_select(db, full_in_table,

            # Recurse

            pkeys_loc = put_table_(node, in_row_ct_ref)

            sql.empty_temp(db, in_table)

            # Work around PostgreSQL's temp table disk space leak

        return pkeys_loc

    is_func = xml_func.is_func(node)

    out_table = name_of(node)

    row = {}

        else:

            if is_func or is_ptr(child_name) or xml_func.is_func(child_value):

                row[child_name] = child_value

            else: children.append(child)

        # Parse args

        try:

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

            require = row['require']

        except KeyError, e: wrap_e(e)

        return put_table_(path)

    no_empty = set([in_table])

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

    # Remove any explicit pkey

    try: del row[pkey(out_table)]

    except KeyError: pass

    # Add fkey to parent

        no_empty.add(parent_ids_loc.table)

        parent_ptr = node.getAttribute('fkey')

    # Divide fields into input columns and literal values

    parent_ids_loc = None # applies to this section

    in_tables = [in_table]

    for out_col, value in row.iteritems():

        # Handle forward pointers

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

        elif util.is_str(value) and value.startswith(input_col_prefix):

            row[out_col] = sql_gen.Col(strings.remove_prefix(input_col_prefix,

                value), in_table)

        else: # value is literal value; should only be string or None

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

    pkeys_loc = sql_io.put_table(db, out_table, in_tables, row, row_ins_ct_ref,

        None, next, is_func, on_error)

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

    # Insert children with fkeys to parent

    parent_ids_loc = pkeys_loc # applies to this section

    # Count # rows and update in_row_ct_ref once all columns have been processed

        in_row_ct_ref[0] += sql.table_row_count(db, in_table)

    return pkeys_loc