- fixed BiblioSpec DiaNN parser to support mass-only mods (#3173)

* fixed BlibException to avoid potential buffer overruns by switching to boost::format
- verified by Fengchao Yu

pwiz_tools/BiblioSpec/src/BlibException.h

@@ -27,34 +27,48 @@
  * decide what to add to the string.
  */
 
-#include <cstdarg>
-#include <cstdio>
 #include <exception>
 #include <string>
+#include <boost/format.hpp>
 
 namespace BiblioSpec{
 
     class BlibException : public std::exception{
     protected:
         std::string msgStr_;
         bool hasFilename_;
-
+
+        static std::string formatStringForward(boost::format& message)
+        {
+            return message.str();
+        }
+
+        template <typename TValue, typename... TArgs>
+        std::string formatStringForward(boost::format& message, TValue&& arg, TArgs&&... args)
+        {
+            message % std::forward<TValue>(arg);
+            return formatStringForward(message, std::forward<TArgs>(args)...);
+        }
+
     public:
         BlibException() 
             : hasFilename_(false)
         {
             msgStr_ = "BiblioSpec exception thrown.";
         }
-
-        BlibException(bool filename, const char* format, ...) 
+
+        template <typename... TArgs>
+        BlibException(bool filename, const char* format, TArgs&&... args)
             : hasFilename_(filename)
         {
-            va_list args;
-            va_start(args, format);
-            char buffer[4096];
-
-            vsprintf(buffer, format, args);
-            msgStr_ = buffer;
+            boost::format message(format);
+            msgStr_ = formatStringForward(message, std::forward<TArgs>(args)...);
+        }
+
+        BlibException(bool filename, const std::string& message)
+            : hasFilename_(filename)
+        {
+            msgStr_ = message;
         }
 
         ~BlibException()throw(){}
@@ -63,17 +77,15 @@ namespace BiblioSpec{
             hasFilename_ = hasIt;
         }
 
-       virtual bool hasFilename(){
+        virtual bool hasFilename(){
             return hasFilename_;
         }
 
-        virtual void addMessage(const char* format, ...){
-            va_list args;
-            va_start(args, format);
-            char buffer[4096];
-
-            vsprintf(buffer, format, args);
-            msgStr_ += buffer;
+        template <typename... TArgs>
+        void addMessage(const char* format, TArgs&&... args)
+        {
+            boost::format message(format);
+            msgStr_ += formatStringForward(message, std::forward<TArgs>(args)...);
         }
 
         virtual const char* what() const throw()

pwiz_tools/BiblioSpec/src/BuildParser.cpp

@@ -176,7 +176,7 @@ void BuildParser::setSpecFileName(
     if( curSpecFileName_.empty() ) {
         string extString = fileNotFoundMessage(specfileroot,
                                                extensions, localDirectories);
-        throw BlibException(true, extString.c_str());
+        throw BlibException(true, extString);
     }// else we found a file and set the name
 
     Verbosity::comment(V_DETAIL, "spectrum filename set to %s", curSpecFileName_.c_str());

pwiz_tools/BiblioSpec/src/DiaNNSpecLibReader.cpp

@@ -20,6 +20,7 @@
 #include "pwiz/data/common/Unimod.hpp"
 #include "libraries/csv.h"
 #include <boost/type_index.hpp>
+#include <pwiz/data/proteome/AminoAcid.hpp>
 
 namespace BiblioSpec
 {
@@ -184,14 +185,27 @@ inline std::string get_aas(const std::string &name, vector<SeqMod>& mods)
         if (symbol < 'A' || symbol > 'Z') {
             if (symbol != '(' && symbol != '[') continue;
             end = closing_bracket(name, symbol, i);
+            int position = max(1, static_cast<int>(j));
 
             if (name.find("(UniMod:", i, modPrefixLength) != i)
-                throw std::runtime_error("unable to handle mod in library entry: " + name);
+            {
+                string potentialMass = name.substr(i + 1, end - i - 1);
+                if (potentialMass.find_first_not_of("01234567890.") != string::npos)
+                    throw std::runtime_error("unable to handle mod in library entry as either a UniMod id or delta mass: " + potentialMass + " in " + name);
+                double modMass = lexical_cast<double>(potentialMass);
+                if (i > 0)
+                    modMass -= pwiz::proteome::AminoAcid::Info::record(name[i - 1]).residueFormula.monoisotopicMass();
+
+                if (!mods.empty() && mods.back().position == position)
+                    mods.back().deltaMass += modMass;
+                else
+                    mods.emplace_back(position, modMass);
+                continue;
+            }
 
             i += modPrefixLength;
-            mod = name.substr(i, end-i);
+            mod = name.substr(i, end - i);
             CVID unimodCvid = (CVID) (UNIMOD_unimod_root_node + lexical_cast<int>(mod));
-            int position = max(1, static_cast<int>(j));
             if (!mods.empty() && mods.back().position == position)
                 mods.back().deltaMass += unimod::modification(unimodCvid).deltaMonoisotopicMass();
             else 

pwiz_tools/BiblioSpec/src/saxhandler.cpp

@@ -136,7 +136,7 @@ bool SAXHandler::parse()
     if (!success) {
         string error = generateError(message.empty() ? getParserError() : message);
         Verbosity::debug(error.c_str());
-        throw BlibException(true, error.c_str());
+        throw BlibException(true, error);
     }
 
     return true;

pwiz_tools/BiblioSpec/tests/Jamfile.jam

@@ -332,6 +332,7 @@ blib-test-build metamorpheus-mzid : --unicode -o : output/metamorpheus.blib : me
 blib-test-build diann-speclib : -o : output/diann-speclib.blib : diann-speclib.check : $(TEST_INPUTS_PATH)/diann-swath.speclib ;
 blib-test-build diann-speclib-diapasef : -o : output/diann-hela-diapasef.blib : diann-hela-diapasef.check : $(TEST_INPUTS_PATH)/diann-hela-diapasef-lib.speclib ;
 blib-test-build-basic diann-mod-test : -o : $(TEST_INPUTS_PATH)/diann-mod-test.tsv.speclib ;
+blib-test-build-basic diann-mass-mod-test : -o : $(TEST_INPUTS_PATH)/diann-mass-mods.tsv.speclib ;
 blib-test-build-basic msfragger-diann : -o : $(TEST_INPUTS_PATH)/library.tsv.speclib ;
 blib-test-build-basic msfragger-diann-predicted : -o : $(TEST_INPUTS_PATH)/diann-predicted/lib.predicted.speclib ;