First tests with parsing of lhe file. Need to incorp EDM lhe reader

GeneratorInterface/LHEInterface/interface/LHEWeightGroupReaderHelper.h

@@ -0,0 +1,128 @@
+#ifndef GeneratorInterface_LHEInterface_LHEWeightGroupReaderHelper_h
+#define GeneratorInterface_LHEInterface_LHEWeightGroupReaderHelper_h
+
+#include <string>
+#include <vector>
+#include <map>
+#include <regex>
+
+#include "SimDataFormats/GeneratorProducts/interface/WeightGroupInfo.h"
+#include "SimDataFormats/GeneratorProducts/interface/PdfWeightGroupInfo.h"
+#include "SimDataFormats/GeneratorProducts/interface/ScaleWeightGroupInfo.h"
+
+
+class LHEWeightGroupReaderHelper {
+public:
+    LHEWeightGroupReaderHelper() : curGroup(gen::kUnknownWeights) {}
+
+    //// possibly add more versions of this functions for different inputs
+    void parseLHEFile(std::string filename);
+
+
+    gen::WeightGroupInfo* getScaleInfo() {return scaleInfo;}
+    edm::OwnVector<gen::WeightGroupInfo> getPdfVector() {return pdfVector;}
+
+private:
+    // Functions
+    std::regex createRegexSearch(std::vector<std::string>);
+    std::map<std::string, std::string> getTagsMap(std::string, std::regex);
+
+    // Variables
+    gen::WeightType curWeight;
+    gen::WeightGroupInfo* scaleInfo;
+    edm::OwnVector<gen::WeightGroupInfo> pdfVector;
+    std::regex weightStart(".*<weightgroup.+>.*");
+    std::regex weightEnd(".*</weightgroup>.*");
+    std::regex weightContent("<weight.*>\\s*(.+)</weight>");
+
+};
+
+void
+LHEWeightGroupReaderHelper::parseLHEFile(std::string filename) {
+    ifstream file;
+    file.open(filename);
+
+
+    //// may put in constructor, can have flag to specify these values
+    ////     To make this class a little more flexible
+    std::vector<std::string> weightGroup = {"name|type", "combine"};
+    std::vector<std::string> weightInfo = {"MUF", "id", "MUR", "PDF"};
+
+    std::regex groupTags = createRegexSearch(weightGroup);
+    std::regex infoTags = createRegexSearch(weightInfo);
+    /// end that comment
+
+
+    std::string line;
+    std::smatch m;
+    int index = 0;
+    while(getline(file, line)) {
+	if(std::regex_match(line, weightStart)) {
+	    std::string groupLine = line;
+	    std::string name = getTagsMap(line, groupTags)["name"];
+
+	    if(name == "Central scale variation")
+		    curWeight = gen::kScaleWeights;
+		else 
+		    curWeight = gen::kPdfWeights;
+
+	    /// file weights
+
+	    while(getline(file, line) && !std::regex_match(line, weightEnd)) {
+		auto tmp = getTagsMap(line, infoTags);
+		std::regex_search(line, m, weightContent);
+		std::string content = m[1].str();
+
+		gen::WeightGroupInfo* tmpWeight = nullptr;
+		if(curWeight == gen::kScaleWeights) 
+		    tmpWeight = new gen::ScaleWeightGroupInfo(groupLine);
+		else if(curWeight == gen::kPdfWeights) 
+		    tmpWeight = new gen::PdfWeightGroupInfo(groupLine);
+
+		tmpWeight->setWeightType(curWeight);
+		tmpWeight->addContainedId(index, tmp["id"], line);
+                index++;
+	    }
+	    curWeight = gen::kUnknownWeights;
+	}
+    }
+
+}
+
+
+std::map<std::string, std::string>
+LHEWeightGroupReaderHelper::getTagsMap(std::string s, std::regex r) {
+    std::smatch m;
+
+    std::map<std::string, std::string> retMap;
+    while(std::regex_search(s, m, r)) {
+	for(int i = 1; i < m.length() - 1; i += 2) {
+	    if(m[i] != "") {
+		retMap[m[i]] = m[i + 1];                
+	    }
+	}
+	s = m.suffix().str();
+    }
+
+    return retMap;
+}
+
+
+
+std::regex
+LHEWeightGroupReaderHelper::createRegexSearch(std::vector<std::string> names) {
+    std::string s = "(?:";
+    size_t numNames = names.size();
+    for(int i=0; i < numNames; ++i) {
+	s += "\\s*(" + names[i] + ")=\"([^\"]+)\"";
+	if(i != numNames - 1) {
+	    s += "|";
+	}
+    }
+    s += ")";
+
+    return std::regex(s);
+}
+
+
+#endif

GeneratorInterface/LHEInterface/plugins/ExternalLHEProducer.cc

@@ -58,6 +58,7 @@ Description: [one line class summary]
 #include "GeneratorInterface/LHEInterface/interface/LHEEvent.h"
 #include "GeneratorInterface/LHEInterface/interface/LHEReader.h"
 #include "GeneratorInterface/LHEInterface/interface/TestWeightInfo.h"
+#include "GeneratorInterface/LHEInterface/interface/LHEWeightGroupReaderHelper.h"
 
 #include "FWCore/ServiceRegistry/interface/Service.h"
 #include "FWCore/Utilities/interface/RandomNumberGenerator.h"
@@ -350,38 +351,50 @@ ExternalLHEProducer::beginRunProduce(edm::Run& run, edm::EventSetup const& es)
   unsigned int skip = 0;
   reader_ = std::make_unique<lhef::LHEReader>(infiles, skip);
 
+
+
+
   std::unique_ptr<LHEWeightInfoProduct> weightInfoProduct(new LHEWeightInfoProduct);
-  gen::WeightGroupInfo scaleInfo = getExampleScaleWeights();
+  gen::WeightGroupInfo scaleInfo;// = getExampleScaleWeights();
+  edm::OwnVector<gen::WeightGroupInfo> pdfSets;// = getExamplePdfWeights();
   //gen::WeightGroupInfo scaleInfo = getExampleScaleWeightsOutOfOrder();
-  edm::OwnVector<gen::WeightGroupInfo> pdfSets = getExamplePdfWeights();
+
+  // setup file reader
+  string LHEfilename ="cmsgrid_final.lhe";
+  LHEWeightGroupReaderHelper reader;
+  reader.parseLHEFile(LHEfilename);
+  scaleInfo = *reader.getScaleInfo();
+  pdfSet = reader.getPdfVector();
+
+
 
   weightInfoProduct->addWeightGroupInfo(scaleInfo);
   for (auto pdfSet : pdfSets)
-    weightInfoProduct->addWeightGroupInfo(pdfSet);
+      weightInfoProduct->addWeightGroupInfo(pdfSet);
   weightGroups_ = weightInfoProduct->allWeightGroupsInfo();
   run.put(std::move(weightInfoProduct));
 
   nextEvent();
   if (runInfoLast) {
-    runInfo = runInfoLast;
+      runInfo = runInfoLast;
 
-    std::unique_ptr<LHERunInfoProduct> product(new LHERunInfoProduct(*runInfo->getHEPRUP()));
-    std::for_each(runInfo->getHeaders().begin(),
-                  runInfo->getHeaders().end(),
-                  boost::bind(&LHERunInfoProduct::addHeader,
-                              product.get(), _1));
-    std::for_each(runInfo->getComments().begin(),
-                  runInfo->getComments().end(),
-                  boost::bind(&LHERunInfoProduct::addComment,
-                              product.get(), _1));
+      std::unique_ptr<LHERunInfoProduct> product(new LHERunInfoProduct(*runInfo->getHEPRUP()));
+      std::for_each(runInfo->getHeaders().begin(),
+		    runInfo->getHeaders().end(),
+		    boost::bind(&LHERunInfoProduct::addHeader,
+				product.get(), _1));
+      std::for_each(runInfo->getComments().begin(),
+		    runInfo->getComments().end(),
+		    boost::bind(&LHERunInfoProduct::addComment,
+				product.get(), _1));
 
-    // keep a copy around in case of merging
-    runInfoProducts.push_back(new LHERunInfoProduct(*product));
-    wasMerged = false;
+      // keep a copy around in case of merging
+      runInfoProducts.push_back(new LHERunInfoProduct(*product));
+      wasMerged = false;
 
-    run.put(std::move(product));
+      run.put(std::move(product));
 
-    runInfo.reset();
+      runInfo.reset();
   }
 }