1*b1cdbd2cSJim Jagielski /**************************************************************
2*b1cdbd2cSJim Jagielski *
3*b1cdbd2cSJim Jagielski * Licensed to the Apache Software Foundation (ASF) under one
4*b1cdbd2cSJim Jagielski * or more contributor license agreements. See the NOTICE file
5*b1cdbd2cSJim Jagielski * distributed with this work for additional information
6*b1cdbd2cSJim Jagielski * regarding copyright ownership. The ASF licenses this file
7*b1cdbd2cSJim Jagielski * to you under the Apache License, Version 2.0 (the
8*b1cdbd2cSJim Jagielski * "License"); you may not use this file except in compliance
9*b1cdbd2cSJim Jagielski * with the License. You may obtain a copy of the License at
10*b1cdbd2cSJim Jagielski *
11*b1cdbd2cSJim Jagielski * http://www.apache.org/licenses/LICENSE-2.0
12*b1cdbd2cSJim Jagielski *
13*b1cdbd2cSJim Jagielski * Unless required by applicable law or agreed to in writing,
14*b1cdbd2cSJim Jagielski * software distributed under the License is distributed on an
15*b1cdbd2cSJim Jagielski * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
16*b1cdbd2cSJim Jagielski * KIND, either express or implied. See the License for the
17*b1cdbd2cSJim Jagielski * specific language governing permissions and limitations
18*b1cdbd2cSJim Jagielski * under the License.
19*b1cdbd2cSJim Jagielski *
20*b1cdbd2cSJim Jagielski *************************************************************/
21*b1cdbd2cSJim Jagielski
22*b1cdbd2cSJim Jagielski
23*b1cdbd2cSJim Jagielski
24*b1cdbd2cSJim Jagielski // MARKER(update_precomp.py): autogen include statement, do not remove
25*b1cdbd2cSJim Jagielski #include "precompiled_i18npool.hxx"
26*b1cdbd2cSJim Jagielski
27*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/CTLScriptType.hpp>
28*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/ScriptDirection.hpp>
29*b1cdbd2cSJim Jagielski #include <com/sun/star/i18n/UnicodeScript.hpp>
30*b1cdbd2cSJim Jagielski #include <scripttypedetector.hxx>
31*b1cdbd2cSJim Jagielski #include <i18nutil/unicode.hxx>
32*b1cdbd2cSJim Jagielski
33*b1cdbd2cSJim Jagielski // ----------------------------------------------------
34*b1cdbd2cSJim Jagielski // class ScriptTypeDetector
35*b1cdbd2cSJim Jagielski // ----------------------------------------------------;
36*b1cdbd2cSJim Jagielski
37*b1cdbd2cSJim Jagielski using namespace com::sun::star::i18n;
38*b1cdbd2cSJim Jagielski
ScriptTypeDetector()39*b1cdbd2cSJim Jagielski ScriptTypeDetector::ScriptTypeDetector()
40*b1cdbd2cSJim Jagielski {
41*b1cdbd2cSJim Jagielski }
42*b1cdbd2cSJim Jagielski
~ScriptTypeDetector()43*b1cdbd2cSJim Jagielski ScriptTypeDetector::~ScriptTypeDetector()
44*b1cdbd2cSJim Jagielski {
45*b1cdbd2cSJim Jagielski }
46*b1cdbd2cSJim Jagielski
47*b1cdbd2cSJim Jagielski static sal_Int16 scriptDirection[] = {
48*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_LEFT_TO_RIGHT = 0,
49*b1cdbd2cSJim Jagielski ScriptDirection::RIGHT_TO_LEFT, // DirectionProperty_RIGHT_TO_LEFT = 1,
50*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_EUROPEAN_NUMBER = 2,
51*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_EUROPEAN_NUMBER_SEPARATOR = 3,
52*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_EUROPEAN_NUMBER_TERMINATOR = 4,
53*b1cdbd2cSJim Jagielski ScriptDirection::RIGHT_TO_LEFT, // DirectionProperty_ARABIC_NUMBER = 5,
54*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_COMMON_NUMBER_SEPARATOR = 6,
55*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_BLOCK_SEPARATOR = 7,
56*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_SEGMENT_SEPARATOR = 8,
57*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_WHITE_SPACE_NEUTRAL = 9,
58*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_OTHER_NEUTRAL = 10,
59*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_LEFT_TO_RIGHT_EMBEDDING = 11,
60*b1cdbd2cSJim Jagielski ScriptDirection::LEFT_TO_RIGHT, // DirectionProperty_LEFT_TO_RIGHT_OVERRIDE = 12,
61*b1cdbd2cSJim Jagielski ScriptDirection::RIGHT_TO_LEFT, // DirectionProperty_RIGHT_TO_LEFT_ARABIC = 13,
62*b1cdbd2cSJim Jagielski ScriptDirection::RIGHT_TO_LEFT, // DirectionProperty_RIGHT_TO_LEFT_EMBEDDING = 14,
63*b1cdbd2cSJim Jagielski ScriptDirection::RIGHT_TO_LEFT, // DirectionProperty_RIGHT_TO_LEFT_OVERRIDE = 15,
64*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_POP_DIRECTIONAL_FORMAT = 16,
65*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_DIR_NON_SPACING_MARK = 17,
66*b1cdbd2cSJim Jagielski ScriptDirection::NEUTRAL, // DirectionProperty_BOUNDARY_NEUTRAL = 18,
67*b1cdbd2cSJim Jagielski };
68*b1cdbd2cSJim Jagielski
69*b1cdbd2cSJim Jagielski sal_Int16 SAL_CALL
getScriptDirection(const::rtl::OUString & Text,sal_Int32 nPos,sal_Int16 defaultScriptDirection)70*b1cdbd2cSJim Jagielski ScriptTypeDetector::getScriptDirection( const ::rtl::OUString& Text, sal_Int32 nPos, sal_Int16 defaultScriptDirection ) throw (::com::sun::star::uno::RuntimeException)
71*b1cdbd2cSJim Jagielski {
72*b1cdbd2cSJim Jagielski sal_Int16 dir = scriptDirection[unicode::getUnicodeDirection(Text[nPos])];
73*b1cdbd2cSJim Jagielski return (dir == ScriptDirection::NEUTRAL) ? defaultScriptDirection : dir;
74*b1cdbd2cSJim Jagielski }
75*b1cdbd2cSJim Jagielski
76*b1cdbd2cSJim Jagielski // return value '-1' means either the direction on nPos is not same as scriptDirection or nPos is out of range.
77*b1cdbd2cSJim Jagielski sal_Int32 SAL_CALL
beginOfScriptDirection(const::rtl::OUString & Text,sal_Int32 nPos,sal_Int16 direction)78*b1cdbd2cSJim Jagielski ScriptTypeDetector::beginOfScriptDirection( const ::rtl::OUString& Text, sal_Int32 nPos, sal_Int16 direction ) throw (::com::sun::star::uno::RuntimeException)
79*b1cdbd2cSJim Jagielski {
80*b1cdbd2cSJim Jagielski sal_Int32 cPos = nPos;
81*b1cdbd2cSJim Jagielski
82*b1cdbd2cSJim Jagielski if (cPos < Text.getLength()) {
83*b1cdbd2cSJim Jagielski for (; cPos >= 0; cPos--) {
84*b1cdbd2cSJim Jagielski if (direction != getScriptDirection(Text, cPos, direction))
85*b1cdbd2cSJim Jagielski break;
86*b1cdbd2cSJim Jagielski }
87*b1cdbd2cSJim Jagielski }
88*b1cdbd2cSJim Jagielski return cPos == nPos ? -1 : cPos + 1;
89*b1cdbd2cSJim Jagielski }
90*b1cdbd2cSJim Jagielski
91*b1cdbd2cSJim Jagielski sal_Int32 SAL_CALL
endOfScriptDirection(const::rtl::OUString & Text,sal_Int32 nPos,sal_Int16 direction)92*b1cdbd2cSJim Jagielski ScriptTypeDetector::endOfScriptDirection( const ::rtl::OUString& Text, sal_Int32 nPos, sal_Int16 direction ) throw (::com::sun::star::uno::RuntimeException)
93*b1cdbd2cSJim Jagielski {
94*b1cdbd2cSJim Jagielski sal_Int32 cPos = nPos;
95*b1cdbd2cSJim Jagielski sal_Int32 len = Text.getLength();
96*b1cdbd2cSJim Jagielski
97*b1cdbd2cSJim Jagielski if (cPos >=0) {
98*b1cdbd2cSJim Jagielski for (; cPos < len; cPos++) {
99*b1cdbd2cSJim Jagielski if (direction != getScriptDirection(Text, cPos, direction))
100*b1cdbd2cSJim Jagielski break;
101*b1cdbd2cSJim Jagielski }
102*b1cdbd2cSJim Jagielski }
103*b1cdbd2cSJim Jagielski return cPos == nPos ? -1 : cPos;
104*b1cdbd2cSJim Jagielski }
105*b1cdbd2cSJim Jagielski
106*b1cdbd2cSJim Jagielski sal_Int16 SAL_CALL
getCTLScriptType(const::rtl::OUString & Text,sal_Int32 nPos)107*b1cdbd2cSJim Jagielski ScriptTypeDetector::getCTLScriptType( const ::rtl::OUString& Text, sal_Int32 nPos ) throw (::com::sun::star::uno::RuntimeException)
108*b1cdbd2cSJim Jagielski {
109*b1cdbd2cSJim Jagielski static ScriptTypeList typeList[] = {
110*b1cdbd2cSJim Jagielski { UnicodeScript_kHebrew, UnicodeScript_kHebrew, CTLScriptType::CTL_HEBREW }, // 10
111*b1cdbd2cSJim Jagielski { UnicodeScript_kArabic, UnicodeScript_kArabic, CTLScriptType::CTL_ARABIC }, // 11
112*b1cdbd2cSJim Jagielski { UnicodeScript_kDevanagari, UnicodeScript_kDevanagari, CTLScriptType::CTL_INDIC }, // 14
113*b1cdbd2cSJim Jagielski { UnicodeScript_kThai, UnicodeScript_kThai, CTLScriptType::CTL_THAI }, // 24
114*b1cdbd2cSJim Jagielski { UnicodeScript_kScriptCount, UnicodeScript_kScriptCount, CTLScriptType::CTL_UNKNOWN } // 88
115*b1cdbd2cSJim Jagielski };
116*b1cdbd2cSJim Jagielski
117*b1cdbd2cSJim Jagielski return unicode::getUnicodeScriptType(Text[nPos], typeList, CTLScriptType::CTL_UNKNOWN);
118*b1cdbd2cSJim Jagielski }
119*b1cdbd2cSJim Jagielski
120*b1cdbd2cSJim Jagielski // Begin of Script Type is inclusive.
121*b1cdbd2cSJim Jagielski sal_Int32 SAL_CALL
beginOfCTLScriptType(const::rtl::OUString & Text,sal_Int32 nPos)122*b1cdbd2cSJim Jagielski ScriptTypeDetector::beginOfCTLScriptType( const ::rtl::OUString& Text, sal_Int32 nPos ) throw (::com::sun::star::uno::RuntimeException)
123*b1cdbd2cSJim Jagielski {
124*b1cdbd2cSJim Jagielski if (nPos < 0)
125*b1cdbd2cSJim Jagielski return 0;
126*b1cdbd2cSJim Jagielski else if (nPos >= Text.getLength())
127*b1cdbd2cSJim Jagielski return Text.getLength();
128*b1cdbd2cSJim Jagielski else {
129*b1cdbd2cSJim Jagielski sal_Int16 cType = getCTLScriptType(Text, nPos);
130*b1cdbd2cSJim Jagielski for (nPos--; nPos >= 0; nPos--) {
131*b1cdbd2cSJim Jagielski if (cType != getCTLScriptType(Text, nPos))
132*b1cdbd2cSJim Jagielski break;
133*b1cdbd2cSJim Jagielski }
134*b1cdbd2cSJim Jagielski return nPos + 1;
135*b1cdbd2cSJim Jagielski }
136*b1cdbd2cSJim Jagielski }
137*b1cdbd2cSJim Jagielski
138*b1cdbd2cSJim Jagielski // End of the Script Type is exclusive, the return value pointing to the begin of next script type
139*b1cdbd2cSJim Jagielski sal_Int32 SAL_CALL
endOfCTLScriptType(const::rtl::OUString & Text,sal_Int32 nPos)140*b1cdbd2cSJim Jagielski ScriptTypeDetector::endOfCTLScriptType( const ::rtl::OUString& Text, sal_Int32 nPos ) throw (::com::sun::star::uno::RuntimeException)
141*b1cdbd2cSJim Jagielski {
142*b1cdbd2cSJim Jagielski if (nPos < 0)
143*b1cdbd2cSJim Jagielski return 0;
144*b1cdbd2cSJim Jagielski else if (nPos >= Text.getLength())
145*b1cdbd2cSJim Jagielski return Text.getLength();
146*b1cdbd2cSJim Jagielski else {
147*b1cdbd2cSJim Jagielski sal_Int16 cType = getCTLScriptType(Text, nPos);
148*b1cdbd2cSJim Jagielski sal_Int32 len = Text.getLength();
149*b1cdbd2cSJim Jagielski for (nPos++; nPos < len; nPos++) {
150*b1cdbd2cSJim Jagielski if (cType != getCTLScriptType(Text, nPos))
151*b1cdbd2cSJim Jagielski break;
152*b1cdbd2cSJim Jagielski }
153*b1cdbd2cSJim Jagielski return nPos;
154*b1cdbd2cSJim Jagielski }
155*b1cdbd2cSJim Jagielski }
156*b1cdbd2cSJim Jagielski
157*b1cdbd2cSJim Jagielski const sal_Char sDetector[] = "draft.com.sun.star.i18n.ScriptTypeDetector";
158*b1cdbd2cSJim Jagielski
159*b1cdbd2cSJim Jagielski rtl::OUString SAL_CALL
getImplementationName()160*b1cdbd2cSJim Jagielski ScriptTypeDetector::getImplementationName() throw( ::com::sun::star::uno::RuntimeException )
161*b1cdbd2cSJim Jagielski {
162*b1cdbd2cSJim Jagielski return ::rtl::OUString::createFromAscii(sDetector);
163*b1cdbd2cSJim Jagielski }
164*b1cdbd2cSJim Jagielski
165*b1cdbd2cSJim Jagielski sal_Bool SAL_CALL
supportsService(const rtl::OUString & ServiceName)166*b1cdbd2cSJim Jagielski ScriptTypeDetector::supportsService(const rtl::OUString& ServiceName) throw( ::com::sun::star::uno::RuntimeException )
167*b1cdbd2cSJim Jagielski {
168*b1cdbd2cSJim Jagielski return !ServiceName.compareToAscii(sDetector);
169*b1cdbd2cSJim Jagielski }
170*b1cdbd2cSJim Jagielski
171*b1cdbd2cSJim Jagielski ::com::sun::star::uno::Sequence< rtl::OUString > SAL_CALL
getSupportedServiceNames()172*b1cdbd2cSJim Jagielski ScriptTypeDetector::getSupportedServiceNames() throw( ::com::sun::star::uno::RuntimeException )
173*b1cdbd2cSJim Jagielski {
174*b1cdbd2cSJim Jagielski ::com::sun::star::uno::Sequence< ::rtl::OUString > aRet(1);
175*b1cdbd2cSJim Jagielski aRet[0] = ::rtl::OUString::createFromAscii(sDetector);
176*b1cdbd2cSJim Jagielski return aRet;
177*b1cdbd2cSJim Jagielski }
178*b1cdbd2cSJim Jagielski
179