Mercurial > hg > orthanc

changeset 6606:29388610f3e2

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
adding support for UTF-8 in ZipWriter and HierarchicalZipWriter
author Sebastien Jodogne <s.jodogne@gmail.com>
date Fri, 27 Feb 2026 18:37:52 +0100
parents 6e17d039332d
children f30191e9ffbf
files OrthancFramework/Sources/Compression/HierarchicalZipWriter.cpp OrthancFramework/Sources/Compression/HierarchicalZipWriter.h OrthancFramework/Sources/Compression/ZipWriter.cpp OrthancFramework/Sources/Compression/ZipWriter.h OrthancFramework/Sources/Toolbox.cpp OrthancFramework/Sources/Toolbox.h OrthancFramework/UnitTestsSources/ZipTests.cpp
diffstat 7 files changed, 159 insertions(+), 108 deletions(-) [+]
line wrap: on
line diff
--- a/OrthancFramework/Sources/Compression/HierarchicalZipWriter.cpp	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Compression/HierarchicalZipWriter.cpp	Fri Feb 27 18:37:52 2026 +0100
@@ -32,43 +32,6 @@
 
 namespace Orthanc
 {
-  std::string HierarchicalZipWriter::Index::KeepAlphanumeric(const std::string& source)
-  {
-    std::string result;
-
-    bool lastSpace = false;
-
-    result.reserve(source.size());
-    for (size_t i = 0; i < source.size(); i++)
-    {
-      char c = source[i];
-      if (c == '^')
-        c = ' ';
-
-      if (c <= 127 && 
-          c >= 0)
-      {
-        if (isspace(c)) 
-        {
-          if (!lastSpace)
-          {
-            lastSpace = true;
-            result.push_back(' ');
-          }
-        }
-        else if (isalnum(c) || 
-                 c == '.' || 
-                 c == '_')
-        {
-          result.push_back(c);
-          lastSpace = false;
-        }
-      }
-    }
-
-    return Toolbox::StripSpaces(result);
-  }
-
   std::string HierarchicalZipWriter::Index::GetCurrentDirectoryPath() const
   {
     std::string result;
@@ -85,9 +48,15 @@
     return result;
   }
 
-  std::string HierarchicalZipWriter::Index::EnsureUniqueFilename(const std::string& filename)
+  std::string HierarchicalZipWriter::Index::EnsureUniqueFilename(const std::string& filename,
+                                                                 bool allowUtf8)
   {
-    std::string standardized = KeepAlphanumeric(filename);
+    const std::string standardized = Toolbox::NormalizePath(filename, allowUtf8);
+
+    if (standardized.find('/') != std::string::npos)
+    {
+      throw Orthanc::OrthancException(Orthanc::ErrorCode_ParameterOutOfRange, "Your filename must not contain slashes or backslashes");
+    }
 
     Directory& d = *stack_.back();
     Directory::Content::iterator it = d.content_.find(standardized);
@@ -122,14 +91,16 @@
     return stack_.size() == 1;
   }
 
-  std::string HierarchicalZipWriter::Index::OpenFile(const std::string& name)
+  std::string HierarchicalZipWriter::Index::OpenFile(const std::string& name,
+                                                     bool allowUtf8)
   {
-    return GetCurrentDirectoryPath() + EnsureUniqueFilename(name);
+    return GetCurrentDirectoryPath() + EnsureUniqueFilename(name, allowUtf8);
   }
 
-  void HierarchicalZipWriter::Index::OpenDirectory(const std::string& name)
+  void HierarchicalZipWriter::Index::OpenDirectory(const std::string& name,
+                                                   bool allowUtf8)
   {
-    std::string d = EnsureUniqueFilename(name);
+    std::string d = EnsureUniqueFilename(name, allowUtf8);
 
     // Push the new directory onto the stack
     stack_.push_back(new Directory);
@@ -201,13 +172,13 @@
 
   void HierarchicalZipWriter::OpenFile(const std::string& name)
   {
-    std::string p = indexer_.OpenFile(name);
+    std::string p = indexer_.OpenFile(name, IsAllowUtf8());
     writer_.OpenFile(p.c_str());
   }
 
   void HierarchicalZipWriter::OpenDirectory(const std::string& name)
   {
-    indexer_.OpenDirectory(name);
+    indexer_.OpenDirectory(name, IsAllowUtf8());
   }
 
   void HierarchicalZipWriter::CloseDirectory()
--- a/OrthancFramework/Sources/Compression/HierarchicalZipWriter.h	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Compression/HierarchicalZipWriter.h	Fri Feb 27 18:37:52 2026 +0100
@@ -41,7 +41,6 @@
   {
 #if ORTHANC_BUILD_UNIT_TESTS == 1
     FRIEND_TEST(HierarchicalZipWriter, Index);
-    FRIEND_TEST(HierarchicalZipWriter, Filenames);
 #endif
 
   private:
@@ -60,7 +59,8 @@
   
       Stack stack_;
 
-      std::string EnsureUniqueFilename(const std::string& name);
+      std::string EnsureUniqueFilename(const std::string& name,
+                                       bool allowUtf8);
 
     public:
       Index();
@@ -69,15 +69,15 @@
 
       bool IsRoot() const;
 
-      std::string OpenFile(const std::string& name);
+      std::string OpenFile(const std::string& name,
+                           bool allowUtf8);
 
-      void OpenDirectory(const std::string& name);
+      void OpenDirectory(const std::string& name,
+                         bool allowUtf8);
 
       void CloseDirectory();
 
       std::string GetCurrentDirectoryPath() const;
-
-      static std::string KeepAlphanumeric(const std::string& source);
     };
 
     Index indexer_;
@@ -95,6 +95,16 @@
 
     bool IsZip64() const;
 
+    void SetAllowUtf8(bool allowUtf8)
+    {
+      writer_.SetAllowUtf8(allowUtf8);
+    }
+
+    bool IsAllowUtf8() const
+    {
+      return writer_.IsAllowUtf8();
+    }
+
     void SetCompressionLevel(uint8_t level);
 
     uint8_t GetCompressionLevel() const;
--- a/OrthancFramework/Sources/Compression/ZipWriter.cpp	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Compression/ZipWriter.cpp	Fri Feb 27 18:37:52 2026 +0100
@@ -441,6 +441,7 @@
   ZipWriter::ZipWriter() :
     pimpl_(new PImpl),
     isZip64_(false),
+    allowUtf8_(false),
     hasFileInZip_(false),
     append_(false),
     compressionLevel_(6)
@@ -608,29 +609,7 @@
   {
     Open();
 
-    std::string utf8;
-
-#if 0
-    if (Toolbox::IsValidUtf8(filename))
-    {
-      /**
-       * Starting with minizip 1.3.2 (released on 17 Feb 2026), bit 11
-       * of the ZIP header flags is appropriately set if the filename
-       * is encoded using UTF-8. This enables compatibility with UTF-8.
-       * https://zlib.net/ChangeLog.txt
-       * https://discourse.orthanc-server.org/t/seriesdescription-characters-and-removed-during-oe2-zip-export/6397
-       **/
-      utf8 = filename;
-    }
-    else
-    {
-      // Fallback if the filename is not encoded using valid UTF-8
-      utf8 = Toolbox::ConvertToAscii(filename);
-    }
-#else
-    // This was the code used in Orthanc <= 1.12.10
-    utf8 = Toolbox::ConvertToAscii(filename);
-#endif
+    const std::string normalized = Toolbox::NormalizePath(filename, allowUtf8_);
 
     zip_fileinfo zfi;
     PrepareFileInfo(zfi);
@@ -639,7 +618,7 @@
 
     if (isZip64_)
     {
-      result = zipOpenNewFileInZip64(pimpl_->file_, utf8.c_str(),
+      result = zipOpenNewFileInZip64(pimpl_->file_, normalized.c_str(),
                                      &zfi,
                                      NULL,   0,
                                      NULL,   0,
@@ -649,7 +628,7 @@
     }
     else
     {
-      result = zipOpenNewFileInZip(pimpl_->file_, utf8.c_str(),
+      result = zipOpenNewFileInZip(pimpl_->file_, normalized.c_str(),
                                    &zfi,
                                    NULL,   0,
                                    NULL,   0,
--- a/OrthancFramework/Sources/Compression/ZipWriter.h	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Compression/ZipWriter.h	Fri Feb 27 18:37:52 2026 +0100
@@ -130,6 +130,7 @@
     boost::shared_ptr<PImpl> pimpl_;
 
     bool isZip64_;
+    bool allowUtf8_;
     bool hasFileInZip_;
     bool append_;
     uint8_t compressionLevel_;
@@ -146,6 +147,26 @@
 
     bool IsZip64() const;
 
+    /**
+     * Starting with minizip 1.3.2 (released on 17 Feb 2026), bit 11
+     * of the ZIP header flags is appropriately set if the path is
+     * encoded using UTF-8. This enables compatibility with UTF-8.
+     * The behavior of Orthanc <= 1.12.10 corresponds to
+     * "SetAllowUtf8(false)".
+     *
+     * https://zlib.net/ChangeLog.txt
+     * https://discourse.orthanc-server.org/t/seriesdescription-characters-and-removed-during-oe2-zip-export/6397
+     **/
+    void SetAllowUtf8(bool allowUtf8)
+    {
+      allowUtf8_ = allowUtf8;
+    }
+
+    bool IsAllowUtf8() const
+    {
+      return allowUtf8_;
+    }
+
     void SetCompressionLevel(uint8_t level);
 
     uint8_t GetCompressionLevel() const;
--- a/OrthancFramework/Sources/Toolbox.cpp	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Toolbox.cpp	Fri Feb 27 18:37:52 2026 +0100
@@ -2958,6 +2958,67 @@
   }
 
 
+  std::string Toolbox::NormalizePath(const std::string& utf8,
+                                     bool allowUtf8)
+  {
+    std::string converted;
+
+    if (allowUtf8 &&
+        IsValidUtf8(utf8))
+    {
+      converted = utf8;
+    }
+    else
+    {
+      converted = Toolbox::ConvertToAscii(utf8);
+    }
+
+    bool previousSpace = true;
+
+    std::string result;
+    result.reserve(converted.size());
+
+    for (size_t i = 0; i < converted.size(); i++)
+    {
+      unsigned char c = converted[i];
+
+      if (isspace(c) ||
+          c <= 31 ||  // Control characters
+
+          // Remove characters that could break some filesystems
+          c == ':' ||
+          c == '*' ||
+          c == '?' ||
+          c == '"' ||
+          c == '<' ||
+          c == '>' ||
+          c == '|')
+      {
+        if (!previousSpace)
+        {
+          previousSpace = true;
+          result.push_back(' ');
+        }
+      }
+      else
+      {
+        previousSpace = false;
+
+        if (c == '\\')
+        {
+          result.push_back('/');  // Replace backslashes by forward slashes
+        }
+        else
+        {
+          result.push_back(c);
+        }
+      }
+    }
+
+    return Toolbox::StripSpaces(result);
+  }
+
+
   Toolbox::ElapsedTimer::ElapsedTimer()
   {
     Restart();
--- a/OrthancFramework/Sources/Toolbox.h	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/Sources/Toolbox.h	Fri Feb 27 18:37:52 2026 +0100
@@ -407,6 +407,9 @@
 
     static bool IsValidUtf8(const std::string& s);
 
+    static std::string NormalizePath(const std::string& utf8,
+                                     bool allowUtf8);
+
     class ORTHANC_PUBLIC ElapsedTimer : public boost::noncopyable
     {
     private:
--- a/OrthancFramework/UnitTestsSources/ZipTests.cpp	Fri Feb 27 16:57:00 2026 +0100
+++ b/OrthancFramework/UnitTestsSources/ZipTests.cpp	Fri Feb 27 18:37:52 2026 +0100
@@ -43,7 +43,7 @@
 
 TEST(ZipWriter, Basic)
 {
-  Orthanc::ZipWriter w;
+  ZipWriter w;
   w.SetOutputPath("UnitTestsResults/hello.zip");
   w.Open();
   w.OpenFile("world/hello");
@@ -53,7 +53,7 @@
 
 TEST(ZipWriter, Basic64)
 {
-  Orthanc::ZipWriter w;
+  ZipWriter w;
   w.SetOutputPath("UnitTestsResults/hello64.zip");
   w.SetZip64(true);
   w.Open();
@@ -64,18 +64,18 @@
 
 TEST(ZipWriter, Exceptions)
 {
-  Orthanc::ZipWriter w;
-  ASSERT_THROW(w.Open(), Orthanc::OrthancException);
+  ZipWriter w;
+  ASSERT_THROW(w.Open(), OrthancException);
   w.SetOutputPath("UnitTestsResults/hello3.zip");
   w.Open();
-  ASSERT_THROW(w.Write("hello world"), Orthanc::OrthancException);
+  ASSERT_THROW(w.Write("hello world"), OrthancException);
 }
 
 
 TEST(ZipWriter, Append)
 {
   {
-    Orthanc::ZipWriter w;
+    ZipWriter w;
     w.SetAppendToExisting(false);
     w.SetOutputPath("UnitTestsResults/append.zip");
     w.Open();
@@ -84,7 +84,7 @@
   }
 
   {
-    Orthanc::ZipWriter w;
+    ZipWriter w;
     w.SetAppendToExisting(true);
     w.SetOutputPath("UnitTestsResults/append.zip");
     w.Open();
@@ -105,40 +105,46 @@
   TEST(HierarchicalZipWriter, Index)
   {
     HierarchicalZipWriter::Index i;
-    ASSERT_EQ("hello", i.OpenFile("hello"));
-    ASSERT_EQ("hello-2", i.OpenFile("hello"));
-    ASSERT_EQ("coucou", i.OpenFile("coucou"));
-    ASSERT_EQ("hello-3", i.OpenFile("hello"));
+    ASSERT_EQ("hello", i.OpenFile("hello", true));
+    ASSERT_EQ("hello-2", i.OpenFile("hello", true));
+    ASSERT_EQ("coucou", i.OpenFile("coucou", true));
+    ASSERT_EQ("hello-3", i.OpenFile("hello", true));
+    ASSERT_EQ("trE hell", i.OpenFile("    ÊtrE hellô  ", false));
+    ASSERT_EQ("ÊtrE hellô", i.OpenFile("    ÊtrE hellô  ", true));
+    ASSERT_THROW(i.OpenFile("hel/lo", true), OrthancException);
+    ASSERT_THROW(i.OpenFile("hel\\lo", true), OrthancException);
 
-    i.OpenDirectory("coucou");
+    i.OpenDirectory("coucou", true);
 
-    ASSERT_EQ("coucou-2/world", i.OpenFile("world"));
-    ASSERT_EQ("coucou-2/world-2", i.OpenFile("world"));
+    ASSERT_EQ("coucou-2/world", i.OpenFile("world", true));
+    ASSERT_EQ("coucou-2/world-2", i.OpenFile("world", true));
 
-    i.OpenDirectory("world");
+    ASSERT_THROW(i.OpenDirectory("hel/lo", true), OrthancException);
+    ASSERT_THROW(i.OpenDirectory("hel\\lo", true), OrthancException);
+
+    i.OpenDirectory("world", true);
   
-    ASSERT_EQ("coucou-2/world-3/hello", i.OpenFile("hello"));
-    ASSERT_EQ("coucou-2/world-3/hello-2", i.OpenFile("hello"));
+    ASSERT_EQ("coucou-2/world-3/hello", i.OpenFile("hello", true));
+    ASSERT_EQ("coucou-2/world-3/hello-2", i.OpenFile("hello", true));
 
     i.CloseDirectory();
 
-    ASSERT_EQ("coucou-2/world-4", i.OpenFile("world"));
+    ASSERT_EQ("coucou-2/world-4", i.OpenFile("world", true));
 
     i.CloseDirectory();
 
-    ASSERT_EQ("coucou-3", i.OpenFile("coucou"));
+    ASSERT_EQ("coucou-3", i.OpenFile("coucou", true));
 
     ASSERT_THROW(i.CloseDirectory(), OrthancException);
   }
+}
 
 
-  TEST(HierarchicalZipWriter, Filenames)
-  {
-    ASSERT_EQ("trE hell", HierarchicalZipWriter::Index::KeepAlphanumeric("    ÊtrE hellô  "));
-
-    // The "^" character is considered as a space in DICOM
-    ASSERT_EQ("Hel lo world", HierarchicalZipWriter::Index::KeepAlphanumeric("    Hel^^  ^\r\n\t^^lo  \t  <world>  "));
-  }
+TEST(HierarchicalZipWriter, Filenames)
+{
+  ASSERT_EQ("trE hell", Toolbox::NormalizePath("    ÊtrE hellô  ", false));
+  ASSERT_EQ("ÊtrE hellô", Toolbox::NormalizePath("    ÊtrE hellô  ", true));
+  ASSERT_EQ("Hel^^ ^ ^^lo world", Toolbox::NormalizePath("    Hel^^   ^\r\n\t^^lo  \t  <world>  ", true));
 }
 
 
@@ -193,7 +199,7 @@
   TemporaryFile f;
   
   {
-    Orthanc::ZipWriter w;
+    ZipWriter w;
     ASSERT_EQ(0u, w.GetArchiveSize());
 
     w.SetOutputPath(f.GetPath());
@@ -233,7 +239,7 @@
   for (int i = 0; i < 2; i++)
   {
     {
-      Orthanc::ZipWriter w;
+      ZipWriter w;
       w.SetMemoryOutput(memory, (i == 0) /* ZIP64? */);
       w.Open();
 
@@ -242,13 +248,13 @@
       w.CancelStream();
     }
 
-    ASSERT_THROW(ZipReader::CreateFromMemory(memory), Orthanc::OrthancException);
+    ASSERT_THROW(ZipReader::CreateFromMemory(memory), OrthancException);
 
     memory.clear();
     uint64_t archiveSize;
     
     {
-      Orthanc::ZipWriter w;
+      ZipWriter w;
       ASSERT_EQ(0u, w.GetArchiveSize());
       
       w.SetMemoryOutput(memory, (i == 0) /* ZIP64? */);
@@ -426,14 +432,14 @@
   static const std::string path = "UnitTestsResults/hello-utf8.zip";
 
   {
-    Orthanc::ZipWriter w;
+    ZipWriter w;
     w.SetOutputPath(path);
 
     for (unsigned int i = 0; i < testEncodingsCount; i++)
     {
       std::string source(testEncodingsEncoded[i]);
       std::string s = Toolbox::ConvertToUtf8(source, testEncodings[i], false, false);
-      ASSERT_TRUE(Orthanc::Toolbox::IsValidUtf8(s));
+      ASSERT_TRUE(Toolbox::IsValidUtf8(s));
       printf("[%s]\n", s.c_str());
       w.OpenFile(s);
       w.Write("Hello");