#!/usr/bin/env python

# Maps one datasource to another, using a map spreadsheet if needed

# For outputting an XML file to a PostgreSQL database, use the general format of

# http://vegbank.org/vegdocs/xml/vegbank_example_ver1.0.2.xml

import os.path

import sys

import traceback

import xml.dom.minidom

sys.path.append(os.path.dirname(__file__)+"/../lib")

import opts

import Parser

import sql

import xml_dom

import xml_func

def metadata_value(name):

    if type(name) == str and name.startswith(':'): return name[1:]

    else: return None

def main():

    env_names = []

    def usage_err():

        raise SystemExit('Usage: '+opts.env_usage(env_names, True)

            +' [commit=1] '+sys.argv[0]+' [map_path] [<input] [>output]')

    limit = opts.get_env_var('n', None, env_names)

    if limit != None: limit = int(limit)

    commit = opts.env_flag('commit')

    # Get db config from env vars

    db_config_names = ['engine', 'host', 'user', 'password', 'database']

    def get_db_config(prefix):

        return opts.get_env_vars(db_config_names, prefix, env_names)

    in_db_config = get_db_config('in')

    out_db_config = get_db_config('out')

    in_is_db = 'engine' in in_db_config

    out_is_db = 'engine' in out_db_config

    # Parse args

    map_path = None

    try: _prog_name, map_path = sys.argv

    except ValueError:

        if in_is_db: usage_err()

    # Load map header

    in_is_xpaths = True

    if map_path != None:

        import copy

        import csv

        import xpath

        metadata = []

        mappings = []

        stream = open(map_path, 'rb')

        reader = csv.reader(stream)

        in_label, out_label = reader.next()[:2]

        def split_col_name(name):

            name, sep, root = name.partition(':')

            return name, sep != '', root

        in_label, in_is_xpaths, in_root = split_col_name(in_label)

        out_label, out_is_xpaths, out_root = split_col_name(out_label)

        assert out_is_xpaths # CSV output not supported yet

        has_types = out_root.startswith('/*s/') # outer elements are types

        for row in reader:

            in_, out = row[:2]

            if out != '':

                if out_is_xpaths: out = out_root+out

                mappings.append((in_, out))

        stream.close()

    in_is_xml = in_is_xpaths and not in_is_db

    if in_is_xml: doc0 = xml.dom.minidom.parse(sys.stdin)

    def process_xml(use_row):

        '''Maps datasource to XML tree'''

        doc1 = xml_dom.create_doc(out_label)

        root = doc1.documentElement

        if in_is_db:

            assert in_is_xpaths

            import db_xml

            in_root_xml = xpath.path2xml(in_root)

            for i, mapping in enumerate(mappings):

                in_, out = mapping

                if metadata_value(in_) == None:

                    mappings[i] = (xpath.path2xml(in_root+'/'+in_), out)

            in_db = sql.connect(in_db_config)

            in_pkeys = {}

            for row_idx, row in enumerate(sql.rows(db_xml.get(in_db,

                in_root_xml, in_pkeys, limit))):

                row_id = str(row_idx)

                pkey, = row

                for in_, out in mappings:

                    value = metadata_value(in_)

                    if value == None:

                        in_ = in_.cloneNode(True) # don't modify orig value!

                        xml_dom.set_id(xpath.get(in_, in_root), pkey)

                        value = sql.value_or_none(db_xml.get(in_db, in_,

                            in_pkeys))

                    if value != None:

                        try: xpath.put_obj(root, out, row_id, has_types,

                            str(value))

                        except Exception: traceback.print_exc()

            in_db.close()

        elif in_is_xml:

            row = xpath.get(doc0.documentElement, in_root)

            for row_idx, row in enumerate(xml_dom.NodeElemIter(row.parentNode)):

                if not (limit == None or row_idx < limit): break

                row_id = str(row_idx)

                for in_, out in mappings:

                    value = metadata_value(in_)

                    if value == None:

                        node = xpath.get(row, in_)

                        if node != None: value = xml_dom.value(node)

                    if value != None:

                        try: xpath.put_obj(root, out, row_id, has_types, value)

                        except Exception: traceback.print_exc()

        else: # input is CSV

            map_ = dict(mappings)

            reader = csv.reader(sys.stdin)

            cols = reader.next()

            col_idxs = dict([(value, idx) for idx, value in enumerate(cols)])

            for i, mapping in enumerate(mappings):

                in_, out = mapping

                if metadata_value(in_) == None:

                    try: mappings[i] = (col_idxs[in_], out)

                    except KeyError: pass

            for row_idx, row in enumerate(reader):

                if not (limit == None or row_idx < limit): break

                row_id = str(row_idx)

                for in_, out in mappings:

                    value = metadata_value(in_)

                    if value == None:

                        value = row[in_]

                        if value == '': value = None

                    if value != None:

                        try: xpath.put_obj(root, out, row_id, has_types, value)

                        except Exception: traceback.print_exc()

        xml_func.process(root)

        return doc1

    def get_xml(use_row):

        '''Inputs datasource to XML tree, mapping if needed'''

        if map_path != None: return process_xml(use_row)

        else: return doc0

    # Output XML tree

    if out_is_db:

        from psycopg2.extensions import ISOLATION_LEVEL_SERIALIZABLE

        import db_xml

        out_db = sql.connect(out_db_config)

        out_db.set_isolation_level(ISOLATION_LEVEL_SERIALIZABLE)

        try:

            row_ct_ref = [0]

            def use_row(root): db_xml.xml2db(out_db, root, row_ct_ref)

            db_xml.xml2db(out_db, get_xml(use_row).documentElement, row_ct_ref)

            print 'Inserted '+str(row_ct_ref[0])+' rows'

            if commit: out_db.commit()

        finally:

            out_db.rollback()

            out_db.close()

    else:

        def use_row(root): pass # TODO: implement this

        xml_dom.writexml(sys.stdout, get_xml(use_row)) # output is XML

try: main()

except Parser.SyntaxException, e: raise SystemExit(str(e))