# Date: April 25, 2013
# Author:Shivani Rao (raoshivani@gmail.com)
# Description: This extracts original source files for the files that are parsed an cleaned in moreBugs. 
# Requirements: In order to run this code you need to have R installed on your computer and need to have the
# moreBugs and the original git repsository cloned. You can use the steps given below to do this:
# # Step 1: Clone the git repository of AspectJ
# <srcDir>$git clone https://github.com/eclipse/org.aspectj.git
# 
# Step 2: Download moreBugs from https://engineering.purdue.edu/RVL/Database/moreBugs/ 
# into a folder called moreBugs
# moreBugs/AspectJ$ls
# 
# bugs  metaInfo  revisions  tags
# 
# Step3:  Use this tool to extract original source files
# Example using AspectJ revision number 4154
# 
# Rscript extract_original_sourceFiles.r <dir_containig_dataset>/moreBugs/AspectJ <srcDir> <destDir> revision 4154

CopyFile <- function(srcDir,destDir,srcFile){
  srcFile = gsub("^\\s+","",srcFile)
  for (i in (1:length(srcFile))){
    dirtocreate = dirname(paste(destDir,srcFile[i],sep=""))
    if (!file.exists(dirtocreate))
      dir.create(dirtocreate,recursive=T)
    file.copy(paste(srcDir,srcFile[i],sep=""),paste(destDir,srcFile[i],sep=""))
  } 
}

ExtractOriginalSourceFiles <- function(basedir,srcDir,destDir,type,value) {
  if (!file.exists(basedir)){
    print("looks like you have not downloaded moreBugs... exiting")
    # download moreBugs from the link provided and provide the directory of the software you want to work with as basedir
    # For example if you want to extract original source files of AspectJ then basedir should be "<pathwheremorebugswasdownloaded>/AspectJ/"
    return
  }
  gitDir = paste(srcDir,".git",sep="")
  
  if (!file.exists(gitDir)) {
    print("sorry you need clone your git first. This source directory you provided does not have a .git directory.. exiting")
    return
  }

  if (type =="revision"){ # this means you just want to copy the source files that have changed in this revision
    revID <- value
  } else { 
    if (type =="bug"){ # you want all the source files that were present in the prefix snapshot of this bug
      bugID <- value 
    } else {
      if (type =="tag"){ # you want all the source files that were present in this tag 
        tagName <- value
      } else{
        print("sorry wrong input... exiting")
        return
      }
    }
  }

  if (!file.exists(destDir)){
    print("destination directory was not found, we created one")
    dir.create(destDir)
  }
  
  commit_revs = scan(file=paste(basedir,"metaInfo/commit_revs.txt",sep=""),what=list(commitid="",revID=0))
  tags_revs = scan(file=paste(basedir,"metaInfo/tags_revs.txt",sep=""),what=list(revID=0,tagName=""))
  bug_prefix = scan(file=paste(basedir,"metaInfo/bug_prefix.txt",sep=""),what=list(bugID=0,commitid="",revID=0))
  revList = as.numeric(unlist(readLines(paste(basedir,"metaInfo/revList.txt",sep=""),n=-1)))
  bugList = sort(as.numeric(unlist(readLines(paste(basedir,"metaInfo/bugList.txt",sep=""),n=-1))))
  
  if (type=="revision") {
    # extraction of source files that were changed in this revision.
    if (length(which(commit_revs$revID==revID))>0){
      changefile = paste(destDir,"changelist.txt",sep="")
      ind = which(commit_revs$revID==revID)
      commitid = commit_revs$commitid[ind]
      command = paste("cd ",srcDir,"\n git checkout ",commitid,"\n",sep="")
      system(command)
      command = paste("cd ",srcDir,"\ngit diff --name-status -C ",commit_revs$commitid[ind],"..",commit_revs$commitid[ind+1]," >> ",changefile,sep="")
      system(command)
      print("You may now copy the files specified in changelist.txt into your destination folder")
      ChangeInfo = gsub(" $","",readLines(changefile,n=-1))
      print(length(ChangeInfo))
      cvsFiles = grep("^\\.",unlist(lapply(gsub("^[AMD]\t","",ChangeInfo),function(x)basename(x))))  
      if (length(cvsFiles)>0) {
        ChangeInfo = ChangeInfo[setdiff((1:length(ChangeInfo)),cvsFiles)]
      }
      cvsFiles = grep("CVS",gsub("^[AMD]\t","",ChangeInfo))
      if (length(cvsFiles)>0){
        ChangeInfo = ChangeInfo[setdiff((1:length(ChangeInfo)),cvsFiles)]
      }
      
      if (length(ChangeInfo)>0){
        listOfrevA = grep("^[AMC]",ChangeInfo)
        if (length(listOfrevA)>0){
          srcFile = gsub("^[AMC]\t","",ChangeInfo[grep("^[AMC]",ChangeInfo)])
          print(srcFile)
          CopyFile(srcDir,destDir,srcFile)
        }
      }
    }  
  } else { 
    if (type=="bug"){
      # extraction of all source files that were present in the prefix snapshot of this bug
      ind = which(bug_prefix$bugID==bugID)
      if (length(ind)==0){
        print("wrong bugID we could not find it")
        return
      }
      commitid = bug_prefix$commitid[ind]
      command = paste("cd ",srcDir,"\n git checkout ",commitid,"\n",sep="")
      system(command)
      system(paste("cp -r",srcDir,destDir))
    } else {
      if (type=="tag"){
        # extraction of source files that were present in this tag
        ind = which(tags_revs$tagName==tagName)
        if (length(ind)==0){
          print("wrong bugID we could not find it")
          return
        }
        commitid = tags_revs$commitid[ind]
        command = paste("cd ",srcDir,"\n git checkout ",commitid,"\n",sep="")
        system(command)
        system(paste("cp -r",srcDir,destDir))
      } 
    }
  }
}


args <- commandArgs(trailingOnly = TRUE)
print(length(args))
basedir <- args[1] # the location where the metaInfo directory of the dataset can be found
srcDir <- args[2] # the location where .git can be found. Basically location of the clone
destDir <- args[3] # location where you want your original source files
type <- args[4] # the type of information you are seeking, it could be about revisions, bugs or tags
if (type =="revision"){ # this means you just want to copy the source files that have changed in this revision
  value <- as.numeric(args[5])
} else { 
  if (type =="bug"){ # you want all the source files that were present in the prefix snapshot of this bug
    value <- as.numeric(args[5])  
  } else {
    if (type =="tag"){ # you want all the source files that were present in this tag 
      value <- args[5]
    } else{
      print("sorry wrong input... exiting")
      quit()
    }
  }
}

ExtractOriginalSourceFiles(basedir,srcDir,destDir,type,value)