From 9e2ea7ce77659e7c57c3252cb257f8e39183be8e Mon Sep 17 00:00:00 2001
From: Vincent van Hees <v.vanhees@accelting.com>
Date: Thu, 26 Sep 2024 17:08:09 +0200
Subject: [PATCH] reduce code duplication #68

---
 R/readActiGraphCount.R  | 14 +++-----------
 R/readActiwatchCount.R  |  9 +--------
 R/utils_for_countdata.R | 21 ++++++++++++++++++++-
 3 files changed, 24 insertions(+), 20 deletions(-)

diff --git a/R/readActiGraphCount.R b/R/readActiGraphCount.R
index 20060f4..1455c73 100644
--- a/R/readActiGraphCount.R
+++ b/R/readActiGraphCount.R
@@ -107,7 +107,7 @@ readActiGraphCount = function(filename = file, desiredEpochSize = NULL,
   keep = c(acccol, vmcol)[!is.na(c(acccol, vmcol))]
   D = D[, keep, drop = FALSE]
   if (ncol(D) == 3 & is.na(vmcol)) {
-    D$vm = sqrt(D[,1]^2 + D[,2]^2 + D[,3]^2)
+    D$vm = sqrt(D[, 1] ^ 2 + D[, 2] ^ 2 + D[, 3] ^ 2)
   }
   # Extract information from header, if present
   if (headerAvailable == TRUE) {
@@ -129,12 +129,7 @@ readActiGraphCount = function(filename = file, desiredEpochSize = NULL,
     if (colnames == TRUE) {
       datecol = grep("date", colnames(tmp), ignore.case = TRUE)
       timecol = grep("time|epoch", colnames(tmp), ignore.case = TRUE)
-      time = tmp[, timecol]
-      date = tmp[, datecol]
-      
-      starttime = time[1]
-      starttime = date[1]
-      timestamp = paste0(date, " ", time)
+      timestamp = paste0(tmp[, datecol], " ", tmp[1, timecol])
       format = timeformat
       timestamp_POSIX = as.POSIXlt(timestamp, tz = tz, format = format)
       if (all(is.na(timestamp_POSIX))) {
@@ -159,10 +154,7 @@ readActiGraphCount = function(filename = file, desiredEpochSize = NULL,
   if (!is.null(desiredEpochSize)) {
     if (desiredEpochSize > epSizeShort) {
       step = desiredEpochSize %/% epSizeShort
-      D = rbind(rep(0, ncol(D)), D)
-      D = apply(D, 2, cumsum)
-      D = D[seq(1, nrow(D), by = step), , drop = FALSE]
-      D = apply(D, 2, diff)
+      D = sumAggregate(D, step)
       epSizeShort = epSizeShort * step
     }
     checkEpochMatch(desiredEpochSize, epSizeShort)
diff --git a/R/readActiwatchCount.R b/R/readActiwatchCount.R
index 0a351bf..e5e8416 100644
--- a/R/readActiwatchCount.R
+++ b/R/readActiwatchCount.R
@@ -111,14 +111,7 @@ readActiwatchCount = function(filename = file, desiredEpochSize = NULL,
   if (!is.null(desiredEpochSize)) {
     if (desiredEpochSize > epSizeShort) {
       step = desiredEpochSize %/% epSizeShort
-      D = rbind(rep(0, ncol(D)), D)
-      cumsum2 = function(x) {
-        x = cumsum(ifelse(is.na(x), 0, x)) + x*0
-        return(x)
-      }
-      D = apply(D, 2, cumsum2)
-      D = D[seq(1, nrow(D), by = step), , drop = FALSE]
-      D = apply(D, 2, diff)
+      D = sumAggregate(D, step)
       epSizeShort = epSizeShort * step
     }
     checkEpochMatch(desiredEpochSize, epSizeShort)
diff --git a/R/utils_for_countdata.R b/R/utils_for_countdata.R
index 34ce7ce..49a778a 100644
--- a/R/utils_for_countdata.R
+++ b/R/utils_for_countdata.R
@@ -1,6 +1,10 @@
-# Collection of short function used in functions readActigraphCount, readActiwatchCount
+# Collection of short function used in functions
+# - readActigraphCount
+# - readActiwatchCount
+#-----------------------------------------------------------------------------------------
 checkTimeFormat = function(timestamp_POSIX, rawValue = " ?? ", timeformat = " ?? ",
                            timeformatName = NULL) {
+  # If timestamp_POSIX is NA gieve error message to inform user that something went wrong.
   if (is.na(timestamp_POSIX)) {
     stop(paste0("\nTime format in data ", rawValue, 
                 " does not match with time format ", timeformat,
@@ -10,6 +14,7 @@ checkTimeFormat = function(timestamp_POSIX, rawValue = " ?? ", timeformat = " ??
 }
 
 checkEpochMatch = function(desiredEpochSize, epSizeShort) {
+  # Check whether desired and derived epoch size match
   if (!is.null(desiredEpochSize) && epSizeShort != desiredEpochSize) {
     stop(paste0("\nThe short epoch size as specified by the user (",
                 desiredEpochSize, " seconds) does NOT match the short",
@@ -39,7 +44,21 @@ detectQuote = function(fn, index) {
 }
 
 getExtension <- function(filename){ 
+  # Extract file extension
   ex <- unlist(strsplit(basename(filename), split = "[.]"))
   if (length(ex) < 2) stop(paste0("Cannot recognise extension from '", filename, "' as filename, please check"), call. = FALSE)
   return(ex[-1])
+}
+
+sumAggregate = function(mat, step) {
+  # Aggregate matrix mat by taking the sum over step number of rows
+  mat = rbind(rep(0, ncol(mat)), mat)
+  cumsum2 = function(x) {
+    x = cumsum(ifelse(is.na(x), 0, x)) + x*0
+    return(x)
+  }
+  mat = apply(mat, 2, cumsum2)
+  mat = mat[seq(1, nrow(mat), by = step), , drop = FALSE]
+  mat = apply(mat, 2, diff)
+  return(mat)
 }
\ No newline at end of file