entries <- readLines("~/r-book/ed2/index/index.tex")
entries <- entries[-(1:41)]
z <- regexpr("\\\\indexspace", entries)
entries <- entries[z<0]
entries <- trim.String(entries)
entries <- entries[entries!=""]


makeone <- function(ent=entries, z = grep( "^[0-9]", ent)){
z <- sort(unique(c(z-1,z)))
omit <- logical(length(ent))
repeat {
  if(length(z)==0) break
  n1 <- seq(along=z)
  m1 <- 1:min(n1[c(diff(z),2)>1])
  d <- z[m1]
  ent[d[1]] <- paste(ent[d], collapse=" ")
  omit[d[-1]] <- TRUE
  z <- z[-m1]
}
ent <- ent[!omit]
ent
}

entries[1046] <- paste("\\", entries[1046], sep="")
st <- substring(entries, 1, 5)
table(st)
entries <- entries[st!="\\clea"]

nond <- seq(along=entries)[-grep("^\\\\",entries)]
newe <- makeone(z=nond)
st <- substring(newe, 1, 5)
table(st)

nond <- seq(along=newe)[grep("^\\\\text", newe)]
newe <- makeone(ent=newe,z=nond)  
grep("\\\\end", entries)
ifun <- entries[2:495]
ifun <- substring(ifun,7)
iterms <- entries[498:1067]
iauthors <- entries[1070:1242]

z <- regexpr(", +[0-9]", ifun)
ifun <- ifun[z>=0]
z <- regexpr(", +[0-9]", ifun)
pages <- substring(ifun, z+attr(z,"match.length")-1)
fundex <- substring(ifun, 1, z-1)
funmat <- cbind(item=fundex, pages=pages)

bypage <- function(tx=funmat){
  pag <- character(0)
  for(i in 1:dim(tx)[1]){
    nam <- trim.String(tx[i,1])
    pages <- trim.String(strsplit(tx[i,2], ",")[[1]])
    names(pages) <- rep(nam, length(pages))
    pag <- c(pag,pages)
  }
  sort(pag)
}

funpages <- bypage()

iauthors <- substring(iauthors, 7)      ## Omit \\item
z <- regexpr(", +[0-9ivx]", iauthors)
authdex <- substring(iauthors, 1, z-3)  ## All end in \\
authdex <- sub("{","", authdex, fixed=T)
authdex <- sub("}","", authdex, fixed=T)
pages <- substring(iauthors, z+attr(z,"match.length")-1)
authmat <- cbind(item=authdex, pages=pages)
authpages <- bypage(authmat)

iterms <- entries[498:1067]
z <- regexpr("m ", iterms)
iterms <- trim.String(substring(iterms, z+1))
iterms[352] <- "mean, 5, 18, {\\itshape passim}"
iterms[397] <- "DAAG, 1, 9, {\\itshape passim}"
iterms[399] <- "dichromat, 460, 492"
iterms[407] <- "KernSmooth, 240"
iterms[441] <- "histogram, 6, 38, 44--46, 59, 74, 92, 121"
iterms[517] <- "data collection, 276"
iterms[520] <- "martingale, 282"
iterms[521] <- "Schoenfeld, 281"
iterms[522] <- "survival estimate, 277--278"
iterms[523] <- "tree-based, 372--373"
iterms[525] <- "of frequencies, 19--20, 60--63, 95--96, 115-119, 141--142"
iterms[526] <- "adding across tables (Simpson's paradox), 61, 63, 95--96, 141--142"

[16,] "399" "dichromat 460"                              
[17,] "407" "KernSmooth 240"                             
[19,] "441" "histogram 6"                                
[20,] "517" "data collection 276"                        
[21,] "520" "martingale 282"                             
[22,] "521" "Schoenfeld 281"                             
[23,] "522" "survival estimate 277--278"                 
[24,] "523" "tree-based  372--373"                       
[25,] "525" "of frequencies 19--20"                      
[26,] "526" "adding across tables (Simpson's paradox) 61"

sum(substring(termdex,1,3)=="see")
iterms <- iterms[substring(iterms,1,3)!="see",]
mterm <- matrix("", nrow=length(z), ncol=4)
mterm[,4] <- z
id <- c("5"=1, "8"=2, "11"=3)[paste(z)]

z <- regexpr(", +[0-9]", iterms)
z[z==-1] <- nchar(iterms[z==-1])+1
termdex <- trim.String(substring(iterms,1,z-1))
seq(along=iterms)[substring(termdex,nchar(termdex))%in%c("i","x")]
z[c(8,291,422)] <- regexpr(", +[ix]", iterms[c(8,291,422)])
termdex <- trim.String(substring(iterms,1,z-1))
termdex[substring(termdex,nchar(termdex))%in%c(paste(0-9),"i","v","x")]
table(substring(termdex,nchar(termdex)))

pages <- trim.String(substring(iterms,z+1))
mterm[cbind(1:570, id)] <- termdex
mterm <- cbind(mterm, pages)
u <- (1:570)[nchar(termdex)>50]
mterm <- mterm[-u[-11],]
termdex <- termdex[-u[-11]]]
pages <- pages[-u[-11]]
iterms <- iterms[-u[-11]]

make.termindex <- function(mterm, completecol=2){
nrow <- dim(mterm)[1]
k <- completecol
irow <- 0
jmax <- max((1:nrow)[mterm[,k+1]!=""])
repeat {
  lookatrows <- (irow+1):nrow
j <- min(lookatrows[mterm[lookatrows,k+1]!=""])
  cat(" ",j)
  belowj <- 1:(j-1)
if(mterm[j,k]=="")
  mterm[j,k] <- mterm[max(belowj[mterm[belowj,k]!=""]), k]
irow <- j
  if(j==jmax)break
}
mterm
}

termsmat <- cbind(paste(mterm[,1], mterm[,2], mterm[,3], sep="! "), mterm[,5])

termspages <- bypage(termsmat)