removing file reads and vocab string

IITH-Compilers · Apr 3, 2024 · c508c6f · c508c6f
1 parent 9d997fb
commit c508c6f
Show file tree

Hide file tree

Showing 7 changed files with 25 additions and 55 deletions.
diff --git a/Manylinux2014_Compliant_Source/pkg/regen-oracle.sh b/Manylinux2014_Compliant_Source/pkg/regen-oracle.sh
@@ -28,13 +28,12 @@ mkdir -p ${DEST_FOLDER_SYM_P}
 mkdir -p ${DEST_FOLDER_FA_P}
 
 IR2VEC_PATH=../../build/bin/ir2vec
-VOCAB_PATH="../../vocabulary/seedEmbeddingVocab.txt"
 
 while IFS= read -r d; do
 	echo "Generating embeddings for ${d}"
-	${IR2VEC_PATH} -sym -vocab=${VOCAB_PATH} -o ${DEST_FOLDER_SYM}/ir2vec.txt -level f ${d} &>/dev/null
-	${IR2VEC_PATH} -fa -vocab=${VOCAB_PATH} -o ${DEST_FOLDER_FA}/ir2vec.txt -level f ${d} &>/dev/null
-	${IR2VEC_PATH} -sym -vocab=${VOCAB_PATH} -o ${DEST_FOLDER_SYM_P}/ir2vec.txt -level p ${d} >/dev/null
-	${IR2VEC_PATH} -fa -vocab=${VOCAB_PATH} -o ${DEST_FOLDER_FA_P}/ir2vec.txt -level p ${d} >/dev/null
+	${IR2VEC_PATH} -sym -o ${DEST_FOLDER_SYM}/ir2vec.txt -level f ${d} &>/dev/null
+	${IR2VEC_PATH} -fa -o ${DEST_FOLDER_FA}/ir2vec.txt -level f ${d} &>/dev/null
+	${IR2VEC_PATH} -sym -o ${DEST_FOLDER_SYM_P}/ir2vec.txt -level p ${d} >/dev/null
+	${IR2VEC_PATH} -fa -o ${DEST_FOLDER_FA_P}/ir2vec.txt -level p ${d} >/dev/null
 done <index-${SEED_VERSION}.files
 wait
diff --git a/src/IR2Vec.cpp b/src/IR2Vec.cpp
@@ -32,11 +32,6 @@ cl::opt<bool> cl_collectIR(
     "collectIR", cl::Optional,
     cl::desc("Generate triplets for training seed embedding vocabulary"),
     cl::init(false), cl::cat(category));
-
-cl::opt<std::string> cl_vocab("vocab", cl::Optional, cl::init(""),
-                              cl::desc("Use embeddings from file path"),
-                              cl::cat(category));
-
 cl::opt<std::string> cl_iname(cl::Positional, cl::desc("Input file path"),
                               cl::Required, cl::cat(category));
 
@@ -81,7 +76,6 @@ int main(int argc, char **argv) {
   fa = cl_fa;
   sym = cl_sym;
   collectIR = cl_collectIR;
-  vocab = cl_vocab;
   iname = cl_iname;
   oname = cl_oname;
   // newly added
@@ -105,18 +99,12 @@ int main(int argc, char **argv) {
       errs() << "Invalid level specified: Use either p or f\n";
       failed = true;
     }
-    if (vocab.empty()) {
-      errs() << "Should specify vocab pointing to the path of vocabulary\n";
-      failed = true;
-    }
   } else {
     if (!collectIR) {
       errs() << "Either of sym, fa or collectIR should be specified\n";
       failed = true;
     } else if (level)
       errs() << "[WARNING] level would not be used in collectIR mode\n";
-    else if (!vocab.empty())
-      errs() << "[WARNING] vocab would not be used in collectIR mode\n";
   }
 
   if (failed)

diff --git a/src/include/IR2Vec.h b/src/include/IR2Vec.h
@@ -19,30 +19,29 @@ using Vector = llvm::SmallVector<double, DIM>;
 enum IR2VecMode { FlowAware, Symbolic };
 
 class Embeddings {
-  int generateEncodings(llvm::Module &M, IR2VecMode mode, std::string vocab,
-                        char level = '\0', std::string funcName = "",
-                        std::ostream *o = nullptr, int cls = -1, float WO = 1,
-                        float WA = 0.2, float WT = 0.5);
+  int generateEncodings(llvm::Module &M, IR2VecMode mode, char level = '\0',
+                        std::string funcName = "", std::ostream *o = nullptr,
+                        int cls = -1, float WO = 1, float WA = 0.2,
+                        float WT = 0.5);
 
   llvm::SmallMapVector<const llvm::Instruction *, Vector, 128> instVecMap;
   llvm::SmallMapVector<const llvm::Function *, Vector, 16> funcVecMap;
   Vector pgmVector;
 
 public:
   Embeddings() = default;
-  Embeddings(llvm::Module &M, IR2VecMode mode, std::string vocab,
-             std::string funcName = "", float WO = 1, float WA = 0.2,
-             float WT = 0.5) {
-    generateEncodings(M, mode, vocab, '\0', funcName, nullptr, -1, WO, WA, WT);
+  Embeddings(llvm::Module &M, IR2VecMode mode, std::string funcName = "",
+             float WO = 1, float WA = 0.2, float WT = 0.5) {
+    generateEncodings(M, mode, '\0', funcName, nullptr, -1, WO, WA, WT);
   }
 
   // Use this constructor if the representations ought to be written to a
   // file. Analogous to the command line options that are being used in IR2Vec
   // binary.
-  Embeddings(llvm::Module &M, IR2VecMode mode, std::string vocab, char level,
-             std::ostream *o, std::string funcName = "", float WO = 1,
-             float WA = 0.2, float WT = 0.5) {
-    generateEncodings(M, mode, vocab, level, funcName, o, -1, WO, WA, WT);
+  Embeddings(llvm::Module &M, IR2VecMode mode, char level, std::ostream *o,
+             std::string funcName = "", float WO = 1, float WA = 0.2,
+             float WT = 0.5) {
+    generateEncodings(M, mode, level, funcName, o, -1, WO, WA, WT);
   }
 
   // Returns a map containing instructions and the corresponding vector

diff --git a/src/libIR2Vec.cpp b/src/libIR2Vec.cpp
@@ -14,12 +14,11 @@
 #include "llvm/Support/CommandLine.h"
 
 int IR2Vec::Embeddings::generateEncodings(llvm::Module &M,
-                                          IR2Vec::IR2VecMode mode,
-                                          std::string vocab, char level,
+                                          IR2Vec::IR2VecMode mode, char level,
                                           std::string funcName, std::ostream *o,
                                           int cls, float WO, float WA,
                                           float WT) {
-  IR2Vec::vocab = vocab;
+
   IR2Vec::level = level;
   IR2Vec::cls = cls;
   IR2Vec::WO = WO;

diff --git a/src/test-suite/generateOracle.sh b/src/test-suite/generateOracle.sh
@@ -28,15 +28,6 @@ LLVM_BUILD="/usr"
 #Update IR2Vec Path to use
 IR2VEC_PATH="../../build/bin/ir2vec"
 
-#Update Vocabulary Path to use
-VOCAB_PATH="../../vocabulary/seedEmbeddingVocab.txt"
-
-# does the VOCAB_PATH exist?
-if [ ! -f ${VOCAB_PATH} ]; then
-	echo "Vocabulary file does not exist.."
-	exit
-fi
-
 if [ -z ${LLVM_BUILD} ]; then
 	echo "Enter the llvm build path.."
 	exit
@@ -45,10 +36,6 @@ if [ -z ${IR2VEC_PATH} ]; then
 	echo "Enter the ir2vec path.."
 	exit
 fi
-if [ -z ${VOCAB_PATH} ]; then
-	echo "Enter the vocabulary path.."
-	exit
-fi
 
 functions=("main" "buildMatchingMachine" "search" "BellamFord" "BFS" "isBCUtil" "insertionSort" "binomialCoeff" "find" "countParenth" "boruvkaMST" "maxStackHeight" "badCharHeuristic" "bpm"
 	"count" "getMaxUtil" "buildSuffixArray" "countOnes" "countStrings" "countRec" "countWays" "AP" "cutRod" "isCyclic" "isDivisible" "DFS" "editDist" "eggDrop" "isSC" "isConnected" "printClosest"
@@ -61,24 +48,24 @@ functions=("main" "buildMatchingMachine" "search" "BellamFord" "BFS" "isBCUtil"
 cat index-${SEED_VERSION}.files | wc -l
 echo "generating P level files"
 while IFS= read -r d; do
-	${IR2VEC_PATH} -sym -vocab=${VOCAB_PATH} -level p -o ${DEST_FOLDER_SYM_P}/ir2vec.txt ${d} &>/dev/null
-	${IR2VEC_PATH} -fa -vocab=${VOCAB_PATH} -level p -o ${DEST_FOLDER_FA_P}/ir2vec.txt ${d} &>/dev/null
+	${IR2VEC_PATH} -sym -level p -o ${DEST_FOLDER_SYM_P}/ir2vec.txt ${d} &>/dev/null
+	${IR2VEC_PATH} -fa -level p -o ${DEST_FOLDER_FA_P}/ir2vec.txt ${d} &>/dev/null
 done <index-${SEED_VERSION}.files
 wait
 
 echo "generating F level files"
 while IFS= read -r d; do
-	${IR2VEC_PATH} -sym -vocab=${VOCAB_PATH} -level f -o ${DEST_FOLDER_SYM}/ir2vec.txt ${d} &>/dev/null
-	${IR2VEC_PATH} -fa -vocab=${VOCAB_PATH} -level f -o ${DEST_FOLDER_FA}/ir2vec.txt ${d} &>/dev/null
+	${IR2VEC_PATH} -sym -level f -o ${DEST_FOLDER_SYM}/ir2vec.txt ${d} &>/dev/null
+	${IR2VEC_PATH} -fa -level f -o ${DEST_FOLDER_FA}/ir2vec.txt ${d} &>/dev/null
 done <index-${SEED_VERSION}.files
 wait
 
 echo "generating onDemand level files"
 while IFS= read -r d; do
 	echo $d
 	for func in "${functions[@]}"; do
-		${IR2VEC_PATH} -sym -vocab=${VOCAB_PATH} -level f -funcName=$func -o ${DEST_FOLDER_SYM_ONDEMAND}/ir2vec.txt ${d} &>/dev/null
-		${IR2VEC_PATH} -fa -vocab=${VOCAB_PATH} -level f -funcName=$func -o ${DEST_FOLDER_FA_ONDEMAND}/ir2vec.txt ${d} &>/dev/null
+		${IR2VEC_PATH} -sym -level f -funcName=$func -o ${DEST_FOLDER_SYM_ONDEMAND}/ir2vec.txt ${d} &>/dev/null
+		${IR2VEC_PATH} -fa -level f -funcName=$func -o ${DEST_FOLDER_FA_ONDEMAND}/ir2vec.txt ${d} &>/dev/null
 	done
 done <index-${SEED_VERSION}.files
 wait
diff --git a/src/test-suite/sanity_check.sh.cmake b/src/test-suite/sanity_check.sh.cmake
@@ -24,7 +24,6 @@ else
 fi
 
 SEED_VERSION=$2
-VOCAB_PATH="./vocabulary/seedEmbeddingVocab.txt"
 IR2VEC_PATH="../../bin/ir2vec"
 
 functions=("main" "buildMatchingMachine" "search" "BellamFord" "BFS" "isBCUtil" "insertionSort" "binomialCoeff" "find" "countParenth" "boruvkaMST" "maxStackHeight" "badCharHeuristic" "bpm"
@@ -47,15 +46,15 @@ perform_vector_comparison() {
 
     if [[ "$FILE_PREFIX" == "p" || "$FILE_PREFIX" == "f" ]]; then
         while IFS= read -r d; do
-            ${IR2VEC_PATH} -${PASS} -vocab=${VOCAB_PATH} -level ${LEVEL} -o ${VIR_FILE} ${d} &> /dev/null
+            ${IR2VEC_PATH} -${PASS} -level ${LEVEL} -o ${VIR_FILE} ${d} &> /dev/null
         done < index-${SEED_VERSION}.files
         wait
     else
         while IFS= read -r d_on
         do
             for func in "${functions[@]}"
             do
-                ${IR2VEC_PATH} -${PASS} -vocab=${VOCAB_PATH} -level ${LEVEL} -funcName=$func -o ${VIR_FILE} ${d_on} &> /dev/null
+                ${IR2VEC_PATH} -${PASS} -level ${LEVEL} -funcName=$func -o ${VIR_FILE} ${d_on} &> /dev/null
             done
         done < index-${SEED_VERSION}.files
         wait

diff --git a/src/utils.cpp b/src/utils.cpp
@@ -17,7 +17,6 @@ bool IR2Vec::fa;
 bool IR2Vec::sym;
 bool IR2Vec::printTime;
 bool IR2Vec::collectIR;
-std::string IR2Vec::vocab;
 std::string IR2Vec::iname;
 std::string IR2Vec::oname;
 std::string IR2Vec::funcName;