sportdb-quick 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/CHANGELOG.md +3 -0
- data/Manifest.txt +13 -0
- data/README.md +5 -0
- data/Rakefile +32 -0
- data/bin/fbt +121 -0
- data/lib/sportdb/quick/linter.rb +149 -0
- data/lib/sportdb/quick/match_parser.rb +735 -0
- data/lib/sportdb/quick/opts.rb +70 -0
- data/lib/sportdb/quick/outline_reader.rb +97 -0
- data/lib/sportdb/quick/quick_league_outline_reader.rb +85 -0
- data/lib/sportdb/quick/quick_match_reader.rb +95 -0
- data/lib/sportdb/quick/version.rb +24 -0
- data/lib/sportdb/quick.rb +31 -0
- metadata +137 -0
@@ -0,0 +1,735 @@
|
|
1
|
+
|
2
|
+
module SportDb
|
3
|
+
|
4
|
+
class MatchParser ## simple match parser for team match schedules
|
5
|
+
|
6
|
+
|
7
|
+
def self.parse( lines, start: )
|
8
|
+
## todo/fix: add support for txt and lines
|
9
|
+
## check if lines_or_txt is an array or just a string
|
10
|
+
## use teams: like start: why? why not?
|
11
|
+
parser = new( lines, start )
|
12
|
+
parser.parse
|
13
|
+
end
|
14
|
+
|
15
|
+
|
16
|
+
include Logging ## e.g. logger#debug, logger#info, etc.
|
17
|
+
|
18
|
+
def self.debug=(value) @@debug = value; end
|
19
|
+
def self.debug?() @@debug ||= false; end ## note: default is FALSE
|
20
|
+
def debug?() self.class.debug?; end
|
21
|
+
|
22
|
+
def _read_lines( txt ) ## todo/check: add alias preproc_lines or build_lines or prep_lines etc. - why? why not?
|
23
|
+
## returns an array of lines with comments and empty lines striped / removed
|
24
|
+
lines = []
|
25
|
+
txt.each_line do |line| ## preprocess
|
26
|
+
line = line.strip
|
27
|
+
|
28
|
+
next if line.empty? || line.start_with?('#') ### skip empty lines and comments
|
29
|
+
line = line.sub( /#.*/, '' ).strip ### cut-off end-of line comments too
|
30
|
+
lines << line
|
31
|
+
end
|
32
|
+
lines
|
33
|
+
end
|
34
|
+
|
35
|
+
|
36
|
+
## note: colon (:) MUST be followed by one (or more) spaces
|
37
|
+
## make sure mon feb 12 18:10 will not match
|
38
|
+
## allow 1. FC Köln etc.
|
39
|
+
## Mainz 05:
|
40
|
+
## limit to 30 chars max
|
41
|
+
## only allow chars incl. intl buut (NOT ()[]/;)
|
42
|
+
##
|
43
|
+
## Group A:
|
44
|
+
## Group B: - remove colon
|
45
|
+
## or lookup first
|
46
|
+
|
47
|
+
ATTRIB_RE = %r{^
|
48
|
+
[ ]*? # slurp leading spaces
|
49
|
+
(?<key>[^:|\]\[()\/; -]
|
50
|
+
[^:|\]\[()\/;]{0,30}
|
51
|
+
)
|
52
|
+
[ ]*? # slurp trailing spaces
|
53
|
+
:[ ]+
|
54
|
+
(?<value>.+)
|
55
|
+
[ ]*? # slurp trailing spaces
|
56
|
+
$
|
57
|
+
}ix
|
58
|
+
|
59
|
+
#
|
60
|
+
# todo/fix: change start to start: too!!!
|
61
|
+
# might be optional in the future!! - why? why not?
|
62
|
+
|
63
|
+
def initialize( lines, start )
|
64
|
+
# for convenience split string into lines
|
65
|
+
## note: removes/strips empty lines
|
66
|
+
## todo/check: change to text instead of array of lines - why? why not?
|
67
|
+
|
68
|
+
## note - wrap in enumerator/iterator a.k.a lines reader
|
69
|
+
@lines = lines.is_a?( String ) ?
|
70
|
+
_read_lines( lines ) : lines
|
71
|
+
|
72
|
+
@start = start
|
73
|
+
end
|
74
|
+
|
75
|
+
|
76
|
+
def parse
|
77
|
+
@last_date = nil
|
78
|
+
@last_time = nil
|
79
|
+
@last_round = nil
|
80
|
+
@last_group = nil
|
81
|
+
|
82
|
+
## last_goals - rename to (longer) @last_team_goals or such - why? why not?
|
83
|
+
@last_goals = 1 ## toggle between 1|2 - hacky (quick & dirty) support for multi-line goals, fix soon!
|
84
|
+
|
85
|
+
@teams = Hash.new(0) ## track counts (only) for now for (interal) team stats - why? why not?
|
86
|
+
@rounds = {}
|
87
|
+
@groups = {}
|
88
|
+
@matches = []
|
89
|
+
|
90
|
+
@warns = [] ## track list of warnings (unmatched lines) too - why? why not?
|
91
|
+
|
92
|
+
|
93
|
+
|
94
|
+
@parser = Parser.new
|
95
|
+
|
96
|
+
@errors = []
|
97
|
+
@tree = []
|
98
|
+
|
99
|
+
attrib_found = false
|
100
|
+
|
101
|
+
@lines.each_with_index do |line,i|
|
102
|
+
|
103
|
+
if debug?
|
104
|
+
puts
|
105
|
+
puts "line >#{line}<"
|
106
|
+
end
|
107
|
+
|
108
|
+
## skip new (experimental attrib syntax)
|
109
|
+
if attrib_found == false &&
|
110
|
+
ATTRIB_RE.match?( line )
|
111
|
+
## note: check attrib regex AFTER group def e.g.:
|
112
|
+
## Group A:
|
113
|
+
## Group B: etc.
|
114
|
+
## todo/fix - change Group A: to Group A etc.
|
115
|
+
## Group B: to Group B
|
116
|
+
attrib_found = true
|
117
|
+
## logger.debug "skipping key/value line - >#{line}<"
|
118
|
+
next
|
119
|
+
end
|
120
|
+
|
121
|
+
if attrib_found
|
122
|
+
## check if line ends with dot
|
123
|
+
## if not slurp up lines to the next do!!!
|
124
|
+
## logger.debug "skipping key/value line - >#{line}<"
|
125
|
+
attrib_found = false if line.end_with?( '.' )
|
126
|
+
# logger.debug "skipping key/value line (cont.) - >#{line}<"
|
127
|
+
next
|
128
|
+
end
|
129
|
+
|
130
|
+
t, error_messages = @parser.parse_with_errors( line )
|
131
|
+
|
132
|
+
|
133
|
+
if error_messages.size > 0
|
134
|
+
## add to "global" error list
|
135
|
+
## make a triplet tuple (file / msg / line text)
|
136
|
+
error_messages.each do |msg|
|
137
|
+
@errors << [ '<file>', ## add filename here
|
138
|
+
msg,
|
139
|
+
line
|
140
|
+
]
|
141
|
+
end
|
142
|
+
end
|
143
|
+
|
144
|
+
pp t if debug?
|
145
|
+
|
146
|
+
@tree << t
|
147
|
+
end # each lines
|
148
|
+
|
149
|
+
## pp @tree
|
150
|
+
|
151
|
+
## report parse errors here - why? why not?
|
152
|
+
|
153
|
+
|
154
|
+
|
155
|
+
@tree.each do |nodes|
|
156
|
+
|
157
|
+
node_type = nodes[0][0] ## get node type of first/head node
|
158
|
+
|
159
|
+
if node_type == :round_def
|
160
|
+
## todo/fix: add round definition (w begin n end date)
|
161
|
+
## todo: do not patch rounds with definition (already assume begin/end date is good)
|
162
|
+
## -- how to deal with matches that get rescheduled/postponed?
|
163
|
+
parse_round_def( nodes )
|
164
|
+
elsif node_type == :group_def ## NB: group goes after round (round may contain group marker too)
|
165
|
+
### todo: add pipe (|) marker (required)
|
166
|
+
parse_group_def( nodes )
|
167
|
+
|
168
|
+
elsif node_type == :player ||
|
169
|
+
node_type == :none # e.g [[:none], [:";"], [:player, "Xhaka"],...]
|
170
|
+
## note - for now goals line MUST start with player!!
|
171
|
+
parse_goals( nodes )
|
172
|
+
else
|
173
|
+
## try to be liberal/flexible
|
174
|
+
## eat-up nodes as we go
|
175
|
+
## assume match with group / round header
|
176
|
+
## etc. on its own line or not
|
177
|
+
|
178
|
+
## preprocess possible before match nodes
|
179
|
+
|
180
|
+
while !nodes.empty? do
|
181
|
+
node_type = nodes[0][0] ## get node type of first/head node
|
182
|
+
if node_type == :round
|
183
|
+
node = nodes.shift ## eat-up
|
184
|
+
parse_round_header( node )
|
185
|
+
elsif node_type == :leg
|
186
|
+
node = nodes.shift ## eat-up
|
187
|
+
## ignore (round) leg for now - add later leg - 1|2|3 etc!!!
|
188
|
+
## needs to get added to db/schema too!!!!
|
189
|
+
## add @last_leg = nil or 1|2|3 etc.
|
190
|
+
elsif node_type == :group
|
191
|
+
## -- lets you set group e.g. Group A etc.
|
192
|
+
node = nodes.shift ## eat-up
|
193
|
+
parse_group_header( node )
|
194
|
+
elsif node_type == :date
|
195
|
+
node = nodes.shift ## eat-up
|
196
|
+
parse_date_header( node )
|
197
|
+
## add time here too - why? why not?
|
198
|
+
## add skip comma separator here too - why? why not?
|
199
|
+
## "slurp-up" in upstream parser?
|
200
|
+
## e.g. round, group or group, round ?
|
201
|
+
else
|
202
|
+
break
|
203
|
+
end
|
204
|
+
end
|
205
|
+
next if nodes.empty?
|
206
|
+
|
207
|
+
## rename to try_parse_match - why? why not?
|
208
|
+
parse_match( nodes )
|
209
|
+
end
|
210
|
+
|
211
|
+
end # tree.each
|
212
|
+
|
213
|
+
## note - team keys are names and values are "internal" stats!!
|
214
|
+
## and NOT team/club/nat_team structs!!
|
215
|
+
[@teams.keys, @matches, @rounds.values, @groups.values]
|
216
|
+
end # method parse
|
217
|
+
|
218
|
+
|
219
|
+
|
220
|
+
def parse_group_header( node )
|
221
|
+
logger.debug "parsing group header: >#{node}<"
|
222
|
+
|
223
|
+
# note: group header resets (last) round (allows, for example):
|
224
|
+
# e.g.
|
225
|
+
# Group Playoffs/Replays -- round header
|
226
|
+
# team1 team2 -- match
|
227
|
+
# Group B -- group header
|
228
|
+
# team1 team2 - match (will get new auto-matchday! not last round)
|
229
|
+
@last_round = nil
|
230
|
+
|
231
|
+
name = node[1]
|
232
|
+
|
233
|
+
group = @groups[ name ]
|
234
|
+
if group.nil?
|
235
|
+
puts "!! PARSE ERROR - no group def found for >#{name}<"
|
236
|
+
exit 1
|
237
|
+
end
|
238
|
+
|
239
|
+
# set group for games
|
240
|
+
@last_group = group
|
241
|
+
end
|
242
|
+
|
243
|
+
|
244
|
+
def parse_group_def( nodes )
|
245
|
+
logger.debug "parsing group def: >#{nodes}<"
|
246
|
+
|
247
|
+
## e.g
|
248
|
+
## [:group_def, "Group A"],
|
249
|
+
## [:team, "Germany"],
|
250
|
+
## [:team, "Scotland"],
|
251
|
+
## [:team, "Hungary"],
|
252
|
+
## [:team, "Switzerland"]
|
253
|
+
|
254
|
+
node = nodes[0]
|
255
|
+
name = node[1] ## group name
|
256
|
+
|
257
|
+
teams = nodes[1..-1].map do |node|
|
258
|
+
if node[0] == :team
|
259
|
+
team = node[1]
|
260
|
+
@teams[ team ] += 1
|
261
|
+
team
|
262
|
+
else
|
263
|
+
puts "!! PARSE ERROR - only teams expected in group def; got:"
|
264
|
+
pp nodes
|
265
|
+
exit 1
|
266
|
+
end
|
267
|
+
end
|
268
|
+
|
269
|
+
## todo/check/fix: add back group key - why? why not?
|
270
|
+
group = Import::Group.new( name: name,
|
271
|
+
teams: teams )
|
272
|
+
|
273
|
+
@groups[ name ] = group
|
274
|
+
end
|
275
|
+
|
276
|
+
|
277
|
+
def _build_date( m:, d:, y:, start: )
|
278
|
+
|
279
|
+
|
280
|
+
## quick debug hack
|
281
|
+
if m == 2 && d == 29
|
282
|
+
puts "quick check feb/29 dates"
|
283
|
+
pp [d,m,y]
|
284
|
+
pp start
|
285
|
+
end
|
286
|
+
|
287
|
+
if y.nil? ## try to calculate year
|
288
|
+
y = if m > start.month ||
|
289
|
+
(m == start.month && d >= start.day)
|
290
|
+
# assume same year as start_at event (e.g. 2013 for 2013/14 season)
|
291
|
+
start.year
|
292
|
+
else
|
293
|
+
# assume year+1 as start_at event (e.g. 2014 for 2013/14 season)
|
294
|
+
start.year+1
|
295
|
+
end
|
296
|
+
end
|
297
|
+
|
298
|
+
|
299
|
+
|
300
|
+
Date.new( y,m,d ) ## y,m,d
|
301
|
+
end
|
302
|
+
|
303
|
+
def parse_round_def( nodes )
|
304
|
+
logger.debug "parsing round def: >#{nodes}<"
|
305
|
+
|
306
|
+
## e.g. [[:round_def, "Matchday 1"], [:duration, "Fri Jun/14 - Tue Jun/18"]]
|
307
|
+
## [[:round_def, "Matchday 2"], [:duration, "Wed Jun/19 - Sat Jun/22"]]
|
308
|
+
## [[:round_def, "Matchday 3"], [:duration, "Sun Jun/23 - Wed Jun/26"]]
|
309
|
+
|
310
|
+
node = nodes[0]
|
311
|
+
name = node[1]
|
312
|
+
# NB: use extracted round name for knockout check
|
313
|
+
# knockout_flag = is_knockout_round?( name )
|
314
|
+
|
315
|
+
node = nodes[1]
|
316
|
+
node_type = node[0]
|
317
|
+
if node_type == :date
|
318
|
+
start_date = end_date = _build_date( m: node[2][:m],
|
319
|
+
d: node[2][:d],
|
320
|
+
y: node[2][:y],
|
321
|
+
start: @start)
|
322
|
+
elsif node_type == :duration
|
323
|
+
start_date = _build_date( m: node[2][:start][:m],
|
324
|
+
d: node[2][:start][:d],
|
325
|
+
y: node[2][:start][:y],
|
326
|
+
start: @start)
|
327
|
+
end_date = _build_date( m: node[2][:end][:m],
|
328
|
+
d: node[2][:end][:d],
|
329
|
+
y: node[2][:end][:y],
|
330
|
+
start: @start)
|
331
|
+
else
|
332
|
+
puts "!! PARSE ERROR - expected date or duration for round def; got:"
|
333
|
+
pp nodes
|
334
|
+
exit 1
|
335
|
+
end
|
336
|
+
|
337
|
+
# note: - NOT needed; start_at and end_at are saved as date only (NOT datetime)
|
338
|
+
# set hours,minutes,secs to beginning and end of day (do NOT use default 12.00)
|
339
|
+
# e.g. use 00.00 and 23.59
|
340
|
+
# start_at = start_at.beginning_of_day
|
341
|
+
# end_at = end_at.end_of_day
|
342
|
+
|
343
|
+
# note: make sure start_at/end_at is date only (e.g. use start_at.to_date)
|
344
|
+
# sqlite3 saves datetime in date field as datetime, for example (will break date compares later!)
|
345
|
+
|
346
|
+
# note - _build_date always returns Date for now - no longer needed!!
|
347
|
+
# start_date = start_date.to_date
|
348
|
+
# end_date = end_date.to_date
|
349
|
+
|
350
|
+
|
351
|
+
## fix:
|
352
|
+
## remove knockout_flag - why? why not?
|
353
|
+
knockout_flag = false
|
354
|
+
|
355
|
+
|
356
|
+
logger.debug " start_date: #{start_date}"
|
357
|
+
logger.debug " end_date: #{end_date}"
|
358
|
+
logger.debug " name: >#{name}<"
|
359
|
+
logger.debug " knockout_flag: #{knockout_flag}"
|
360
|
+
|
361
|
+
round = Import::Round.new( name: name,
|
362
|
+
start_date: start_date,
|
363
|
+
end_date: end_date,
|
364
|
+
knockout: knockout_flag,
|
365
|
+
auto: false )
|
366
|
+
|
367
|
+
@rounds[ name ] = round
|
368
|
+
end
|
369
|
+
|
370
|
+
|
371
|
+
def parse_round_header( node )
|
372
|
+
logger.debug "parsing round header: >#{node}<"
|
373
|
+
|
374
|
+
name = node[1]
|
375
|
+
|
376
|
+
# name = name.sub( ROUND_EXTRA_WORDS_RE, '' )
|
377
|
+
# name = name.strip
|
378
|
+
|
379
|
+
round = @rounds[ name ]
|
380
|
+
if round.nil? ## auto-add / create if missing
|
381
|
+
## todo/check: add num (was pos) if present - why? why not?
|
382
|
+
round = Import::Round.new( name: name )
|
383
|
+
@rounds[ name ] = round
|
384
|
+
end
|
385
|
+
|
386
|
+
## todo/check: if pos match (MUST always match for now)
|
387
|
+
@last_round = round
|
388
|
+
@last_group = nil # note: reset group to no group - why? why not?
|
389
|
+
|
390
|
+
## todo/fix/check
|
391
|
+
## make round a scope for date(time) - why? why not?
|
392
|
+
## reset date/time e.g. @last_date = nil !!!!
|
393
|
+
end
|
394
|
+
|
395
|
+
def parse_date_header( node )
|
396
|
+
logger.debug( "date header: >#{node}<")
|
397
|
+
|
398
|
+
date = _build_date( m: node[2][:m],
|
399
|
+
d: node[2][:d],
|
400
|
+
y: node[2][:y],
|
401
|
+
start: @start )
|
402
|
+
|
403
|
+
logger.debug( " date: #{date} with start: #{@start}")
|
404
|
+
|
405
|
+
@last_date = date # keep a reference for later use
|
406
|
+
@last_time = nil
|
407
|
+
|
408
|
+
### quick "corona" hack - support seasons going beyond 12 month (see swiss league 2019/20 and others!!)
|
409
|
+
## find a better way??
|
410
|
+
## set @start date to full year (e.g. 1.1.) if date.year is @start.year+1
|
411
|
+
## todo/fix: add to linter to check for chronological dates!! - warn if NOT chronological
|
412
|
+
### todo/check: just turn on for 2019/20 season or always? why? why not?
|
413
|
+
|
414
|
+
## todo/fix: add switch back to old @start_org
|
415
|
+
## if year is date.year == @start.year-1 -- possible when full date with year set!!!
|
416
|
+
=begin
|
417
|
+
if @start.month != 1
|
418
|
+
if date.year == @start.year+1
|
419
|
+
logger.debug( "!! hack - extending start date to full (next/end) year; assumes all dates are chronologigal - always moving forward" )
|
420
|
+
@start_org = @start ## keep a copy of the original (old) start date - why? why not? - not used for now
|
421
|
+
@start = Date.new( @start.year+1, 1, 1 )
|
422
|
+
end
|
423
|
+
end
|
424
|
+
=end
|
425
|
+
end
|
426
|
+
|
427
|
+
def parse_minutes( nodes )
|
428
|
+
## parse goals by player
|
429
|
+
## may have multiple minutes!!
|
430
|
+
goals = []
|
431
|
+
|
432
|
+
node = nodes.shift ## get player
|
433
|
+
name = node[1]
|
434
|
+
|
435
|
+
loop do
|
436
|
+
goal = {}
|
437
|
+
goal[:name] = name
|
438
|
+
|
439
|
+
node_type = nodes[0][0]
|
440
|
+
if node_type != :minute
|
441
|
+
puts "!! PARSE ERROR - minute expected to follow player (in goal); got #{node_type}:"
|
442
|
+
pp nodes
|
443
|
+
exit 1
|
444
|
+
end
|
445
|
+
|
446
|
+
node = nodes.shift
|
447
|
+
goal[:minute] = node[2][:m]
|
448
|
+
goal[:offset] = node[2][:offset] if node[2][:offset]
|
449
|
+
|
450
|
+
## check for own goal or penalty or such
|
451
|
+
if !nodes.empty?
|
452
|
+
node_type = nodes[0][0]
|
453
|
+
if node_type == :og
|
454
|
+
nodes.shift
|
455
|
+
goal[:og] = true
|
456
|
+
elsif node_type == :pen
|
457
|
+
nodes.shift
|
458
|
+
goal[:pen] = true
|
459
|
+
else
|
460
|
+
# do nothing
|
461
|
+
end
|
462
|
+
end
|
463
|
+
|
464
|
+
goals << goal
|
465
|
+
|
466
|
+
## check if another minute ahead; otherwise break
|
467
|
+
break if nodes.empty?
|
468
|
+
|
469
|
+
node_type = nodes[0][0]
|
470
|
+
|
471
|
+
## Kane 39', 62', 67'
|
472
|
+
## consume/eat-up (optional?) commas
|
473
|
+
if node_type == :','
|
474
|
+
nodes.shift
|
475
|
+
node_type = nodes[0][0]
|
476
|
+
end
|
477
|
+
|
478
|
+
break if node_type != :minute
|
479
|
+
end
|
480
|
+
|
481
|
+
|
482
|
+
goals
|
483
|
+
end
|
484
|
+
|
485
|
+
|
486
|
+
def parse_goals( nodes )
|
487
|
+
logger.debug "parse goals: >#{nodes}<"
|
488
|
+
|
489
|
+
goals1 = []
|
490
|
+
goals2 = []
|
491
|
+
|
492
|
+
while !nodes.empty?
|
493
|
+
node_type = nodes[0][0]
|
494
|
+
if node_type == :player
|
495
|
+
more_goals = parse_minutes( nodes )
|
496
|
+
## hacky multi-line support for goals
|
497
|
+
## using last_goal (1|2)
|
498
|
+
@last_goals == 2 ? goals2 += more_goals :
|
499
|
+
goals1 += more_goals
|
500
|
+
elsif node_type == :';' ## team separator
|
501
|
+
nodes.shift # eat-up
|
502
|
+
@last_goals = 2
|
503
|
+
elsif node_type == :none
|
504
|
+
nodes.shift # eat-up
|
505
|
+
else
|
506
|
+
puts "!! PARSE ERROR - unexpected node type in goals;; got #{node_type}:"
|
507
|
+
pp nodes
|
508
|
+
exit 1
|
509
|
+
end
|
510
|
+
end
|
511
|
+
|
512
|
+
pp [goals1,goals2]
|
513
|
+
|
514
|
+
## wrap in struct andd add/append to match
|
515
|
+
=begin
|
516
|
+
class GoalStruct
|
517
|
+
######
|
518
|
+
# flat struct for goals - one entry per goals
|
519
|
+
attr_accessor :name
|
520
|
+
attr_accessor :team # 1 or 2 ? check/todo: add team1 or team2 flag?
|
521
|
+
attr_accessor :minute, :offset
|
522
|
+
attr_accessor :penalty, :owngoal
|
523
|
+
attr_accessor :score1, :score2 # gets calculated
|
524
|
+
=end
|
525
|
+
|
526
|
+
goals = []
|
527
|
+
goals1.each do |rec|
|
528
|
+
goal = Import::Goal.new(
|
529
|
+
player: rec[:name],
|
530
|
+
team: 1,
|
531
|
+
minute: rec[:minute],
|
532
|
+
offset: rec[:offset],
|
533
|
+
penalty: rec[:pen] || false, # note: pass along/use false NOT nil
|
534
|
+
owngoal: rec[:og] || false
|
535
|
+
)
|
536
|
+
goals << goal
|
537
|
+
end
|
538
|
+
goals2.each do |rec|
|
539
|
+
goal = Import::Goal.new(
|
540
|
+
player: rec[:name],
|
541
|
+
team: 2,
|
542
|
+
minute: rec[:minute],
|
543
|
+
offset: rec[:offset],
|
544
|
+
penalty: rec[:pen] || false, # note: pass along/use false NOT nil
|
545
|
+
owngoal: rec[:og] || false
|
546
|
+
)
|
547
|
+
goals << goal
|
548
|
+
end
|
549
|
+
|
550
|
+
pp goals
|
551
|
+
|
552
|
+
## quick & dirty - auto add goals to last match
|
553
|
+
## note - for hacky (quick& dirty) multi-line support
|
554
|
+
## always append for now
|
555
|
+
match = @matches[-1]
|
556
|
+
match.goals ||= []
|
557
|
+
match.goals += goals
|
558
|
+
|
559
|
+
## todo/fix
|
560
|
+
## sort by minute
|
561
|
+
## PLUS auto-fill score1,score2 - why? why not?
|
562
|
+
end
|
563
|
+
|
564
|
+
|
565
|
+
def parse_match( nodes )
|
566
|
+
logger.debug( "parse match: >#{nodes}<" )
|
567
|
+
|
568
|
+
## collect (possible) nodes by type
|
569
|
+
num = nil
|
570
|
+
date = nil
|
571
|
+
time = nil
|
572
|
+
teams = []
|
573
|
+
score = nil
|
574
|
+
more = []
|
575
|
+
|
576
|
+
while !nodes.empty?
|
577
|
+
node = nodes.shift
|
578
|
+
node_type = node[0]
|
579
|
+
|
580
|
+
if node_type == :num
|
581
|
+
num = node[1]
|
582
|
+
elsif node_type == :date
|
583
|
+
## note: date wipes out/clear time
|
584
|
+
## time MUST always come after date
|
585
|
+
time = nil
|
586
|
+
date = _build_date( m: node[2][:m],
|
587
|
+
d: node[2][:d],
|
588
|
+
y: node[2][:y],
|
589
|
+
start: @start )
|
590
|
+
elsif node_type == :time
|
591
|
+
## note - there's no time (-only) type in ruby
|
592
|
+
## use string (e.g. '14:56', '1:44')
|
593
|
+
## use 01:44 or 1:44 ?
|
594
|
+
## check for 0:00 or 24:00 possible?
|
595
|
+
time = '%d:%02d' % [node[2][:h], node[2][:m]]
|
596
|
+
elsif node_type == :team
|
597
|
+
teams << node[1]
|
598
|
+
elsif node_type == :score
|
599
|
+
### todo/fix
|
600
|
+
## add keywords (e.g. ht, ft or such) to Score.new - why? why not?
|
601
|
+
## or use new Score.build( ht:, ft:, ) or such - why? why not?
|
602
|
+
ht = node[2][:ht] || [nil,nil]
|
603
|
+
ft = node[2][:ft] || [nil,nil]
|
604
|
+
et = node[2][:et] || [nil,nil]
|
605
|
+
p = node[2][:p] || [nil,nil]
|
606
|
+
values = [*ht, *ft, *et, *p]
|
607
|
+
## pp values
|
608
|
+
|
609
|
+
score = Score.new( *values )
|
610
|
+
## pp score
|
611
|
+
elsif node_type == :vs
|
612
|
+
## skip; do nothing
|
613
|
+
##
|
614
|
+
## todo - add ## find (optional) match status e.g. [abandoned] or [replay] or [awarded]
|
615
|
+
## or [cancelled] or [postponed] etc.
|
616
|
+
## status = find_status!( line ) ## todo/check: allow match status also in geo part (e.g. after @) - why? why not?
|
617
|
+
|
618
|
+
elsif node_type == :'@' ||
|
619
|
+
node_type == :',' ||
|
620
|
+
node_type == :geo ||
|
621
|
+
node_type == :timezone
|
622
|
+
## e.g.
|
623
|
+
## [:"@"], [:geo, "Stade de France"], [:","], [:geo, "Saint-Denis"]]
|
624
|
+
## [:"@"], [:geo, "Arena de São Paulo"], [:","], [:geo, "São Paulo"], [:timezone, "(UTC-3)"]
|
625
|
+
more << node[1] if node_type == :geo
|
626
|
+
else
|
627
|
+
puts "!! PARSE ERROR - unexpected node type #{node_type} in match line; got:"
|
628
|
+
pp node
|
629
|
+
exit 1
|
630
|
+
end
|
631
|
+
end
|
632
|
+
|
633
|
+
|
634
|
+
if teams.size != 2
|
635
|
+
puts "!! PARSE ERROR - expected two teams; got #{teams.size}:"
|
636
|
+
pp teams
|
637
|
+
exit 1
|
638
|
+
end
|
639
|
+
|
640
|
+
team1 = teams[0]
|
641
|
+
team2 = teams[1]
|
642
|
+
|
643
|
+
@teams[ team1 ] += 1
|
644
|
+
@teams[ team2 ] += 1
|
645
|
+
|
646
|
+
|
647
|
+
###
|
648
|
+
# check if date found?
|
649
|
+
# note: ruby falsey is nil & false only (not 0 or empty array etc.)
|
650
|
+
if date
|
651
|
+
### check: use date_v2 if present? why? why not?
|
652
|
+
@last_date = date # keep a reference for later use
|
653
|
+
@last_time = nil
|
654
|
+
# @last_time = nil
|
655
|
+
else
|
656
|
+
date = @last_date # no date found; (re)use last seen date
|
657
|
+
end
|
658
|
+
|
659
|
+
if time
|
660
|
+
@last_time = time
|
661
|
+
else
|
662
|
+
time = @last_time
|
663
|
+
end
|
664
|
+
|
665
|
+
|
666
|
+
round = nil
|
667
|
+
if @last_round
|
668
|
+
round = @last_round
|
669
|
+
else
|
670
|
+
## find (first) matching round by date if rounds / matchdays defined
|
671
|
+
## if not rounds / matchdays defined - YES, allow matches WITHOUT rounds!!!
|
672
|
+
if @rounds.size > 0
|
673
|
+
@rounds.values.each do |round_rec|
|
674
|
+
## note: convert date to date only (no time) with to_date!!!
|
675
|
+
if (round_rec.start_date && round_rec.end_date) &&
|
676
|
+
(date.to_date >= round_rec.start_date &&
|
677
|
+
date.to_date <= round_rec.end_date)
|
678
|
+
round = round_rec
|
679
|
+
break
|
680
|
+
end
|
681
|
+
end
|
682
|
+
if round.nil?
|
683
|
+
puts "!! PARSE ERROR - no matching round found for match date:"
|
684
|
+
pp date
|
685
|
+
exit 1
|
686
|
+
end
|
687
|
+
end
|
688
|
+
end
|
689
|
+
|
690
|
+
## todo/check: scores are integers or strings?
|
691
|
+
|
692
|
+
## todo/check: pass along round and group refs or just string (canonical names) - why? why not?
|
693
|
+
|
694
|
+
## split date in date & time if DateTime
|
695
|
+
=begin
|
696
|
+
time_str = nil
|
697
|
+
date_str = nil
|
698
|
+
if date.is_a?( DateTime )
|
699
|
+
date_str = date.strftime('%Y-%m-%d')
|
700
|
+
time_str = date.strftime('%H:%M')
|
701
|
+
elsif date.is_a?( Date )
|
702
|
+
date_str = date.strftime('%Y-%m-%d')
|
703
|
+
else # assume date is nil
|
704
|
+
end
|
705
|
+
=end
|
706
|
+
|
707
|
+
time_str = nil
|
708
|
+
date_str = nil
|
709
|
+
|
710
|
+
date_str = date.strftime('%Y-%m-%d') if date
|
711
|
+
time_str = time if date && time
|
712
|
+
|
713
|
+
|
714
|
+
status = nil
|
715
|
+
ground = nil
|
716
|
+
|
717
|
+
@matches << Import::Match.new( num: num,
|
718
|
+
date: date_str,
|
719
|
+
time: time_str,
|
720
|
+
team1: team1, ## note: for now always use mapping value e.g. rec (NOT string e.g. team1.name)
|
721
|
+
team2: team2, ## note: for now always use mapping value e.g. rec (NOT string e.g. team2.name)
|
722
|
+
score: score,
|
723
|
+
round: round ? round.name : nil, ## note: for now always use string (assume unique canonical name for event)
|
724
|
+
group: @last_group ? @last_group.name : nil, ## note: for now always use string (assume unique canonical name for event)
|
725
|
+
status: status,
|
726
|
+
ground: ground )
|
727
|
+
### todo: cache team lookups in hash?
|
728
|
+
|
729
|
+
## hacky goals support
|
730
|
+
### reset/toggle 1/2
|
731
|
+
@last_goals = 1
|
732
|
+
end
|
733
|
+
end # class MatchParser
|
734
|
+
end # module SportDb
|
735
|
+
|