tripPlanner.Rmd

---
title: "Travel time wales"
output: html_notebook
---


```{r}
library("propeR")
library("progress")
library("MASS")
library("foreach")
library("doParallel")
library("doSNOW")
```


connect to OTP
```{r}
otpcon<-otpConnect()
```

change importLocationData to not drop point if no lat lon
```{r}
insertSource("~/Documents/GitHub/access-to-services/propeR/R/data.R", package = "propeR", functions = "importLocationData") 
```



Read all the locations
```{r}
originPoints <- importLocationData("./origin.csv")
destinationPoints <-importLocationData("./destination.csv")
```

Here's a function that takes just one argument which is the row number of the locations to process. The mode, starttime, arriveby, date are all hard coded.

We've done it this way so we can process stuff in parallel
```{r}
fx<-function(k){
  message("processing number ",k)
  point_to_point<-propeR::otpTripTime(otpcon,  from = originPoints[k,]$lat_lon,to=destinationPoints[k,]$lat_lon,modes="CAR",  detail = T,date = "2019-03-25",time = "09:00am", arriveBy = 'true')
  
from=originPoints[k,]
to=destinationPoints[k,]

if (!is.null(point_to_point$errorId)){
        if (point_to_point$errorId == "OK") {
          point_to_point_table_overview_tmp <- point_to_point$itineraries
          point_to_point_table_overview_tmp["origin"] <- from$name
          point_to_point_table_overview_tmp["destination"] <- to$name
          point_to_point_table_overview_tmp["distance_km"] <- round(sum(point_to_point$output_table$distance) / 1000, digits = 2)
          point_to_point_table_overview_tmp["journey_details"] <- jsonlite::toJSON(point_to_point$output_table)
        } else {
          point_to_point_table_overview_tmp <- data.frame(
            "start" = NA,
            "end" = NA,
            "duration" = NA,
            "driveTime" = NA,
            "transitTime" = NA,
            "waitingTime" = NA,
            "transfers" = NA,
            "origin" = from$name,
            "destination" = to$name,
            "distance_km" = NA,
            "journey_details" = NA)
        }
        return(point_to_point_table_overview_tmp)
      }
message(k)
} #end of function
```


Do things in parallel
```{r}
numCores<-detectCores()
cl<-makeCluster(numCores)
#registerDoParallel(numCores)
registerDoSNOW(cl)
totalRows <- nrow(destinationPoints)

#make a loop for to spit out csvs periodically
iterations = ceiling(totalRows/1000)

startingSequence<-seq(from=1,to=totalRows,by=1000)
endSequence<-if(iterations==1){}else{seq(from=1000,to=totalRows,by=1000)}
endSequence<-append(endSequence,totalRows,after = length(endSequence))


for(i in 1:iterations){
  if(iterations>1){message("\n Starting iteration ",i," of ",iterations,"\n")}
  message("processing rows ", startingSequence[i]," to ",endSequence[i])

  stamp <- format(Sys.time(), "%Y_%m_%d_%H_%M_%S")



  pb <- txtProgressBar(max=if(i==iterations){if(totalRows%%1000==0){1000}else{totalRows%%1000}}else{1000},style=3)
  progress <- function(n) setTxtProgressBar(pb,n)
  opts <- list(progress=progress)
  res <- foreach(k=startingSequence[i]:endSequence[i],.combine=rbind, .inorder = FALSE,.options.snow=opts) %dopar% {
    fx(k)
  }
  
  point_to_point_table_overview_out <- res[, c(8, 9, 1, 2, 10, 3, 4, 5, 6, 7, 11)] 
  colnames(point_to_point_table_overview_out) <-
              c(
                "origin",
                "destination",
                "start_time",
                "end_time",
                "distance_km",
                "duration_mins",
                "drive_time_mins",
                "transit_time_mins",
                "waiting_time_mins",
                "transfers",
                "journey_details")
  
  write.csv(
              point_to_point_table_overview_out,
              file = paste0("./", "/pointToPointLoop-", stamp, ".csv"),
              row.names = F)
    
  
  
#copy the time to complete a loop thing here  
}


            
#clean up the cluster
stopImplicitCluster()
close(pb)
stopCluster(cl)
```

Read all csvs in folder and merge
```{r}
folder <- "./"
filenames <- list.files(folder, pattern="^pointToPoint")

all_files <- Reduce(rbind, lapply(filenames, read.csv))
write.csv(all_files,'all_postcodes.csv', row.names=FALSE)
```