Merge pull request #1279 from wadpac/issue1278_split_recordings

Add option to split recordings into multiple recordings

Author: vincentvanhees

NAMESPACE

@@ -42,7 +42,8 @@ export(g.analyse, g.calibrate,
        g.part5_analyseSegment, g.part5_initialise_ts,
        g.part5.analyseRest, part6AlignIndividuals,
        part6PairwiseAggregation, g.part6, g.report.part6, 
-       check_log, g.report.part5_dictionary, DFA, ABI, SSP, visualReport)
+       check_log, g.report.part5_dictionary, DFA, ABI, SSP, visualReport,
+       splitRecords, getSplitNames, addSplitNames)
 
 importFrom("grDevices", "colors", "dev.off", "pdf",
            "rgb", "rainbow", "palette", "adjustcolor", "gray.colors")

NEWS.md

@@ -6,6 +6,8 @@
 
 - Part 5 and visualreport: When wake up time at start of recording is added in part 5 no log was kept of the guider and bedend column was not recognised, this is now fixed. #1273
 
+- General: Add functionality to split recordings based on recording specific timestamps or dates, for example useful when studying effect of interventions. This new functionality comes with new parameters recording_split_times, recording_split_overlap, recording_split_timeformat #1278.
+
 # CHANGES IN GGIR VERSION 3.2-0
 
 - Part 2:

R/GGIR.R

@@ -269,6 +269,12 @@ GGIR = function(mode = 1:5, datadir = c(), outputdir = c(),
                        params_general = params_general,
                        verbose = verbose)
     }
+    if (!is.null(params_general[["recording_split_times"]])) {
+      # Split recordings based on user specified time points
+      splitRecords(metadatadir = metadatadir,
+                   params_general = params_general)
+    }
+    
     if (!is.null(params_general[["maxRecordingInterval"]])) {
       # Append recordings when ID and brand match and gap between
       # recordings does not exceed maxRecordingInterval,

R/addSplitNames.R

@@ -0,0 +1,26 @@
+addSplitNames = function(x) {
+  if (nrow(x) != 0) {
+    x$filename = as.character(x$filename)
+    for (spi in 1:nrow(x)) {
+      splitnames = getSplitNames(as.character(x$filename[spi]))
+      segment_names = splitnames$segment_names
+      filename = splitnames$filename
+      if (!is.null(segment_names)) {
+        x$filename[spi] = filename
+        if ("split1_name" %in% colnames(x) == FALSE) {
+          x$split1_name = NA
+          x$split2_name = NA
+          col_index_filename = which(colnames(x) == "filename")
+          
+          x = x[,c(1:col_index_filename, 
+                   which(colnames(x) %in% c("split1_name", "split2_name")),
+                   (col_index_filename + 1):(ncol(x) - 2))] 
+        }
+        x$split1_name[spi] = segment_names[2]
+        x$split2_name[spi] = segment_names[3]
+      }
+    }
+    x$filename = as.factor(x$filename)
+  }
+  return(x)
+}

R/appendRecords.R

@@ -1,20 +1,6 @@
 appendRecords = function(metadatadir, desiredtz = "", idloc = 1, maxRecordingInterval = NULL) {
 
   # Declare local functions:
-  getInfo = function(fn, idloc, tz) {
-    load(fn)
-    if (is.null(M$metashort)) return()
-    hvars = g.extractheadervars(I)
-    if (exists("Clist")) {
-      ID = NA # If Clist exists then ignore this file as it was previously appended
-    } else {
-      ID = extractID(hvars, idloc, fname = I$filename)
-    }
-    start = as.POSIXct(x = M$metashort$timestamp[1], format = "%Y-%m-%dT%H:%M:%S%z", tz = tz)
-    end = as.POSIXct(x = M$metashort$timestamp[nrow(M$metashort)], format = "%Y-%m-%dT%H:%M:%S%z", tz = tz)
-    info = data.frame(ID = ID, start = start, end = end, filename = fn, brand = I$monn)
-    return(info)
-  }
   mergePair = function(M1, M2, overlap, tz) {
     # overlap in epochs
     if (overlap < 0) { 
@@ -77,7 +63,7 @@ appendRecords = function(metadatadir, desiredtz = "", idloc = 1, maxRecordingInt
   # Create overview of all recordings ID, start time, end time, and filename
   fns = dir(paste0(metadatadir, "/meta/basic"), full.names = TRUE)
 
-  S = do.call("rbind", lapply(X = fns, FUN = getInfo, idloc = idloc, tz = desiredtz)) 
+  S = do.call("rbind", lapply(X = fns, FUN = getPart1BasicInfo, idloc = idloc, tz = desiredtz)) 
   if (length(S) > 0 & !is.null(maxRecordingInterval)) {
     S = S[!is.na(S$ID),]
     S = S[order(S$ID, S$start), ]

R/check_params.R

@@ -131,10 +131,12 @@ check_params = function(params_sleep = c(), params_metrics = c(),
   }
   if (length(params_general) > 0) {
     numeric_params = c("maxNcores", "windowsizes", "idloc", "dayborder",
-                       "expand_tail_max_hours", "maxRecordingInterval")
+                       "expand_tail_max_hours", "maxRecordingInterval",
+                       "recording_split_overlap")
     boolean_params = c("overwrite", "print.filename", "do.parallel", "part5_agg2_60seconds")
     character_params = c("acc.metric", "desiredtz", "configtz", "sensor.location", 
-                         "dataFormat", "extEpochData_timeformat")
+                         "dataFormat", "extEpochData_timeformat", "recording_split_times",
+                         "recording_split_timeformat")
     check_class("general", params = params_general, parnames = numeric_params, parclass = "numeric")
     check_class("general", params = params_general, parnames = boolean_params, parclass = "boolean")
     check_class("general", params = params_general, parnames = character_params, parclass = "character")
@@ -557,10 +559,20 @@ check_params = function(params_sleep = c(), params_metrics = c(),
     }
   }
 
-  if (!is.null(params_general[["maxRecordingInterval"]])) {
-    if (params_general[["maxRecordingInterval"]] > 24 * 21) {
-      stop(paste0("A maxRecordingInterval value higher than 21 days (504 hours) is permitted,",
-                  " please specify a lower value."), call. = FALSE)
+  if (length(params_general) > 0) {
+    if (!is.null(params_general[["maxRecordingInterval"]])) {
+      if (params_general[["maxRecordingInterval"]] > 24 * 21) {
+        stop(paste0("A maxRecordingInterval value higher than 21 days (504 hours) is not permitted,",
+                    " please specify a lower value."), call. = FALSE)
+      }
+    }
+    if (!is.null(params_general[["recording_split_times"]])) {
+      if (!file.exists(params_general[["recording_split_times"]])) {
+        stop(paste0("File .../", basename(params_general[["recording_split_times"]]),
+                    " as specified with parameter recording_split_times does not exist, ",
+                    " please fix."), call. = FALSE)
+        
+      }
     }
   }
 

R/g.analyse.R

@@ -220,16 +220,17 @@ g.analyse =  function(I, C, M, IMP, params_247 = c(), params_phyact = c(),
                                      doiglevels = doiglevels, nfulldays = nfulldays,
                                      lastmidnight = lastmidnight,
                                      ws3 = ws3, ws2 = ws2, qcheck = qcheck, fname = fname,
-                                     idloc = idloc, sensor.location = sensor.location,
+                                     sensor.location = sensor.location,
                                      wdayname = wdayname,
                                      tooshort = tooshort, includedaycrit = includedaycrit,
                                      quantiletype = quantiletype, doilevels = doilevels, 
                                      domvpa = domvpa,
                                      mvpanames = mvpanames, wdaycode = wdaycode, ID = ID, 
                                      deviceSerialNumber = deviceSerialNumber,
                                      doquan = doquan,  ExtFunColsi = ExtFunColsi,
-                                     myfun = myfun, desiredtz = desiredtz,
-                                     params_247 = params_247, params_phyact = params_phyact)
+                                     myfun = myfun,
+                                     params_247 = params_247, params_phyact = params_phyact,
+                                     params_general = params_general)
   }
   #metashort is shortened from midgnight to midnight if requested (data_masking_strategy 2)
   if (data_masking_strategy == 2) {

R/g.analyse.perday.R

@@ -1,12 +1,14 @@
 g.analyse.perday = function(ndays, firstmidnighti, time, nfeatures,
                             midnightsi, metashort, averageday,
                             doiglevels, nfulldays,lastmidnight, ws3, ws2, qcheck,
-                            fname, idloc, sensor.location, wdayname, tooshort, includedaycrit,
+                            fname, sensor.location, wdayname, tooshort, includedaycrit,
                             doquan, quantiletype, doilevels, domvpa,
                             mvpanames, wdaycode, ID,
-                            deviceSerialNumber, ExtFunColsi, myfun, desiredtz = "",
+                            deviceSerialNumber, ExtFunColsi, myfun, 
                             params_247 = c(), params_phyact = c(),
+                            params_general = c(),
                             ...) {
+
   #get input variables
   input = list(...)
 
@@ -50,7 +52,7 @@ g.analyse.perday = function(ndays, firstmidnighti, time, nfeatures,
   }
   unique_dates_recording = unique(as.Date(iso8601chartime2POSIX(time[c(seq(1, length(time),
                                                                            by = (3600/ws2) * 12),
-                                                                       length(time))], tz = desiredtz)))
+                                                                       length(time))], tz = params_general[["desiredtz"]])))
   ExtFunColsi = ExtFunColsi - 1 # subtract 1 because code ignores timestamp
 
   for (di in 1:ndays) { #run through days
@@ -198,8 +200,23 @@ g.analyse.perday = function(ndays, firstmidnighti, time, nfeatures,
     daysummary[di,fi] = ID
     idremember = daysummary[di,fi]
     ds_names[fi] = "ID";      fi = fi + 1
-    daysummary[di,fi] = fname
+    if (!is.null(params_general[["recording_split_times"]])) {
+      filename = fname
+      splitnames = getSplitNames(filename)
+      segment_names = splitnames$segment_names
+      filename = splitnames$filename
+    } else {
+      segment_names = NULL
+      filename = fname
+    }
+    daysummary[di,fi] = filename
     ds_names[fi] = "filename";  fi = fi + 1
+    if (!is.null(params_general[["recording_split_times"]]) && !is.null(segment_names)) {
+      daysummary[di,fi] = segment_names[2]
+      daysummary[di,fi + 1] = segment_names[3]
+      ds_names[fi:(fi + 1)] = c("split1_name","split2_name")
+      fi = fi + 2
+    }
     calendardate = unlist(strsplit(as.character(vari[1,1])," "))[1]
     daysummary[di,fi] = calendardate
     daysummary[di,(fi + 1)] = sensor.location
@@ -342,17 +359,17 @@ g.analyse.perday = function(ndays, firstmidnighti, time, nfeatures,
                 }
               } else {
                 # When it is not the first day of recording
-                if (NRV == 23) { # day has 23 hours (assuming DST)
+                if (NRV == 23 * 60 * (60/ws3)) { # day has 23 hours (assuming DST)
                   # Append data after 2nd hour
                   startMissingHour = 2 * 60 * (60/ws3) + 1
                   enMissingHour = 3 * 60 * (60/ws3)
                   vari = rbind(vari[1:(startMissingHour - 1)], averageday[startMissingHour:enMissingHour, ],
-                           vari[startMissingHour:nrow(vari),])
+                               vari[startMissingHour:nrow(vari), , drop = FALSE])
                 } else { # day has less than 24 hours for another reason
                   # Append the average day to the end
                   a56 = nrow(averageday) - abs(deltaLength) + 1
                   a57 = nrow(averageday)
-                  vari = rbind(vari,averageday[a56:a57,])
+                  vari = rbind(vari, averageday[a56:a57, , drop = FALSE])
                 }
               }
             } else if (deltaLength > 0) { # 25 hour days, assuming DST

R/g.analyse.perfile.R

@@ -37,10 +37,26 @@ g.analyse.perfile = function(I, C, metrics_nav,
   vi = vi + 2
   # starttime of measurement, body location, filename
   filesummary[vi] = file_summary$sensor.location
-  filesummary[(vi + 1)] = file_summary$fname
+  if (!is.null(params_general[["recording_split_times"]])) {
+    filename = file_summary$fname
+    splitnames = getSplitNames(filename)
+    segment_names = splitnames$segment_names
+    filename = splitnames$filename
+  } else {
+    segment_names = NULL
+    filename = file_summary$fname
+  }
+  filesummary[(vi + 1)] = filename
   filesummary[(vi + 2)] = file_summary$startt # starttime of measurement
   s_names[vi:(vi + 2)] = c("bodylocation","filename","start_time")
   vi = vi + 3
+  
+  if (!is.null(params_general[["recording_split_times"]]) && !is.null(segment_names)) {
+    filesummary[vi] = segment_names[2]
+    filesummary[vi + 1] = segment_names[3]
+    s_names[vi:(vi + 1)] = c("split1_name","split2_name")
+    vi = vi + 2
+  }
   # weekday on which measurement started, sample frequency and device
   filesummary[vi] = file_summary$wdayname
   filesummary[(vi + 1)] = I$sf

R/g.report.part2.R

@@ -309,7 +309,7 @@ g.report.part2 = function(metadatadir = c(), f0 = c(), f1 = c(), maxdur = 0,
       # reorder to have starttime next to calendar_date
       old_vars = which(colnames(daySUMMARY_clean) != "start_time")
       new_var = which(colnames(daySUMMARY_clean) == "start_time")
-      daySUMMARY_clean = daySUMMARY_clean[, c(old_vars[1:3], new_var, old_vars[4:length(old_vars)])]
+      daySUMMARY_clean = daySUMMARY_clean[, c(old_vars[1:2], new_var, old_vars[3:length(old_vars)])]
       # format calendar dates
       dd = iso8601chartime2POSIX(daySUMMARY_clean$calendar_date, tz = desiredtz) 
       daySUMMARY_clean$calendar_date = format(dd, format = "%Y-%m-%d")