Fixed buffer overflow with inaccurate exon length

closes #26
getzlab · Apr 17, 2019 · fe0ded9 · fe0ded9
1 parent 58d57ef
commit fe0ded9
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 1 deletion.
diff --git a/src/GTF.cpp b/src/GTF.cpp
@@ -9,6 +9,7 @@
 #include "GTF.h"
 #include <exception>
 #include <stdexcept>
+#include <unordered_set>
 #include <boost/regex.hpp>
 
 using std::ifstream;
@@ -27,6 +28,7 @@ namespace rnaseqc {
 
     ifstream& operator>>(ifstream &in, Feature &out)
     {
+        static std::unordered_set<std::string> geneIds, exonIds;
         try{
             string line;
             while(getline(in, line))
@@ -76,6 +78,8 @@ namespace rnaseqc {
                     std::cout << "Bad fead feature range:" << out.start << " - " << out.end << std::endl;
                 if (out.type == FeatureType::Gene && attributes.find("gene_id") != attributes.end())
                 {
+                    if (geneIds.count(out.feature_id)) throw gtfException(std::string("Detected non-unique Gene ID: "+out.feature_id));
+                    geneIds.insert(out.feature_id);
                     //Parse gene attributes
                     out.feature_id = attributes["gene_id"];
                     geneLengths[out.feature_id] = out.end - out.start + 1;
@@ -85,6 +89,8 @@ namespace rnaseqc {
                 if (attributes.find("gene_id") != attributes.end()) out.gene_id = attributes["gene_id"];
                 if (out.type == FeatureType::Exon)
                 {
+                    if (exonIds.count(out.feature_id)) throw gtfException(std::string("Detected non-unique Exon ID: "+out.feature_id));
+                    exonIds.insert(out.feature_id);
                     //Parse exon attributes
                     if (attributes.find("exon_id") != attributes.end())
                     {
@@ -107,6 +113,10 @@ namespace rnaseqc {
             }
 
         }
+        catch(gtfException &e)
+        {
+            throw e;
+        }
         catch(std::invalid_argument &e)
         {
             throw gtfException(std::string("GTF is in an invalid format: ") + e.what());

diff --git a/src/Metrics.cpp b/src/Metrics.cpp
@@ -251,7 +251,9 @@ namespace rnaseqc {
 
     void add_range(std::vector<unsigned long> &coverage, coord offset, unsigned int length)
     {
-        for (coord i = offset; i < offset + length; ++i) coverage[i] += 1ul;
+        const size_t size = coverage.size();
+        for (coord i = offset; i < offset + length && i < size; ++i) coverage[i] += 1ul;
+        if (offset + length > size) std::cerr << "Error: Attempted to write more coverage than present on exon. Coverage-based metrics may be inaccurate. This may be a sign of an invalid bam or gtf entry" << std::endl;
     }
 
     //Compute exon coverage metrics, then stich exons together and compute gene coverage metrics