summaryrefslogtreecommitdiff
path: root/bidding_data.py
blob: 0ee025a0da7d5a4d6ff452775092a51638fd86b1 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
import csv, sys, json, glob, re
from os import path
from bs4 import BeautifulSoup as bs4

class JFRBidding:

    # alignment of the bidding table
    __directions = ['W', 'N', 'E', 'S']

    # converts csv file to list of row lists
    def __csv_to_list(self, file_path):
        file_data = []
        with file(file_path) as csv_file:
            csv_data = csv.reader(csv_file)
            for line in csv_data:
                file_data.append(line)
        return file_data

    # converts CSV lineup data to {round}.{sector}_{table}.{pair numbers} structure
    def __parse_lineup_data(self, sitting_data):
        round_lineups = {}
        for sitting in sitting_data[1:]:
            round_no = int(sitting[2])
            table_no = sitting[0] + '_' + sitting[1]
            if not round_lineups.has_key(round_no):
                round_lineups[round_no] = {}
            round_lineups[round_no][table_no] = sorted([int(sitting[3]), int(sitting[4])])
        return round_lineups

    # converts CSV bidding to {board}.{sector}_{table}.{round}.{bidding}[] structure,
    # including erased calls
    def __parse_bidding_data(self, bidding_data):
        bids = {}
        for bid in bidding_data[1:]:
            board_no = int(bid[4])
            round_no = int(bid[3])
            table_no = bid[1] + '_' + bid[2]
            bid_counter = int(bid[5])
            bid_erased = int(bid[10])
            if not bids.has_key(board_no):
                bids[board_no] = {}
            if not bids[board_no].has_key(table_no):
                bids[board_no][table_no] = {}
            if not bids[board_no][table_no].has_key(round_no):
                bids[board_no][table_no][round_no] = {}
            if bid_erased == 1:
                if bids[board_no][table_no][round_no].has_key(bid_counter):
                    if bids[board_no][table_no][round_no][bid_counter]['direction'] == bid[6]:
                        bids[board_no][table_no][round_no].pop(bid_counter, None)
            else:
                bids[board_no][table_no][round_no][bid_counter] = {'direction': bid[6], 'bid': bid[7] }
        return bids

    # converts bidding data into HTML table
    def __format_bidding(self, bidding):
        html_output = '<table>'
        html_output = html_output + '<tr>'
        for dir in self.__directions:
            html_output = html_output + '<th>' + dir + '</th>'
        html_output = html_output + '</tr>'
        for bid_round in bidding:
            html_output = html_output + '<tr>'
            for bid in bid_round:
                bid_match = re.match(r'(\d)([SHDCN])', bid)
                if bid_match:
		    bid = bid_match.group(1) + '<img src="images/' + bid_match.group(2) + '.gif" />'
                html_output = html_output + '<td>' + bid + '</td>'
            html_output = html_output + '</tr>'
        html_output = html_output + '</table>'
        return html_output

    # returns file path for bidding HTML output
    def __get_bidding_file_output_path(self, board_no, round_no=None, table_no=None, pair_numbers=None):
        return '{0}_bidding_{1:03}_{2}.txt'.format(self.__tournament_prefix,
                                                   board_no,
                                                   '_'.join(map(str, self.__round_lineups[round_no][table_no] if pair_numbers is None else pair_numbers)))

    def __map_board_numbers(self, custom_mapping=None):
        self.__tournament_files = [f for f
                                   in glob.glob(self.__tournament_prefix + '*.html')
                                   if re.search(self.__tournament_files_match, f)]
        if custom_mapping is not None:
            custom_files = []
            for jfr_number in range(custom_mapping[0], custom_mapping[1]+1):
                board_files = [f for f in self.__tournament_files if f.endswith('{0:03}.html'.format(jfr_number))]
                if len(board_files):
                    self.__board_number_mapping[jfr_number - custom_mapping[0] + custom_mapping[2]] = jfr_number
                    custom_files = custom_files + board_files
            self.__tournament_files = custom_files
        else:
            for tournament_file in self.__tournament_files:
                file_number = re.match(self.__tournament_files_match, tournament_file).group(1)
                with file(tournament_file, 'r+') as board_html:
                    board_content = bs4(board_html, from_encoding='utf-8')
                    board_number = board_content.select('h4')[0].contents[0].strip().replace('ROZDANIE ', '')
                    self.__board_number_mapping[int(board_number)] = int(file_number, 10)

    __round_lineups = {}
    __bids = {}

    __tournament_prefix = ''
    __tournament_files_match = None
    __tournament_files = []

    # BWS number -> JFR number
    __board_number_mapping = {}

    def __init__(self, bidding_file, lineup_file, file_prefix, board_mapping):
        self.__round_lineups = self.__parse_lineup_data(self.__csv_to_list(lineup_file))
        self.__bids = self.__parse_bidding_data(self.__csv_to_list(bidding_file))
        self.__tournament_prefix = path.splitext(path.realpath(file_prefix + '.html'))[0]
        self.__tournament_files_match = re.compile(re.escape(self.__tournament_prefix) + '([0-9]{3})\.html')
        self.__map_board_numbers(board_mapping)

    def write_bidding_tables(self):
        for board_no, board_data in self.__bids.items():
            if board_no in self.__board_number_mapping:
                for table_no, table_data in board_data.items():
                    for round_no, round_data in table_data.items():
                        bidding = sorted(round_data)
                        dealer = round_data[bidding[0]]['direction']
                        bidding_table = [[], [], [], []]
                        for bid_index in bidding:
                            bid = round_data[bid_index]
                            bidding_table[self.__directions.index(bid['direction'])].append(bid['bid'])
                            last_bidder = bid['direction']
                        for pos in range(0, self.__directions.index(dealer)):
                            bidding_table[pos].insert(0, '')
                        for pos in range(self.__directions.index(last_bidder), len(self.__directions)):
                            bidding_table[pos].append('')
                        bidding_table = map(list, zip(*bidding_table))
                        bidding_file_path = self.__get_bidding_file_output_path(self.__board_number_mapping[board_no], round_no, table_no)
                        with file(bidding_file_path, 'w') as bidding_file:
                            bidding_file.write(self.__format_bidding(bidding_table))

    def write_bidding_scripts(self):
        for tournament_file in self.__tournament_files:
            with file(tournament_file, 'r+') as board_html:
                board_content = bs4(board_html, from_encoding='utf-8')
                header_scripts = board_content.select('head script')
                jquery_scripts = [script for script in header_scripts if script['src'] == 'javas/jquery.js']
                if not len(jquery_scripts):
                    board_content.head.append(bs4('<script src="javas/jquery.js" type="text/javascript"></script>').script)
                bidding_scripts = [script for script in header_scripts if script['src'] == 'javas/bidding.js']
                if not len(bidding_scripts):
                    board_content.head.append(bs4('<script src="javas/bidding.js" type="text/javascript"></script>').script)
                board_html.seek(0)
                board_html.write(board_content.prettify('utf-8', formatter='html'))
                board_html.truncate()

    def write_bidding_links(self):
        for tournament_file in self.__tournament_files:
            file_number = re.match(self.__tournament_files_match, tournament_file).group(1)
            board_text_path = path.splitext(tournament_file)[0] + '.txt'
            with file(board_text_path, 'r+') as board_text:
                board_text_content = bs4(board_text, from_encoding='iso-8859-2')
                for row in board_text_content.select('tr'):
                    cells = row.select('td')
                    if len(cells) == 11:
                        pair_numbers = sorted([int(cells[1].contents[0]), int(cells[2].contents[0])])
                        bidding_link = bs4('<a href="#" class="biddingLink">[lic]</a>')
                        bidding_link.a['data-bidding-link'] = path.basename(self.__get_bidding_file_output_path(
                            int(file_number, 10),
                            pair_numbers=pair_numbers
                        ))
                        for link in cells[3].select('a.biddingLink'):
                            link.extract()
                        cells[3].append(bidding_link)
                board_text.seek(0)
                board_text.write(board_text_content.table.prettify('iso-8859-2', formatter='html'))

bidding_parser = JFRBidding(sys.argv[1], sys.argv[2], sys.argv[3], map(int, sys.argv[4:]) if len(sys.argv) > 4 else None)
bidding_parser.write_bidding_tables()
bidding_parser.write_bidding_scripts()
bidding_parser.write_bidding_links()