From c1450bc43086e301b2f0ea5b20eb817973951ce4 Mon Sep 17 00:00:00 2001
From: Dhirendra Singh <dhi.singh@gmail.com>
Date: Wed, 3 Mar 2021 16:03:01 +1100
Subject: [PATCH] Updated expected data files; #4

---
 R/checksum.R    | 14 +++++++++-----
 data/.gitignore |  4 +++-
 2 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/R/checksum.R b/R/checksum.R
index 2f357d2..a5714a9 100644
--- a/R/checksum.R
+++ b/R/checksum.R
@@ -8,6 +8,7 @@ checksum <- function() {
     '../data/SA1attributed.csv.gz' = '980faffc7c90996510123ce4d69f140e', 
     '../data/SA1centroids.csv.gz' = 'b7f27ff1e176b70cee71723f528f800a', 
     '../data/addresses.csv.gz' = 'ed761018dfdf4e00fdf2054aa25fe3ba',
+    '../data/vistaCohorts.csv.gz' = 'd9f0c9ad76fc31513afbc9348b747e32',
     '../data/expectedDistances.rds' = '91e137c02459e21a2fabf541e8cc36de',
     '../data/vistaSummaries/destinationProbabilitiesSA3.rds' = '7dbacc829b312d608b6f24df7235b028',
     '../data/vistaSummaries/distanceDistributionsSA3.rds' = '9ded1a901e720c3f670b5ecda6c105f8',
@@ -21,12 +22,15 @@ checksum <- function() {
   for (file in names(expected)) {
     e <- expected[file]
     a <- actual[file]
-    if (a != e) {
-      cat("MD5 hash of input data file did not match expectation. File may be outdated.\n")
-      cat(paste0("File:          ", file, "\n"))
-      cat(paste0("Expected hash: ", e, "\n"))
-      cat(paste0("Actual hash:   ", a, "\n"))
+    cat(paste0("Checking for ", file, ": "))
+    if (!file.exists(file)) {
+      cat("NOT FOUND\n")
       result <- FALSE
+    } else if (a != e) {
+      cat(paste0("MISMATCH\n!!! expected ", e, ", actual ", a, "\n"))
+      result <- FALSE
+    } else {
+      cat("OK\n")
     }
   }
   return(result)
diff --git a/data/.gitignore b/data/.gitignore
index a4606f0..a9fdcb9 100644
--- a/data/.gitignore
+++ b/data/.gitignore
@@ -1,6 +1,6 @@
 distanceMatrix.rds
 distanceMatrixIndex.csv
-expectedDistances.gz
+expectedDistances.rds
 melbourne-2016-population.zip*
 sa1_2016_aust.csv.gz
 SA1attributed.csv.gz
@@ -8,3 +8,5 @@ SA1centroids.csv.gz
 addresses.csv.gz
 vista_2012_16_v1_sa1_csv.zip*
 VISTA_12_18_CSV.zip*
+vistaCohorts.csv.gz
+vistaSummaries/