# R program to create .csv file from cgiwrite output
# This program is indended to be called from the shell file "cgiread".
# It creates cgireadlog.tmp with comments.

# Shell file "cgiread" writes user's request to "cgireadDBfile.tmp"
# Read filename from this file and verify that it exists.
if (!file.exists("cgireadDBfile.tmp")) {
  write("Can't open cgireadDBfile.tmp", file="cgireadlog.tmp")
  quit(save="no")
}
filename=scan("cgireadDBfile.tmp", what="", quiet=TRUE)
if (!file.exists(filename)) {
  write(paste("Can't open",filename), file="cgireadlog.tmp")
  quit(save="no")
}

# Read data from file (one field per line)
lines=scan(filename, sep="\n", strip.white=TRUE, what="", quiet=TRUE)
lines=lines[nchar(lines)>0]
nl=length(lines)
if (nl==0) {
  write(paste(filename, "is empty."), file="cgireadlog.tmp")
  quit(save="no")
}

# Record lines with equal signs (start of field) and positions of equal signs
eqloc=as.numeric(regexpr("\=",lines))
while (length(eqloc)>1 && eqloc[1]==-1) {
  eqloc=eqloc[-1]
  lines=lines[-1]
}
nl=length(lines)
if (length(eqloc[1])==0) {
  write(paste(filename,"appears to have the wrong format"), file="cgireadlog.tmp")
  quit(save="no")
}

# Assume first observed field name starts all records
firstname=substring(lines[1],1,eqloc[1])
startrec=regexpr(paste("^",firstname,sep=""),lines)
startrec=seq(along=startrec)[startrec>0]
nrec=length(startrec)

# Get all field names
allnames=unique(substring(lines[eqloc>0],1,eqloc[eqloc>0]))
nfields=length(allnames)

# Construct matrix of input, concatenating as needed
# First drop trailing characters of field names
trimnames=allnames
for (i in seq(along=trimnames)) {
  tmp=trimnames[i]
  j=nchar(tmp)
  while (j>1 && (substring(tmp,j,j)==" " || substring(tmp,j,j)=="=")) {
    tmp=substring(tmp,1,j-1)
    j=j-1
  }
  trimnames[i]=tmp
}
mat=matrix("", nrec, nfields, dimnames=list(1:nrec, trimnames))
startlines=seq(along=eqloc)[eqloc>0] # named fields start here
startlinesX=c(startlines,nl+1)
startrecX=c(startrec,nl+1)
for (row in seq(along=startrec)) {
  n1=startlines[startlines>=startrec[row] & startlines<startrecX[row+1]]
  nn=c(n1[-1]-1,startrecX[row+1]-1)
  for (field in seq(along=n1)) {
    this1=lines[n1[field]]
    e=as.numeric(regexpr("=",this1))
    if (e<=0) {
      write(paste("Bad logic (#1) -- see Howard"), file="cgireadlog.tmp")
      #quit(save="no")
    }
    col=match(substring(this1,1,e),allnames)
    if (is.na(col) || col<1 || col>nfields) {
      write(paste("Bad logic (#2) -- see Howard"), file="cgireadlog.tmp")
      #quit(save="no")
    }
    if (nchar(this1)==e) {
      tmp="" } else {
      tmp=substring(this1,1+e)
    }
    while (nchar(tmp)>0 && substring(tmp,1,1)==" ") tmp=substring(tmp,2)
    if (nn[field]>n1[field]) tmp=paste(tmp,paste(lines[(n1[field]+1):(nn[field])],collapse=" "),collapse=" ")
    mat[row,col]=tmp
  }
}

# Get basename for creating new filenames
basename=filename
i=nc=nchar(filename)
while (i>1) {
  if (substring(filename,i,i)==".") {
    basename=substring(filename,1,i-1)
    break
  }
  i=i-1
}

# Write csv file
csvname=paste(basename,".csv",sep="")
write.table(data.frame(mat), file=csvname, sep=",",
            row=FALSE, col=TRUE, quote=TRUE)
write(paste("Wrote",csvname), file="cgireadlog.tmp")