#!/usr/bin/env python

# Inner-joins two map spreadsheets A->B and B->C to A->C

# Multi-safe (supports an input appearing multiple times). If multiple outputs

# should be considered ambiguous, they can be discarded by setting only_one.

# Case- and punctuation-insensitive.

import csv

import operator

import os.path

import re

import sys

import warnings

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

import maps

import opts

import util

def main():

    env_names = []

    def usage_err():

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

            +sys.argv[0]+' <map_0 map_1 [| '+sys.argv[0]+' map_2]... >out_map')

    only_one = opts.env_flag('only_one', False, env_names)

    # Uses only columns 0 and 1 of map_1

    map_1_core_only = opts.env_flag('map_1_core_only', False, env_names)

    # Passes through terms with no input mapping or no join mapping

    passthru = opts.env_flag('passthru', False, env_names)

    # Turns off "No input mapping" errors

    quiet = passthru or opts.env_flag('quiet', False, env_names)

    try: _prog_name, map_1_path = sys.argv

    except ValueError: usage_err()

    # Get map 1

    map_1 = {}

    stream = open(map_1_path, 'rb')

    reader = csv.reader(stream)

    map_1_cols = reader.next()

    for row in reader:

        if map_1_core_only: row = row[:2]

        if row[0] != '': map_1.setdefault(maps.simplify(row[0]), []).append(row)

    stream.close()

    # Join map 1 to map 0

    reader = csv.reader(sys.stdin)

    writer = csv.writer(sys.stdout)

    map_0_cols = reader.next()

    if not maps.join_combinable(map_0_cols, map_1_cols):

        raise SystemExit('Map error: Map 0 output column name "'+map_0_cols[1]

        +'" doesn\'t match map 1 input column name "'+map_1_cols[0]+'"')

    writer.writerow(maps.merge_mappings(map_0_cols, map_1_cols))

    for row in reader:

        def set_error(msg):

            if not passthru: row[1] = ''

            elif row[1] == '': row[1] = row[0] # no input mapping

            if not quiet:

                warnings.warn(UserWarning(msg))

                row[2] = '** '+msg+' ** '+util.list_setdefault(row, 2, '')

        row_written = False

        if row[1] != '':

            out_orig = row[1] # used in "No join mapping" error msg

            # Look for a match

            out_rows = []

            suffix = ''

            while True:

                try:

                    out_rows = map_1[maps.simplify(row[1])]

                    break

                except KeyError:

                    # Heuristically look for a match on a parent path.

                    # If this produces a syntactically invalid parent path (e.g.

                    # there is a / within []), it will be ignored since there

                    # wouldn't be a an entry in map_1.

                    row[1], sep, new_suffix = row[1].rpartition('/')

                    if sep == '': break

                    suffix = sep+new_suffix+suffix # prepend new suffix

            is_empty = len(out_rows) == 1 and out_rows[0][1] == ''

            if is_empty:

                if passthru: out_rows[0][1] = row[1] # for maps.merge_mappings()

                set_error('No non-empty join mapping for '+out_orig)

            # Write new mapping

            if only_one and len(out_rows) > 1: # multiple outputs are ambiguous

                set_error('Ambiguous mapping for '+row[1]) # discards mapping

            elif out_rows: # found mapping(s)

                for out_row in out_rows:

                    row_ = row[:] # don't modify row, since it will be reused

                    row_ = maps.merge_mappings(row_, out_row)

                    if row_[1] != '': row_[1] += suffix # don't modify out_row!

                    writer.writerow(row_)

                row_written = True

            else: set_error('No join mapping for '+out_orig)

        elif row[2] == '': # also no comment explaining why no input mapping

            set_error('No input mapping for '+row[0])

        if not row_written: writer.writerow(row)

main()