diff OrthancFramework/UnitTestsSources/DicomMapTests.cpp @ 4220:92a21efa5c96

reorganization of DicomStreamReader
author Sebastien Jodogne <s.jodogne@gmail.com>
date Wed, 30 Sep 2020 15:33:47 +0200
parents b8ed2852a35d
children 3d6f14a05db1
line wrap: on
line diff
--- a/OrthancFramework/UnitTestsSources/DicomMapTests.cpp	Wed Sep 30 12:18:43 2020 +0200
+++ b/OrthancFramework/UnitTestsSources/DicomMapTests.cpp	Wed Sep 30 15:33:47 2020 +0200
@@ -34,6 +34,7 @@
 #include "../Sources/Compatibility.h"
 #include "../Sources/OrthancException.h"
 #include "../Sources/DicomFormat/DicomMap.h"
+#include "../Sources/DicomFormat/DicomStreamReader.h"
 #include "../Sources/DicomParsing/FromDcmtkBridge.h"
 #include "../Sources/DicomParsing/ToDcmtkBridge.h"
 #include "../Sources/DicomParsing/ParsedDicomFile.h"
@@ -798,721 +799,6 @@
 
 namespace
 {
-  class StreamBlockReader : public boost::noncopyable
-  {
-  private:
-    std::istream&  stream_;
-    std::string    block_;
-    size_t         blockPos_;
-    uint64_t       processedBytes_;
-
-  public:
-    StreamBlockReader(std::istream& stream) :
-      stream_(stream),
-      blockPos_(0),
-      processedBytes_(0)
-    {
-    }
-
-    void Schedule(size_t blockSize)
-    {
-      if (!block_.empty())
-      {
-        throw OrthancException(ErrorCode_BadSequenceOfCalls);
-      }
-      else
-      {
-        block_.resize(blockSize);
-        blockPos_ = 0;
-      }
-    }
-
-    bool Read(std::string& block)
-    {
-      if (block_.empty())
-      {
-        if (blockPos_ != 0)
-        {
-          throw OrthancException(ErrorCode_BadSequenceOfCalls);
-        }
-        
-        block.clear();
-        return true;
-      }
-      else
-      {
-        while (blockPos_ < block_.size())
-        {
-#if 0          
-          char c;
-          stream_.get(c);
-
-          if (stream_.good())
-          {
-            block_[blockPos_] = c;
-            blockPos_++;
-          }
-          else
-          {
-            return false;
-          }
-#else
-          size_t n = block_.size() - blockPos_;
-          std::streamsize r = stream_.readsome(&block_[blockPos_], n);
-          if (r == 0)
-          {
-            return false;
-          }
-          else
-          {
-            blockPos_ += r;
-          }
-#endif
-        }
-
-        processedBytes_ += block_.size();
-
-        block.swap(block_);
-        block_.clear();
-        return true;
-      }
-    }
-
-    uint64_t GetProcessedBytes() const
-    {
-      return processedBytes_;
-    }
-  };
-
-
-
-
-  /**
-   * This class parses a stream containing a DICOM instance. It does
-   * *not* support the visit of sequences (it only works at the first
-   * level of the hierarchy), and it stops the processing once pixel
-   * data is reached in compressed transfer syntaxes.
-   **/
-  class DicomStreamReader : public boost::noncopyable
-  {
-  public:
-    class IVisitor : public boost::noncopyable
-    {
-    public:
-      virtual ~IVisitor()
-      {
-      }
-
-      // The data from this function will always be Little Endian (as
-      // specified by the DICOM standard)
-      virtual void VisitMetaHeaderTag(const DicomTag& tag,
-                                      const ValueRepresentation& vr,
-                                      const std::string& value) = 0;
-
-      // Return "false" to stop processing
-      virtual bool VisitDatasetTag(const DicomTag& tag,
-                                   const ValueRepresentation& vr,
-                                   DicomTransferSyntax transferSyntax,
-                                   const std::string& value,
-                                   bool isLittleEndian) = 0;
-    };
-    
-  private:
-    enum State
-    {
-      State_Preamble,
-      State_MetaHeader,
-      State_DatasetTag,
-      State_SequenceExplicitLength,
-      State_SequenceExplicitValue,
-      State_DatasetExplicitLength,
-      State_DatasetValue,
-      State_Done
-    };
-
-    StreamBlockReader    reader_;
-    State                state_;
-    DicomTransferSyntax  transferSyntax_;
-    DicomTag             previousTag_;
-    DicomTag             danglingTag_;  // Root-level tag
-    ValueRepresentation  danglingVR_;
-    unsigned int         sequenceDepth_;
-    
-    static uint16_t ReadUnsignedInteger16(const char* dicom,
-                                          bool littleEndian)
-    {
-      const uint8_t* p = reinterpret_cast<const uint8_t*>(dicom);
-
-      if (littleEndian)
-      {
-        return (static_cast<uint16_t>(p[0]) |
-                (static_cast<uint16_t>(p[1]) << 8));
-      }
-      else
-      {
-        return (static_cast<uint16_t>(p[1]) |
-                (static_cast<uint16_t>(p[0]) << 8));
-      }
-    }
-
-
-    static uint32_t ReadUnsignedInteger32(const char* dicom,
-                                          bool littleEndian)
-    {
-      const uint8_t* p = reinterpret_cast<const uint8_t*>(dicom);
-
-      if (littleEndian)
-      {
-        return (static_cast<uint32_t>(p[0]) |
-                (static_cast<uint32_t>(p[1]) << 8) |
-                (static_cast<uint32_t>(p[2]) << 16) |
-                (static_cast<uint32_t>(p[3]) << 24));
-      }
-      else
-      {
-        return (static_cast<uint32_t>(p[3]) |
-                (static_cast<uint32_t>(p[2]) << 8) |
-                (static_cast<uint32_t>(p[1]) << 16) |
-                (static_cast<uint32_t>(p[0]) << 24));
-      }
-    }
-
-
-    static DicomTag ReadTag(const char* dicom,
-                            bool littleEndian)
-    {
-      return DicomTag(ReadUnsignedInteger16(dicom, littleEndian),
-                      ReadUnsignedInteger16(dicom + 2, littleEndian));
-    }
-
-
-    static bool IsShortExplicitTag(ValueRepresentation vr)
-    {
-      /**
-       * Are we in the case of Table 7.1-2? "Data Element with
-       * Explicit VR of AE, AS, AT, CS, DA, DS, DT, FL, FD, IS, LO,
-       * LT, PN, SH, SL, SS, ST, TM, UI, UL and US"
-       * http://dicom.nema.org/medical/dicom/current/output/chtml/part05/chapter_7.html#sect_7.1.2
-       **/
-      return (vr == ValueRepresentation_ApplicationEntity   /* AE */ ||
-              vr == ValueRepresentation_AgeString           /* AS */ ||
-              vr == ValueRepresentation_AttributeTag        /* AT */ ||
-              vr == ValueRepresentation_CodeString          /* CS */ ||
-              vr == ValueRepresentation_Date                /* DA */ ||
-              vr == ValueRepresentation_DecimalString       /* DS */ ||
-              vr == ValueRepresentation_DateTime            /* DT */ ||
-              vr == ValueRepresentation_FloatingPointSingle /* FL */ ||
-              vr == ValueRepresentation_FloatingPointDouble /* FD */ ||
-              vr == ValueRepresentation_IntegerString       /* IS */ ||
-              vr == ValueRepresentation_LongString          /* LO */ ||
-              vr == ValueRepresentation_LongText            /* LT */ ||
-              vr == ValueRepresentation_PersonName          /* PN */ ||
-              vr == ValueRepresentation_ShortString         /* SH */ ||
-              vr == ValueRepresentation_SignedLong          /* SL */ ||
-              vr == ValueRepresentation_SignedShort         /* SS */ ||
-              vr == ValueRepresentation_ShortText           /* ST */ ||
-              vr == ValueRepresentation_Time                /* TM */ ||
-              vr == ValueRepresentation_UniqueIdentifier    /* UI */ ||
-              vr == ValueRepresentation_UnsignedLong        /* UL */ ||
-              vr == ValueRepresentation_UnsignedShort       /* US */);
-    }
-
-
-    bool IsLittleEndian() const
-    {
-      return (transferSyntax_ != DicomTransferSyntax_BigEndianExplicit);
-    }
-
-
-    void PrintBlock(const std::string& block)
-    {
-      for (size_t i = 0; i < block.size(); i++)
-      {
-        printf("%02x ", static_cast<uint8_t>(block[i]));
-        if (i % 16 == 15)
-          printf("\n");
-      }
-      printf("\n");
-    }
-    
-    void HandlePreamble(IVisitor& visitor,
-                        const std::string& block)
-    {
-      //printf("PREAMBLE:\n");
-      //PrintBlock(block);
-
-      assert(block.size() == 144u);
-      assert(reader_.GetProcessedBytes() == 144u);
-
-      /**
-       * The "DICOM file meta information" is always encoded using
-       * "Explicit VR Little Endian Transfer Syntax"
-       * http://dicom.nema.org/medical/dicom/current/output/chtml/part10/chapter_7.html
-       **/
-      if (block[128] != 'D' ||
-          block[129] != 'I' ||
-          block[130] != 'C' ||
-          block[131] != 'M' ||
-          ReadTag(block.c_str() + 132, true) != DicomTag(0x0002, 0x0000) ||
-          block[136] != 'U' ||
-          block[137] != 'L' ||
-          ReadUnsignedInteger16(block.c_str() + 138, true) != 4)
-      {
-        throw OrthancException(ErrorCode_BadFileFormat);
-      }
-
-      uint32_t length = ReadUnsignedInteger32(block.c_str() + 140, true);
-
-      reader_.Schedule(length);
-      state_ = State_MetaHeader;
-    }
-
-    
-    void HandleMetaHeader(IVisitor& visitor,
-                          const std::string& block)
-    {
-      //printf("META-HEADER:\n");
-      //PrintBlock(block);
-
-      size_t pos = 0;
-      const char* p = block.c_str();
-
-      bool hasTransferSyntax = false;
-
-      while (pos + 8 <= block.size())
-      {
-        DicomTag tag = ReadTag(p + pos, true);
-        
-        ValueRepresentation vr = StringToValueRepresentation(std::string(p + pos + 4, 2), true);
-
-        if (IsShortExplicitTag(vr))
-        {
-          uint16_t length = ReadUnsignedInteger16(p + pos + 6, true);
-
-          std::string value;
-          value.assign(p + pos + 8, length);
-
-          if (tag.GetGroup() == 0x0002)
-          {
-            visitor.VisitMetaHeaderTag(tag, vr, value);
-          }                  
-
-          if (tag == DICOM_TAG_TRANSFER_SYNTAX_UID)
-          {
-            // Remove possible padding byte
-            if (!value.empty() &&
-                value[value.size() - 1] == '\0')
-            {
-              value.resize(value.size() - 1);
-            }
-            
-            if (LookupTransferSyntax(transferSyntax_, value))
-            {
-              hasTransferSyntax = true;
-            }
-            else
-            {
-              throw OrthancException(ErrorCode_NotImplemented, "Unsupported transfer syntax: " + value);
-            }
-          }
-          
-          pos += length + 8;
-        }
-        else if (pos + 12 <= block.size())
-        {
-          uint16_t reserved = ReadUnsignedInteger16(p + pos + 6, true);
-          if (reserved != 0)
-          {
-            break;
-          }
-          
-          uint32_t length = ReadUnsignedInteger32(p + pos + 8, true);
-
-          std::string value;
-          value.assign(p + pos + 12, length);
-
-          if (tag.GetGroup() == 0x0002)
-          {
-            visitor.VisitMetaHeaderTag(tag, vr, value);
-          }                  
-          
-          pos += length + 12;
-        }
-      }
-
-      if (pos != block.size())
-      {
-        throw OrthancException(ErrorCode_BadFileFormat);
-      }
-
-      if (!hasTransferSyntax)
-      {
-        throw OrthancException(ErrorCode_BadFileFormat, "DICOM file meta-header without transfer syntax UID");
-      }
-
-      reader_.Schedule(8);
-      state_ = State_DatasetTag;
-    }
-    
-
-    void HandleDatasetTag(const std::string& block,
-                          const DicomTag& untilTag)
-    {
-      static const DicomTag DICOM_TAG_SEQUENCE_ITEM(0xfffe, 0xe000);
-      static const DicomTag DICOM_TAG_SEQUENCE_DELIMITATION_ITEM(0xfffe, 0xe00d);
-      static const DicomTag DICOM_TAG_SEQUENCE_DELIMITATION_SEQUENCE(0xfffe, 0xe0dd);
-
-      assert(block.size() == 8u);
-
-      const bool littleEndian = IsLittleEndian();
-      DicomTag tag = ReadTag(block.c_str(), littleEndian);
-
-      if (sequenceDepth_ == 0 &&
-          tag >= untilTag)
-      {
-        state_ = State_Done;
-        return;
-      }
-      
-      if (tag == DICOM_TAG_SEQUENCE_ITEM ||
-          tag == DICOM_TAG_SEQUENCE_DELIMITATION_ITEM ||
-          tag == DICOM_TAG_SEQUENCE_DELIMITATION_SEQUENCE)
-      {
-        //printf("SEQUENCE TAG:\n");
-        //PrintBlock(block);
-
-        // The special sequence items are encoded like "Implicit VR"
-        uint32_t length = ReadUnsignedInteger32(block.c_str() + 4, littleEndian);
-
-        if (tag == DICOM_TAG_SEQUENCE_ITEM)
-        {
-          for (unsigned int i = 0; i <= sequenceDepth_; i++)
-            printf("  ");
-          if (length == 0xffffffffu)
-          {
-            // Undefined length: Need to loop over the tags of the nested dataset
-            printf("...next dataset in sequence...\n");
-            reader_.Schedule(8);
-            state_ = State_DatasetTag;
-          }
-          else
-          {
-            // Explicit length: Can skip the full sequence at once
-            printf("...next dataset in sequence... %u bytes\n", length);
-            reader_.Schedule(length);
-            state_ = State_DatasetValue;
-          }
-        }
-        else if (tag == DICOM_TAG_SEQUENCE_DELIMITATION_ITEM ||
-                 tag == DICOM_TAG_SEQUENCE_DELIMITATION_SEQUENCE)
-        {
-          if (length != 0 ||
-              sequenceDepth_ == 0)
-          {
-            throw OrthancException(ErrorCode_BadFileFormat);
-          }
-
-          if (tag == DICOM_TAG_SEQUENCE_DELIMITATION_SEQUENCE)
-          {
-            for (unsigned int i = 0; i < sequenceDepth_; i++)
-              printf("  ");
-            printf("...leaving sequence...\n");
-
-            sequenceDepth_ --;
-          }
-          else
-          {
-            if (sequenceDepth_ == 0)
-            {
-              throw OrthancException(ErrorCode_BadFileFormat);
-            }
-          }
-
-          reader_.Schedule(8);
-          state_ = State_DatasetTag;          
-        }
-        else
-        {
-          throw OrthancException(ErrorCode_InternalError);
-        }
-      }
-      else
-      {
-        //printf("DATASET TAG:\n");
-        //PrintBlock(block);
-
-        previousTag_ = tag;
-          
-        ValueRepresentation vr = ValueRepresentation_Unknown;
-        
-        if (transferSyntax_ == DicomTransferSyntax_LittleEndianImplicit)
-        {
-          if (sequenceDepth_ == 0)
-          {
-            danglingTag_ = tag;
-            danglingVR_ = vr;
-          }
-
-          uint32_t length = ReadUnsignedInteger32(block.c_str() + 4, true /* little endian */);
-          HandleDatasetExplicitLength(length);
-        }
-        else
-        {
-          // This in an explicit transfer syntax
-
-          vr = StringToValueRepresentation(
-            std::string(block.c_str() + 4, 2), false /* ignore unknown VR */);
-
-          if (vr != ValueRepresentation_Sequence &&
-              sequenceDepth_ > 0)
-          {
-            for (unsigned int i = 0; i <= sequenceDepth_; i++)
-              printf("  ");
-            printf("%s\n", tag.Format().c_str());
-          }
-          
-          if (vr == ValueRepresentation_Sequence)
-          {
-            for (unsigned int i = 0; i <= sequenceDepth_; i++)
-              printf("  ");
-            printf("...entering sequence... %s\n", tag.Format().c_str());
-            sequenceDepth_ ++;
-            reader_.Schedule(4);
-            state_ = State_SequenceExplicitLength;
-          }
-          else if (IsShortExplicitTag(vr))
-          {
-            uint16_t length = ReadUnsignedInteger16(block.c_str() + 6, littleEndian);
-
-            reader_.Schedule(length);
-            state_ = State_DatasetValue;
-          }
-          else
-          {
-            uint16_t reserved = ReadUnsignedInteger16(block.c_str() + 6, littleEndian);
-            if (reserved != 0)
-            {
-              throw OrthancException(ErrorCode_BadFileFormat);
-            }
-
-            reader_.Schedule(4);
-            state_ = State_DatasetExplicitLength;
-          }
-
-          if (sequenceDepth_ == 0)
-          {
-            danglingTag_ = tag;
-            danglingVR_ = vr;
-          }
-        }
-      }
-    }
-
-
-    void HandleDatasetExplicitLength(uint32_t length)
-    {
-      if (length == 0xffffffffu)
-      {
-        /**
-         * This is the case of pixel data with compressed transfer
-         * syntaxes. Schedule the reading of the first tag of the
-         * nested dataset.
-         * http://dicom.nema.org/medical/dicom/current/output/chtml/part05/sect_7.5.html
-         **/
-
-        for (unsigned int i = 0; i <= sequenceDepth_; i++)
-          printf("  ");
-        printf("...entering sequence... %s\n", previousTag_.Format().c_str());
-        
-        state_ = State_DatasetTag;
-        reader_.Schedule(8);
-        sequenceDepth_ ++;
-      }
-      else
-      {
-        reader_.Schedule(length);
-        state_ = State_DatasetValue;
-      }
-    }    
-    
-    void HandleDatasetExplicitLength(const std::string& block)
-    {
-      //printf("DATASET TAG LENGTH:\n");
-      //PrintBlock(block);
-
-      assert(block.size() == 4);
-
-      uint32_t length = ReadUnsignedInteger32(block.c_str(), IsLittleEndian());
-      HandleDatasetExplicitLength(length);
-    }
-
-    void HandleSequenceExplicitLength(const std::string& block)
-    {
-      //printf("DATASET TAG LENGTH:\n");
-      //PrintBlock(block);
-
-      assert(block.size() == 4);
-
-      uint32_t length = ReadUnsignedInteger32(block.c_str(), IsLittleEndian());
-      if (length == 0xffffffffu)
-      {
-        state_ = State_DatasetTag;
-        reader_.Schedule(8);
-      }
-      else
-      {
-        for (unsigned int i = 0; i <= sequenceDepth_; i++)
-          printf("  ");
-        printf("...skipping sequence thanks to explicit length... %d\n", length);
-
-        reader_.Schedule(length);
-        state_ = State_SequenceExplicitValue;
-      }
-    }
-
-    void HandleSequenceExplicitValue()
-    {
-      if (sequenceDepth_ == 0)
-      {
-        throw OrthancException(ErrorCode_InternalError);
-      }
-
-      sequenceDepth_ --;
-
-      state_ = State_DatasetTag;
-      reader_.Schedule(8);
-    }
-
-
-    void HandleDatasetValue(IVisitor& visitor,
-                            const std::string& block)
-    {
-      if (sequenceDepth_ == 0)
-      {
-        bool c;
-        
-        if (!block.empty() &&
-            (block[block.size() - 1] == ' ' ||
-             block[block.size() - 1] == '\0') &&
-            (danglingVR_ == ValueRepresentation_ApplicationEntity ||
-             danglingVR_ == ValueRepresentation_AgeString ||
-             danglingVR_ == ValueRepresentation_CodeString ||
-             danglingVR_ == ValueRepresentation_DecimalString ||
-             danglingVR_ == ValueRepresentation_IntegerString ||
-             danglingVR_ == ValueRepresentation_LongString ||
-             danglingVR_ == ValueRepresentation_LongText ||
-             danglingVR_ == ValueRepresentation_PersonName ||
-             danglingVR_ == ValueRepresentation_ShortString ||
-             danglingVR_ == ValueRepresentation_ShortText ||
-             danglingVR_ == ValueRepresentation_UniqueIdentifier ||
-             danglingVR_ == ValueRepresentation_UnlimitedText))
-        {
-          std::string s(block.begin(), block.end() - 1);
-          c = visitor.VisitDatasetTag(danglingTag_, danglingVR_, transferSyntax_, s, IsLittleEndian());
-        }
-        else
-        {
-          c = visitor.VisitDatasetTag(danglingTag_, danglingVR_, transferSyntax_, block, IsLittleEndian());
-        }
-        
-        if (!c)
-        {
-          state_ = State_Done;
-          return;
-        }
-      }
-
-      reader_.Schedule(8);
-      state_ = State_DatasetTag;
-    }
-    
-    
-  public:
-    DicomStreamReader(std::istream& stream) :
-      reader_(stream),
-      state_(State_Preamble),
-      transferSyntax_(DicomTransferSyntax_LittleEndianImplicit),  // Dummy
-      previousTag_(0x0000, 0x0000),  // Dummy
-      danglingTag_(0x0000, 0x0000),  // Dummy
-      danglingVR_(ValueRepresentation_Unknown),  // Dummy
-      sequenceDepth_(0)
-    {
-      reader_.Schedule(128 /* empty header */ +
-                       4 /* "DICM" magic value */ +
-                       4 /* (0x0002, 0x0000) tag */ +
-                       2 /* value representation of (0x0002, 0x0000) == "UL" */ +
-                       2 /* length of "UL" value == 4 */ +
-                       4 /* actual length of the meta-header */);
-    }
-
-    void Consume(IVisitor& visitor,
-                 const DicomTag& untilTag)
-    {
-      while (state_ != State_Done)
-      {
-        std::string block;
-        if (reader_.Read(block))
-        {
-          switch (state_)
-          {
-            case State_Preamble:
-              HandlePreamble(visitor, block);
-              break;
-
-            case State_MetaHeader:
-              HandleMetaHeader(visitor, block);
-              break;
-
-            case State_DatasetTag:
-              HandleDatasetTag(block, untilTag);
-              break;
-
-            case State_DatasetExplicitLength:
-              HandleDatasetExplicitLength(block);
-              break;
-
-            case State_SequenceExplicitLength:
-              HandleSequenceExplicitLength(block);
-              break;
-
-            case State_SequenceExplicitValue:
-              HandleSequenceExplicitValue();
-              break;
-
-            case State_DatasetValue:
-              HandleDatasetValue(visitor, block);
-              break;
-
-            default:
-              throw OrthancException(ErrorCode_InternalError);
-          }
-        }
-        else
-        {
-          return;  // No more data in the stream
-        }
-      }
-    }
-
-    void Consume(IVisitor& visitor)
-    {
-      DicomTag untilTag(0xffff, 0xffff);
-      Consume(visitor, untilTag);
-    }
-
-    bool IsDone() const
-    {
-      return (state_ == State_Done);
-    }
-
-    uint64_t GetProcessedBytes() const
-    {
-      return reader_.GetProcessedBytes();
-    }
-  };
-
-
-
   class V : public DicomStreamReader::IVisitor
   {
   private:
@@ -1531,9 +817,13 @@
       std::cout << "Header: " << tag.Format() << " [" << Toolbox::ConvertToAscii(value).c_str() << "] (" << value.size() << ")" << std::endl;
     }
 
+    virtual void VisitTransferSyntax(DicomTransferSyntax transferSyntax) ORTHANC_OVERRIDE
+    {
+      printf("TRANSFER SYNTAX: %s\n", GetTransferSyntaxUid(transferSyntax));
+    }
+    
     virtual bool VisitDatasetTag(const DicomTag& tag,
                                  const ValueRepresentation& vr,
-                                 DicomTransferSyntax transferSyntax,
                                  const std::string& value,
                                  bool isLittleEndian) ORTHANC_OVERRIDE
     {
@@ -1648,8 +938,8 @@
             
             try
             {
-              //r.Consume(visitor, DICOM_TAG_PIXEL_DATA);
-              r.Consume(visitor);
+              r.Consume(visitor, DICOM_TAG_PIXEL_DATA);
+              //r.Consume(visitor);
               success++;
             }
             catch (OrthancException& e)