Merge pull request #283 from JanWielemaker/develop-buffer2

mielvds · web-flow · commit 88110cc24d4d · 2025-03-20T13:46:25.000+01:00
Fix potential buffer overflow/truncation, null-ptr error
diff --git a/.gitignore b/.gitignore
@@ -57,7 +57,7 @@ stamp-h1
 .deps/
 
 *.hdt
-*.hdt.index
+*.hdt.index*
 *.nq
 *.nt
 *.rdf
diff --git a/libcds/include/RMQ_succinct.h b/libcds/include/RMQ_succinct.h
@@ -55,7 +55,7 @@ namespace cds_static
 			void  save(ostream & fp);
 			RMQ_succinct * load(istream & fp);
 
-			~RMQ_succinct();
+			virtual ~RMQ_succinct();
 
 		protected:
 			/* size of array a*/
diff --git a/libcds/include/TextIndexCSA.h b/libcds/include/TextIndexCSA.h
@@ -49,7 +49,7 @@ namespace cds_static
 			 **/
 			TextIndexCSA(uchar *text, ulong length, char *build_options);
 
-			~TextIndexCSA();
+			virtual ~TextIndexCSA();
 
 			/* Writes in numocc the number of occurrences of the substring
 			 * pattern[0..length-1] found in the text indexed by index. */
diff --git a/libcds/include/WaveletTreeNoptrsS.h b/libcds/include/WaveletTreeNoptrsS.h
@@ -89,7 +89,7 @@ namespace cds_static
             uint set(uint val, uint ind) const;
 
             /** Recursive function for building the Wavelet Tree. */
-            void build_level(uint **bm, uint *symbols, uint length, uint *occs);
+            void build_level(uint **bm, uint *symbols, uint length);
     };
 };
 #endif
diff --git a/libcds/src/static/bitsequence/BitSequenceRRR.cpp b/libcds/src/static/bitsequence/BitSequenceRRR.cpp
@@ -422,7 +422,7 @@ namespace cds_static
 			ret->create_sampling(ret->sample_rate);
 			return ret;
 		}
-		catch(exception e) {
+		catch(const exception&) {
 			delete ret;
 		}
 		return NULL;
diff --git a/libcds/src/static/sequence/WaveletMatrix.cpp b/libcds/src/static/sequence/WaveletMatrix.cpp
@@ -24,9 +24,11 @@
 namespace cds_static
 {
 
-	WaveletMatrix::WaveletMatrix(const Array &symbols2, BitSequenceBuilder * bmb, Mapper * am) : Sequence(n) {
+	WaveletMatrix::WaveletMatrix(const Array &symbols2, BitSequenceBuilder * bmb, Mapper * am) : Sequence(0) {
 		bmb->use();
 		n = symbols2.getLength();
+		this->length = n; // sets Sequence::length (Sequence(0) in initializer)
+
 		uint *symbols = new uint[n];
 		this->am = am;
 		am->use();
@@ -81,8 +83,6 @@ namespace cds_static
 		delete [] _bm;
 		// delete [] oc;
 		bmb->unuse();
-
-		this->length = n;
 	}
 
 	WaveletMatrix::WaveletMatrix(uint * symbols, size_t n, BitSequenceBuilder * bmb, Mapper * am, bool deleteSymbols) : Sequence(n) {
diff --git a/libcds/src/static/sequence/WaveletTreeNoptrsS.cpp b/libcds/src/static/sequence/WaveletTreeNoptrsS.cpp
@@ -24,9 +24,10 @@
 namespace cds_static
 {
 
-    WaveletTreeNoptrsS::WaveletTreeNoptrsS(const Array &symb, BitSequenceBuilder * bmb, Mapper * am) : Sequence(n) {
+    WaveletTreeNoptrsS::WaveletTreeNoptrsS(const Array &symb, BitSequenceBuilder * bmb, Mapper * am) : Sequence(0) {
         bmb->use();
         this->n=symb.getLength();
+        this->length = this->n; // sets Sequence::length (Sequence(0) in initializer)
         this->am=am;
         bool deleteSymbols = true;
         am->use();
@@ -80,7 +81,7 @@ namespace cds_static
                 _bm[i][j]=0;
         }
 
-        build_level(_bm,new_symb,new_n,occurrences);
+        build_level(_bm,new_symb,new_n);
         bitstring = new BitSequence*[height];
         for(uint i=0;i<height;i++) {
             bitstring[i] = bmb->build(_bm[i],new_n);
@@ -148,7 +149,7 @@ namespace cds_static
                 _bm[i][j]=0;
         }
 
-        build_level(_bm,new_symb,new_n,occurrences);
+        build_level(_bm,new_symb,new_n);
         bitstring = new BitSequence*[height];
         for(uint i=0;i<height;i++) {
             bitstring[i] = bmb->build(_bm[i],new_n);
@@ -328,7 +329,7 @@ namespace cds_static
         return bytesBitstrings+occ->getSize()+ptrs;
     }
 
-    void WaveletTreeNoptrsS::build_level(uint **bm, uint *symbols, uint length, uint *occs) {
+    void WaveletTreeNoptrsS::build_level(uint **bm, uint *symbols, uint length) {
             // for (uint i = 0; i < length; i++)
             //     cout << " " << symbols[i];
             // cout << endl;
diff --git a/libcds/tests/testLCP.cpp b/libcds/tests/testLCP.cpp
@@ -53,8 +53,8 @@ bool testLCP(LCP *s1, LCP *s2, TextIndex *csa){
 
 
 int main(int argc, char *argv[]){
-	char *text;
-	size_t length;
+	char *text = nullptr;
+	size_t length = 0;
 	LCP *lcp_naive=NULL;
 	LCP *lcp=NULL;
 
diff --git a/libcds/tests/testNPR.cpp b/libcds/tests/testNPR.cpp
@@ -71,8 +71,8 @@ bool testNPR(NPR *npr, LCP *lcp, TextIndex *csa,  size_t *naive_nsv, size_t *nai
 
 
 int main(int argc, char *argv[]){
-	char *text;
-	size_t length;
+	char *text = nullptr;
+	size_t length = 0;
 	LCP *lcp = NULL;
 	NPR *npr = NULL;
 
diff --git a/libhdt/src/libdcs/CSD_FMIndex.cpp b/libhdt/src/libdcs/CSD_FMIndex.cpp
@@ -54,7 +54,7 @@ CSD_FMIndex::CSD_FMIndex(hdt::IteratorUCharString *it, bool sparse_bitsequence,
 
   this->type = FMINDEX;
   string element;
-  unsigned char *text;
+  unsigned char *text; // TODO: std::vector<unsigned char>
   uint *bitmap = 0;
   // uint32_t *bitmap = 0;
 
@@ -90,7 +90,7 @@ CSD_FMIndex::CSD_FMIndex(hdt::IteratorUCharString *it, bool sparse_bitsequence,
     // Checking the current size of the encoded
     // sequence: realloc if necessary
     if ((total + currentLength + 1) > reservedSize) {
-      while (((size_t)total + currentLength + 1) > reservedSize) {
+      while ((total + currentLength + 1) > reservedSize) {
         reservedSize <<= 1;
         if (reservedSize == 0) {
           reservedSize = ((size_t)total + currentLength) * 2;
@@ -99,7 +99,7 @@ CSD_FMIndex::CSD_FMIndex(hdt::IteratorUCharString *it, bool sparse_bitsequence,
       text =
           (unsigned char *)realloc(text, reservedSize * sizeof(unsigned char));
     }
-    strncpy((char *)(text + total), (char *)currentStr, currentLength);
+    strncpy((char *)(text + total), (char *)currentStr, reservedSize - total);
 
     total += currentLength;
 
@@ -118,7 +118,7 @@ CSD_FMIndex::CSD_FMIndex(hdt::IteratorUCharString *it, bool sparse_bitsequence,
   textFinal = new char[total + 1];
   //	cout<<"testing:total cpy:"<<total<<endl;
   //	cout<<"testing:text:"<<text<<endl;
-  strncpy((char *)(textFinal), (char *)text, total);
+  strncpy((char *)(textFinal), (char *)text, total + 1);
   textFinal[total] = '\0'; // end of the text
   //	cout<<"testing:textFinal:"<<textFinal<<endl;
 
@@ -149,7 +149,7 @@ CSD_FMIndex::CSD_FMIndex(hdt::IteratorUCharString *it, bool sparse_bitsequence,
     separators = new BitSequenceRG(bitmap, len, 4);
     delete[] bitmap;
   }
-  delete[] text;
+  free(text);
 }
 
 void CSD_FMIndex::build_ssa(unsigned char *text, size_t len,
diff --git a/libhdt/src/libdcs/CSD_HTFC.cpp b/libhdt/src/libdcs/CSD_HTFC.cpp
@@ -25,6 +25,7 @@
  *   Miguel A. Martinez-Prieto:  migumar2@infor.uva.es
  */
 
+#include <string>
 #include "CSD_HTFC.h"
 
 #if HAVE_CDS
@@ -57,8 +58,9 @@ CSD_HTFC::CSD_HTFC(hdt::IteratorUCharString *it, uint32_t blocksize,
 
   vector<uint> xblocks; // Temporal storage for start positions
 
-  unsigned char *previousStr = NULL, *currentStr = NULL;
-  uint previousLength = 0, currentLength = 0;
+  std::basic_string<unsigned char> previousStr((const unsigned char*)"");
+  unsigned char *currentStr = NULL;
+  uint currentLength = 0;
 
   while (it->hasNext()) {
     currentStr = it->next();
@@ -91,16 +93,16 @@ CSD_HTFC::CSD_HTFC(hdt::IteratorUCharString *it, uint32_t blocksize,
 
       // The string is explicitly copied to the
       // encoded sequence.
-      strncpy((char *)(textfc + bytesfc), (char *)currentStr, currentLength);
+      strncpy((char *)(textfc + bytesfc), (char *)currentStr, reservedSize - bytesfc);
       bytesfc += currentLength;
 
       // cout << nblocks-1 << "," << length << " => " << currentStr << endl;
     } else {
       // Regular string
 
       // Calculating the length of the long common prefix
-      uint delta = longest_common_prefix(previousStr, currentStr,
-                                         previousLength, currentLength);
+      uint delta = longest_common_prefix(previousStr.data(), currentStr,
+                                         previousStr.length(), currentLength);
 
       // cout << "Block: " << nblocks << " Pos: "<< length << endl;
       // cout << previousStr << endl << currentStr << endl << " Delta: " <<
@@ -111,7 +113,7 @@ CSD_HTFC::CSD_HTFC(hdt::IteratorUCharString *it, uint32_t blocksize,
 
       // The suffix is copied to the sequence
       strncpy((char *)(textfc + bytesfc), (char *)currentStr + delta,
-              currentLength - delta);
+              reservedSize - bytesfc);
       bytesfc += currentLength - delta;
       // cout << nblocks-1 << "," << length << " => " << currentStr << endl;
     }
@@ -121,8 +123,7 @@ CSD_HTFC::CSD_HTFC(hdt::IteratorUCharString *it, uint32_t blocksize,
 
     // New string processed
     numstrings++;
-    memcpy(previousStr, currentStr, currentLength);
-    previousLength = currentLength;
+    previousStr.assign(currentStr, currentLength);
 
     it->freeStr(currentStr);
     // NOTIFYCOND(listener, "Converting dictionary to HTFC", length,
@@ -332,7 +333,7 @@ void CSD_HTFC::dumpBlock(uint block) {
   uint idInBlock = 0;
 
   // Reading the first string
-  strncpy((char *)string, (char *)(text + pos), slen);
+  strncpy((char *)string, (char *)(text + pos), maxlength + 1);
   string[slen] = '\0';
   pos += slen;
 
@@ -351,7 +352,7 @@ void CSD_HTFC::dumpBlock(uint block) {
 
     // Copying the suffix
     slen = strlen((char *)text + pos) + 1;
-    strncpy((char *)(string + delta), (char *)(text + pos), slen);
+    strncpy((char *)(string + delta), (char *)(text + pos), maxlength - delta + 1);
 
     cout << block * blocksize + idInBlock << " (" << idInBlock << ") => "
          << string << " Delta=" << delta << " Len=" << slen << endl;
diff --git a/libhdt/src/libdcs/CSD_PFC.cpp b/libhdt/src/libdcs/CSD_PFC.cpp
@@ -81,7 +81,7 @@ CSD_PFC::CSD_PFC(hdt::IteratorUCharString *it, uint32_t blocksize,
       nblocks++;
 
       // The string is explicitly copied to the encoded sequence.
-      strncpy((char *)(text + bytes), (char *)currentStr, currentLength);
+      strncpy((char *)(text + bytes), (char *)currentStr, reservedSize - bytes);
       bytes += currentLength;
     } else {
       // Regular string
@@ -96,7 +96,7 @@ CSD_PFC::CSD_PFC(hdt::IteratorUCharString *it, uint32_t blocksize,
 
       // The suffix is copied to the sequence
       strncpy((char *)(text + bytes), (char *)currentStr + delta,
-              currentLength - delta);
+              reservedSize - bytes);
       bytes += currentLength - delta;
     }
 
diff --git a/libhdt/src/rdf/RDFParser.cpp b/libhdt/src/rdf/RDFParser.cpp
@@ -24,6 +24,7 @@ RDFParserCallback *RDFParserCallback::getParserCallback(RDFNotation notation) {
 #else
 		throw ParseException("No Parser available for input RDF Format");
 #endif
+    return nullptr;
 }
 
 }
diff --git a/libhdt/src/rdf/RDFParserSerd.cpp b/libhdt/src/rdf/RDFParserSerd.cpp
@@ -185,7 +185,7 @@ void RDFParserSerd::doParse(const char *fileName, const char *baseUri, RDFNotati
 
 	serd_reader_set_error_sink(reader, hdtserd_on_error, NULL);
 
-	const uint8_t* input=serd_uri_to_path((const uint8_t *)fileName);
+	uint8_t* input=serd_file_uri_parse((const uint8_t *)fileName, NULL);
 
 	if(fileUtil::str_ends_with(fileName,".gz")){
 
@@ -216,6 +216,7 @@ void RDFParserSerd::doParse(const char *fileName, const char *baseUri, RDFNotati
 
 	serd_env_free(env);
 	serd_node_free(&base);
+        serd_free(input);
 }
 
 }
diff --git a/libhdt/src/triples/BitmapTriples.cpp b/libhdt/src/triples/BitmapTriples.cpp
@@ -204,7 +204,7 @@ struct sort_pred {
 };
 
 // function object
-class PredicateComparator : public std::binary_function<unsigned int,unsigned int,bool>
+class PredicateComparator
 {
 private:
 	IntSequence *array;
diff --git a/libhdt/src/triples/TripleListDisk.cpp b/libhdt/src/triples/TripleListDisk.cpp
@@ -303,7 +303,7 @@ void TripleListDisk::insert(TripleID &triple)
 
 	//cout << "Insert: " <<&pointer[numTotalTriples] << "* "<< triple << " "<<sizeof(TripleID) << endl;
 
-	memcpy(&arrayTriples[numTotalTriples], &triple, sizeof(TripleID));
+	arrayTriples[numTotalTriples] = triple;
 	numTotalTriples++;
 	numValidTriples++;
 	//cout << "Inserted: "<< numTotalTriples << endl;
diff --git a/libhdt/src/util/Histogram.h b/libhdt/src/util/Histogram.h
@@ -28,7 +28,7 @@ class Histogram {
 
 	Histogram() :
 			Start(0), nBins_by_interval(0), nBins(0),
-                    freq(new size_t[0]) {
+			freq(new size_t[1]) { // size_t[0] is undefined behavior
 			reset();
 		}
 
@@ -69,7 +69,7 @@ class Histogram {
 		delete[] freq;
 	}
 
-	/** operator=
+  /** operator=
 	 * Set this histogram equal to another.
 	 * @param other Description of the param.
 	 * @return The expected result.
@@ -81,9 +81,9 @@ class Histogram {
 			if (nBins != other.nBins) {
 				nBins = other.nBins;
 				delete[] freq;
-                freq = new size_t[nBins];
+				freq = new size_t[nBins];
 			}
-            for (size_t i(0); i < nBins; ++i)
+			for (size_t i(0); i < nBins; ++i)
 				freq[i] = other.freq[i];
 		}
 		return *this;
diff --git a/libhdt/src/util/StopWatch.cpp b/libhdt/src/util/StopWatch.cpp
@@ -31,6 +31,7 @@
 #include <sstream>
 
 #include <string.h> // For memcpy()
+#include <cstdint>  // for uint64_t etc
 
 #include "StopWatch.hpp"
 
diff --git a/libhdt/tools/hdtSearch.cpp b/libhdt/tools/hdtSearch.cpp
@@ -98,7 +98,7 @@ void iterate(HDT *hdt, char *query, ostream &out, bool measure, uint32_t offset)
                 it->skip(offset);
                 offset = 0;
             }
-            catch (const runtime_error error) {
+            catch (const runtime_error&) {
                 /*invalid offset*/
                 interruptSignal = 1;
             }

Original file line number	Diff line number	Diff line change
`@@ -422,7 +422,7 @@ namespace cds_static`
`422`	`422`	`ret->create_sampling(ret->sample_rate);`
`423`	`423`	`return ret;`
`424`	`424`	`}`
`425`		`- catch(exception e) {`
	`425`	`+ catch(const exception&) {`
`426`	`426`	`delete ret;`
`427`	`427`	`}`
`428`	`428`	`return NULL;`