ffi-icu 0.0.4 → 0.0.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/lib/ffi-icu/lib.rb +57 -53
- data/lib/ffi-icu/transliteration.rb +3 -3
- data/lib/ffi-icu/version.rb +1 -1
- data/spec/transliteration_spec.rb +3 -3
- metadata +1 -1
data/lib/ffi-icu/lib.rb
CHANGED
|
@@ -10,7 +10,9 @@ module ICU
|
|
|
10
10
|
|
|
11
11
|
VERSIONS = {
|
|
12
12
|
"42" => "_4_2",
|
|
13
|
-
"44" => "_44"
|
|
13
|
+
"44" => "_44",
|
|
14
|
+
"45" => "_45",
|
|
15
|
+
"46" => "_46"
|
|
14
16
|
}
|
|
15
17
|
|
|
16
18
|
# FIXME: this is incredibly ugly, figure out some better way
|
|
@@ -37,8 +39,8 @@ module ICU
|
|
|
37
39
|
ffi_lib "icucore"
|
|
38
40
|
when :linux
|
|
39
41
|
versions = VERSIONS.keys
|
|
40
|
-
libs = ffi_lib versions.map { |v| "libicui18n.so.#{v}"},
|
|
41
|
-
versions.map { |v| "libicutu.so.#{v}"}
|
|
42
|
+
libs = ffi_lib versions.map { |v| "libicui18n.so.#{v}" },
|
|
43
|
+
versions.map { |v| "libicutu.so.#{v}" }
|
|
42
44
|
|
|
43
45
|
VERSIONS.find do |so_version, func_version|
|
|
44
46
|
if libs.first.name =~ /#{so_version}$/
|
|
@@ -46,10 +48,12 @@ module ICU
|
|
|
46
48
|
end
|
|
47
49
|
end
|
|
48
50
|
else
|
|
49
|
-
raise
|
|
51
|
+
raise LoadError
|
|
50
52
|
end
|
|
51
53
|
|
|
52
54
|
suffix
|
|
55
|
+
rescue LoadError => ex
|
|
56
|
+
raise LoadError, "no idea how to load ICU on #{ICU.platform}, patches appreciated! (#{ex.message})"
|
|
53
57
|
end
|
|
54
58
|
|
|
55
59
|
def self.check_error
|
|
@@ -92,47 +96,47 @@ module ICU
|
|
|
92
96
|
|
|
93
97
|
suffix = find_icu()
|
|
94
98
|
|
|
95
|
-
attach_function :u_errorName,
|
|
96
|
-
attach_function :uenum_count,
|
|
97
|
-
attach_function :uenum_close,
|
|
98
|
-
attach_function :uenum_next,
|
|
99
|
-
attach_function :u_charsToUChars, "u_charsToUChars#{suffix}", [:string,
|
|
100
|
-
attach_function :u_UCharsToChars, "u_UCharsToChars#{suffix}", [:pointer,
|
|
99
|
+
attach_function :u_errorName, "u_errorName#{suffix}", [:int], :string
|
|
100
|
+
attach_function :uenum_count, "uenum_count#{suffix}", [:pointer, :pointer], :int
|
|
101
|
+
attach_function :uenum_close, "uenum_close#{suffix}", [:pointer], :void
|
|
102
|
+
attach_function :uenum_next, "uenum_next#{suffix}", [:pointer, :pointer, :pointer], :string
|
|
103
|
+
attach_function :u_charsToUChars, "u_charsToUChars#{suffix}", [:string, :pointer, :int32_t], :void
|
|
104
|
+
attach_function :u_UCharsToChars, "u_UCharsToChars#{suffix}", [:pointer, :string, :int32_t], :void
|
|
101
105
|
|
|
102
106
|
# CharDet
|
|
103
107
|
#
|
|
104
108
|
# http://icu-project.org/apiref/icu4c/ucsdet_8h.html
|
|
105
109
|
#
|
|
106
110
|
|
|
107
|
-
attach_function :ucsdet_open,
|
|
108
|
-
attach_function :ucsdet_close,
|
|
109
|
-
attach_function :ucsdet_setText,
|
|
110
|
-
attach_function :ucsdet_setDeclaredEncoding,
|
|
111
|
-
attach_function :ucsdet_detect,
|
|
112
|
-
attach_function :ucsdet_detectAll,
|
|
113
|
-
attach_function :ucsdet_getName,
|
|
114
|
-
attach_function :ucsdet_getConfidence,
|
|
115
|
-
attach_function :ucsdet_getLanguage,
|
|
116
|
-
attach_function :ucsdet_getAllDetectableCharsets, "ucsdet_getAllDetectableCharsets#{suffix}", [:pointer,
|
|
117
|
-
attach_function :ucsdet_isInputFilterEnabled,
|
|
118
|
-
attach_function :ucsdet_enableInputFilter,
|
|
111
|
+
attach_function :ucsdet_open, "ucsdet_open#{suffix}", [:pointer], :pointer
|
|
112
|
+
attach_function :ucsdet_close, "ucsdet_close#{suffix}", [:pointer], :void
|
|
113
|
+
attach_function :ucsdet_setText, "ucsdet_setText#{suffix}", [:pointer, :string, :int32_t, :pointer], :void
|
|
114
|
+
attach_function :ucsdet_setDeclaredEncoding, "ucsdet_setDeclaredEncoding#{suffix}", [:pointer, :string, :int32_t, :pointer], :void
|
|
115
|
+
attach_function :ucsdet_detect, "ucsdet_detect#{suffix}", [:pointer, :pointer], :pointer
|
|
116
|
+
attach_function :ucsdet_detectAll, "ucsdet_detectAll#{suffix}", [:pointer, :pointer, :pointer], :pointer
|
|
117
|
+
attach_function :ucsdet_getName, "ucsdet_getName#{suffix}", [:pointer, :pointer], :string
|
|
118
|
+
attach_function :ucsdet_getConfidence, "ucsdet_getConfidence#{suffix}", [:pointer, :pointer], :int32_t
|
|
119
|
+
attach_function :ucsdet_getLanguage, "ucsdet_getLanguage#{suffix}", [:pointer, :pointer], :string
|
|
120
|
+
attach_function :ucsdet_getAllDetectableCharsets, "ucsdet_getAllDetectableCharsets#{suffix}", [:pointer, :pointer], :pointer
|
|
121
|
+
attach_function :ucsdet_isInputFilterEnabled, "ucsdet_isInputFilterEnabled#{suffix}", [:pointer], :bool
|
|
122
|
+
attach_function :ucsdet_enableInputFilter, "ucsdet_enableInputFilter#{suffix}", [:pointer, :bool], :bool
|
|
119
123
|
|
|
120
124
|
# Collation
|
|
121
125
|
#
|
|
122
126
|
# http://icu-project.org/apiref/icu4c/ucol_8h.html
|
|
123
127
|
#
|
|
124
128
|
|
|
125
|
-
attach_function :ucol_open,
|
|
126
|
-
attach_function :ucol_close,
|
|
127
|
-
attach_function :ucol_strcoll,
|
|
128
|
-
attach_function :ucol_getKeywords,
|
|
129
|
-
attach_function :ucol_getKeywordValues, "ucol_getKeywordValues#{suffix}", [:string,
|
|
130
|
-
attach_function :ucol_getAvailable,
|
|
131
|
-
attach_function :ucol_countAvailable,
|
|
132
|
-
attach_function :ucol_getLocale,
|
|
133
|
-
attach_function :ucol_greater,
|
|
134
|
-
attach_function :ucol_greaterOrEqual,
|
|
135
|
-
attach_function :ucol_equal,
|
|
129
|
+
attach_function :ucol_open, "ucol_open#{suffix}", [:string, :pointer], :pointer
|
|
130
|
+
attach_function :ucol_close, "ucol_close#{suffix}", [:pointer], :void
|
|
131
|
+
attach_function :ucol_strcoll, "ucol_strcoll#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :int
|
|
132
|
+
attach_function :ucol_getKeywords, "ucol_getKeywords#{suffix}", [:pointer], :pointer
|
|
133
|
+
attach_function :ucol_getKeywordValues, "ucol_getKeywordValues#{suffix}", [:string, :pointer], :pointer
|
|
134
|
+
attach_function :ucol_getAvailable, "ucol_getAvailable#{suffix}", [:int32_t], :string
|
|
135
|
+
attach_function :ucol_countAvailable, "ucol_countAvailable#{suffix}", [], :int32_t
|
|
136
|
+
attach_function :ucol_getLocale, "ucol_getLocale#{suffix}", [:pointer, :int, :pointer], :string
|
|
137
|
+
attach_function :ucol_greater, "ucol_greater#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
|
|
138
|
+
attach_function :ucol_greaterOrEqual, "ucol_greaterOrEqual#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
|
|
139
|
+
attach_function :ucol_equal, "ucol_equal#{suffix}", [:pointer, :pointer, :int32_t, :pointer, :int32_t], :bool
|
|
136
140
|
|
|
137
141
|
# Transliteration
|
|
138
142
|
#
|
|
@@ -160,11 +164,11 @@ module ICU
|
|
|
160
164
|
|
|
161
165
|
enum :trans_direction, [:forward, :reverse]
|
|
162
166
|
|
|
163
|
-
attach_function :utrans_openIDs,
|
|
164
|
-
attach_function :utrans_openU,
|
|
165
|
-
attach_function :utrans_open,
|
|
166
|
-
attach_function :utrans_close,
|
|
167
|
-
attach_function :utrans_transUChars, "utrans_transUChars#{suffix}", [:pointer,
|
|
167
|
+
attach_function :utrans_openIDs, "utrans_openIDs#{suffix}", [:pointer], :pointer
|
|
168
|
+
attach_function :utrans_openU, "utrans_openU#{suffix}", [:pointer, :int32_t, :trans_direction, :pointer, :int32_t, :pointer, :pointer], :pointer
|
|
169
|
+
attach_function :utrans_open, "utrans_open#{suffix}", [:string, :trans_direction, :pointer, :int32_t, :pointer, :pointer], :pointer
|
|
170
|
+
attach_function :utrans_close, "utrans_close#{suffix}", [:pointer], :void
|
|
171
|
+
attach_function :utrans_transUChars, "utrans_transUChars#{suffix}", [:pointer, :pointer, :pointer, :int32_t, :int32_t, :pointer, :pointer], :void
|
|
168
172
|
|
|
169
173
|
# Normalization
|
|
170
174
|
#
|
|
@@ -199,21 +203,21 @@ module ICU
|
|
|
199
203
|
:ideo_limit, 400
|
|
200
204
|
]
|
|
201
205
|
|
|
202
|
-
attach_function :ubrk_countAvailable, "ubrk_countAvailable#{suffix}", [],
|
|
203
|
-
attach_function :ubrk_getAvailable,
|
|
204
|
-
|
|
205
|
-
attach_function :ubrk_open,
|
|
206
|
-
attach_function :ubrk_close,
|
|
207
|
-
attach_function :ubrk_setText,
|
|
208
|
-
attach_function :ubrk_current,
|
|
209
|
-
attach_function :ubrk_next,
|
|
210
|
-
attach_function :ubrk_previous,
|
|
211
|
-
attach_function :ubrk_first,
|
|
212
|
-
attach_function :ubrk_last,
|
|
213
|
-
|
|
214
|
-
attach_function :ubrk_preceding,
|
|
215
|
-
attach_function :ubrk_following,
|
|
216
|
-
attach_function :ubrk_isBoundary,
|
|
206
|
+
attach_function :ubrk_countAvailable, "ubrk_countAvailable#{suffix}", [], :int32_t
|
|
207
|
+
attach_function :ubrk_getAvailable, "ubrk_getAvailable#{suffix}", [:int32_t], :string
|
|
208
|
+
|
|
209
|
+
attach_function :ubrk_open, "ubrk_open#{suffix}", [:iterator_type, :string, :pointer, :int32_t, :pointer], :pointer
|
|
210
|
+
attach_function :ubrk_close, "ubrk_close#{suffix}", [:pointer], :void
|
|
211
|
+
attach_function :ubrk_setText, "ubrk_setText#{suffix}", [:pointer, :pointer, :int32_t, :pointer], :void
|
|
212
|
+
attach_function :ubrk_current, "ubrk_current#{suffix}", [:pointer], :int32_t
|
|
213
|
+
attach_function :ubrk_next, "ubrk_next#{suffix}", [:pointer], :int32_t
|
|
214
|
+
attach_function :ubrk_previous, "ubrk_previous#{suffix}", [:pointer], :int32_t
|
|
215
|
+
attach_function :ubrk_first, "ubrk_first#{suffix}", [:pointer], :int32_t
|
|
216
|
+
attach_function :ubrk_last, "ubrk_last#{suffix}", [:pointer], :int32_t
|
|
217
|
+
|
|
218
|
+
attach_function :ubrk_preceding, "ubrk_preceding#{suffix}", [:pointer, :int32_t], :int32_t
|
|
219
|
+
attach_function :ubrk_following, "ubrk_following#{suffix}", [:pointer, :int32_t], :int32_t
|
|
220
|
+
attach_function :ubrk_isBoundary, "ubrk_isBoundary#{suffix}", [:pointer, :int32_t], :int32_t
|
|
217
221
|
|
|
218
222
|
end # Lib
|
|
219
223
|
end # ICU
|
|
@@ -26,18 +26,18 @@ module ICU
|
|
|
26
26
|
class Transliterator
|
|
27
27
|
|
|
28
28
|
def initialize(id, rules = nil, direction = :forward)
|
|
29
|
+
rules_length = 0
|
|
30
|
+
|
|
29
31
|
if rules
|
|
30
32
|
rules_length = rules.length + 1
|
|
31
33
|
rules = UCharPointer.from_string(rules)
|
|
32
|
-
else
|
|
33
|
-
rules_length = 0
|
|
34
34
|
end
|
|
35
35
|
|
|
36
36
|
parse_error = Lib::UParseError.new
|
|
37
37
|
begin
|
|
38
38
|
Lib.check_error do |status|
|
|
39
39
|
# couldn't get utrans_openU to work properly, so using deprecated utrans_open for now
|
|
40
|
-
@tr = Lib.
|
|
40
|
+
@tr = Lib.utrans_openU(UCharPointer.from_string(id), id.length, direction, rules, rules_length, @parse_error, status)
|
|
41
41
|
end
|
|
42
42
|
rescue ICU::Error => ex
|
|
43
43
|
raise ex, "#{ex.message} (#{parse_error})"
|
data/lib/ffi-icu/version.rb
CHANGED
|
@@ -30,9 +30,9 @@ module ICU
|
|
|
30
30
|
ids.should_not be_empty
|
|
31
31
|
end
|
|
32
32
|
|
|
33
|
-
|
|
34
|
-
|
|
35
|
-
|
|
33
|
+
it "should transliterate custom rules" do
|
|
34
|
+
ICU::Transliteration.translit("NFD; [:Nonspacing Mark:] Remove; NFC", "âêîôû").should == "aeiou"
|
|
35
|
+
end
|
|
36
36
|
|
|
37
37
|
end # Transliteration
|
|
38
38
|
end # ICU
|