sportdb-models 1.18.0 → 1.18.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/{HISTORY.md → CHANGELOG.md} +0 -0
- data/Manifest.txt +1 -16
- data/Rakefile +6 -7
- data/lib/sportdb/models.rb +8 -30
- data/lib/sportdb/version.rb +1 -1
- data/test/test_assoc_reader.rb +1 -3
- data/test/test_changes.rb +9 -5
- data/test/test_cursor.rb +6 -1
- data/test/test_event_meta_reader.rb +0 -3
- data/test/test_event_reader.rb +2 -6
- data/test/test_event_table_reader.rb +1 -5
- metadata +24 -25
- data/config/fixtures/de.yml +0 -46
- data/config/fixtures/en.yml +0 -54
- data/config/fixtures/es.yml +0 -48
- data/config/fixtures/fr.yml +0 -53
- data/config/fixtures/it.yml +0 -55
- data/config/fixtures/pt.yml +0 -46
- data/config/fixtures/ro.yml +0 -55
- data/data/seasons.txt +0 -74
- data/data/setups/all.txt +0 -5
- data/lib/sportdb/csv_reader.rb +0 -241
- data/lib/sportdb/finders/date.rb +0 -446
- data/lib/sportdb/lang.rb +0 -216
- data/test/test_csv_reader.rb +0 -45
- data/test/test_date.rb +0 -100
- data/test/test_lang.rb +0 -130
data/lib/sportdb/csv_reader.rb
DELETED
@@ -1,241 +0,0 @@
|
|
1
|
-
# encoding: UTF-8
|
2
|
-
|
3
|
-
|
4
|
-
module SportDb
|
5
|
-
|
6
|
-
|
7
|
-
class CsvGameReader
|
8
|
-
|
9
|
-
include LogUtils::Logging
|
10
|
-
|
11
|
-
## make models available by default with namespace
|
12
|
-
# e.g. lets you use Usage instead of Model::Usage
|
13
|
-
include Models
|
14
|
-
|
15
|
-
##
|
16
|
-
## todo: add from_file and from_zip too
|
17
|
-
|
18
|
-
def self.from_string( event_key, text )
|
19
|
-
### fix - fix -fix:
|
20
|
-
## change event to event_or_event_key !!!!! - allow event_key as string passed in
|
21
|
-
self.new( event_key, text )
|
22
|
-
end
|
23
|
-
|
24
|
-
def initialize( event_key, text )
|
25
|
-
### fix - fix -fix:
|
26
|
-
## change event to event_or_event_key !!!!! - allow event_key as string passed in
|
27
|
-
|
28
|
-
## todo/fix: how to add opts={} ???
|
29
|
-
@event_key = event_key
|
30
|
-
@text = text
|
31
|
-
end
|
32
|
-
|
33
|
-
|
34
|
-
def read
|
35
|
-
## note: assume active activerecord connection
|
36
|
-
@event = Event.find_by!( key: @event_key )
|
37
|
-
|
38
|
-
logger.debug "Event #{@event.key} >#{@event.title}<"
|
39
|
-
|
40
|
-
@team_mapper = TextUtils::TitleMapper.new( @event.teams, 'team' )
|
41
|
-
|
42
|
-
## reset cached values
|
43
|
-
@patch_round_ids = []
|
44
|
-
|
45
|
-
@last_round = nil ## remove last round ?? - always required - why? why not?
|
46
|
-
@last_date = nil ## remove last date ?? - always required - why? why not?
|
47
|
-
|
48
|
-
parse_fixtures
|
49
|
-
end # method load_fixtures
|
50
|
-
|
51
|
-
|
52
|
-
def handle_round( round_pos_str )
|
53
|
-
|
54
|
-
round_pos = round_pos_str.to_i
|
55
|
-
|
56
|
-
round_attribs = { }
|
57
|
-
|
58
|
-
round = Round.find_by( event_id: @event.id,
|
59
|
-
pos: round_pos )
|
60
|
-
|
61
|
-
if round.present?
|
62
|
-
logger.debug "update round #{round.id}:"
|
63
|
-
else
|
64
|
-
logger.debug "create round:"
|
65
|
-
round = Round.new
|
66
|
-
|
67
|
-
round_attribs = round_attribs.merge( {
|
68
|
-
event_id: @event.id,
|
69
|
-
pos: round_pos,
|
70
|
-
title: "Round #{round_pos}",
|
71
|
-
title2: nil,
|
72
|
-
knockout: false,
|
73
|
-
start_at: Date.parse('1911-11-11'),
|
74
|
-
end_at: Date.parse('1911-11-11')
|
75
|
-
})
|
76
|
-
end
|
77
|
-
|
78
|
-
logger.debug round_attribs.to_json
|
79
|
-
|
80
|
-
round.update_attributes!( round_attribs )
|
81
|
-
|
82
|
-
### store list of round ids for patching start_at/end_at at the end
|
83
|
-
@patch_round_ids << round.id
|
84
|
-
@last_round = round ## keep track of last seen round for matches that follow etc.
|
85
|
-
end
|
86
|
-
|
87
|
-
def handle_game( date_str, team1_str, team2_str, ft_str, ht_str )
|
88
|
-
|
89
|
-
## todo/fix: make more "efficient"
|
90
|
-
## - e.g. add new support method for mapping single team/reference - why? why not??
|
91
|
-
line = "#{team1_str} - #{team2_str}"
|
92
|
-
@team_mapper.map_titles!( line )
|
93
|
-
team1_key = @team_mapper.find_key!( line )
|
94
|
-
team2_key = @team_mapper.find_key!( line )
|
95
|
-
|
96
|
-
## note: if we do NOT find two teams; return false - no match found
|
97
|
-
if team1_key.nil? || team2_key.nil?
|
98
|
-
fail " no game match (two teams required) found for line: >#{line}<"
|
99
|
-
end
|
100
|
-
|
101
|
-
if date_str
|
102
|
-
date = DateTime.strptime( date_str, '%Y-%m-%d' ) ## (always) use DateTime - why? why not??
|
103
|
-
@last_date = date # keep a reference for later use
|
104
|
-
else
|
105
|
-
date = @last_date # no date found; (re)use last seen date
|
106
|
-
end
|
107
|
-
|
108
|
-
##
|
109
|
-
## todo: support for awarded, abadoned, a.e.t, pen. etc. - why?? why not??
|
110
|
-
##
|
111
|
-
|
112
|
-
if ht_str ## half time scores
|
113
|
-
scoresi_str = ht_str.gsub(/ /, '').split('-') ## note: remove all (inline) spaces first
|
114
|
-
score1i = scoresi_str[0].to_i
|
115
|
-
score2i = scoresi_str[1].to_i
|
116
|
-
else
|
117
|
-
score1i = nil
|
118
|
-
score2i = nil
|
119
|
-
end
|
120
|
-
|
121
|
-
if ft_str ## full time scores
|
122
|
-
scores_str = ft_str.gsub(/ /, '').split('-') ## note: remove all (inline) spaces first
|
123
|
-
score1 = scores_str[0].to_i
|
124
|
-
score2 = scores_str[1].to_i
|
125
|
-
else
|
126
|
-
score1 = nil
|
127
|
-
score2 = nil
|
128
|
-
end
|
129
|
-
|
130
|
-
### todo: cache team lookups in hash? - why? why not??
|
131
|
-
team1 = Team.find_by!( key: team1_key )
|
132
|
-
team2 = Team.find_by!( key: team2_key )
|
133
|
-
|
134
|
-
round = @last_round
|
135
|
-
|
136
|
-
### check if games exists
|
137
|
-
## with this teams in this round if yes only update
|
138
|
-
game = Game.find_by( round_id: round.id,
|
139
|
-
team1_id: team1.id,
|
140
|
-
team2_id: team2.id )
|
141
|
-
|
142
|
-
game_attribs = {
|
143
|
-
score1: score1,
|
144
|
-
score2: score2,
|
145
|
-
score1i: score1i,
|
146
|
-
score2i: score2i,
|
147
|
-
play_at: date,
|
148
|
-
play_at_v2: nil,
|
149
|
-
postponed: false,
|
150
|
-
knockout: false, ## round.knockout, ## note: for now always use knockout flag from round - why? why not??
|
151
|
-
ground_id: nil,
|
152
|
-
group_id: nil
|
153
|
-
}
|
154
|
-
|
155
|
-
if game.present?
|
156
|
-
logger.debug "update game #{game.id}:"
|
157
|
-
else
|
158
|
-
logger.debug "create game:"
|
159
|
-
game = Game.new
|
160
|
-
|
161
|
-
## Note: use round.games.count for pos
|
162
|
-
## lets us add games out of order if later needed
|
163
|
-
more_game_attribs = {
|
164
|
-
round_id: round.id,
|
165
|
-
team1_id: team1.id,
|
166
|
-
team2_id: team2.id,
|
167
|
-
pos: round.games.count+1
|
168
|
-
}
|
169
|
-
game_attribs = game_attribs.merge( more_game_attribs )
|
170
|
-
end
|
171
|
-
|
172
|
-
logger.debug game_attribs.to_json
|
173
|
-
game.update_attributes!( game_attribs )
|
174
|
-
|
175
|
-
end # method handle_game
|
176
|
-
|
177
|
-
|
178
|
-
def parse_fixtures
|
179
|
-
|
180
|
-
CSV.parse( @text, headers: true ) do |row|
|
181
|
-
puts row.inspect
|
182
|
-
|
183
|
-
pp round = row['Round']
|
184
|
-
pp date = row['Date']
|
185
|
-
pp team1 = row['Team 1']
|
186
|
-
pp team2 = row['Team 2']
|
187
|
-
pp ft = row['FT']
|
188
|
-
pp ht = row['HT']
|
189
|
-
|
190
|
-
## find round by pos
|
191
|
-
if round
|
192
|
-
handle_round( round )
|
193
|
-
handle_game( date, team1, team2, ft, ht )
|
194
|
-
else
|
195
|
-
fail "round required for import; sorry"
|
196
|
-
end
|
197
|
-
end
|
198
|
-
|
199
|
-
###########################
|
200
|
-
# backtrack and patch round dates (start_at/end_at)
|
201
|
-
|
202
|
-
unless @patch_round_ids.empty?
|
203
|
-
###
|
204
|
-
# note: use uniq - to allow multiple round headers (possible?)
|
205
|
-
|
206
|
-
Round.find( @patch_round_ids.uniq ).each do |r|
|
207
|
-
logger.debug "patch round start_at/end_at date for #{r.title}:"
|
208
|
-
|
209
|
-
## note:
|
210
|
-
## will add "scope" pos first e.g
|
211
|
-
#
|
212
|
-
## SELECT "games".* FROM "games" WHERE "games"."round_id" = ?
|
213
|
-
# ORDER BY pos, play_at asc [["round_id", 7]]
|
214
|
-
# thus will NOT order by play_at but by pos first!!!
|
215
|
-
# =>
|
216
|
-
# need to unscope pos!!! or use unordered_games - games_by_play_at_date etc.??
|
217
|
-
# thus use reorder()!!! - not just order('play_at asc')
|
218
|
-
|
219
|
-
games = r.games.reorder( 'play_at asc' ).all
|
220
|
-
|
221
|
-
## skip rounds w/ no games
|
222
|
-
|
223
|
-
## todo/check/fix: what's the best way for checking assoc w/ 0 recs?
|
224
|
-
next if games.size == 0
|
225
|
-
|
226
|
-
# note: make sure start_at/end_at is date only (e.g. use play_at.to_date)
|
227
|
-
# sqlite3 saves datetime in date field as datetime, for example (will break date compares later!)
|
228
|
-
|
229
|
-
round_attribs = {
|
230
|
-
start_at: games[0].play_at.to_date, # use games.first ?
|
231
|
-
end_at: games[-1].play_at.to_date # use games.last ? why? why not?
|
232
|
-
}
|
233
|
-
|
234
|
-
logger.debug round_attribs.to_json
|
235
|
-
r.update_attributes!( round_attribs )
|
236
|
-
end
|
237
|
-
end
|
238
|
-
end # method parse_fixtures
|
239
|
-
|
240
|
-
end # class CsvGameReader
|
241
|
-
end # module SportDb
|
data/lib/sportdb/finders/date.rb
DELETED
@@ -1,446 +0,0 @@
|
|
1
|
-
# encoding: utf-8
|
2
|
-
|
3
|
-
#### fix: move to textutils for reuse !!!!!!!!!! - why?? why not ??
|
4
|
-
|
5
|
-
|
6
|
-
module SportDb
|
7
|
-
|
8
|
-
|
9
|
-
class DateFinderBase
|
10
|
-
|
11
|
-
MONTH_EN_TO_MM = {
|
12
|
-
'Jan' => '1', 'January' => '1',
|
13
|
-
'Feb' => '2', 'February' => '2',
|
14
|
-
'Mar' => '3', 'March' => '3',
|
15
|
-
'Apr' => '4', 'April' => '4',
|
16
|
-
'May' => '5',
|
17
|
-
'Jun' => '6', 'June' => '6',
|
18
|
-
'Jul' => '7', 'July' => '7',
|
19
|
-
'Aug' => '8', 'August' => '8',
|
20
|
-
'Sep' => '9', 'Sept' => '9', 'September' => '9',
|
21
|
-
'Oct' => '10', 'October' => '10',
|
22
|
-
'Nov' => '11', 'November' => '11',
|
23
|
-
'Dec' => '12', 'December' =>'12' }
|
24
|
-
|
25
|
-
MONTH_FR_TO_MM = {
|
26
|
-
'Janvier' => '1', 'Janv' => '1', 'Jan' => '1', ## check janv in use??
|
27
|
-
'Février' => '2', 'Févr' => '2', 'Fév' => '2', ## check fevr in use???
|
28
|
-
'Mars' => '3', 'Mar' => '3',
|
29
|
-
'Avril' => '4', 'Avri' => '4', 'Avr' => '4', ## check avri in use??? if not remove
|
30
|
-
'Mai' => '5',
|
31
|
-
'Juin' => '6',
|
32
|
-
'Juillet' => '7', 'Juil' => '7',
|
33
|
-
'Août' => '8',
|
34
|
-
'Septembre' => '9', 'Sept' => '9',
|
35
|
-
'Octobre' => '10', 'Octo' => '10', 'Oct' => '10', ### check octo in use??
|
36
|
-
'Novembre' => '11', 'Nove' => '11', 'Nov' => '11', ## check nove in use??
|
37
|
-
'Décembre' => '12', 'Déce' => '12', 'Déc' => '12' } ## check dece in use??
|
38
|
-
|
39
|
-
MONTH_ES_TO_MM = {
|
40
|
-
'Ene' => '1', 'Enero' => '1',
|
41
|
-
'Feb' => '2',
|
42
|
-
'Mar' => '3', 'Marzo' => '3',
|
43
|
-
'Abr' => '4', 'Abril' => '4',
|
44
|
-
'May' => '5', 'Mayo' => '5',
|
45
|
-
'Jun' => '6', 'Junio' => '6',
|
46
|
-
'Jul' => '7', 'Julio' => '7',
|
47
|
-
'Ago' => '8', 'Agosto' => '8',
|
48
|
-
'Sep' => '9', 'Set' => '9', 'Sept' => '9',
|
49
|
-
'Oct' => '10',
|
50
|
-
'Nov' => '11',
|
51
|
-
'Dic' => '12' }
|
52
|
-
|
53
|
-
private
|
54
|
-
def calc_year( month, day, opts )
|
55
|
-
start_at = opts[:start_at]
|
56
|
-
|
57
|
-
logger.debug " [calc_year] ????-#{month}-#{day} -- start_at: #{start_at}"
|
58
|
-
|
59
|
-
if month >= start_at.month
|
60
|
-
# assume same year as start_at event (e.g. 2013 for 2013/14 season)
|
61
|
-
start_at.year
|
62
|
-
else
|
63
|
-
# assume year+1 as start_at event (e.g. 2014 for 2013/14 season)
|
64
|
-
start_at.year+1
|
65
|
-
end
|
66
|
-
end
|
67
|
-
|
68
|
-
def parse_date_time( match_data, opts={} )
|
69
|
-
|
70
|
-
# convert regex match_data captures to hash
|
71
|
-
# - note: cannont use match_data like a hash (e.g. raises exception if key/name not present/found)
|
72
|
-
h = {}
|
73
|
-
# - note: do NOT forget to turn name into symbol for lookup in new hash (name.to_sym)
|
74
|
-
match_data.names.each { |name| h[name.to_sym] = match_data[name] } # or use match_data.names.zip( match_data.captures ) - more cryptic but "elegant"??
|
75
|
-
|
76
|
-
## puts "[parse_date_time] match_data:"
|
77
|
-
## pp h
|
78
|
-
logger.debug " [parse_date_time] hash: >#{h.inspect}<"
|
79
|
-
|
80
|
-
h[ :month ] = MONTH_EN_TO_MM[ h[:month_en] ] if h[:month_en]
|
81
|
-
h[ :month ] = MONTH_ES_TO_MM[ h[:month_es] ] if h[:month_es]
|
82
|
-
h[ :month ] = MONTH_FR_TO_MM[ h[:month_fr] ] if h[:month_fr]
|
83
|
-
|
84
|
-
month = h[:month]
|
85
|
-
day = h[:day]
|
86
|
-
year = h[:year] || calc_year( month.to_i, day.to_i, opts ).to_s
|
87
|
-
|
88
|
-
hours = h[:hours] || '12' # default to 12:00 for HH:MM (hours:minutes)
|
89
|
-
minutes = h[:minutes] || '00'
|
90
|
-
|
91
|
-
value = '%d-%02d-%02d %02d:%02d' % [year.to_i, month.to_i, day.to_i, hours.to_i, minutes.to_i]
|
92
|
-
logger.debug " date: >#{value}<"
|
93
|
-
|
94
|
-
DateTime.strptime( value, '%Y-%m-%d %H:%M' )
|
95
|
-
end
|
96
|
-
|
97
|
-
end # class DateFinderBase
|
98
|
-
|
99
|
-
|
100
|
-
class DateFinder < DateFinderBase
|
101
|
-
|
102
|
-
include LogUtils::Logging
|
103
|
-
|
104
|
-
# todo: make more generic for reuse
|
105
|
-
### fix:
|
106
|
-
### move to textutils
|
107
|
-
## date/fr.yml en.yml etc. ???
|
108
|
-
## why? why not?
|
109
|
-
|
110
|
-
MONTH_FR = 'Janvier|Janv|Jan|' +
|
111
|
-
'Février|Févr|Fév|' +
|
112
|
-
'Mars|Mar|' +
|
113
|
-
'Avril|Avri|Avr|' +
|
114
|
-
'Mai|' +
|
115
|
-
'Juin|' +
|
116
|
-
'Juillet|Juil|' +
|
117
|
-
'Août|' +
|
118
|
-
'Septembre|Sept|' +
|
119
|
-
'Octobre|Octo|Oct|' +
|
120
|
-
'Novembre|Nove|Nov|' +
|
121
|
-
'Décembre|Déce|Déc'
|
122
|
-
|
123
|
-
WEEKDAY_FR = 'Lundi|Lun|L|' +
|
124
|
-
'Mardi|Mar|Ma|' +
|
125
|
-
'Mercredi|Mer|Me|' +
|
126
|
-
'Jeudi|Jeu|J|' +
|
127
|
-
'Vendredi|Ven|V|' +
|
128
|
-
'Samedi|Sam|S|' +
|
129
|
-
'Dimanche|Dim|D|'
|
130
|
-
|
131
|
-
|
132
|
-
MONTH_EN = 'January|Jan|'+
|
133
|
-
'February|Feb|'+
|
134
|
-
'March|Mar|'+
|
135
|
-
'April|Apr|'+
|
136
|
-
'May|'+
|
137
|
-
'June|Jun|'+
|
138
|
-
'July|Jul|'+
|
139
|
-
'August|Aug|'+
|
140
|
-
'September|Sept|Sep|'+
|
141
|
-
'October|Oct|'+
|
142
|
-
'November|Nov|'+
|
143
|
-
'December|Dec'
|
144
|
-
|
145
|
-
###
|
146
|
-
## todo: add days
|
147
|
-
## 1. Sunday - Sun. 2. Monday - Mon.
|
148
|
-
## 3. Tuesday - Tu., Tue., or Tues. 4. Wednesday - Wed.
|
149
|
-
## 5. Thursday - Th., Thu., Thur., or Thurs. 6. Friday - Fri.
|
150
|
-
## 7. Saturday - Sat.
|
151
|
-
|
152
|
-
|
153
|
-
MONTH_ES = 'Enero|Ene|'+
|
154
|
-
'Feb|'+
|
155
|
-
'Marzo|Mar|'+
|
156
|
-
'Abril|Abr|'+
|
157
|
-
'Mayo|May|'+
|
158
|
-
'Junio|Jun|'+
|
159
|
-
'Julio|Jul|'+
|
160
|
-
'Agosto|Ago|'+
|
161
|
-
'Sept|Set|Sep|'+
|
162
|
-
'Oct|'+
|
163
|
-
'Nov|'+
|
164
|
-
'Dic'
|
165
|
-
|
166
|
-
# todo/fix - add de and es too!!
|
167
|
-
# note: in Austria - Jänner - in Deutschland Januar allow both ??
|
168
|
-
# MONTH_DE = 'J[aä]n|Feb|Mär|Apr|Mai|Jun|Jul|Aug|Sep|Okt|Nov|Dez'
|
169
|
-
|
170
|
-
|
171
|
-
# e.g. 2012-09-14 20:30 => YYYY-MM-DD HH:MM
|
172
|
-
# nb: allow 2012-9-3 7:30 e.g. no leading zero required
|
173
|
-
# regex_db
|
174
|
-
DB__DATE_TIME_REGEX = /\b
|
175
|
-
(?<year>\d{4})
|
176
|
-
-
|
177
|
-
(?<month>\d{1,2})
|
178
|
-
-
|
179
|
-
(?<day>\d{1,2})
|
180
|
-
\s+
|
181
|
-
(?<hours>\d{1,2})
|
182
|
-
:
|
183
|
-
(?<minutes>\d{2})
|
184
|
-
\b/x
|
185
|
-
|
186
|
-
# e.g. 2012-09-14 w/ implied hours (set to 12:00)
|
187
|
-
# nb: allow 2012-9-3 e.g. no leading zero required
|
188
|
-
# regex_db2
|
189
|
-
DB__DATE_REGEX = /\b
|
190
|
-
(?<year>\d{4})
|
191
|
-
-
|
192
|
-
(?<month>\d{1,2})
|
193
|
-
-
|
194
|
-
(?<day>\d{1,2})
|
195
|
-
\b/x
|
196
|
-
|
197
|
-
# e.g. 14.09.2012 20:30 => DD.MM.YYYY HH:MM
|
198
|
-
# nb: allow 2.3.2012 e.g. no leading zero required
|
199
|
-
# nb: allow hour as 20.30
|
200
|
-
# regex_de
|
201
|
-
DD_MM_YYYY__DATE_TIME_REGEX = /\b
|
202
|
-
(?<day>\d{1,2})
|
203
|
-
\.
|
204
|
-
(?<month>\d{1,2})
|
205
|
-
\.
|
206
|
-
(?<year>\d{4})
|
207
|
-
\s+
|
208
|
-
(?<hours>\d{1,2})
|
209
|
-
[:.]
|
210
|
-
(?<minutes>\d{2})
|
211
|
-
\b/x
|
212
|
-
|
213
|
-
# e.g. 14.09. 20:30 => DD.MM. HH:MM
|
214
|
-
# nb: allow 2.3.2012 e.g. no leading zero required
|
215
|
-
# nb: allow hour as 20.30 or 3.30 instead of 03.30
|
216
|
-
# regex_de2
|
217
|
-
DD_MM__DATE_TIME_REGEX = /\b
|
218
|
-
(?<day>\d{1,2})
|
219
|
-
\.
|
220
|
-
(?<month>\d{1,2})
|
221
|
-
\.
|
222
|
-
\s+
|
223
|
-
(?<hours>\d{1,2})
|
224
|
-
[:.]
|
225
|
-
(?<minutes>\d{2})
|
226
|
-
\b/x
|
227
|
-
|
228
|
-
# e.g. 14.09.2012 => DD.MM.YYYY w/ implied hours (set to 12:00)
|
229
|
-
# regex_de3
|
230
|
-
DD_MM_YYYY__DATE_REGEX = /\b
|
231
|
-
(?<day>\d{1,2})
|
232
|
-
\.
|
233
|
-
(?<month>\d{1,2})
|
234
|
-
\.
|
235
|
-
(?<year>\d{4})
|
236
|
-
\b/x
|
237
|
-
|
238
|
-
# e.g. 14.09. => DD.MM. w/ implied year and implied hours (set to 12:00)
|
239
|
-
# note: allow end delimiter ] e.g. [Sa 12.01.] or end-of-string ($) too
|
240
|
-
# note: we use a lookahead for last part e.g. (?:\s+|$|[\]]) - do NOT cosume
|
241
|
-
# regex_de4 (use lookahead assert)
|
242
|
-
DD_MM__DATE_REGEX = /\b
|
243
|
-
(?<day>\d{1,2})
|
244
|
-
\.
|
245
|
-
(?<month>\d{1,2})
|
246
|
-
\.
|
247
|
-
(?=\s+|$|[\]])/x ## note: allow end-of-string/line too
|
248
|
-
|
249
|
-
|
250
|
-
# e.g. 12 May 2013 14:00 => D|DD.MMM.YYYY H|HH:MM
|
251
|
-
EN__DD_MONTH_YYYY__DATE_TIME_REGEX = /\b
|
252
|
-
(?<day>\d{1,2})
|
253
|
-
\s
|
254
|
-
(?<month_en>#{MONTH_EN})
|
255
|
-
\s
|
256
|
-
(?<year>\d{4})
|
257
|
-
\s+
|
258
|
-
(?<hours>\d{1,2})
|
259
|
-
:
|
260
|
-
(?<minutes>\d{2})
|
261
|
-
\b/x
|
262
|
-
|
263
|
-
###
|
264
|
-
# fix: pass in lang (e.g. en or es)
|
265
|
-
# only process format for lang plus fallback to en?
|
266
|
-
# e.g. EN__DD_MONTH and ES__DD_MONTH depend on order for match (first listed will match)
|
267
|
-
|
268
|
-
# e.g. 12 May => D|DD.MMM w/ implied year and implied hours
|
269
|
-
EN__DD_MONTH__DATE_REGEX = /\b
|
270
|
-
(?<day>\d{1,2})
|
271
|
-
\s
|
272
|
-
(?<month_en>#{MONTH_EN})
|
273
|
-
\b/x
|
274
|
-
|
275
|
-
|
276
|
-
# e.g. Jun/12 2011 14:00
|
277
|
-
EN__MONTH_DD_YYYY__DATE_TIME_REGEX = /\b
|
278
|
-
(?<month_en>#{MONTH_EN})
|
279
|
-
\/
|
280
|
-
(?<day>\d{1,2})
|
281
|
-
\s
|
282
|
-
(?<year>\d{4})
|
283
|
-
\s+
|
284
|
-
(?<hours>\d{1,2})
|
285
|
-
:
|
286
|
-
(?<minutes>\d{2})
|
287
|
-
\b/x
|
288
|
-
|
289
|
-
# e.g. Jun/12 14:00 w/ implied year H|HH:MM
|
290
|
-
EN__MONTH_DD__DATE_TIME_REGEX = /\b
|
291
|
-
(?<month_en>#{MONTH_EN})
|
292
|
-
\/
|
293
|
-
(?<day>\d{1,2})
|
294
|
-
\s+
|
295
|
-
(?<hours>\d{1,2})
|
296
|
-
:
|
297
|
-
(?<minutes>\d{2})
|
298
|
-
\b/x
|
299
|
-
|
300
|
-
# e.g. Jun/12 2013 w/ implied hours (set to 12:00)
|
301
|
-
EN__MONTH_DD_YYYY__DATE_REGEX = /\b
|
302
|
-
(?<month_en>#{MONTH_EN})
|
303
|
-
\/
|
304
|
-
(?<day>\d{1,2})
|
305
|
-
\s
|
306
|
-
(?<year>\d{4})
|
307
|
-
\b/x
|
308
|
-
|
309
|
-
# e.g. Jun/12 w/ implied year and implied hours (set to 12:00)
|
310
|
-
# note: allow space too e.g Jun 12 -- check if conflicts w/ other formats??? (added for rsssf reader)
|
311
|
-
# -- fix: might eat french weekday mar 12 is mardi (mar) !!! see FR__ pattern
|
312
|
-
# fix: remove space again for now - and use simple en date reader or something!!!
|
313
|
-
## was [\/ ] changed back to \/
|
314
|
-
EN__MONTH_DD__DATE_REGEX = /\b
|
315
|
-
(?<month_en>#{MONTH_EN})
|
316
|
-
\/
|
317
|
-
(?<day>\d{1,2})
|
318
|
-
\b/x
|
319
|
-
|
320
|
-
|
321
|
-
# e.g. 12 Ene w/ implied year and implied hours (set to 12:00)
|
322
|
-
ES__DD_MONTH__DATE_REGEX = /\b
|
323
|
-
(?<day>\d{1,2})
|
324
|
-
\s
|
325
|
-
(?<month_es>#{MONTH_ES})
|
326
|
-
\b/x
|
327
|
-
|
328
|
-
# e.g. Ven 8 Août or [Ven 8 Août] or Ven 8. Août or [Ven 8. Août]
|
329
|
-
### note: do NOT consume [] in regex (use lookahead assert)
|
330
|
-
FR__WEEKDAY_DD_MONTH__DATE_REGEX = /\b
|
331
|
-
(?:#{WEEKDAY_FR}) # note: skip weekday for now; do NOT capture
|
332
|
-
\s+
|
333
|
-
(?<day>\d{1,2})
|
334
|
-
\.? # note: make dot optional
|
335
|
-
\s+
|
336
|
-
(?<month_fr>#{MONTH_FR})
|
337
|
-
(?=\s+|$|[\]])/x ## note: allow end-of-string/line too
|
338
|
-
|
339
|
-
|
340
|
-
|
341
|
-
|
342
|
-
# map table - 1) tag, 2) regex - note: order matters; first come-first matched/served
|
343
|
-
FORMATS = [
|
344
|
-
[ '[YYYY_MM_DD_hh_mm]', DB__DATE_TIME_REGEX ],
|
345
|
-
[ '[YYYY_MM_DD]', DB__DATE_REGEX ],
|
346
|
-
[ '[DD_MM_YYYY_hh_mm]', DD_MM_YYYY__DATE_TIME_REGEX ],
|
347
|
-
[ '[DD_MM_hh_mm]', DD_MM__DATE_TIME_REGEX ],
|
348
|
-
[ '[DD_MM_YYYY]', DD_MM_YYYY__DATE_REGEX ],
|
349
|
-
[ '[DD_MM]', DD_MM__DATE_REGEX ],
|
350
|
-
[ '[FR_WEEKDAY_DD_MONTH]', FR__WEEKDAY_DD_MONTH__DATE_REGEX ],
|
351
|
-
[ '[EN_DD_MONTH_YYYY_hh_mm]', EN__DD_MONTH_YYYY__DATE_TIME_REGEX ],
|
352
|
-
[ '[EN_MONTH_DD_YYYY_hh_mm]', EN__MONTH_DD_YYYY__DATE_TIME_REGEX ],
|
353
|
-
[ '[EN_MONTH_DD_hh_mm]', EN__MONTH_DD__DATE_TIME_REGEX ],
|
354
|
-
[ '[EN_MONTH_DD_YYYY]', EN__MONTH_DD_YYYY__DATE_REGEX ],
|
355
|
-
[ '[EN_MONTH_DD]', EN__MONTH_DD__DATE_REGEX ],
|
356
|
-
[ '[EN_DD_MONTH]', EN__DD_MONTH__DATE_REGEX ],
|
357
|
-
[ '[ES_DD_MONTH]', ES__DD_MONTH__DATE_REGEX ]
|
358
|
-
]
|
359
|
-
|
360
|
-
|
361
|
-
|
362
|
-
def initialize
|
363
|
-
# nothing here for now
|
364
|
-
end
|
365
|
-
|
366
|
-
def find!( line, opts={} )
|
367
|
-
# fix: use more lookahead for all required trailing spaces!!!!!
|
368
|
-
# fix: use <name capturing group> for month,day,year etc.!!!
|
369
|
-
|
370
|
-
#
|
371
|
-
# fix: !!!!
|
372
|
-
# date in [] will become [[DATE.DE4]] - when getting removed will keep ]!!!!
|
373
|
-
# fix: change regex to \[[A-Z0-9.]\] !!!!!! plus add unit test too!!!
|
374
|
-
#
|
375
|
-
|
376
|
-
md = nil
|
377
|
-
FORMATS.each do |format|
|
378
|
-
tag = format[0]
|
379
|
-
pattern = format[1]
|
380
|
-
md=pattern.match( line )
|
381
|
-
if md
|
382
|
-
date = parse_date_time( md, opts )
|
383
|
-
## fix: use md[0] e.g. match for sub! instead of using regex again - why? why not???
|
384
|
-
## fix: use md.begin(0), md.end(0)
|
385
|
-
line.sub!( md[0], tag )
|
386
|
-
## todo/fix: make sure match data will not get changed (e.g. using sub! before parse_date_time)
|
387
|
-
return date
|
388
|
-
end
|
389
|
-
# no match; continue; try next pattern
|
390
|
-
end
|
391
|
-
|
392
|
-
return nil # no match found
|
393
|
-
end
|
394
|
-
|
395
|
-
end # class DateFinder
|
396
|
-
|
397
|
-
|
398
|
-
class RsssfDateFinder < DateFinderBase
|
399
|
-
|
400
|
-
include LogUtils::Logging
|
401
|
-
|
402
|
-
MONTH_EN = 'Jan|'+
|
403
|
-
'Feb|'+
|
404
|
-
'March|Mar|'+
|
405
|
-
'April|Apr|'+
|
406
|
-
'May|'+
|
407
|
-
'June|Jun|'+
|
408
|
-
'July|Jul|'+
|
409
|
-
'Aug|'+
|
410
|
-
'Sept|Sep|'+
|
411
|
-
'Oct|'+
|
412
|
-
'Nov|'+
|
413
|
-
'Dec'
|
414
|
-
|
415
|
-
## e.g.
|
416
|
-
## [Jun 7] or [Aug 12] etc. - not MUST include brackets e.g. []
|
417
|
-
##
|
418
|
-
## check add \b at the beginning and end - why?? why not?? working??
|
419
|
-
EN__MONTH_DD__DATE_REGEX = /\[
|
420
|
-
(?<month_en>#{MONTH_EN})
|
421
|
-
\s
|
422
|
-
(?<day>\d{1,2})
|
423
|
-
\]/x
|
424
|
-
|
425
|
-
def find!( line, opts={} )
|
426
|
-
# fix: use more lookahead for all required trailing spaces!!!!!
|
427
|
-
# fix: use <name capturing group> for month,day,year etc.!!!
|
428
|
-
|
429
|
-
tag = '[EN_MONTH_DD]'
|
430
|
-
pattern = EN__MONTH_DD__DATE_REGEX
|
431
|
-
md = pattern.match( line )
|
432
|
-
if md
|
433
|
-
date = parse_date_time( md, opts )
|
434
|
-
## fix: use md[0] e.g. match for sub! instead of using regex again - why? why not???
|
435
|
-
## fix: use md.begin(0), md.end(0)
|
436
|
-
line.sub!( md[0], tag )
|
437
|
-
## todo/fix: make sure match data will not get changed (e.g. using sub! before parse_date_time)
|
438
|
-
return date
|
439
|
-
end
|
440
|
-
return nil # no match found
|
441
|
-
end
|
442
|
-
|
443
|
-
|
444
|
-
end ## class RsssfDateFinder
|
445
|
-
|
446
|
-
end # module SportDb
|