Bound text value counts in ID3v2 and APE tags (#1458)

Null-separated text values create a ByteVector and String for each
tiny field. A 1 MB tag can allocate over 100 MB and abort
memory-limited applications.

Limit decoded values based on input size and stop before creating
String objects when the value count exceeds the budget.

taglib/ape/apeitem.cpp

@@ -25,6 +25,7 @@
 
 #include "apeitem.h"
 
+#include <algorithm>
 #include <utility>
 #include <numeric>
 
@@ -232,8 +233,20 @@ void APE::Item::parse(const ByteVector &data)
   setReadOnly(flags & 1);
   setType(static_cast<ItemTypes>((flags >> 1) & 3));
 
-  if(Text == d->type)
-    d->text = StringList(ByteVectorList::split(val, '\0'), String::UTF8);
+  if(Text == d->type) {
+    // Short, null-separated values otherwise allocate far more memory than
+    // the item occupies on disk.
+    static constexpr unsigned int MAX_TEXT_VALUE_COUNT = 50000;
+    const unsigned int maxValues = std::min(MAX_TEXT_VALUE_COUNT,
+      std::max(1024U, val.size() / 32));
+    const ByteVectorList values = ByteVectorList::split(
+      val, '\0', 1, static_cast<int>(maxValues + 1));
+    if(values.size() > maxValues) {
+      debug("APE::Item::parse() -- Maximum text value count exceeded");
+      return;
+    }
+    d->text = StringList(values, String::UTF8);
+  }
   else
     d->value = val;
 }

taglib/mpeg/id3v2/frames/textidentificationframe.cpp

@@ -31,6 +31,7 @@
 
 #include "tutils.h"
 #include "tpropertymap.h"
+#include "tdebug.h"
 #include "id3v1genres.h"
 #include "id3v2tag.h"
 
@@ -236,7 +237,18 @@ void TextIdentificationFrame::parseFields(const ByteVector &data)
   while(dataLength % byteAlign != 0)
     dataLength++;
 
-  const ByteVectorList l = ByteVectorList::split(data.mid(1, dataLength), textDelimiter(d->textEncoding), byteAlign);
+  // Short, null-separated values otherwise allocate far more memory than the
+  // frame occupies on disk.  Allow at least 1024 fields in small frames.
+  static constexpr unsigned int MAX_TEXT_FIELD_COUNT = 50000;
+  const unsigned int maxFields = std::min(MAX_TEXT_FIELD_COUNT,
+    std::max(1024U, static_cast<unsigned int>(dataLength) / 32));
+  const ByteVectorList l = ByteVectorList::split(
+    data.mid(1, dataLength), textDelimiter(d->textEncoding), byteAlign,
+    static_cast<int>(maxFields + 1));
+  if(l.size() > maxFields) {
+    debug("ID3v2: Maximum text field count exceeded");
+    return;
+  }
 
   d->fieldList.clear();