Bugfix remove ccs Parameter from title after mgf file was read

pwiz/data/msdata/Diff.cpp

@@ -547,25 +547,6 @@ void diff(const Spectrum& a,
         }
     }
 
-    if (config.ignoreSpectrumTitle)
-    {
-        vector<CVParam>::const_iterator it =
-            find_if(a_b.cvParams.begin(), a_b.cvParams.end(), CVParamIs(MS_spectrum_title));
-
-        if (it != a_b.cvParams.end())
-        {
-            a_b.cvParams.erase(it);
-        }
-
-        it =
-            find_if(b_a.cvParams.begin(), b_a.cvParams.end(), CVParamIs(MS_spectrum_title));
-
-        if (it != b_a.cvParams.end())
-        {
-            b_a.cvParams.erase(it);
-        }
-    }
-
     // provide context
     if (!a_b.empty() || !b_a.empty()) 
     {

pwiz/data/msdata/Diff.hpp

@@ -228,17 +228,14 @@ struct PWIZ_API_DECL DiffConfig : public pwiz::data::BaseDiffConfig
 
     bool ignoreDataProcessing;
 
-    bool ignoreSpectrumTitle;
-
     DiffConfig()
     :   pwiz::data::BaseDiffConfig(),
         ignoreIdentity(false),
         ignoreMetadata(false),
         ignoreExtraBinaryDataArrays(false),
         ignoreSpectra(false),
         ignoreChromatograms(false),
-        ignoreDataProcessing(false),
-        ignoreSpectrumTitle(false)
+        ignoreDataProcessing(false)
     {}
 };
 

pwiz/data/msdata/Serializer_MGF.cpp

@@ -122,7 +122,7 @@ void Serializer_MGF::Impl::write(ostream& os, const MSData& msd,
             CVParam collisionalCrossSectionalArea = scan.cvParam(MS_collisional_cross_sectional_area);
             if (!collisionalCrossSectionalArea.empty())
             {
-                os << "," << "ccs=" << collisionalCrossSectionalArea.valueFixedNotation() << '\n';
+                os << ", " << "ccs=" << collisionalCrossSectionalArea.valueFixedNotation() << '\n';
             }
             else
             {

pwiz/data/msdata/Serializer_MGF_Test.cpp

@@ -156,8 +156,7 @@ void testWriteRead(const MSData& msd)
     DiffConfig diffConfig;
     diffConfig.ignoreIdentity = true;
     diffConfig.ignoreChromatograms = true;
-    diffConfig.ignoreSpectrumTitle = true;
-
+
     Diff<MSData, DiffConfig> diff(msd, msd2, diffConfig);
     if (os_ && diff) *os_ << diff << endl; 
     unit_assert(!diff);
@@ -206,5 +205,4 @@ int main(int argc, char* argv[])
     }
 
     TEST_EPILOG
-}
-
+}

pwiz/data/msdata/SpectrumList_MGF.cpp

@@ -40,6 +40,9 @@ using namespace pwiz::util;
 
 namespace {
 
+const char* startTagCCSInTitle = "ccs=";
+const char* endTagsCCSInTitle[] = { " ", ",",";","\t" };
+
 class SpectrumList_MGFImpl : public SpectrumList_MGF
 {
     public:
@@ -218,6 +221,7 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
                                 // Some formats omit RTINSECONDS and store the retention time
                                 // in the title field instead.
                                 double scanTimeMin = getRetentionTimeFromTitle(value);
+
                                 if (scanTimeMin > 0)
                                     scan.set(MS_scan_start_time, scanTimeMin * 60, UO_second);
 
@@ -228,7 +232,7 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
                                     scan.cvParams.push_back(CVParam(MS_collisional_cross_sectional_area, ccs, UO_square_angstrom));
                                 }
 
-                                spectrum.set(MS_spectrum_title, value);
+                                spectrum.set(MS_spectrum_title, removeCCSFromTitle(value));
                             }
                             else if (name == "PEPMASS")
                             {
@@ -344,49 +348,59 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
         spectrum.set(MS_base_peak_intensity, basePeakIntensity);
     }
 
-    /**
-     * Parse the spectrum title to look for CCS.
-     */
-    static double getCCSFromTitle(const string& title)
+    static size_t FindEndTag(const string& title, size_t start)
     {
-        // text to search for preceeding and following ccs
-        return getCCS(title, "ccs=");
+        for (int i = 0; i < sizeof(endTagsCCSInTitle) / sizeof(endTagsCCSInTitle[0]); i++)
+        {
+            const size_t tmpEnd = title.find(endTagsCCSInTitle[i], start);
+            if (tmpEnd != string::npos)
+            {
+                return tmpEnd;
+            }
+        }
+
+        return title.length();
     }
 
     /**
-     * Helper function to parse a double from the given string
-     * found between the two tags.  Search for number after position
-     * Update position to the end of the parsed double.
-     */
-    static double getCCS(const string& title, const char* startTag)
+    * Parse the spectrum title to look for CCS.
+    */
+    static double getCCSFromTitle(const string& title)
     {
-        size_t start = title.find(startTag, 0);
+	    size_t start = title.find(startTagCCSInTitle, 0);
+
         if (start == string::npos)
             return -1; // not found
 
-        start += strlen(startTag);
-        const string endTags[] = { " ", ",",";","\t" };
+        start += strlen(startTagCCSInTitle);
 
-        size_t end = title.length();
-        for (int i = 0; i < sizeof(endTags) / sizeof(endTags[0]); i++)
+        try
         {
-            const size_t tmpEnd = title.find(endTags[i], start);
-            if (tmpEnd != string::npos)
-            {
-                end = tmpEnd;
-                break;
-            }
+            return boost::lexical_cast<double>(title.substr(start, FindEndTag(title, start + strlen(startTagCCSInTitle)) - start));
+        }
+        catch (...)
+        {
+            return -1;
         }
+    }
+
+    /**
+     * Parse the spectrum title to look for CCS.
+    */
+    static string removeCCSFromTitle(const string& title)
+    {
+        size_t start = title.find(startTagCCSInTitle, 0);
 
-        const string ccsStr = title.substr(start, end - start);
+        if (start == string::npos)
+            return title; // not found
 
         try
         {
-            return boost::lexical_cast<double>(ccsStr);
+            return title.substr(0, start - 2) + title.substr(FindEndTag(title, start));
         }
         catch (...)
         {
-            return 0;
+            return title;
         }
     }
 
@@ -406,7 +420,6 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
         double secondTime = 0;
         for(int format_idx = 0; format_idx < 2; format_idx++)
         {
-
             size_t position = 0;
             firstTime = getTime(title, startTags[format_idx], 
                                 endTags[format_idx], position);
@@ -463,7 +476,6 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
         size_t lineCount = 0;
         bool inBeginIons = false;
         vector<SpectrumIdentity>::iterator curIdentityItr;
-        map<string, size_t>::iterator curIdToIndexItr;
 
         while (std::getline(*is_, lineStr)) // need accurate line length, so do not use pwiz::util convenience wrapper
         {
@@ -481,7 +493,7 @@ class SpectrumList_MGFImpl : public SpectrumList_MGF
                 curIdentityItr->index = index_.size()-1;
                 curIdentityItr->id = "index=" + lexical_cast<string>(index_.size()-1);
                 curIdentityItr->sourceFilePosition = size_t(is_->tellg())-lineStr.length()-1;
-                curIdToIndexItr = idToIndex_.insert(pair<string, size_t>(curIdentityItr->id, index_.size()-1)).first;
+                idToIndex_.insert(pair<string, size_t>(curIdentityItr->id, index_.size() - 1)).first;
                 inBeginIons = true;
             }
             else if (lineStr.find("TITLE=") == 0)