xref: /trunk/main/sal/qa/rtl/uri/rtl_Uri.cxx (revision 71cbe377)
187d2adbcSAndrew Rist /**************************************************************
2cdf0e10cSrcweir  *
387d2adbcSAndrew Rist  * Licensed to the Apache Software Foundation (ASF) under one
487d2adbcSAndrew Rist  * or more contributor license agreements.  See the NOTICE file
587d2adbcSAndrew Rist  * distributed with this work for additional information
687d2adbcSAndrew Rist  * regarding copyright ownership.  The ASF licenses this file
787d2adbcSAndrew Rist  * to you under the Apache License, Version 2.0 (the
887d2adbcSAndrew Rist  * "License"); you may not use this file except in compliance
987d2adbcSAndrew Rist  * with the License.  You may obtain a copy of the License at
1087d2adbcSAndrew Rist  *
1187d2adbcSAndrew Rist  *   http://www.apache.org/licenses/LICENSE-2.0
1287d2adbcSAndrew Rist  *
1387d2adbcSAndrew Rist  * Unless required by applicable law or agreed to in writing,
1487d2adbcSAndrew Rist  * software distributed under the License is distributed on an
1587d2adbcSAndrew Rist  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
1687d2adbcSAndrew Rist  * KIND, either express or implied.  See the License for the
1787d2adbcSAndrew Rist  * specific language governing permissions and limitations
1887d2adbcSAndrew Rist  * under the License.
1987d2adbcSAndrew Rist  *
2087d2adbcSAndrew Rist  *************************************************************/
2187d2adbcSAndrew Rist 
2287d2adbcSAndrew 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 
34*71cbe377SDamjan Jovanovic #include "gtest/gtest.h"
35cdf0e10cSrcweir 
36cdf0e10cSrcweir // -----------------------------------------------------------------------------
37cdf0e10cSrcweir 
38cdf0e10cSrcweir namespace Stringtest
39cdf0e10cSrcweir {
toHex(unsigned char _c)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 
escapeString(rtl::OString const & _sStr)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 
74*71cbe377SDamjan Jovanovic     class Convert : public ::testing::Test
75cdf0e10cSrcweir     {
76*71cbe377SDamjan Jovanovic     protected:
77cdf0e10cSrcweir         rtl::OUString m_aStr;
78cdf0e10cSrcweir     public:
79cdf0e10cSrcweir         /*
80cdf0e10cSrcweir           rtl::OString toUTF8(rtl::OUString const& _suStr)
81cdf0e10cSrcweir             {
82cdf0e10cSrcweir                 rtl::OString sStrAsUTF8 = rtl::OUStringToOString(_suStr, RTL_TEXTENCODING_UTF8);
83*71cbe377SDamjan Jovanovic                 printf("%s\n", escapeString(sStrAsUTF8).getStr());
84cdf0e10cSrcweir                 return sStrAsUTF8;
85cdf0e10cSrcweir             }
86cdf0e10cSrcweir         */
fromUTF8(rtl::OString const & _suStr)87cdf0e10cSrcweir         rtl::OUString fromUTF8(rtl::OString const& _suStr)
88cdf0e10cSrcweir             {
89cdf0e10cSrcweir                 rtl::OUString suStr = rtl::OStringToOUString(_suStr, RTL_TEXTENCODING_UTF8);
90cdf0e10cSrcweir                 return suStr;
91cdf0e10cSrcweir             }
92cdf0e10cSrcweir 
convertToOString(rtl::OUString const & _suStr)93cdf0e10cSrcweir         rtl::OString convertToOString(rtl::OUString const& _suStr)
94cdf0e10cSrcweir             {
95cdf0e10cSrcweir                 return rtl::OUStringToOString(_suStr, osl_getThreadTextEncoding()/*RTL_TEXTENCODING_ASCII_US*/);
96cdf0e10cSrcweir             }
97cdf0e10cSrcweir 
showContent(rtl::OUString const & _suStr)98cdf0e10cSrcweir         void showContent(rtl::OUString const& _suStr)
99cdf0e10cSrcweir             {
100cdf0e10cSrcweir                 rtl::OString sStr = convertToOString(_suStr);
101*71cbe377SDamjan Jovanovic                 printf("%s\n", sStr.getStr());
102cdf0e10cSrcweir             }
103cdf0e10cSrcweir 
toUTF8_mech(rtl::OUString const & _suStr,rtl_UriEncodeMechanism _eMechanism)104cdf0e10cSrcweir         void toUTF8_mech(rtl::OUString const& _suStr, rtl_UriEncodeMechanism _eMechanism)
105cdf0e10cSrcweir             {
106cdf0e10cSrcweir                 rtl::OUString suStr;
107cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassNone,          _eMechanism, RTL_TEXTENCODING_UTF8);
108cdf0e10cSrcweir                 showContent(suStr);
109cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUric,          _eMechanism, RTL_TEXTENCODING_UTF8);
110cdf0e10cSrcweir                 showContent(suStr);
111cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUricNoSlash,   _eMechanism, RTL_TEXTENCODING_UTF8);
112cdf0e10cSrcweir                 showContent(suStr);
113cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRelSegment,    _eMechanism, RTL_TEXTENCODING_UTF8);
114cdf0e10cSrcweir                 showContent(suStr);
115cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassRegName,       _eMechanism, RTL_TEXTENCODING_UTF8);
116cdf0e10cSrcweir                 showContent(suStr);
117cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUserinfo,      _eMechanism, RTL_TEXTENCODING_UTF8);
118cdf0e10cSrcweir                 showContent(suStr);
119cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassPchar,         _eMechanism, RTL_TEXTENCODING_UTF8);
120cdf0e10cSrcweir                 showContent(suStr);
121cdf0e10cSrcweir                 suStr = rtl::Uri::encode(_suStr, rtl_UriCharClassUnoParamValue, _eMechanism, RTL_TEXTENCODING_UTF8);
122cdf0e10cSrcweir                 showContent(suStr);
123cdf0e10cSrcweir             }
124cdf0e10cSrcweir 
toUTF8(rtl::OUString const & _suStr)125cdf0e10cSrcweir         void toUTF8(rtl::OUString const& _suStr)
126cdf0e10cSrcweir             {
127*71cbe377SDamjan Jovanovic                 printf("rtl_UriEncodeIgnoreEscapes \n");
128cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeIgnoreEscapes);
129*71cbe377SDamjan Jovanovic                 printf("\n");
130*71cbe377SDamjan Jovanovic                 printf("# rtl_UriEncodeKeepEscapes\n");
131cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeKeepEscapes);
132*71cbe377SDamjan Jovanovic                 printf("\n");
133*71cbe377SDamjan Jovanovic                 printf("# rtl_UriEncodeCheckEscapes\n");
134cdf0e10cSrcweir                 toUTF8_mech(_suStr, rtl_UriEncodeCheckEscapes);
135*71cbe377SDamjan Jovanovic                 printf("\n");
136cdf0e10cSrcweir             }
137cdf0e10cSrcweir 
138cdf0e10cSrcweir         // "%C3%84qypten";
139cdf0e10cSrcweir         // testshl2 ../../../unxlngi4.pro/lib/libConvert.so "-onlyerrors"
140cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%89%8B%E6%9C%BA%E5%8F%B7%E7%A0%81'
141cdf0e10cSrcweir         // # Type: 'Directory' file name '%E6%9C%AA%E5%91%BD%E5%90%8Dzhgb18030'
142cdf0e10cSrcweir         // # Type: 'Regular file' file name '%E5%B7%A5%E4%BD%9C'
143cdf0e10cSrcweir         // # 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'
144cdf0e10cSrcweir         // ls
getFileTypeName(osl::FileStatus const & _aStatus)145cdf0e10cSrcweir         rtl::OString getFileTypeName(osl::FileStatus const& _aStatus)
146cdf0e10cSrcweir             {
147cdf0e10cSrcweir                 rtl::OString sType;
148cdf0e10cSrcweir                 if (_aStatus.isValid(osl_FileStatus_Mask_Type))
149cdf0e10cSrcweir                 {
150cdf0e10cSrcweir                     osl::FileStatus::Type aType = _aStatus.getFileType();
151cdf0e10cSrcweir                     if (aType == osl::FileStatus::Directory)
152cdf0e10cSrcweir                     {
153cdf0e10cSrcweir                         sType = "Directory";
154cdf0e10cSrcweir                     }
155cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Regular)
156cdf0e10cSrcweir                     {
157cdf0e10cSrcweir                         sType = "Regular file";
158cdf0e10cSrcweir                     }
159cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Volume)
160cdf0e10cSrcweir                     {
161cdf0e10cSrcweir                         sType = "Volume";
162cdf0e10cSrcweir                     }
163cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Fifo)
164cdf0e10cSrcweir                     {
165cdf0e10cSrcweir                         sType = "Fifo";
166cdf0e10cSrcweir                     }
167cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Socket)
168cdf0e10cSrcweir                     {
169cdf0e10cSrcweir                         sType = "Socket";
170cdf0e10cSrcweir                     }
171cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Link)
172cdf0e10cSrcweir                     {
173cdf0e10cSrcweir                         sType = "Link";
174cdf0e10cSrcweir                     }
175cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Special)
176cdf0e10cSrcweir                     {
177cdf0e10cSrcweir                         sType = "Special";
178cdf0e10cSrcweir                     }
179cdf0e10cSrcweir                     else if (aType == osl::FileStatus::Unknown)
180cdf0e10cSrcweir                     {
181cdf0e10cSrcweir                         sType = "Unknown";
182cdf0e10cSrcweir                     }
183cdf0e10cSrcweir                     else
184cdf0e10cSrcweir                     {
185cdf0e10cSrcweir                         sType = "Not handled yet";
186cdf0e10cSrcweir                     }
187cdf0e10cSrcweir                 }
188cdf0e10cSrcweir                 else
189cdf0e10cSrcweir                 {
190cdf0e10cSrcweir                     sType = "ERROR: osl_FileStatus_Mask_Type not set for FileStatus!";
191cdf0e10cSrcweir                 }
192cdf0e10cSrcweir                 return sType;
193cdf0e10cSrcweir             }
194*71cbe377SDamjan Jovanovic 	};
195cdf0e10cSrcweir 
TEST_F(Convert,test_FromUTF8_001)196*71cbe377SDamjan Jovanovic     TEST_F(Convert, test_FromUTF8_001)
197*71cbe377SDamjan Jovanovic     {
198*71cbe377SDamjan Jovanovic         // string --> ustring
199*71cbe377SDamjan Jovanovic         rtl::OString sStrUTF8("h%C3%A4llo");
200*71cbe377SDamjan Jovanovic         rtl::OUString suStrUTF8 = rtl::OStringToOUString(sStrUTF8, RTL_TEXTENCODING_ASCII_US);
201cdf0e10cSrcweir 
202*71cbe377SDamjan Jovanovic         // UTF8 --> real ustring
203*71cbe377SDamjan Jovanovic         rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStrUTF8, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
204*71cbe377SDamjan Jovanovic         showContent(suStr_UriDecodeToIuri);
205cdf0e10cSrcweir 
206*71cbe377SDamjan Jovanovic         // string --> ustring
207*71cbe377SDamjan Jovanovic         rtl::OString sStr("h\xE4llo");
208*71cbe377SDamjan Jovanovic         rtl::OUString suString = rtl::OStringToOUString(sStr, RTL_TEXTENCODING_ISO_8859_15);
209cdf0e10cSrcweir 
210*71cbe377SDamjan Jovanovic         ASSERT_TRUE(suString.equals(suStr_UriDecodeToIuri) == sal_True) << "Strings must be equal";
211cdf0e10cSrcweir 
212*71cbe377SDamjan Jovanovic         // ustring --> ustring (UTF8)
213*71cbe377SDamjan Jovanovic         rtl::OUString suStr2 = rtl::Uri::encode(suStr_UriDecodeToIuri, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
214*71cbe377SDamjan Jovanovic         showContent(suStr2);
215cdf0e10cSrcweir 
216*71cbe377SDamjan Jovanovic         ASSERT_TRUE(suStr2.equals(suStrUTF8) == sal_True) << "Strings must be equal";
217*71cbe377SDamjan Jovanovic         // suStr should be equal to suStr2
218*71cbe377SDamjan Jovanovic     }
219cdf0e10cSrcweir 
220*71cbe377SDamjan Jovanovic // These tests were commented out in the pre-gtest code:
221*71cbe377SDamjan Jovanovic //    TEST_F(Convert, test_UTF8_files)
222*71cbe377SDamjan Jovanovic //    {
223*71cbe377SDamjan Jovanovic //#ifdef UNX
224*71cbe377SDamjan Jovanovic //        rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///tmp/atestdir"));
225*71cbe377SDamjan Jovanovic //#else /* Windows */
226*71cbe377SDamjan Jovanovic //        rtl::OUString suDirURL(rtl::OUString::createFromAscii("file:///c:/temp/atestdir"));
227*71cbe377SDamjan Jovanovic //#endif
228*71cbe377SDamjan Jovanovic //        osl::Directory aDir(suDirURL);
229*71cbe377SDamjan Jovanovic //        aDir.open();
230*71cbe377SDamjan Jovanovic //        if (aDir.isOpen())
231*71cbe377SDamjan Jovanovic //        {
232*71cbe377SDamjan Jovanovic //            osl::DirectoryItem aItem;
233*71cbe377SDamjan Jovanovic //            osl::FileStatus aStatus(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes | osl_FileStatus_Mask_Type);
234*71cbe377SDamjan Jovanovic //            while (aDir.getNextItem(aItem) == ::osl::FileBase::E_None)
235*71cbe377SDamjan Jovanovic //            {
236*71cbe377SDamjan Jovanovic //                if (osl::FileBase::E_None == aItem.getFileStatus(aStatus) &&
237*71cbe377SDamjan Jovanovic //                    aStatus.isValid(osl_FileStatus_Mask_FileName | osl_FileStatus_Mask_Attributes))
238*71cbe377SDamjan Jovanovic //                {
239*71cbe377SDamjan Jovanovic //                    rtl::OString sType = getFileTypeName(aStatus);
240*71cbe377SDamjan Jovanovic //
241*71cbe377SDamjan Jovanovic //                    rtl::OUString suFilename = aStatus.getFileName();
242*71cbe377SDamjan Jovanovic //                    // rtl::OUString suFullFileURL;
243*71cbe377SDamjan Jovanovic //
244*71cbe377SDamjan Jovanovic //                    rtl::OUString suStrUTF8 = rtl::Uri::encode(suFilename, rtl_UriCharClassUnoParamValue, rtl_UriEncodeKeepEscapes, RTL_TEXTENCODING_UTF8);
245*71cbe377SDamjan Jovanovic //                    rtl::OString sStrUTF8 = convertToOString(suStrUTF8);
246*71cbe377SDamjan Jovanovic //                    printf("Type: '%s' file name '%s'\n", sType.getStr(), sStrUTF8.getStr());
247*71cbe377SDamjan Jovanovic //                }
248*71cbe377SDamjan Jovanovic //            }
249*71cbe377SDamjan Jovanovic //            aDir.close();
250*71cbe377SDamjan Jovanovic //        }
251*71cbe377SDamjan Jovanovic //        else
252*71cbe377SDamjan Jovanovic //        {
253*71cbe377SDamjan Jovanovic //            rtl::OString sStr;
254*71cbe377SDamjan Jovanovic //            sStr = rtl::OUStringToOString(suDirURL, osl_getThreadTextEncoding());
255*71cbe377SDamjan Jovanovic //            printf("can't open dir:'%s'\n", sStr.getStr());
256*71cbe377SDamjan Jovanovic //        }
257*71cbe377SDamjan Jovanovic //    }
258*71cbe377SDamjan Jovanovic //
259*71cbe377SDamjan Jovanovic //    TEST_F(Convert, test_FromUTF8)
260*71cbe377SDamjan Jovanovic //    {
261*71cbe377SDamjan Jovanovic //        rtl::OString sStr("h%C3%A4llo");
262*71cbe377SDamjan Jovanovic //        rtl::OUString suStr = rtl::OStringToOUString(sStr, osl_getThreadTextEncoding());
263*71cbe377SDamjan Jovanovic //
264*71cbe377SDamjan Jovanovic ////    rtl_UriEncodeIgnoreEscapes,
265*71cbe377SDamjan Jovanovic ////    rtl_UriEncodeKeepEscapes,
266*71cbe377SDamjan Jovanovic ////     rtl_UriEncodeCheckEscapes,
267*71cbe377SDamjan Jovanovic ////                rtl::OUString suStr2 = rtl::Uri::encode(suStr, rtl_UriCharClassRegName, rtl_UriEncodeCheckEscapes, RTL_TEXTENCODING_UTF8);
268*71cbe377SDamjan Jovanovic //        rtl::OUString suStr_UriDecodeNone        = rtl::Uri::decode(suStr, rtl_UriDecodeNone, RTL_TEXTENCODING_UTF8);
269*71cbe377SDamjan Jovanovic //        showContent(suStr_UriDecodeNone);
270*71cbe377SDamjan Jovanovic //        toUTF8(suStr_UriDecodeNone);
271*71cbe377SDamjan Jovanovic //
272*71cbe377SDamjan Jovanovic //        rtl::OUString suStr_UriDecodeToIuri      = rtl::Uri::decode(suStr, rtl_UriDecodeToIuri, RTL_TEXTENCODING_UTF8);
273*71cbe377SDamjan Jovanovic //        showContent(suStr_UriDecodeToIuri);
274*71cbe377SDamjan Jovanovic //        toUTF8(suStr_UriDecodeToIuri);
275*71cbe377SDamjan Jovanovic //
276*71cbe377SDamjan Jovanovic //        rtl::OUString suStr_UriDecodeWithCharset = rtl::Uri::decode(suStr, rtl_UriDecodeWithCharset, RTL_TEXTENCODING_UTF8);
277*71cbe377SDamjan Jovanovic //        showContent(suStr_UriDecodeWithCharset);
278*71cbe377SDamjan Jovanovic //        toUTF8(suStr_UriDecodeWithCharset);
279*71cbe377SDamjan Jovanovic //    }
280cdf0e10cSrcweir }
281cdf0e10cSrcweir 
main(int argc,char ** argv)282*71cbe377SDamjan Jovanovic int main(int argc, char **argv)
283*71cbe377SDamjan Jovanovic {
284*71cbe377SDamjan Jovanovic     ::testing::InitGoogleTest(&argc, argv);
285*71cbe377SDamjan Jovanovic     return RUN_ALL_TESTS();
286*71cbe377SDamjan Jovanovic }
287