xref: /aoo42x/main/sal/qa/rtl/uri/rtl_Uri.cxx (revision cdf0e10c)
1*cdf0e10cSrcweir /*************************************************************************
2*cdf0e10cSrcweir  *
3*cdf0e10cSrcweir  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4*cdf0e10cSrcweir  *
5*cdf0e10cSrcweir  * Copyright 2000, 2010 Oracle and/or its affiliates.
6*cdf0e10cSrcweir  *
7*cdf0e10cSrcweir  * OpenOffice.org - a multi-platform office productivity suite
8*cdf0e10cSrcweir  *
9*cdf0e10cSrcweir  * This file is part of OpenOffice.org.
10*cdf0e10cSrcweir  *
11*cdf0e10cSrcweir  * OpenOffice.org is free software: you can redistribute it and/or modify
12*cdf0e10cSrcweir  * it under the terms of the GNU Lesser General Public License version 3
13*cdf0e10cSrcweir  * only, as published by the Free Software Foundation.
14*cdf0e10cSrcweir  *
15*cdf0e10cSrcweir  * OpenOffice.org is distributed in the hope that it will be useful,
16*cdf0e10cSrcweir  * but WITHOUT ANY WARRANTY; without even the implied warranty of
17*cdf0e10cSrcweir  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
18*cdf0e10cSrcweir  * GNU Lesser General Public License version 3 for more details
19*cdf0e10cSrcweir  * (a copy is included in the LICENSE file that accompanied this code).
20*cdf0e10cSrcweir  *
21*cdf0e10cSrcweir  * You should have received a copy of the GNU Lesser General Public License
22*cdf0e10cSrcweir  * version 3 along with OpenOffice.org.  If not, see
23*cdf0e10cSrcweir  * <http://www.openoffice.org/license.html>
24*cdf0e10cSrcweir  * for a copy of the LGPLv3 License.
25*cdf0e10cSrcweir  *
26*cdf0e10cSrcweir  ************************************************************************/
27*cdf0e10cSrcweir 
28*cdf0e10cSrcweir 
29*cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove
30*cdf0e10cSrcweir #include "precompiled_sal.hxx"
31*cdf0e10cSrcweir #include <stdlib.h>
32*cdf0e10cSrcweir #include <rtl/ustring.hxx>
33*cdf0e10cSrcweir #include <rtl/strbuf.hxx>
34*cdf0e10cSrcweir #include <rtl/uri.hxx>
35*cdf0e10cSrcweir #include <osl/thread.h>
36*cdf0e10cSrcweir #include <osl/file.hxx>
37*cdf0e10cSrcweir 
38*cdf0e10cSrcweir #include <testshl/simpleheader.hxx>
39*cdf0e10cSrcweir 
40*cdf0e10cSrcweir // -----------------------------------------------------------------------------
41*cdf0e10cSrcweir 
42*cdf0e10cSrcweir namespace Stringtest
43*cdf0e10cSrcweir {
44*cdf0e10cSrcweir     rtl::OString toHex(unsigned char _c)
45*cdf0e10cSrcweir     {
46*cdf0e10cSrcweir         rtl::OStringBuffer sStrBuf;
47*cdf0e10cSrcweir         static char cHex[] = "0123456789ABCDEF";
48*cdf0e10cSrcweir 
49*cdf0e10cSrcweir         int nhigh = int(_c) >> 4 & 0xf;
50*cdf0e10cSrcweir         int nlow  = int(_c) & 0xf;
51*cdf0e10cSrcweir         sStrBuf.append( cHex[nhigh] );
52*cdf0e10cSrcweir         sStrBuf.append( cHex[nlow] );
53*cdf0e10cSrcweir         return sStrBuf.makeStringAndClear();
54*cdf0e10cSrcweir     }
55*cdf0e10cSrcweir 
56*cdf0e10cSrcweir     rtl::OString escapeString(rtl::OString const& _sStr)
57*cdf0e10cSrcweir     {
58*cdf0e10cSrcweir         rtl::OStringBuffer sStrBuf;
59*cdf0e10cSrcweir         sal_Int32 nLength = _sStr.getLength();
60*cdf0e10cSrcweir         for(int i=0;i<nLength;++i)
61*cdf0e10cSrcweir         {
62*cdf0e10cSrcweir             unsigned char c = (unsigned char)_sStr[i];
63*cdf0e10cSrcweir             if (c > 127)
64*cdf0e10cSrcweir             {
65*cdf0e10cSrcweir                 sStrBuf.append("%");
66*cdf0e10cSrcweir                 sStrBuf.append(toHex(c));
67*cdf0e10cSrcweir             }
68*cdf0e10cSrcweir             else
69*cdf0e10cSrcweir             {
70*cdf0e10cSrcweir                 sStrBuf.append((char)c);
71*cdf0e10cSrcweir             }
72*cdf0e10cSrcweir         }
73*cdf0e10cSrcweir         return sStrBuf.makeStringAndClear();
74*cdf0e10cSrcweir     }
75*cdf0e10cSrcweir 
76*cdf0e10cSrcweir     // -----------------------------------------------------------------------------
77*cdf0e10cSrcweir 
78*cdf0e10cSrcweir     class Convert : public CppUnit::TestFixture
79*cdf0e10cSrcweir     {
80*cdf0e10cSrcweir         rtl::OUString m_aStr;
81*cdf0e10cSrcweir     public:
82*cdf0e10cSrcweir         /*
83*cdf0e10cSrcweir           rtl::OString toUTF8(rtl::OUString const& _suStr)
84*cdf0e10cSrcweir             {
85*cdf0e10cSrcweir                 rtl::OString sStrAsUTF8 = rtl::OUStringToOString(_suStr, RTL_TEXTENCODING_UTF8);
86*cdf0e10cSrcweir                 t_print("%s\n", escapeString(sStrAsUTF8).getStr());
87*cdf0e10cSrcweir                 return sStrAsUTF8;
88*cdf0e10cSrcweir             }
89*cdf0e10cSrcweir         */
90*cdf0e10cSrcweir         rtl::OUString fromUTF8(rtl::OString const& _suStr)
91*cdf0e10cSrcweir             {
92*cdf0e10cSrcweir                 rtl::OUString suStr = rtl::OStringToOUString(_suStr, RTL_TEXTENCODING_UTF8);
93*cdf0e10cSrcweir                 return suStr;
94*cdf0e10cSrcweir             }
95*cdf0e10cSrcweir 
96*cdf0e10cSrcweir         rtl::OString convertToOString(rtl::OUString const& _suStr)
97*cdf0e10cSrcweir             {
98*cdf0e10cSrcweir                 return rtl::OUStringToOString(_suStr, osl_getThreadTextEncoding()/*RTL_TEXTENCODING_ASCII_US*/);
99*cdf0e10cSrcweir             }
100*cdf0e10cSrcweir 
101*cdf0e10cSrcweir         void showContent(rtl::OUString const& _suStr)
102*cdf0e10cSrcweir             {
103*cdf0e10cSrcweir                 rtl::OString sStr = convertToOString(_suStr);
104*cdf0e10cSrcweir                 t_print("%s\n", sStr.getStr());
105*cdf0e10cSrcweir             }
106*cdf0e10cSrcweir 
107*cdf0e10cSrcweir         void toUTF8_mech(rtl::OUString const& _suStr, rtl_UriEncodeMechanism _eMechanism)
108*cdf0e10cSrcweir             {
109*cdf0e10cSrcweir                 rtl::OUString suStr;
110*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassNone,          _eMechanism, RTL_TEXTENCODING_UTF8);
111*cdf0e10cSrcweir                 showContent(suStr);
112*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUric,          _eMechanism, RTL_TEXTENCODING_UTF8);
113*cdf0e10cSrcweir                 showContent(suStr);
114*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUricNoSlash,   _eMechanism, RTL_TEXTENCODING_UTF8);
115*cdf0e10cSrcweir                 showContent(suStr);
116*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRelSegment,    _eMechanism, RTL_TEXTENCODING_UTF8);
117*cdf0e10cSrcweir                 showContent(suStr);
118*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRegName,       _eMechanism, RTL_TEXTENCODING_UTF8);
119*cdf0e10cSrcweir                 showContent(suStr);
120*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUserinfo,      _eMechanism, RTL_TEXTENCODING_UTF8);
121*cdf0e10cSrcweir                 showContent(suStr);
122*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassPchar,         _eMechanism, RTL_TEXTENCODING_UTF8);
123*cdf0e10cSrcweir                 showContent(suStr);
124*cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUnoParamValue, _eMechanism, RTL_TEXTENCODING_UTF8);
125*cdf0e10cSrcweir                 showContent(suStr);
126*cdf0e10cSrcweir             }
127*cdf0e10cSrcweir 
128*cdf0e10cSrcweir         void toUTF8(rtl::OUString const& _suStr)
129*cdf0e10cSrcweir             {
130*cdf0e10cSrcweir                 t_print("rtl_UriEncodeIgnoreEscapes \n");
131*cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeIgnoreEscapes);
132*cdf0e10cSrcweir                 t_print("\n");
133*cdf0e10cSrcweir                 t_print("# rtl_UriEncodeKeepEscapes\n");
134*cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeKeepEscapes);
135*cdf0e10cSrcweir                 t_print("\n");
136*cdf0e10cSrcweir                 t_print("# rtl_UriEncodeCheckEscapes\n");
137*cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeCheckEscapes);
138*cdf0e10cSrcweir                 t_print("\n");
139*cdf0e10cSrcweir             }
140*cdf0e10cSrcweir 
141*cdf0e10cSrcweir         void test_FromUTF8_001()
142*cdf0e10cSrcweir             {
143*cdf0e10cSrcweir                 // string --> ustring
144*cdf0e10cSrcweir                 rtl::OString sStrUTF8("h%C3%A4llo");
145*cdf0e10cSrcweir                 rtl::OUString suStrUTF8 = rtl::OStringToOUString(sStrUTF8, RTL_TEXTENCODING_ASCII_US);
146*cdf0e10cSrcweir 
147*cdf0e10cSrcweir                 // UTF8 --> real ustring
148*cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStrUTF8, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
149*cdf0e10cSrcweir                 showContent(suStr_UriDecodeToIuri);
150*cdf0e10cSrcweir 
151*cdf0e10cSrcweir                 // string --> ustring
152*cdf0e10cSrcweir                 rtl::OString sStr("h\xE4llo");
153*cdf0e10cSrcweir                 rtl::OUString suString = rtl::OStringToOUString(sStr, RTL_TEXTENCODING_ISO_8859_15);
154*cdf0e10cSrcweir 
155*cdf0e10cSrcweir                 CPPUNIT_ASSERT_MESSAGE("Strings must be equal", suString.equals(suStr_UriDecodeToIuri) == sal_True);
156*cdf0e10cSrcweir 
157*cdf0e10cSrcweir                 // ustring --> ustring (UTF8)
158*cdf0e10cSrcweir                 rtl::OUString suStr2 = rtl::Uri::encode(suStr_UriDecodeToIuri, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
159*cdf0e10cSrcweir                 showContent(suStr2);
160*cdf0e10cSrcweir 
161*cdf0e10cSrcweir                 CPPUNIT_ASSERT_MESSAGE("Strings must be equal", suStr2.equals(suStrUTF8) == sal_True);
162*cdf0e10cSrcweir                 // suStr should be equal to suStr2
163*cdf0e10cSrcweir             }
164*cdf0e10cSrcweir 
165*cdf0e10cSrcweir         // "%C3%84qypten";
166*cdf0e10cSrcweir         // testshl2 ../../../unxlngi4.pro/lib/libConvert.so "-onlyerrors"
167*cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%89%8B%E6%9C%BA%E5%8F%B7%E7%A0%81'
168*cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%9C%AA%E5%91%BD%E5%90%8Dzhgb18030'
169*cdf0e10cSrcweir         // # Type: 'Regular file' file name '%E5%B7%A5%E4%BD%9C'
170*cdf0e10cSrcweir         // # Type: 'Regular file' file name '%E4%BA%8C%E6%89%8B%E6%88%BF%E4%B9%B0%E5%8D%96%E5%90%88%E5%90%8C%E8%8D%89%E7%A8%BF.doc'
171*cdf0e10cSrcweir         // ls
172*cdf0e10cSrcweir         rtl::OString getFileTypeName(osl::FileStatus const& _aStatus)
173*cdf0e10cSrcweir             {
174*cdf0e10cSrcweir                 rtl::OString sType;
175*cdf0e10cSrcweir                 if (_aStatus.isValid(osl_FileStatus_Mask_Type))
176*cdf0e10cSrcweir                 {
177*cdf0e10cSrcweir                     osl::FileStatus::Type aType = _aStatus.getFileType();
178*cdf0e10cSrcweir                     if (aType == osl::FileStatus::Directory)
179*cdf0e10cSrcweir                     {
180*cdf0e10cSrcweir                         sType = "Directory";
181*cdf0e10cSrcweir                     }
182*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Regular)
183*cdf0e10cSrcweir                     {
184*cdf0e10cSrcweir                         sType = "Regular file";
185*cdf0e10cSrcweir                     }
186*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Volume)
187*cdf0e10cSrcweir                     {
188*cdf0e10cSrcweir                         sType = "Volume";
189*cdf0e10cSrcweir                     }
190*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Fifo)
191*cdf0e10cSrcweir                     {
192*cdf0e10cSrcweir                         sType = "Fifo";
193*cdf0e10cSrcweir                     }
194*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Socket)
195*cdf0e10cSrcweir                     {
196*cdf0e10cSrcweir                         sType = "Socket";
197*cdf0e10cSrcweir                     }
198*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Link)
199*cdf0e10cSrcweir                     {
200*cdf0e10cSrcweir                         sType = "Link";
201*cdf0e10cSrcweir                     }
202*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Special)
203*cdf0e10cSrcweir                     {
204*cdf0e10cSrcweir                         sType = "Special";
205*cdf0e10cSrcweir                     }
206*cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Unknown)
207*cdf0e10cSrcweir                     {
208*cdf0e10cSrcweir                         sType = "Unknown";
209*cdf0e10cSrcweir                     }
210*cdf0e10cSrcweir                     else
211*cdf0e10cSrcweir                     {
212*cdf0e10cSrcweir                         sType = "Not handled yet";
213*cdf0e10cSrcweir                     }
214*cdf0e10cSrcweir                 }
215*cdf0e10cSrcweir                 else
216*cdf0e10cSrcweir                 {
217*cdf0e10cSrcweir                     sType = "ERROR: osl_FileStatus_Mask_Type not set for FileStatus!";
218*cdf0e10cSrcweir                 }
219*cdf0e10cSrcweir                 return sType;
220*cdf0e10cSrcweir             }
221*cdf0e10cSrcweir 
222*cdf0e10cSrcweir 
223*cdf0e10cSrcweir         void test_UTF8_files()
224*cdf0e10cSrcweir             {
225*cdf0e10cSrcweir #ifdef UNX
226*cdf0e10cSrcweir                 rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///tmp/atestdir"));
227*cdf0e10cSrcweir #else /* Windows */
228*cdf0e10cSrcweir                 rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///c:/temp/atestdir"));
229*cdf0e10cSrcweir #endif
230*cdf0e10cSrcweir                 osl::Directory aDir(suDirURL);
231*cdf0e10cSrcweir                 aDir.open();
232*cdf0e10cSrcweir                 if (aDir.isOpen())
233*cdf0e10cSrcweir                 {
234*cdf0e10cSrcweir                     osl::DirectoryItem aItem;
235*cdf0e10cSrcweir                     osl::FileStatus aStatus(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes | osl_FileStatus_Mask_Type);
236*cdf0e10cSrcweir                     while (aDir.getNextItem(aItem) == ::osl::FileBase::E_None)
237*cdf0e10cSrcweir                     {
238*cdf0e10cSrcweir                         if (osl::FileBase::E_None == aItem.getFileStatus(aStatus) &&
239*cdf0e10cSrcweir                             aStatus.isValid(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes))
240*cdf0e10cSrcweir                         {
241*cdf0e10cSrcweir                             rtl::OString sType = getFileTypeName(aStatus);
242*cdf0e10cSrcweir 
243*cdf0e10cSrcweir                             rtl::OUString suFilename = aStatus.getFileName();
244*cdf0e10cSrcweir                             // rtl::OUString suFullFileURL;
245*cdf0e10cSrcweir 
246*cdf0e10cSrcweir                             rtl::OUString suStrUTF8 = rtl::Uri::encode(suFilename, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
247*cdf0e10cSrcweir                             rtl::OString sStrUTF8 = convertToOString(suStrUTF8);
248*cdf0e10cSrcweir                             t_print("Type: '%s' file name '%s'\n", sType.getStr(), sStrUTF8.getStr());
249*cdf0e10cSrcweir                         }
250*cdf0e10cSrcweir                     }
251*cdf0e10cSrcweir                     aDir.close();
252*cdf0e10cSrcweir                 }
253*cdf0e10cSrcweir                 else
254*cdf0e10cSrcweir                 {
255*cdf0e10cSrcweir                     rtl::OString sStr;
256*cdf0e10cSrcweir                     sStr = rtl::OUStringToOString(suDirURL, osl_getThreadTextEncoding());
257*cdf0e10cSrcweir                     t_print("can't open dir:'%s'\n", sStr.getStr());
258*cdf0e10cSrcweir                 }
259*cdf0e10cSrcweir             }
260*cdf0e10cSrcweir 
261*cdf0e10cSrcweir         void test_FromUTF8()
262*cdf0e10cSrcweir             {
263*cdf0e10cSrcweir                 rtl::OString sStr("h%C3%A4llo");
264*cdf0e10cSrcweir                 rtl::OUString suStr = rtl::OStringToOUString(sStr, osl_getThreadTextEncoding());
265*cdf0e10cSrcweir 
266*cdf0e10cSrcweir //    rtl_UriEncodeIgnoreEscapes,
267*cdf0e10cSrcweir //    rtl_UriEncodeKeepEscapes,
268*cdf0e10cSrcweir //     rtl_UriEncodeCheckEscapes,
269*cdf0e10cSrcweir //                rtl::OUString suStr2 = rtl::Uri::encode(suStr, rtl_UriCharClassRegName, rtl_UriEncodeCheckEscapes, RTL_TEXTENCODING_UTF8);
270*cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeNone        = rtl::Uri::decode(suStr, rtl_UriDecodeNone, RTL_TEXTENCODING_UTF8);
271*cdf0e10cSrcweir                 showContent(suStr_UriDecodeNone);
272*cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeNone);
273*cdf0e10cSrcweir 
274*cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStr, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
275*cdf0e10cSrcweir                 showContent(suStr_UriDecodeToIuri);
276*cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeToIuri);
277*cdf0e10cSrcweir 
278*cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeWithCharset = rtl::Uri::decode(suStr, rtl_UriDecodeWithCharset, RTL_TEXTENCODING_UTF8);
279*cdf0e10cSrcweir                 showContent(suStr_UriDecodeWithCharset);
280*cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeWithCharset);
281*cdf0e10cSrcweir             }
282*cdf0e10cSrcweir 
283*cdf0e10cSrcweir         CPPUNIT_TEST_SUITE( Convert );
284*cdf0e10cSrcweir         CPPUNIT_TEST( test_FromUTF8_001 );
285*cdf0e10cSrcweir //        CPPUNIT_TEST( test_UTF8_files );
286*cdf0e10cSrcweir //      CPPUNIT_TEST( test_FromUTF8 );
287*cdf0e10cSrcweir         CPPUNIT_TEST_SUITE_END( );
288*cdf0e10cSrcweir 	};
289*cdf0e10cSrcweir 
290*cdf0e10cSrcweir }
291*cdf0e10cSrcweir 
292*cdf0e10cSrcweir 
293*cdf0e10cSrcweir CPPUNIT_TEST_SUITE_NAMED_REGISTRATION( Stringtest::Convert, "Stringtest" );
294*cdf0e10cSrcweir 
295*cdf0e10cSrcweir // LLA: doku anpassen!!!
296*cdf0e10cSrcweir 
297*cdf0e10cSrcweir NOADDITIONAL;
298