Merge pull request #508 from fetzerch/pictures-metadata-fixes

Updated metadata parsing for pictures

The relevant (but hard to read) spec would be the Photoshop File Format which references IPTC IIM:
http://www.adobe.com/devnet-apps/photoshop/fileformatashtml/PhotoshopFileFormats.htm#50577409_pgfId-1037504-
http://www.iptc.org/std/IIM/4.1/specification/IIMV4.1.pdf

3 files changed, 88 insertions, 39 deletions

diff --git a/lib/libexif/ExifParse.cpp b/lib/libexif/ExifParse.cpp
index 74ecbbef78..2386fd3dab 100644
--- a/lib/libexif/ExifParse.cpp
+++ b/lib/libexif/ExifParse.cpp
@@ -413,7 +413,13 @@ void CExifParse::ProcessDir(const unsigned char* const DirStart,
     // Extract useful components of tag
     switch(Tag)
     {
-//      case TAG_DESCRIPTION:       strncpy(m_ExifInfo->Description, ValuePtr, 5);    break;
+      case TAG_DESCRIPTION:
+      {
+        int length = max(ByteCount, 0);
+        length = min(length, MAX_COMMENT);
+        strncpy(m_ExifInfo->Description, (char *)ValuePtr, length);
+        break;
+      }
       case TAG_MAKE:              strncpy(m_ExifInfo->CameraMake, (char *)ValuePtr, 32);    break;
       case TAG_MODEL:             strncpy(m_ExifInfo->CameraModel, (char *)ValuePtr, 40);    break;
 //      case TAG_SOFTWARE:          strncpy(m_ExifInfo->Software, ValuePtr, 5);    break;
@@ -446,7 +452,7 @@ void CExifParse::ProcessDir(const unsigned char* const DirStart,
       {
         const int EXIF_COMMENT_HDR_LENGTH = 8;        // All comment tags have 8 bytes of header info
         int length = max(ByteCount - EXIF_COMMENT_HDR_LENGTH, 0);
-        length = min(length, 2000);
+        length = min(length, MAX_COMMENT);
         strncpy(m_ExifInfo->Comments, (char *)ValuePtr+EXIF_COMMENT_HDR_LENGTH, length);
 //        FixComment(comment);                          // Ensure comment is printable
       }
diff --git a/lib/libexif/IptcParse.cpp b/lib/libexif/IptcParse.cpp
index 9b21c64e9d..b58e1f0189 100644
--- a/lib/libexif/IptcParse.cpp
+++ b/lib/libexif/IptcParse.cpp
@@ -37,6 +37,7 @@
 #include "ExifParse.h"
 
 // Supported IPTC entry types
+#define IPTC_RECORD_VERSION         0x00
 #define IPTC_SUPLEMENTAL_CATEGORIES 0x14
 #define IPTC_KEYWORDS               0x19
 #define IPTC_CAPTION                0x78
@@ -58,6 +59,9 @@
 #define IPTC_COPYRIGHT              0x0A
 #define IPTC_COUNTRY_CODE           0x64
 #define IPTC_REFERENCE_SERVICE      0x2D
+#define IPTC_TIME_CREATED           0x3C
+#define IPTC_SUB_LOCATION           0x5C
+#define IPTC_IMAGE_TYPE             0x82
 
 
 //--------------------------------------------------------------------------
@@ -89,66 +93,100 @@ bool CIptcParse::Process (const unsigned char* const Data, const unsigned short
 
   // Check IPTC signatures
   char* pos = (char*)(Data + sizeof(short));  // position data pointer after length field
+  char* maxpos = (char*)(Data+itemlen);
+  unsigned char headerLen = 0;
+  unsigned char dataLen = 0;
+  memset(info, 0, sizeof(IPTCInfo_t));
 
-  if (memcmp(pos, IptcSignature1, strlen(IptcSignature1)) != 0) return false;
-  pos += strlen(IptcSignature1) + 1;          // move data pointer to the next field
+  if (itemlen < 25) return false;
 
-  if (memcmp(pos, IptcSignature2, strlen(IptcSignature2)) != 0) return false;
-  pos += strlen(IptcSignature2);              // move data pointer to the next field
+  if (memcmp(pos, IptcSignature1, strlen(IptcSignature1)-1) != 0) return false;
+  pos += sizeof(IptcSignature1);          // move data pointer to the next field
 
-  if (memcmp(pos, IptcSignature3, sizeof(IptcSignature3)) != 0) return false;
-  pos += sizeof(IptcSignature3);              // move data pointer to the next field
+  if (memcmp(pos, IptcSignature2, strlen(IptcSignature2)-1) != 0) return false;
+  pos += sizeof(IptcSignature2)-1;              // move data pointer to the next field
+
+  while (memcmp(pos, IptcSignature3, sizeof(IptcSignature3)) != 0) { // loop on valid Photoshop blocks
+
+    pos += sizeof(IptcSignature3); // move data pointer to the Header Length
+    // Skip header
+    headerLen = *pos; // get header length and move data pointer to the next field
+    pos += (headerLen & 0xfe) + 2; // move data pointer to the next field (Header is padded to even length, counting the length byte)
+
+    pos += 3; // move data pointer to length, assume only one byte, TODO: use all 4 bytes
+
+    dataLen = *pos++;
+    pos += dataLen; // skip data section
+
+    if (memcmp(pos, IptcSignature2, sizeof(IptcSignature2) - 1) != 0) return false;
+    pos += sizeof(IptcSignature2) - 1; // move data pointer to the next field
+  }
+
+  pos += sizeof(IptcSignature3);          // move data pointer to the next field
+  if (pos >= maxpos) return false;
 
   // IPTC section found
 
   // Skip header
-  unsigned char headerLen = *pos++;           // get header length and move data pointer to the next field
+  headerLen = *pos++;           // get header length and move data pointer to the next field
   pos += headerLen + 1 - (headerLen % 2);     // move data pointer to the next field (Header is padded to even length, counting the length byte)
 
-  // Get length (Motorola format)
-  unsigned long length = CExifParse::Get32(pos);
+  if (pos + 4 >= maxpos) return false;
+
   pos += 4;                                   // move data pointer to the next field
 
   // Now read IPTC data
   while (pos < (char*)(Data + itemlen-5))
   {
+    if (pos + 5 > maxpos) return false;
+
     short signature = (*pos << 8) + (*(pos+1));
 
-    pos += sizeof(short);
-    if (signature != 0x1C02)
+    pos += 2;
+    if (signature != 0x1C01 && signature != 0x1C02)
       break;
 
     unsigned char  type = *pos++;
     unsigned short length  = (*pos << 8) + (*(pos+1));
-    pos += sizeof(short);                   // Skip tag length
+    pos += 2;                   // Skip tag length
+
+    if (pos + length > maxpos) return false;
+
     // Process tag here
     char *tag = NULL;
-    switch (type)
+    if (signature == 0x1C02)
     {
-      case IPTC_SUPLEMENTAL_CATEGORIES:   tag = info->SupplementalCategories;  break;
-      case IPTC_KEYWORDS:                 tag = info->Keywords;                break;
-      case IPTC_CAPTION:                  tag = info->Caption;                 break;
-      case IPTC_AUTHOR:                   tag = info->Author;                  break;
-      case IPTC_HEADLINE:                 tag = info->Headline;                break;
-      case IPTC_SPECIAL_INSTRUCTIONS:     tag = info->SpecialInstructions;     break;
-      case IPTC_CATEGORY:                 tag = info->Category;                break;
-      case IPTC_BYLINE:                   tag = info->Byline;                  break;
-      case IPTC_BYLINE_TITLE:             tag = info->BylineTitle;             break;
-      case IPTC_CREDIT:                   tag = info->Credit;                  break;
-      case IPTC_SOURCE:                   tag = info->Source;                  break;
-      case IPTC_COPYRIGHT_NOTICE:         tag = info->CopyrightNotice;         break;
-      case IPTC_OBJECT_NAME:              tag = info->ObjectName;              break;
-      case IPTC_CITY:                     tag = info->City;                    break;
-      case IPTC_STATE:                    tag = info->State;                   break;
-      case IPTC_COUNTRY:                  tag = info->Country;                 break;
-      case IPTC_TRANSMISSION_REFERENCE:   tag = info->TransmissionReference;   break;
-      case IPTC_DATE:                     tag = info->Date;                    break;
-      case IPTC_COPYRIGHT:                tag = info->Copyright;               break;
-      case IPTC_REFERENCE_SERVICE:        tag = info->ReferenceService;        break;
-      case IPTC_COUNTRY_CODE:             tag = info->CountryCode;             break;
-      default:
-        printf("IptcParse: Unrecognised IPTC tag: 0x%02x", type);
-      break;
+      switch (type)
+      {
+        case IPTC_RECORD_VERSION:           tag = info->RecordVersion;           break;
+        case IPTC_SUPLEMENTAL_CATEGORIES:   tag = info->SupplementalCategories;  break;
+        case IPTC_KEYWORDS:                 tag = info->Keywords;                break;
+        case IPTC_CAPTION:                  tag = info->Caption;                 break;
+        case IPTC_AUTHOR:                   tag = info->Author;                  break;
+        case IPTC_HEADLINE:                 tag = info->Headline;                break;
+        case IPTC_SPECIAL_INSTRUCTIONS:     tag = info->SpecialInstructions;     break;
+        case IPTC_CATEGORY:                 tag = info->Category;                break;
+        case IPTC_BYLINE:                   tag = info->Byline;                  break;
+        case IPTC_BYLINE_TITLE:             tag = info->BylineTitle;             break;
+        case IPTC_CREDIT:                   tag = info->Credit;                  break;
+        case IPTC_SOURCE:                   tag = info->Source;                  break;
+        case IPTC_COPYRIGHT_NOTICE:         tag = info->CopyrightNotice;         break;
+        case IPTC_OBJECT_NAME:              tag = info->ObjectName;              break;
+        case IPTC_CITY:                     tag = info->City;                    break;
+        case IPTC_STATE:                    tag = info->State;                   break;
+        case IPTC_COUNTRY:                  tag = info->Country;                 break;
+        case IPTC_TRANSMISSION_REFERENCE:   tag = info->TransmissionReference;   break;
+        case IPTC_DATE:                     tag = info->Date;                    break;
+        case IPTC_COPYRIGHT:                tag = info->Copyright;               break;
+        case IPTC_REFERENCE_SERVICE:        tag = info->ReferenceService;        break;
+        case IPTC_COUNTRY_CODE:             tag = info->CountryCode;             break;
+        case IPTC_TIME_CREATED:             tag = info->TimeCreated;             break;
+        case IPTC_SUB_LOCATION:             tag = info->SubLocation;             break;
+        case IPTC_IMAGE_TYPE:               tag = info->ImageType;               break;
+        default:
+          printf("IptcParse: Unrecognised IPTC tag: 0x%02x", type);
+          break;
+      }
     }
 
     if (tag)
diff --git a/lib/libexif/libexif.h b/lib/libexif/libexif.h
index e889892a73..f7dc532f0d 100644
--- a/lib/libexif/libexif.h
+++ b/lib/libexif/libexif.h
@@ -48,6 +48,7 @@ extern "C" {
 #define MAX_IPTC_STRING 256
 
 typedef struct {
+  char RecordVersion[MAX_IPTC_STRING];
   char SupplementalCategories[MAX_IPTC_STRING];
   char Keywords[MAX_IPTC_STRING];
   char Caption[MAX_IPTC_STRING];
@@ -69,6 +70,9 @@ typedef struct {
   char Copyright[MAX_IPTC_STRING];
   char ReferenceService[MAX_IPTC_STRING];
   char CountryCode[MAX_IPTC_STRING];
+  char TimeCreated[MAX_IPTC_STRING];
+  char SubLocation[MAX_IPTC_STRING];
+  char ImageType[MAX_IPTC_STRING];
 } IPTCInfo_t;
 
 #define MAX_COMMENT 2000
@@ -98,6 +102,7 @@ typedef struct {
     int   ISOequivalent;
     int   LightSource;
     char  Comments[MAX_COMMENT];
+    char  Description[MAX_COMMENT];
 
     unsigned ThumbnailOffset;          // Exif offset to thumbnail
     unsigned ThumbnailSize;            // Size of thumbnail.