xref: /trunk/main/sal/qa/rtl/uri/rtl_Uri.cxx (revision 87d2adbc)
1*87d2adbcSAndrew Rist /**************************************************************
2cdf0e10cSrcweir  *
3*87d2adbcSAndrew Rist  * Licensed to the Apache Software Foundation (ASF) under one
4*87d2adbcSAndrew Rist  * or more contributor license agreements.  See the NOTICE file
5*87d2adbcSAndrew Rist  * distributed with this work for additional information
6*87d2adbcSAndrew Rist  * regarding copyright ownership.  The ASF licenses this file
7*87d2adbcSAndrew Rist  * to you under the Apache License, Version 2.0 (the
8*87d2adbcSAndrew Rist  * "License"); you may not use this file except in compliance
9*87d2adbcSAndrew Rist  * with the License.  You may obtain a copy of the License at
10*87d2adbcSAndrew Rist  *
11*87d2adbcSAndrew Rist  *   http://www.apache.org/licenses/LICENSE-2.0
12*87d2adbcSAndrew Rist  *
13*87d2adbcSAndrew Rist  * Unless required by applicable law or agreed to in writing,
14*87d2adbcSAndrew Rist  * software distributed under the License is distributed on an
15*87d2adbcSAndrew Rist  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
16*87d2adbcSAndrew Rist  * KIND, either express or implied.  See the License for the
17*87d2adbcSAndrew Rist  * specific language governing permissions and limitations
18*87d2adbcSAndrew Rist  * under the License.
19*87d2adbcSAndrew Rist  *
20*87d2adbcSAndrew Rist  *************************************************************/
21*87d2adbcSAndrew Rist 
22*87d2adbcSAndrew Rist 
23cdf0e10cSrcweir 
24cdf0e10cSrcweir 
25cdf0e10cSrcweir // MARKER(update_precomp.py): autogen include statement, do not remove
26cdf0e10cSrcweir #include "precompiled_sal.hxx"
27cdf0e10cSrcweir #include <stdlib.h>
28cdf0e10cSrcweir #include <rtl/ustring.hxx>
29cdf0e10cSrcweir #include <rtl/strbuf.hxx>
30cdf0e10cSrcweir #include <rtl/uri.hxx>
31cdf0e10cSrcweir #include <osl/thread.h>
32cdf0e10cSrcweir #include <osl/file.hxx>
33cdf0e10cSrcweir 
34cdf0e10cSrcweir #include <testshl/simpleheader.hxx>
35cdf0e10cSrcweir 
36cdf0e10cSrcweir // -----------------------------------------------------------------------------
37cdf0e10cSrcweir 
38cdf0e10cSrcweir namespace Stringtest
39cdf0e10cSrcweir {
40cdf0e10cSrcweir     rtl::OString toHex(unsigned char _c)
41cdf0e10cSrcweir     {
42cdf0e10cSrcweir         rtl::OStringBuffer sStrBuf;
43cdf0e10cSrcweir         static char cHex[] = "0123456789ABCDEF";
44cdf0e10cSrcweir 
45cdf0e10cSrcweir         int nhigh = int(_c) >> 4 & 0xf;
46cdf0e10cSrcweir         int nlow  = int(_c) & 0xf;
47cdf0e10cSrcweir         sStrBuf.append( cHex[nhigh] );
48cdf0e10cSrcweir         sStrBuf.append( cHex[nlow] );
49cdf0e10cSrcweir         return sStrBuf.makeStringAndClear();
50cdf0e10cSrcweir     }
51cdf0e10cSrcweir 
52cdf0e10cSrcweir     rtl::OString escapeString(rtl::OString const& _sStr)
53cdf0e10cSrcweir     {
54cdf0e10cSrcweir         rtl::OStringBuffer sStrBuf;
55cdf0e10cSrcweir         sal_Int32 nLength = _sStr.getLength();
56cdf0e10cSrcweir         for(int i=0;i<nLength;++i)
57cdf0e10cSrcweir         {
58cdf0e10cSrcweir             unsigned char c = (unsigned char)_sStr[i];
59cdf0e10cSrcweir             if (c > 127)
60cdf0e10cSrcweir             {
61cdf0e10cSrcweir                 sStrBuf.append("%");
62cdf0e10cSrcweir                 sStrBuf.append(toHex(c));
63cdf0e10cSrcweir             }
64cdf0e10cSrcweir             else
65cdf0e10cSrcweir             {
66cdf0e10cSrcweir                 sStrBuf.append((char)c);
67cdf0e10cSrcweir             }
68cdf0e10cSrcweir         }
69cdf0e10cSrcweir         return sStrBuf.makeStringAndClear();
70cdf0e10cSrcweir     }
71cdf0e10cSrcweir 
72cdf0e10cSrcweir     // -----------------------------------------------------------------------------
73cdf0e10cSrcweir 
74cdf0e10cSrcweir     class Convert : public CppUnit::TestFixture
75cdf0e10cSrcweir     {
76cdf0e10cSrcweir         rtl::OUString m_aStr;
77cdf0e10cSrcweir     public:
78cdf0e10cSrcweir         /*
79cdf0e10cSrcweir           rtl::OString toUTF8(rtl::OUString const& _suStr)
80cdf0e10cSrcweir             {
81cdf0e10cSrcweir                 rtl::OString sStrAsUTF8 = rtl::OUStringToOString(_suStr, RTL_TEXTENCODING_UTF8);
82cdf0e10cSrcweir                 t_print("%s\n", escapeString(sStrAsUTF8).getStr());
83cdf0e10cSrcweir                 return sStrAsUTF8;
84cdf0e10cSrcweir             }
85cdf0e10cSrcweir         */
86cdf0e10cSrcweir         rtl::OUString fromUTF8(rtl::OString const& _suStr)
87cdf0e10cSrcweir             {
88cdf0e10cSrcweir                 rtl::OUString suStr = rtl::OStringToOUString(_suStr, RTL_TEXTENCODING_UTF8);
89cdf0e10cSrcweir                 return suStr;
90cdf0e10cSrcweir             }
91cdf0e10cSrcweir 
92cdf0e10cSrcweir         rtl::OString convertToOString(rtl::OUString const& _suStr)
93cdf0e10cSrcweir             {
94cdf0e10cSrcweir                 return rtl::OUStringToOString(_suStr, osl_getThreadTextEncoding()/*RTL_TEXTENCODING_ASCII_US*/);
95cdf0e10cSrcweir             }
96cdf0e10cSrcweir 
97cdf0e10cSrcweir         void showContent(rtl::OUString const& _suStr)
98cdf0e10cSrcweir             {
99cdf0e10cSrcweir                 rtl::OString sStr = convertToOString(_suStr);
100cdf0e10cSrcweir                 t_print("%s\n", sStr.getStr());
101cdf0e10cSrcweir             }
102cdf0e10cSrcweir 
103cdf0e10cSrcweir         void toUTF8_mech(rtl::OUString const& _suStr, rtl_UriEncodeMechanism _eMechanism)
104cdf0e10cSrcweir             {
105cdf0e10cSrcweir                 rtl::OUString suStr;
106cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassNone,          _eMechanism, RTL_TEXTENCODING_UTF8);
107cdf0e10cSrcweir                 showContent(suStr);
108cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUric,          _eMechanism, RTL_TEXTENCODING_UTF8);
109cdf0e10cSrcweir                 showContent(suStr);
110cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUricNoSlash,   _eMechanism, RTL_TEXTENCODING_UTF8);
111cdf0e10cSrcweir                 showContent(suStr);
112cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRelSegment,    _eMechanism, RTL_TEXTENCODING_UTF8);
113cdf0e10cSrcweir                 showContent(suStr);
114cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRegName,       _eMechanism, RTL_TEXTENCODING_UTF8);
115cdf0e10cSrcweir                 showContent(suStr);
116cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUserinfo,      _eMechanism, RTL_TEXTENCODING_UTF8);
117cdf0e10cSrcweir                 showContent(suStr);
118cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassPchar,         _eMechanism, RTL_TEXTENCODING_UTF8);
119cdf0e10cSrcweir                 showContent(suStr);
120cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUnoParamValue, _eMechanism, RTL_TEXTENCODING_UTF8);
121cdf0e10cSrcweir                 showContent(suStr);
122cdf0e10cSrcweir             }
123cdf0e10cSrcweir 
124cdf0e10cSrcweir         void toUTF8(rtl::OUString const& _suStr)
125cdf0e10cSrcweir             {
126cdf0e10cSrcweir                 t_print("rtl_UriEncodeIgnoreEscapes \n");
127cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeIgnoreEscapes);
128cdf0e10cSrcweir                 t_print("\n");
129cdf0e10cSrcweir                 t_print("# rtl_UriEncodeKeepEscapes\n");
130cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeKeepEscapes);
131cdf0e10cSrcweir                 t_print("\n");
132cdf0e10cSrcweir                 t_print("# rtl_UriEncodeCheckEscapes\n");
133cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeCheckEscapes);
134cdf0e10cSrcweir                 t_print("\n");
135cdf0e10cSrcweir             }
136cdf0e10cSrcweir 
137cdf0e10cSrcweir         void test_FromUTF8_001()
138cdf0e10cSrcweir             {
139cdf0e10cSrcweir                 // string --> ustring
140cdf0e10cSrcweir                 rtl::OString sStrUTF8("h%C3%A4llo");
141cdf0e10cSrcweir                 rtl::OUString suStrUTF8 = rtl::OStringToOUString(sStrUTF8, RTL_TEXTENCODING_ASCII_US);
142cdf0e10cSrcweir 
143cdf0e10cSrcweir                 // UTF8 --> real ustring
144cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStrUTF8, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
145cdf0e10cSrcweir                 showContent(suStr_UriDecodeToIuri);
146cdf0e10cSrcweir 
147cdf0e10cSrcweir                 // string --> ustring
148cdf0e10cSrcweir                 rtl::OString sStr("h\xE4llo");
149cdf0e10cSrcweir                 rtl::OUString suString = rtl::OStringToOUString(sStr, RTL_TEXTENCODING_ISO_8859_15);
150cdf0e10cSrcweir 
151cdf0e10cSrcweir                 CPPUNIT_ASSERT_MESSAGE("Strings must be equal", suString.equals(suStr_UriDecodeToIuri) == sal_True);
152cdf0e10cSrcweir 
153cdf0e10cSrcweir                 // ustring --> ustring (UTF8)
154cdf0e10cSrcweir                 rtl::OUString suStr2 = rtl::Uri::encode(suStr_UriDecodeToIuri, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
155cdf0e10cSrcweir                 showContent(suStr2);
156cdf0e10cSrcweir 
157cdf0e10cSrcweir                 CPPUNIT_ASSERT_MESSAGE("Strings must be equal", suStr2.equals(suStrUTF8) == sal_True);
158cdf0e10cSrcweir                 // suStr should be equal to suStr2
159cdf0e10cSrcweir             }
160cdf0e10cSrcweir 
161cdf0e10cSrcweir         // "%C3%84qypten";
162cdf0e10cSrcweir         // testshl2 ../../../unxlngi4.pro/lib/libConvert.so "-onlyerrors"
163cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%89%8B%E6%9C%BA%E5%8F%B7%E7%A0%81'
164cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%9C%AA%E5%91%BD%E5%90%8Dzhgb18030'
165cdf0e10cSrcweir         // # Type: 'Regular file' file name '%E5%B7%A5%E4%BD%9C'
166cdf0e10cSrcweir         // # 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'
167cdf0e10cSrcweir         // ls
168cdf0e10cSrcweir         rtl::OString getFileTypeName(osl::FileStatus const& _aStatus)
169cdf0e10cSrcweir             {
170cdf0e10cSrcweir                 rtl::OString sType;
171cdf0e10cSrcweir                 if (_aStatus.isValid(osl_FileStatus_Mask_Type))
172cdf0e10cSrcweir                 {
173cdf0e10cSrcweir                     osl::FileStatus::Type aType = _aStatus.getFileType();
174cdf0e10cSrcweir                     if (aType == osl::FileStatus::Directory)
175cdf0e10cSrcweir                     {
176cdf0e10cSrcweir                         sType = "Directory";
177cdf0e10cSrcweir                     }
178cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Regular)
179cdf0e10cSrcweir                     {
180cdf0e10cSrcweir                         sType = "Regular file";
181cdf0e10cSrcweir                     }
182cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Volume)
183cdf0e10cSrcweir                     {
184cdf0e10cSrcweir                         sType = "Volume";
185cdf0e10cSrcweir                     }
186cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Fifo)
187cdf0e10cSrcweir                     {
188cdf0e10cSrcweir                         sType = "Fifo";
189cdf0e10cSrcweir                     }
190cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Socket)
191cdf0e10cSrcweir                     {
192cdf0e10cSrcweir                         sType = "Socket";
193cdf0e10cSrcweir                     }
194cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Link)
195cdf0e10cSrcweir                     {
196cdf0e10cSrcweir                         sType = "Link";
197cdf0e10cSrcweir                     }
198cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Special)
199cdf0e10cSrcweir                     {
200cdf0e10cSrcweir                         sType = "Special";
201cdf0e10cSrcweir                     }
202cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Unknown)
203cdf0e10cSrcweir                     {
204cdf0e10cSrcweir                         sType = "Unknown";
205cdf0e10cSrcweir                     }
206cdf0e10cSrcweir                     else
207cdf0e10cSrcweir                     {
208cdf0e10cSrcweir                         sType = "Not handled yet";
209cdf0e10cSrcweir                     }
210cdf0e10cSrcweir                 }
211cdf0e10cSrcweir                 else
212cdf0e10cSrcweir                 {
213cdf0e10cSrcweir                     sType = "ERROR: osl_FileStatus_Mask_Type not set for FileStatus!";
214cdf0e10cSrcweir                 }
215cdf0e10cSrcweir                 return sType;
216cdf0e10cSrcweir             }
217cdf0e10cSrcweir 
218cdf0e10cSrcweir 
219cdf0e10cSrcweir         void test_UTF8_files()
220cdf0e10cSrcweir             {
221cdf0e10cSrcweir #ifdef UNX
222cdf0e10cSrcweir                 rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///tmp/atestdir"));
223cdf0e10cSrcweir #else /* Windows */
224cdf0e10cSrcweir                 rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///c:/temp/atestdir"));
225cdf0e10cSrcweir #endif
226cdf0e10cSrcweir                 osl::Directory aDir(suDirURL);
227cdf0e10cSrcweir                 aDir.open();
228cdf0e10cSrcweir                 if (aDir.isOpen())
229cdf0e10cSrcweir                 {
230cdf0e10cSrcweir                     osl::DirectoryItem aItem;
231cdf0e10cSrcweir                     osl::FileStatus aStatus(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes | osl_FileStatus_Mask_Type);
232cdf0e10cSrcweir                     while (aDir.getNextItem(aItem) == ::osl::FileBase::E_None)
233cdf0e10cSrcweir                     {
234cdf0e10cSrcweir                         if (osl::FileBase::E_None == aItem.getFileStatus(aStatus) &&
235cdf0e10cSrcweir                             aStatus.isValid(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes))
236cdf0e10cSrcweir                         {
237cdf0e10cSrcweir                             rtl::OString sType = getFileTypeName(aStatus);
238cdf0e10cSrcweir 
239cdf0e10cSrcweir                             rtl::OUString suFilename = aStatus.getFileName();
240cdf0e10cSrcweir                             // rtl::OUString suFullFileURL;
241cdf0e10cSrcweir 
242cdf0e10cSrcweir                             rtl::OUString suStrUTF8 = rtl::Uri::encode(suFilename, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
243cdf0e10cSrcweir                             rtl::OString sStrUTF8 = convertToOString(suStrUTF8);
244cdf0e10cSrcweir                             t_print("Type: '%s' file name '%s'\n", sType.getStr(), sStrUTF8.getStr());
245cdf0e10cSrcweir                         }
246cdf0e10cSrcweir                     }
247cdf0e10cSrcweir                     aDir.close();
248cdf0e10cSrcweir                 }
249cdf0e10cSrcweir                 else
250cdf0e10cSrcweir                 {
251cdf0e10cSrcweir                     rtl::OString sStr;
252cdf0e10cSrcweir                     sStr = rtl::OUStringToOString(suDirURL, osl_getThreadTextEncoding());
253cdf0e10cSrcweir                     t_print("can't open dir:'%s'\n", sStr.getStr());
254cdf0e10cSrcweir                 }
255cdf0e10cSrcweir             }
256cdf0e10cSrcweir 
257cdf0e10cSrcweir         void test_FromUTF8()
258cdf0e10cSrcweir             {
259cdf0e10cSrcweir                 rtl::OString sStr("h%C3%A4llo");
260cdf0e10cSrcweir                 rtl::OUString suStr = rtl::OStringToOUString(sStr, osl_getThreadTextEncoding());
261cdf0e10cSrcweir 
262cdf0e10cSrcweir //    rtl_UriEncodeIgnoreEscapes,
263cdf0e10cSrcweir //    rtl_UriEncodeKeepEscapes,
264cdf0e10cSrcweir //     rtl_UriEncodeCheckEscapes,
265cdf0e10cSrcweir //                rtl::OUString suStr2 = rtl::Uri::encode(suStr, rtl_UriCharClassRegName, rtl_UriEncodeCheckEscapes, RTL_TEXTENCODING_UTF8);
266cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeNone        = rtl::Uri::decode(suStr, rtl_UriDecodeNone, RTL_TEXTENCODING_UTF8);
267cdf0e10cSrcweir                 showContent(suStr_UriDecodeNone);
268cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeNone);
269cdf0e10cSrcweir 
270cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStr, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
271cdf0e10cSrcweir                 showContent(suStr_UriDecodeToIuri);
272cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeToIuri);
273cdf0e10cSrcweir 
274cdf0e10cSrcweir                 rtl::OUString suStr_UriDecodeWithCharset = rtl::Uri::decode(suStr, rtl_UriDecodeWithCharset, RTL_TEXTENCODING_UTF8);
275cdf0e10cSrcweir                 showContent(suStr_UriDecodeWithCharset);
276cdf0e10cSrcweir                 toUTF8(suStr_UriDecodeWithCharset);
277cdf0e10cSrcweir             }
278cdf0e10cSrcweir 
279cdf0e10cSrcweir         CPPUNIT_TEST_SUITE( Convert );
280cdf0e10cSrcweir         CPPUNIT_TEST( test_FromUTF8_001 );
281cdf0e10cSrcweir //        CPPUNIT_TEST( test_UTF8_files );
282cdf0e10cSrcweir //      CPPUNIT_TEST( test_FromUTF8 );
283cdf0e10cSrcweir         CPPUNIT_TEST_SUITE_END( );
284cdf0e10cSrcweir 	};
285cdf0e10cSrcweir 
286cdf0e10cSrcweir }
287cdf0e10cSrcweir 
288cdf0e10cSrcweir 
289cdf0e10cSrcweir CPPUNIT_TEST_SUITE_NAMED_REGISTRATION( Stringtest::Convert, "Stringtest" );
290cdf0e10cSrcweir 
291cdf0e10cSrcweir // LLA: doku anpassen!!!
292cdf0e10cSrcweir 
293cdf0e10cSrcweir NOADDITIONAL;
294