sportdb-config 0.4.0 → 0.4.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/config/leagues/fr.txt +9 -9
- data/config/leagues/gr.txt +7 -7
- data/config/leagues/sco.txt +19 -19
- data/config/world/eng.txt +162 -162
- data/lib/sportdb/config/club_reader.rb +278 -278
- data/lib/sportdb/config/clubs.rb +7 -0
- data/lib/sportdb/config/config.rb +123 -123
- data/lib/sportdb/config/league.rb +118 -118
- data/lib/sportdb/config/league_reader.rb +65 -65
- data/lib/sportdb/config/league_utils.rb +24 -24
- data/lib/sportdb/config/variants.rb +91 -81
- data/lib/sportdb/config/version.rb +1 -1
- data/test/test_club_reader.rb +150 -150
- data/test/test_league_reader.rb +54 -54
- data/test/test_league_utils.rb +46 -46
- data/test/test_season_utils.rb +29 -29
- data/test/test_variants.rb +14 -0
- metadata +12 -6
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA1:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 5cc9b8eea5116ab13e92c7a5a6b3b7962fa3f1c1
|
|
4
|
+
data.tar.gz: a94f4452d3398743f2e6283cead72ccc552d2ec0
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 6db0746eda38483b0bfcc241730044805c9332a0f2b27ad2aa5b96586371121579c8ef75a1903d482a7c1f68534a4231877a24fedaf87a866399e9f2105439f5
|
|
7
|
+
data.tar.gz: c5803e3626a36b8949d6b774e2b43748c27fa9c819a7ac9984c633f613f8b81ad1c9a026c0bafc3f1c66023d75b937675d6196ea01167b47f89ece4c14dde6fe
|
data/config/leagues/fr.txt
CHANGED
|
@@ -1,9 +1,9 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
1 => ligue1, French Ligue 1
|
|
4
|
-
2 => ligue2, French Ligue 2
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
[2001-02] ## until (including) 2001-02 season
|
|
8
|
-
1 => division1, ? ## use championat or something? check official name
|
|
9
|
-
2 => division2, ?
|
|
1
|
+
|
|
2
|
+
|
|
3
|
+
1 => ligue1, French Ligue 1
|
|
4
|
+
2 => ligue2, French Ligue 2
|
|
5
|
+
|
|
6
|
+
|
|
7
|
+
[2001-02] ## until (including) 2001-02 season
|
|
8
|
+
1 => division1, ? ## use championat or something? check official name
|
|
9
|
+
2 => division2, ?
|
data/config/leagues/gr.txt
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
|
-
|
|
2
|
-
|
|
3
|
-
1 => superleague, Greek Superleague
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
[2005-06] ## until (including) 2005-06 season
|
|
7
|
-
1 => alphaethniki, Greek Alpha Ethniki
|
|
1
|
+
|
|
2
|
+
|
|
3
|
+
1 => superleague, Greek Superleague
|
|
4
|
+
|
|
5
|
+
|
|
6
|
+
[2005-06] ## until (including) 2005-06 season
|
|
7
|
+
1 => alphaethniki, Greek Alpha Ethniki
|
data/config/leagues/sco.txt
CHANGED
|
@@ -1,19 +1,19 @@
|
|
|
1
|
-
|
|
2
|
-
1 => premiership, Scotish Premiership # starting w/ 2013-14 season
|
|
3
|
-
2 => championship, Scotish Championship
|
|
4
|
-
3 => league1, Scotish League One
|
|
5
|
-
4 => league2, Scotish League Two
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
[2012-13] ## until (including) 2012-13 season
|
|
9
|
-
1 => premierleague, Scotish Premiership
|
|
10
|
-
2 => division1, Scotish 1st Division
|
|
11
|
-
3 => division2, Scotish League One
|
|
12
|
-
4 => division3, Scotish League Two
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
[1997-98] ## until (including) season
|
|
16
|
-
1 => premierdivision, Scotish Premier Division
|
|
17
|
-
2 => division1, Scotish 1st Division
|
|
18
|
-
3 => division2, Scotish 2nd Division
|
|
19
|
-
4 => division3, Scotish 3rd Division
|
|
1
|
+
|
|
2
|
+
1 => premiership, Scotish Premiership # starting w/ 2013-14 season
|
|
3
|
+
2 => championship, Scotish Championship
|
|
4
|
+
3 => league1, Scotish League One
|
|
5
|
+
4 => league2, Scotish League Two
|
|
6
|
+
|
|
7
|
+
|
|
8
|
+
[2012-13] ## until (including) 2012-13 season
|
|
9
|
+
1 => premierleague, Scotish Premiership
|
|
10
|
+
2 => division1, Scotish 1st Division
|
|
11
|
+
3 => division2, Scotish League One
|
|
12
|
+
4 => division3, Scotish League Two
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
[1997-98] ## until (including) season
|
|
16
|
+
1 => premierdivision, Scotish Premier Division
|
|
17
|
+
2 => division1, Scotish 1st Division
|
|
18
|
+
3 => division2, Scotish 2nd Division
|
|
19
|
+
4 => division3, Scotish 3rd Division
|
data/config/world/eng.txt
CHANGED
|
@@ -1,162 +1,162 @@
|
|
|
1
|
-
##
|
|
2
|
-
# what name? use regions or maps or geos or zones or __?
|
|
3
|
-
|
|
4
|
-
#
|
|
5
|
-
# england
|
|
6
|
-
# see https://en.wikipedia.org/wiki/Subdivisions_of_England
|
|
7
|
-
# see https://www.bbc.co.uk/news/england/regions
|
|
8
|
-
|
|
9
|
-
##
|
|
10
|
-
# todo: for sort order - allow different sort name
|
|
11
|
-
# e.g. East Sussex => Sussex (East) or Sussex, East or something
|
|
12
|
-
# use <> for marking what counts for sorting
|
|
13
|
-
# e.g East <Sussex> => get auto-converted to Sussex, East - why? why not?
|
|
14
|
-
#
|
|
15
|
-
# more examples:
|
|
16
|
-
# North ‹Yorkshire› => Yorkshire (North) or Yorkshire, North
|
|
17
|
-
# Greater ‹London› => London, Greater
|
|
18
|
-
# Greater ‹Manchaster› => Manchaster, Greater
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
##
|
|
22
|
-
# note: uses all regions following the camra good beer guide (book)
|
|
23
|
-
#
|
|
24
|
-
|
|
25
|
-
======================================
|
|
26
|
-
== North West England ==
|
|
27
|
-
|
|
28
|
-
# Cumbria
|
|
29
|
-
# Lancashire
|
|
30
|
-
# Liverpool
|
|
31
|
-
# Manchester
|
|
32
|
-
|
|
33
|
-
Cheshire
|
|
34
|
-
Cumbria
|
|
35
|
-
Lancashire
|
|
36
|
-
Greater Manchester | Manchester, Greater
|
|
37
|
-
Liverpool & Merseyside | Merseyside ## note: added Liverpool & - why? why not?
|
|
38
|
-
|
|
39
|
-
|
|
40
|
-
===================================
|
|
41
|
-
== North East England
|
|
42
|
-
|
|
43
|
-
# Tees
|
|
44
|
-
# Tyne & Wear
|
|
45
|
-
|
|
46
|
-
Durham
|
|
47
|
-
Newcastle & Northumberland | Northumberland ## note: added Newcastle & - why? why not?
|
|
48
|
-
Tyne and Wear
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
=====================================
|
|
52
|
-
== Yorkshire & Lincolnshire ==
|
|
53
|
-
|
|
54
|
-
# Humberside
|
|
55
|
-
# Leeds & West Yorkshire
|
|
56
|
-
# Lincolnshire
|
|
57
|
-
# Sheffield & South Yorkshire
|
|
58
|
-
# York & North Yorkshire
|
|
59
|
-
|
|
60
|
-
Lincolnshire
|
|
61
|
-
West Yorkshire | Yorkshire, West | Yorkshire (West)
|
|
62
|
-
South Yorkshire | Yorkshire, South | Yorkshire (South)
|
|
63
|
-
North Yorkshire | Yorkshire, North | Yorkshire (North)
|
|
64
|
-
East Yorkshire | Yorkshire, East | Yorkshire (East)
|
|
65
|
-
|
|
66
|
-
|
|
67
|
-
==================================
|
|
68
|
-
== West Midlands
|
|
69
|
-
|
|
70
|
-
# Birmingham & Black Country
|
|
71
|
-
# Coventry & Warwickshire
|
|
72
|
-
# Hereford & Worcester
|
|
73
|
-
# Shropshire
|
|
74
|
-
# Stoke & Staffordshire
|
|
75
|
-
|
|
76
|
-
Herefordshire
|
|
77
|
-
Shropshire
|
|
78
|
-
Staffordshire
|
|
79
|
-
Warwickshire
|
|
80
|
-
Birmingham & West Midlands | West Midlands ## note: added Birmingham - why? why not?
|
|
81
|
-
Worcestershire
|
|
82
|
-
|
|
83
|
-
|
|
84
|
-
==================================
|
|
85
|
-
== East Midlands
|
|
86
|
-
|
|
87
|
-
# Derby
|
|
88
|
-
# Leicester
|
|
89
|
-
# Northampton
|
|
90
|
-
# Nottingham
|
|
91
|
-
|
|
92
|
-
Derbyshire
|
|
93
|
-
Leicestershire
|
|
94
|
-
Northamptonshire
|
|
95
|
-
Nottinghamshire
|
|
96
|
-
Rutland
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
===================================
|
|
100
|
-
== West & South West
|
|
101
|
-
|
|
102
|
-
# Bristol
|
|
103
|
-
# Cornwall
|
|
104
|
-
# Devon
|
|
105
|
-
# Gloucestershire
|
|
106
|
-
# Somerset
|
|
107
|
-
# Wiltshire
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
Cornwall
|
|
111
|
-
Devon
|
|
112
|
-
Bristol & Gloucestershire | Gloucestershire
|
|
113
|
-
Somerset
|
|
114
|
-
Wiltshire
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
=================================
|
|
118
|
-
== East
|
|
119
|
-
|
|
120
|
-
# Beds, Herts & Bucks
|
|
121
|
-
# Cambridgeshire
|
|
122
|
-
# Essex
|
|
123
|
-
# Norfolk
|
|
124
|
-
# Suffolk
|
|
125
|
-
|
|
126
|
-
Bedfordshire
|
|
127
|
-
Hertfordshire
|
|
128
|
-
Buckinghamshire
|
|
129
|
-
Cambridgeshire
|
|
130
|
-
Essex
|
|
131
|
-
Norfolk
|
|
132
|
-
Suffolk
|
|
133
|
-
|
|
134
|
-
|
|
135
|
-
===============================
|
|
136
|
-
== South
|
|
137
|
-
|
|
138
|
-
# Berkshire
|
|
139
|
-
# Dorset
|
|
140
|
-
# Hampshire & Isle of Wight
|
|
141
|
-
# Oxford
|
|
142
|
-
|
|
143
|
-
Berkshire
|
|
144
|
-
Dorset
|
|
145
|
-
Hampshire
|
|
146
|
-
Oxfordshire
|
|
147
|
-
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
===============================
|
|
151
|
-
== London & South East
|
|
152
|
-
|
|
153
|
-
# Kent
|
|
154
|
-
# London
|
|
155
|
-
# Surrey
|
|
156
|
-
# Sussex
|
|
157
|
-
|
|
158
|
-
Greater London | London, Greater
|
|
159
|
-
Kent
|
|
160
|
-
Surrey
|
|
161
|
-
East Sussex | Sussex, East | Sussex (East)
|
|
162
|
-
West Sussex | Sussex, West | Sussex (West)
|
|
1
|
+
##
|
|
2
|
+
# what name? use regions or maps or geos or zones or __?
|
|
3
|
+
|
|
4
|
+
#
|
|
5
|
+
# england
|
|
6
|
+
# see https://en.wikipedia.org/wiki/Subdivisions_of_England
|
|
7
|
+
# see https://www.bbc.co.uk/news/england/regions
|
|
8
|
+
|
|
9
|
+
##
|
|
10
|
+
# todo: for sort order - allow different sort name
|
|
11
|
+
# e.g. East Sussex => Sussex (East) or Sussex, East or something
|
|
12
|
+
# use <> for marking what counts for sorting
|
|
13
|
+
# e.g East <Sussex> => get auto-converted to Sussex, East - why? why not?
|
|
14
|
+
#
|
|
15
|
+
# more examples:
|
|
16
|
+
# North ‹Yorkshire› => Yorkshire (North) or Yorkshire, North
|
|
17
|
+
# Greater ‹London› => London, Greater
|
|
18
|
+
# Greater ‹Manchaster› => Manchaster, Greater
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
##
|
|
22
|
+
# note: uses all regions following the camra good beer guide (book)
|
|
23
|
+
#
|
|
24
|
+
|
|
25
|
+
======================================
|
|
26
|
+
== North West England ==
|
|
27
|
+
|
|
28
|
+
# Cumbria
|
|
29
|
+
# Lancashire
|
|
30
|
+
# Liverpool
|
|
31
|
+
# Manchester
|
|
32
|
+
|
|
33
|
+
Cheshire
|
|
34
|
+
Cumbria
|
|
35
|
+
Lancashire
|
|
36
|
+
Greater Manchester | Manchester, Greater
|
|
37
|
+
Liverpool & Merseyside | Merseyside ## note: added Liverpool & - why? why not?
|
|
38
|
+
|
|
39
|
+
|
|
40
|
+
===================================
|
|
41
|
+
== North East England
|
|
42
|
+
|
|
43
|
+
# Tees
|
|
44
|
+
# Tyne & Wear
|
|
45
|
+
|
|
46
|
+
Durham
|
|
47
|
+
Newcastle & Northumberland | Northumberland ## note: added Newcastle & - why? why not?
|
|
48
|
+
Tyne and Wear
|
|
49
|
+
|
|
50
|
+
|
|
51
|
+
=====================================
|
|
52
|
+
== Yorkshire & Lincolnshire ==
|
|
53
|
+
|
|
54
|
+
# Humberside
|
|
55
|
+
# Leeds & West Yorkshire
|
|
56
|
+
# Lincolnshire
|
|
57
|
+
# Sheffield & South Yorkshire
|
|
58
|
+
# York & North Yorkshire
|
|
59
|
+
|
|
60
|
+
Lincolnshire
|
|
61
|
+
West Yorkshire | Yorkshire, West | Yorkshire (West)
|
|
62
|
+
South Yorkshire | Yorkshire, South | Yorkshire (South)
|
|
63
|
+
North Yorkshire | Yorkshire, North | Yorkshire (North)
|
|
64
|
+
East Yorkshire | Yorkshire, East | Yorkshire (East)
|
|
65
|
+
|
|
66
|
+
|
|
67
|
+
==================================
|
|
68
|
+
== West Midlands
|
|
69
|
+
|
|
70
|
+
# Birmingham & Black Country
|
|
71
|
+
# Coventry & Warwickshire
|
|
72
|
+
# Hereford & Worcester
|
|
73
|
+
# Shropshire
|
|
74
|
+
# Stoke & Staffordshire
|
|
75
|
+
|
|
76
|
+
Herefordshire
|
|
77
|
+
Shropshire
|
|
78
|
+
Staffordshire
|
|
79
|
+
Warwickshire
|
|
80
|
+
Birmingham & West Midlands | West Midlands ## note: added Birmingham - why? why not?
|
|
81
|
+
Worcestershire
|
|
82
|
+
|
|
83
|
+
|
|
84
|
+
==================================
|
|
85
|
+
== East Midlands
|
|
86
|
+
|
|
87
|
+
# Derby
|
|
88
|
+
# Leicester
|
|
89
|
+
# Northampton
|
|
90
|
+
# Nottingham
|
|
91
|
+
|
|
92
|
+
Derbyshire
|
|
93
|
+
Leicestershire
|
|
94
|
+
Northamptonshire
|
|
95
|
+
Nottinghamshire
|
|
96
|
+
Rutland
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
===================================
|
|
100
|
+
== West & South West
|
|
101
|
+
|
|
102
|
+
# Bristol
|
|
103
|
+
# Cornwall
|
|
104
|
+
# Devon
|
|
105
|
+
# Gloucestershire
|
|
106
|
+
# Somerset
|
|
107
|
+
# Wiltshire
|
|
108
|
+
|
|
109
|
+
|
|
110
|
+
Cornwall
|
|
111
|
+
Devon
|
|
112
|
+
Bristol & Gloucestershire | Gloucestershire
|
|
113
|
+
Somerset
|
|
114
|
+
Wiltshire
|
|
115
|
+
|
|
116
|
+
|
|
117
|
+
=================================
|
|
118
|
+
== East
|
|
119
|
+
|
|
120
|
+
# Beds, Herts & Bucks
|
|
121
|
+
# Cambridgeshire
|
|
122
|
+
# Essex
|
|
123
|
+
# Norfolk
|
|
124
|
+
# Suffolk
|
|
125
|
+
|
|
126
|
+
Bedfordshire
|
|
127
|
+
Hertfordshire
|
|
128
|
+
Buckinghamshire
|
|
129
|
+
Cambridgeshire
|
|
130
|
+
Essex
|
|
131
|
+
Norfolk
|
|
132
|
+
Suffolk
|
|
133
|
+
|
|
134
|
+
|
|
135
|
+
===============================
|
|
136
|
+
== South
|
|
137
|
+
|
|
138
|
+
# Berkshire
|
|
139
|
+
# Dorset
|
|
140
|
+
# Hampshire & Isle of Wight
|
|
141
|
+
# Oxford
|
|
142
|
+
|
|
143
|
+
Berkshire
|
|
144
|
+
Dorset
|
|
145
|
+
Hampshire
|
|
146
|
+
Oxfordshire
|
|
147
|
+
|
|
148
|
+
|
|
149
|
+
|
|
150
|
+
===============================
|
|
151
|
+
== London & South East
|
|
152
|
+
|
|
153
|
+
# Kent
|
|
154
|
+
# London
|
|
155
|
+
# Surrey
|
|
156
|
+
# Sussex
|
|
157
|
+
|
|
158
|
+
Greater London | London, Greater
|
|
159
|
+
Kent
|
|
160
|
+
Surrey
|
|
161
|
+
East Sussex | Sussex, East | Sussex (East)
|
|
162
|
+
West Sussex | Sussex, West | Sussex (West)
|
|
@@ -1,278 +1,278 @@
|
|
|
1
|
-
# encoding: utf-8
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
module SportDb
|
|
5
|
-
module Import
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
class ClubReader
|
|
9
|
-
|
|
10
|
-
|
|
11
|
-
def self.read( path ) ## use - rename to read_file or from_file etc. - why? why not?
|
|
12
|
-
txt = File.open( path, 'r:utf-8' ).read
|
|
13
|
-
parse( txt )
|
|
14
|
-
end
|
|
15
|
-
|
|
16
|
-
|
|
17
|
-
def self.parse( txt )
|
|
18
|
-
recs = []
|
|
19
|
-
last_rec = nil
|
|
20
|
-
headings = [] ## headings stack
|
|
21
|
-
|
|
22
|
-
txt.each_line do |line|
|
|
23
|
-
line = line.strip
|
|
24
|
-
|
|
25
|
-
next if line.empty?
|
|
26
|
-
next if line.start_with?( '#' ) ## skip comments too
|
|
27
|
-
|
|
28
|
-
## strip inline (until end-of-line) comments too
|
|
29
|
-
## e.g Eupen => KAS Eupen, ## [de]
|
|
30
|
-
## => Eupen => KAS Eupen,
|
|
31
|
-
line = line.sub( /#.*/, '' ).strip
|
|
32
|
-
pp line
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
next if line =~ /^={1,}$/ ## skip "decorative" only heading e.g. ========
|
|
36
|
-
|
|
37
|
-
## note: like in wikimedia markup (and markdown) all optional trailing ==== too
|
|
38
|
-
## todo/check: allow === Text =-=-=-=-=-= too - why? why not?
|
|
39
|
-
if line =~ /^(={1,}) ## leading ======
|
|
40
|
-
([^=]+?) ## text (note: for now no "inline" = allowed)
|
|
41
|
-
=* ## (optional) trailing ====
|
|
42
|
-
$/x
|
|
43
|
-
heading_marker = $1
|
|
44
|
-
heading_level = $1.length ## count number of = for heading level
|
|
45
|
-
heading = $2.strip
|
|
46
|
-
|
|
47
|
-
puts "heading #{heading_level} >#{heading}<"
|
|
48
|
-
|
|
49
|
-
## 1) first pop headings if present
|
|
50
|
-
while headings.size+1 > heading_level
|
|
51
|
-
headings.pop
|
|
52
|
-
end
|
|
53
|
-
|
|
54
|
-
## 2) add missing (hierarchy) level if
|
|
55
|
-
while headings.size+1 < heading_level
|
|
56
|
-
## todo/fix: issue warning about "skipping" hierarchy level
|
|
57
|
-
puts "!!! warn [team reader] - skipping hierarchy level in headings "
|
|
58
|
-
headings.push( nil )
|
|
59
|
-
end
|
|
60
|
-
|
|
61
|
-
if heading =~ /^\?+$/ ## note: use ? or ?? or ?? to reset level to nil
|
|
62
|
-
## keep level empty
|
|
63
|
-
else
|
|
64
|
-
|
|
65
|
-
## quick hack: if level is 1 assume country for now
|
|
66
|
-
## and extract country code e.g.
|
|
67
|
-
## Austria (at) => at
|
|
68
|
-
## todo/fix: allow code only e.g. at or aut without enclosing () too - why? why not?
|
|
69
|
-
if heading_level == 1
|
|
70
|
-
if heading =~ /\(([a-z]{2,3})\)/i ## note allow (at) or (AUT) too
|
|
71
|
-
country_code = $1
|
|
72
|
-
|
|
73
|
-
## check country code - MUST exist for now!!!!
|
|
74
|
-
country = SportDb::Import.config.countries[ country_code ]
|
|
75
|
-
if country.nil?
|
|
76
|
-
puts "!!! error [team reader] - unknown country with code >#{country_code}< - sorry - add country to config to fix"
|
|
77
|
-
exit 1
|
|
78
|
-
end
|
|
79
|
-
|
|
80
|
-
headings.push( country_code )
|
|
81
|
-
else
|
|
82
|
-
puts "!!! error - heading level 1 - missing country code - >#{heading}<"
|
|
83
|
-
exit 1
|
|
84
|
-
end
|
|
85
|
-
else
|
|
86
|
-
## quick hack:
|
|
87
|
-
## remove known fill/dummy words incl:
|
|
88
|
-
## Provincia San Juan => San Juan (see argentina, for example)
|
|
89
|
-
##
|
|
90
|
-
## use geo tree long term with alternative names - why? why not?
|
|
91
|
-
words = ['Provincia']
|
|
92
|
-
words.each { |word| heading = heading.gsub( word, '' ) }
|
|
93
|
-
heading = heading.strip
|
|
94
|
-
|
|
95
|
-
headings.push( heading )
|
|
96
|
-
end
|
|
97
|
-
|
|
98
|
-
## assert that hierarchy level is ok
|
|
99
|
-
if headings.size != heading_level
|
|
100
|
-
puts "!!! error - headings hierarchy/stack out of order - #{heading.size}<=>#{heading_level}"
|
|
101
|
-
exit 1
|
|
102
|
-
end
|
|
103
|
-
end
|
|
104
|
-
|
|
105
|
-
pp headings
|
|
106
|
-
|
|
107
|
-
elsif line.start_with?( '|' )
|
|
108
|
-
## assume continuation with line of alternative names
|
|
109
|
-
## note: skip leading pipe
|
|
110
|
-
values = line[1..-1].split( '|' ) # team names - allow/use pipe(|)
|
|
111
|
-
## strip and squish (white)spaces
|
|
112
|
-
# e.g. New York FC (2011-) => New York FC (2011-)
|
|
113
|
-
values = values.map { |value| value.strip.gsub( /[ \t]+/, ' ' ) }
|
|
114
|
-
last_rec.alt_names += values
|
|
115
|
-
last_rec.add_variants( values ) # auto-add (possible) auto-generated variant names
|
|
116
|
-
|
|
117
|
-
## check for duplicates
|
|
118
|
-
if last_rec.duplicates?
|
|
119
|
-
duplicates = last_rec.duplicates
|
|
120
|
-
puts "*** !!! WARN !!! - #{duplicates.size} duplicate alt name mapping(s):"
|
|
121
|
-
pp duplicates
|
|
122
|
-
pp last_rec
|
|
123
|
-
##
|
|
124
|
-
## todo/fix: make it only an error with exit 1
|
|
125
|
-
## if (not normalized) names are the same (not unique/uniq)
|
|
126
|
-
## e.g. don't exit on A.F.C. == AFC etc.
|
|
127
|
-
## exit 1
|
|
128
|
-
end
|
|
129
|
-
else
|
|
130
|
-
values = line.split( ',' )
|
|
131
|
-
|
|
132
|
-
rec = Club.new
|
|
133
|
-
value = values.shift ## get first item
|
|
134
|
-
## strip and squish (white)spaces
|
|
135
|
-
# e.g. New York FC (2011-) => New York FC (2011-)
|
|
136
|
-
value = value.strip.gsub( /[ \t]+/, ' ' )
|
|
137
|
-
rec.name = value # canoncial name (global unique "beautiful/long" name)
|
|
138
|
-
rec.add_variants( value ) # auto-add (possible) auto-generated variant names
|
|
139
|
-
|
|
140
|
-
## note:
|
|
141
|
-
## check/todo!!!!!!!!!!!!!!!!!-
|
|
142
|
-
## strip year if to present e.g. (2011-)
|
|
143
|
-
##
|
|
144
|
-
## do NOT strip for defunct / historic clubs e.g.
|
|
145
|
-
## (1899-1910)
|
|
146
|
-
## or (-1914) or (-2011) etc.
|
|
147
|
-
|
|
148
|
-
###
|
|
149
|
-
## todo: move year out of canonical team name - why? why not?
|
|
150
|
-
|
|
151
|
-
## check if canonical name include (2011-) or similar in name
|
|
152
|
-
## if yes, remove (2011-) and add to (alt) names
|
|
153
|
-
## e.g. New York FC (2011) => New York FC
|
|
154
|
-
if rec.name =~ /\(.+?\)/ ## note: use non-greedy (?) match
|
|
155
|
-
name = rec.name.gsub( /\(.+?\)/, '' ).strip
|
|
156
|
-
|
|
157
|
-
if rec.name =~ /\(([0-9]{4})-\)/ ## e.g. (2014-)
|
|
158
|
-
rec.year = $1.to_i
|
|
159
|
-
elsif rec.name =~ /\(-([0-9]{4})\)/ ## e.g. (-2014)
|
|
160
|
-
rec.year_end = $1.to_i
|
|
161
|
-
elsif rec.name =~ /\(([0-9]{4})-([0-9]{4})\)/ ## e.g. (2011-2014)
|
|
162
|
-
rec.year = $1.to_i
|
|
163
|
-
rec.year_end = $2.to_i
|
|
164
|
-
else
|
|
165
|
-
## todo/check: warn about unknown year format
|
|
166
|
-
end
|
|
167
|
-
end
|
|
168
|
-
|
|
169
|
-
## todo/check - check for unknown format values
|
|
170
|
-
## e.g. too many values, duplicate years, etc.
|
|
171
|
-
## check for overwritting, etc.
|
|
172
|
-
while values.size > 0
|
|
173
|
-
value = values.shift
|
|
174
|
-
## strip and squish (white)spaces
|
|
175
|
-
# e.g. León › Guanajuato => León › Guanajuato
|
|
176
|
-
value = value.strip.gsub( /[ \t]+/, ' ' )
|
|
177
|
-
if value =~/^\d{4}$/ # e.g 1904
|
|
178
|
-
## todo/check: issue warning if year is already set!!!!!!!
|
|
179
|
-
if rec.year
|
|
180
|
-
puts "!!! error - year already set to #{rec.year} - CANNOT overwrite with #{value}:"
|
|
181
|
-
pp rec
|
|
182
|
-
exit 1
|
|
183
|
-
end
|
|
184
|
-
rec.year = value.to_i
|
|
185
|
-
elsif value.start_with?( '@' ) # e.g. @ Anfield
|
|
186
|
-
## cut-off leading @ and spaces
|
|
187
|
-
rec.ground = value[1..-1].strip
|
|
188
|
-
else
|
|
189
|
-
## assume city / geo tree
|
|
190
|
-
## split into geo tree
|
|
191
|
-
geos = split_geo( value )
|
|
192
|
-
city = geos[0]
|
|
193
|
-
## check for "embedded" district e.g. London (Fulham) or Hamburg (St. Pauli) etc.
|
|
194
|
-
if city =~ /\((.+?)\)/ ## note: use non-greedy (?) match
|
|
195
|
-
rec.district = $1.strip
|
|
196
|
-
city = city.gsub( /\(.+?\)/, '' ).strip
|
|
197
|
-
end
|
|
198
|
-
rec.city = city
|
|
199
|
-
|
|
200
|
-
if geos.size > 1
|
|
201
|
-
## cut-off city and keep the rest (of geo tree)
|
|
202
|
-
rec.geos = geos[1..-1]
|
|
203
|
-
end
|
|
204
|
-
end
|
|
205
|
-
end ## while values
|
|
206
|
-
|
|
207
|
-
|
|
208
|
-
###############
|
|
209
|
-
## use headings text for geo tree
|
|
210
|
-
|
|
211
|
-
## 1) add country if present
|
|
212
|
-
if headings.size > 0 && headings[0]
|
|
213
|
-
country = SportDb::Import.config.countries[ headings[0] ]
|
|
214
|
-
rec.country = country
|
|
215
|
-
else
|
|
216
|
-
## make it an error - why? why not?
|
|
217
|
-
puts "!!! error - country missing in headings hierarchy - sorry - add to quicklist"
|
|
218
|
-
exit 1
|
|
219
|
-
end
|
|
220
|
-
|
|
221
|
-
## 2) check geo tree with headings hierarchy
|
|
222
|
-
if headings.size > 1 && headings[1]
|
|
223
|
-
geos = split_geo( headings[1] )
|
|
224
|
-
if rec.geos
|
|
225
|
-
if rec.geos[0] != geos[0]
|
|
226
|
-
puts "!!! error - geo tree - headings mismatch >#{rec.geos[0]}< <=> >#{geos[0]}<"
|
|
227
|
-
exit 1
|
|
228
|
-
end
|
|
229
|
-
if rec.geos[1] && rec.geos[1] != geos[1] ## check optional 2nd level too
|
|
230
|
-
puts "!!! error - geo tree - headings mismatch >#{rec.geos[1]}< <=> >#{geos[1]}<"
|
|
231
|
-
exit 1
|
|
232
|
-
end
|
|
233
|
-
else
|
|
234
|
-
## add missing region (state/province) from headings hierarchy
|
|
235
|
-
rec.geos = geos
|
|
236
|
-
end
|
|
237
|
-
end
|
|
238
|
-
|
|
239
|
-
last_rec = rec
|
|
240
|
-
|
|
241
|
-
|
|
242
|
-
### todo/fix:
|
|
243
|
-
## auto-add alt name with dots stripped - why? why not?
|
|
244
|
-
## e.g. D.C. United => DC United
|
|
245
|
-
## e.g. Liverpool F.C. => Liverpool FC
|
|
246
|
-
## e.g. St. Albin => St Albin etc.
|
|
247
|
-
## e.g. 1. FC Köln => 1 FC Köln -- make special case for 1. - why? why not?
|
|
248
|
-
|
|
249
|
-
##
|
|
250
|
-
## todo/fix: unify mapping entries
|
|
251
|
-
## always lowercase !!!! (case insensitive)
|
|
252
|
-
## always strip (2011-) !!!
|
|
253
|
-
## always strip dots (e.g. St., F.C, etc.)
|
|
254
|
-
|
|
255
|
-
recs << rec
|
|
256
|
-
end
|
|
257
|
-
end # each_line
|
|
258
|
-
recs
|
|
259
|
-
end # method read
|
|
260
|
-
|
|
261
|
-
### helpers
|
|
262
|
-
def self.split_geo( str )
|
|
263
|
-
## assume city / geo tree
|
|
264
|
-
## strip and squish (white)spaces
|
|
265
|
-
# e.g. León › Guanajuato => León › Guanajuato
|
|
266
|
-
str = str.strip.gsub( /[ \t]+/, ' ' )
|
|
267
|
-
|
|
268
|
-
## split into geo tree
|
|
269
|
-
geos = str.split( /[<>‹›]/ ) ## note: allow > < or › ‹
|
|
270
|
-
geos = geos.map { |geo| geo.strip } ## remove all whitespaces
|
|
271
|
-
geos
|
|
272
|
-
end
|
|
273
|
-
|
|
274
|
-
end # class ClubReader
|
|
275
|
-
|
|
276
|
-
|
|
277
|
-
end ## module Import
|
|
278
|
-
end ## module SportDb
|
|
1
|
+
# encoding: utf-8
|
|
2
|
+
|
|
3
|
+
|
|
4
|
+
module SportDb
|
|
5
|
+
module Import
|
|
6
|
+
|
|
7
|
+
|
|
8
|
+
class ClubReader
|
|
9
|
+
|
|
10
|
+
|
|
11
|
+
def self.read( path ) ## use - rename to read_file or from_file etc. - why? why not?
|
|
12
|
+
txt = File.open( path, 'r:utf-8' ).read
|
|
13
|
+
parse( txt )
|
|
14
|
+
end
|
|
15
|
+
|
|
16
|
+
|
|
17
|
+
def self.parse( txt )
|
|
18
|
+
recs = []
|
|
19
|
+
last_rec = nil
|
|
20
|
+
headings = [] ## headings stack
|
|
21
|
+
|
|
22
|
+
txt.each_line do |line|
|
|
23
|
+
line = line.strip
|
|
24
|
+
|
|
25
|
+
next if line.empty?
|
|
26
|
+
next if line.start_with?( '#' ) ## skip comments too
|
|
27
|
+
|
|
28
|
+
## strip inline (until end-of-line) comments too
|
|
29
|
+
## e.g Eupen => KAS Eupen, ## [de]
|
|
30
|
+
## => Eupen => KAS Eupen,
|
|
31
|
+
line = line.sub( /#.*/, '' ).strip
|
|
32
|
+
pp line
|
|
33
|
+
|
|
34
|
+
|
|
35
|
+
next if line =~ /^={1,}$/ ## skip "decorative" only heading e.g. ========
|
|
36
|
+
|
|
37
|
+
## note: like in wikimedia markup (and markdown) all optional trailing ==== too
|
|
38
|
+
## todo/check: allow === Text =-=-=-=-=-= too - why? why not?
|
|
39
|
+
if line =~ /^(={1,}) ## leading ======
|
|
40
|
+
([^=]+?) ## text (note: for now no "inline" = allowed)
|
|
41
|
+
=* ## (optional) trailing ====
|
|
42
|
+
$/x
|
|
43
|
+
heading_marker = $1
|
|
44
|
+
heading_level = $1.length ## count number of = for heading level
|
|
45
|
+
heading = $2.strip
|
|
46
|
+
|
|
47
|
+
puts "heading #{heading_level} >#{heading}<"
|
|
48
|
+
|
|
49
|
+
## 1) first pop headings if present
|
|
50
|
+
while headings.size+1 > heading_level
|
|
51
|
+
headings.pop
|
|
52
|
+
end
|
|
53
|
+
|
|
54
|
+
## 2) add missing (hierarchy) level if
|
|
55
|
+
while headings.size+1 < heading_level
|
|
56
|
+
## todo/fix: issue warning about "skipping" hierarchy level
|
|
57
|
+
puts "!!! warn [team reader] - skipping hierarchy level in headings "
|
|
58
|
+
headings.push( nil )
|
|
59
|
+
end
|
|
60
|
+
|
|
61
|
+
if heading =~ /^\?+$/ ## note: use ? or ?? or ?? to reset level to nil
|
|
62
|
+
## keep level empty
|
|
63
|
+
else
|
|
64
|
+
|
|
65
|
+
## quick hack: if level is 1 assume country for now
|
|
66
|
+
## and extract country code e.g.
|
|
67
|
+
## Austria (at) => at
|
|
68
|
+
## todo/fix: allow code only e.g. at or aut without enclosing () too - why? why not?
|
|
69
|
+
if heading_level == 1
|
|
70
|
+
if heading =~ /\(([a-z]{2,3})\)/i ## note allow (at) or (AUT) too
|
|
71
|
+
country_code = $1
|
|
72
|
+
|
|
73
|
+
## check country code - MUST exist for now!!!!
|
|
74
|
+
country = SportDb::Import.config.countries[ country_code ]
|
|
75
|
+
if country.nil?
|
|
76
|
+
puts "!!! error [team reader] - unknown country with code >#{country_code}< - sorry - add country to config to fix"
|
|
77
|
+
exit 1
|
|
78
|
+
end
|
|
79
|
+
|
|
80
|
+
headings.push( country_code )
|
|
81
|
+
else
|
|
82
|
+
puts "!!! error - heading level 1 - missing country code - >#{heading}<"
|
|
83
|
+
exit 1
|
|
84
|
+
end
|
|
85
|
+
else
|
|
86
|
+
## quick hack:
|
|
87
|
+
## remove known fill/dummy words incl:
|
|
88
|
+
## Provincia San Juan => San Juan (see argentina, for example)
|
|
89
|
+
##
|
|
90
|
+
## use geo tree long term with alternative names - why? why not?
|
|
91
|
+
words = ['Provincia']
|
|
92
|
+
words.each { |word| heading = heading.gsub( word, '' ) }
|
|
93
|
+
heading = heading.strip
|
|
94
|
+
|
|
95
|
+
headings.push( heading )
|
|
96
|
+
end
|
|
97
|
+
|
|
98
|
+
## assert that hierarchy level is ok
|
|
99
|
+
if headings.size != heading_level
|
|
100
|
+
puts "!!! error - headings hierarchy/stack out of order - #{heading.size}<=>#{heading_level}"
|
|
101
|
+
exit 1
|
|
102
|
+
end
|
|
103
|
+
end
|
|
104
|
+
|
|
105
|
+
pp headings
|
|
106
|
+
|
|
107
|
+
elsif line.start_with?( '|' )
|
|
108
|
+
## assume continuation with line of alternative names
|
|
109
|
+
## note: skip leading pipe
|
|
110
|
+
values = line[1..-1].split( '|' ) # team names - allow/use pipe(|)
|
|
111
|
+
## strip and squish (white)spaces
|
|
112
|
+
# e.g. New York FC (2011-) => New York FC (2011-)
|
|
113
|
+
values = values.map { |value| value.strip.gsub( /[ \t]+/, ' ' ) }
|
|
114
|
+
last_rec.alt_names += values
|
|
115
|
+
last_rec.add_variants( values ) # auto-add (possible) auto-generated variant names
|
|
116
|
+
|
|
117
|
+
## check for duplicates
|
|
118
|
+
if last_rec.duplicates?
|
|
119
|
+
duplicates = last_rec.duplicates
|
|
120
|
+
puts "*** !!! WARN !!! - #{duplicates.size} duplicate alt name mapping(s):"
|
|
121
|
+
pp duplicates
|
|
122
|
+
pp last_rec
|
|
123
|
+
##
|
|
124
|
+
## todo/fix: make it only an error with exit 1
|
|
125
|
+
## if (not normalized) names are the same (not unique/uniq)
|
|
126
|
+
## e.g. don't exit on A.F.C. == AFC etc.
|
|
127
|
+
## exit 1
|
|
128
|
+
end
|
|
129
|
+
else
|
|
130
|
+
values = line.split( ',' )
|
|
131
|
+
|
|
132
|
+
rec = Club.new
|
|
133
|
+
value = values.shift ## get first item
|
|
134
|
+
## strip and squish (white)spaces
|
|
135
|
+
# e.g. New York FC (2011-) => New York FC (2011-)
|
|
136
|
+
value = value.strip.gsub( /[ \t]+/, ' ' )
|
|
137
|
+
rec.name = value # canoncial name (global unique "beautiful/long" name)
|
|
138
|
+
rec.add_variants( value ) # auto-add (possible) auto-generated variant names
|
|
139
|
+
|
|
140
|
+
## note:
|
|
141
|
+
## check/todo!!!!!!!!!!!!!!!!!-
|
|
142
|
+
## strip year if to present e.g. (2011-)
|
|
143
|
+
##
|
|
144
|
+
## do NOT strip for defunct / historic clubs e.g.
|
|
145
|
+
## (1899-1910)
|
|
146
|
+
## or (-1914) or (-2011) etc.
|
|
147
|
+
|
|
148
|
+
###
|
|
149
|
+
## todo: move year out of canonical team name - why? why not?
|
|
150
|
+
|
|
151
|
+
## check if canonical name include (2011-) or similar in name
|
|
152
|
+
## if yes, remove (2011-) and add to (alt) names
|
|
153
|
+
## e.g. New York FC (2011) => New York FC
|
|
154
|
+
if rec.name =~ /\(.+?\)/ ## note: use non-greedy (?) match
|
|
155
|
+
name = rec.name.gsub( /\(.+?\)/, '' ).strip
|
|
156
|
+
|
|
157
|
+
if rec.name =~ /\(([0-9]{4})-\)/ ## e.g. (2014-)
|
|
158
|
+
rec.year = $1.to_i
|
|
159
|
+
elsif rec.name =~ /\(-([0-9]{4})\)/ ## e.g. (-2014)
|
|
160
|
+
rec.year_end = $1.to_i
|
|
161
|
+
elsif rec.name =~ /\(([0-9]{4})-([0-9]{4})\)/ ## e.g. (2011-2014)
|
|
162
|
+
rec.year = $1.to_i
|
|
163
|
+
rec.year_end = $2.to_i
|
|
164
|
+
else
|
|
165
|
+
## todo/check: warn about unknown year format
|
|
166
|
+
end
|
|
167
|
+
end
|
|
168
|
+
|
|
169
|
+
## todo/check - check for unknown format values
|
|
170
|
+
## e.g. too many values, duplicate years, etc.
|
|
171
|
+
## check for overwritting, etc.
|
|
172
|
+
while values.size > 0
|
|
173
|
+
value = values.shift
|
|
174
|
+
## strip and squish (white)spaces
|
|
175
|
+
# e.g. León › Guanajuato => León › Guanajuato
|
|
176
|
+
value = value.strip.gsub( /[ \t]+/, ' ' )
|
|
177
|
+
if value =~/^\d{4}$/ # e.g 1904
|
|
178
|
+
## todo/check: issue warning if year is already set!!!!!!!
|
|
179
|
+
if rec.year
|
|
180
|
+
puts "!!! error - year already set to #{rec.year} - CANNOT overwrite with #{value}:"
|
|
181
|
+
pp rec
|
|
182
|
+
exit 1
|
|
183
|
+
end
|
|
184
|
+
rec.year = value.to_i
|
|
185
|
+
elsif value.start_with?( '@' ) # e.g. @ Anfield
|
|
186
|
+
## cut-off leading @ and spaces
|
|
187
|
+
rec.ground = value[1..-1].strip
|
|
188
|
+
else
|
|
189
|
+
## assume city / geo tree
|
|
190
|
+
## split into geo tree
|
|
191
|
+
geos = split_geo( value )
|
|
192
|
+
city = geos[0]
|
|
193
|
+
## check for "embedded" district e.g. London (Fulham) or Hamburg (St. Pauli) etc.
|
|
194
|
+
if city =~ /\((.+?)\)/ ## note: use non-greedy (?) match
|
|
195
|
+
rec.district = $1.strip
|
|
196
|
+
city = city.gsub( /\(.+?\)/, '' ).strip
|
|
197
|
+
end
|
|
198
|
+
rec.city = city
|
|
199
|
+
|
|
200
|
+
if geos.size > 1
|
|
201
|
+
## cut-off city and keep the rest (of geo tree)
|
|
202
|
+
rec.geos = geos[1..-1]
|
|
203
|
+
end
|
|
204
|
+
end
|
|
205
|
+
end ## while values
|
|
206
|
+
|
|
207
|
+
|
|
208
|
+
###############
|
|
209
|
+
## use headings text for geo tree
|
|
210
|
+
|
|
211
|
+
## 1) add country if present
|
|
212
|
+
if headings.size > 0 && headings[0]
|
|
213
|
+
country = SportDb::Import.config.countries[ headings[0] ]
|
|
214
|
+
rec.country = country
|
|
215
|
+
else
|
|
216
|
+
## make it an error - why? why not?
|
|
217
|
+
puts "!!! error - country missing in headings hierarchy - sorry - add to quicklist"
|
|
218
|
+
exit 1
|
|
219
|
+
end
|
|
220
|
+
|
|
221
|
+
## 2) check geo tree with headings hierarchy
|
|
222
|
+
if headings.size > 1 && headings[1]
|
|
223
|
+
geos = split_geo( headings[1] )
|
|
224
|
+
if rec.geos
|
|
225
|
+
if rec.geos[0] != geos[0]
|
|
226
|
+
puts "!!! error - geo tree - headings mismatch >#{rec.geos[0]}< <=> >#{geos[0]}<"
|
|
227
|
+
exit 1
|
|
228
|
+
end
|
|
229
|
+
if rec.geos[1] && rec.geos[1] != geos[1] ## check optional 2nd level too
|
|
230
|
+
puts "!!! error - geo tree - headings mismatch >#{rec.geos[1]}< <=> >#{geos[1]}<"
|
|
231
|
+
exit 1
|
|
232
|
+
end
|
|
233
|
+
else
|
|
234
|
+
## add missing region (state/province) from headings hierarchy
|
|
235
|
+
rec.geos = geos
|
|
236
|
+
end
|
|
237
|
+
end
|
|
238
|
+
|
|
239
|
+
last_rec = rec
|
|
240
|
+
|
|
241
|
+
|
|
242
|
+
### todo/fix:
|
|
243
|
+
## auto-add alt name with dots stripped - why? why not?
|
|
244
|
+
## e.g. D.C. United => DC United
|
|
245
|
+
## e.g. Liverpool F.C. => Liverpool FC
|
|
246
|
+
## e.g. St. Albin => St Albin etc.
|
|
247
|
+
## e.g. 1. FC Köln => 1 FC Köln -- make special case for 1. - why? why not?
|
|
248
|
+
|
|
249
|
+
##
|
|
250
|
+
## todo/fix: unify mapping entries
|
|
251
|
+
## always lowercase !!!! (case insensitive)
|
|
252
|
+
## always strip (2011-) !!!
|
|
253
|
+
## always strip dots (e.g. St., F.C, etc.)
|
|
254
|
+
|
|
255
|
+
recs << rec
|
|
256
|
+
end
|
|
257
|
+
end # each_line
|
|
258
|
+
recs
|
|
259
|
+
end # method read
|
|
260
|
+
|
|
261
|
+
### helpers
|
|
262
|
+
def self.split_geo( str )
|
|
263
|
+
## assume city / geo tree
|
|
264
|
+
## strip and squish (white)spaces
|
|
265
|
+
# e.g. León › Guanajuato => León › Guanajuato
|
|
266
|
+
str = str.strip.gsub( /[ \t]+/, ' ' )
|
|
267
|
+
|
|
268
|
+
## split into geo tree
|
|
269
|
+
geos = str.split( /[<>‹›]/ ) ## note: allow > < or › ‹
|
|
270
|
+
geos = geos.map { |geo| geo.strip } ## remove all whitespaces
|
|
271
|
+
geos
|
|
272
|
+
end
|
|
273
|
+
|
|
274
|
+
end # class ClubReader
|
|
275
|
+
|
|
276
|
+
|
|
277
|
+
end ## module Import
|
|
278
|
+
end ## module SportDb
|