Merge pull request #92 from habitus-eu/issue90_integrate_hbGPS

Issue90 integrate hb gps

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: HabitusGUI
 Title: R Shiny App for Processing Behavioural Data
 Description: Shiny app to ease processing behavioural data with research software such as GGIR, activityCounts, PALMSpy,and palmsplusr.
-Version: 0.1.9
-Date: 2022-09-08
+Version: 0.2.0
+Date: 2022-09-30
 Authors@R: 
     c(person(given = "Vincent",
            family = "van Hees",
@@ -13,8 +13,8 @@ Authors@R:
 License: Apache License version 2.0 | file LICENSE
 Imports: shiny, shinyFiles, GGIR, bslib, methods, jsonlite, DT, 
     dplyr, magrittr, shinyjs, sf, readr, tidyr, stringr, callr, palmsplusr,
-    data.table, rlang, purrr, geosphere
-Remotes: vincentvanhees/palmsplusr
+    data.table, rlang, purrr, geosphere, hbGPS
+Remotes: vincentvanhees/palmsplusr, habitus-eu/hbGPS
 LazyData: true
 Suggests: testthat, covr, rmarkdown
 Depends: stats, utils, R (>= 3.5.0)

NAMESPACE

@@ -3,12 +3,13 @@
 export(Counts2csv)
 export(GGIRshiny)
 export(PALMSpyshiny)
-export(checkFile)
+export(checkConfigFile)
 export(check_and_clean_palms_data)
 export(check_params)
 export(cleanPath)
 export(create_test_GGIRconfig)
 export(create_test_files)
+export(hbGPS_shiny)
 export(hbt_build_days)
 export(hbt_build_multimodal)
 export(hbt_build_palmsplus)
@@ -24,6 +25,7 @@ export(update_params)
 exportClasses(toolio)
 import(GGIR)
 import(dplyr)
+import(hbGPS)
 import(palmsplusr)
 import(sf)
 import(shiny)
@@ -39,7 +41,6 @@ importFrom(methods,new)
 importFrom(methods,setClass)
 importFrom(purrr,reduce)
 importFrom(readr,read_csv)
-importFrom(readr,write_csv)
 importFrom(rlang,UQ)
 importFrom(rlang,parse_expr)
 importFrom(stats,aggregate)

R/GGIRshiny.R

@@ -2,8 +2,8 @@
 #'
 #' @param rawaccdir Path to input directory
 #' @param outputdir Path to output directory
-#' @param configfile Configfile path
 #' @param sleepdiary Path to sleep diary
+#' @param configfile Configfile path
 #' @param do.Counts Boolean to indicate whether BrondCounts should be derived
 #' @return no object is returned, only a new file is created in the output directory
 #' @import GGIR
@@ -40,7 +40,7 @@ GGIRshiny = function(rawaccdir, outputdir, sleepdiary = c(), configfile = c(),
   if (.Platform$OS.type == "windows") {
     logFile = paste0(outputdir, "/GGIR.log")
     fileConn = file(logFile)
-    writeLines(c("Hello, this Shiny app is primarily designed for Linux.",
+    writeLines(c("Hello, this Shiny app is primarily designed for Unix.",
                  "In Windows OS live progress of the analysis can be followed in the RStudio console.",
                  "In Unix-like systems the progress would be shown here inside this window in the Shiny app."), fileConn)
     close(fileConn)

R/checkConfigFile.R

@@ -0,0 +1,50 @@
+#' Function to check that the config files have the expected format
+#'
+#' @param file data path to config file
+#' @param tool either PALMSpy, GGIR, or palmsplusr for now.
+#'
+#' @return message with the result of the check (either ok or the description of a problem)
+#' @export
+checkConfigFile = function(file=c(), tool=c()) {
+  # intialize check object, ok by default, it would be overwritten if 
+  # any problem is identified in the config file.
+  check = "ok"
+
+  # General check on file existence and extension 
+  if (!file.exists(file)) {
+    # stop("No config file found at ", path)
+    check = "No config file found at "
+  } else {
+    path_unlist = unlist(strsplit(x = file, split = ".", fixed = TRUE))
+    path_ext = tolower(path_unlist[length(path_unlist)])
+    if (tool %in% c("GGIR", "hbGPS", "palmsplusr") & path_ext != "csv") {
+      check = "The GGIR config file uploaded is not a csv file"
+    } else if (tool == "PALMSpy" & path_ext != "json") {
+      check = "The GGIR config file uploaded is not a json file"
+    }
+  }
+
+  # Specific file content check
+  if (check == "ok") {
+    if (tool == "PALMSpy") {
+      # TO BE DEVELOPED
+    } else if (tool == "GGIR" || tool == "hbGPS") {
+      # read config file if it exists and it is a csv file
+      params = read.csv(file = file)
+      # sanity check 2: colnames of config file ----
+      if (ncol(params) == 3 && tool != "palmsplusr") {
+        check_colnames = all.equal(colnames(params), c("argument", "value", "context"))
+      } else if (ncol(params) == 5 && tool == "palmsplusr") {
+        check_colnames = all.equal(colnames(params), c("context", "name", "formula", "domain_field", "after_conversion"))
+      } else {
+        check_colnames = FALSE
+      }
+      if (check_colnames == FALSE) {
+        # this automatically also checks that csv is separated by commas, as
+        # it would generate just one column (e.g., "argument;value;context") otherwise
+        check = paste0("The csv file uploaded is not a ", tool, " config file")
+      } 
+    }
+  }
+  return(check)
+}

R/check_and_clean_palms_data.R

@@ -2,10 +2,11 @@
 #'
 #' @param palms_to_clean palms_to_clean
 #' @param country_name country_name
+#' @param outputdir outputdir
 #' @return palms_to_clean_lower object
 #' @import dplyr
 #' @export
-check_and_clean_palms_data <- function(palms_to_clean, country_name){
+check_and_clean_palms_data <- function(palms_to_clean, country_name, outputdir = NULL){
   dif = datetime = identifier = tripnumber = triptype = tt1 = tt4 = NULL
   miss_end = miss_start = multi_start = multi_end = error = NULL
   # Create error list -------------------------------------------------------
@@ -40,7 +41,7 @@ check_and_clean_palms_data <- function(palms_to_clean, country_name){
 
   # Saving the new 'clean'  dataset  - %>% ---------------------------------------
   # write_csv(palms, str_replace(link_to_csv, pattern = '.csv', '_cleaned.csv'), na = "") 
-  write_csv(error_list, paste(country_name,"error_list.csv", sep = "_"))
+  data.table::fwrite(error_list, paste(outputdir, country_name,"error_list.csv", sep = "_"))
 
   return(palms_to_clean_lower)
 }

R/check_params.R

@@ -86,7 +86,19 @@ check_params = function(params = c(), tool = c()) {
       }
     }
   }
-
+  # Check parameters which value should be timeformat
+  seti = which(params$class == "timeformat")
+  if (length(seti) > 0) {
+    for (i in seti) {
+      stmp = unlist(strsplit(params$value[i], "%"))
+      stmp = stmp[which(stmp != "")]
+      if (length(stmp) != 6) {
+        blocked_params$name[cnt] = rowNames[i]
+        blocked_params$error[cnt] = "is not a valid R time format specification."
+        cnt = cnt + 1
+      }
+    }
+  }
   # Create messages
   if (cnt <= N) {
     blocked_params = blocked_params[-c(cnt:N),]

R/hbGPS_shiny.R

@@ -0,0 +1,62 @@
+#' hbGPS_shiny
+#'
+#' @param ggiroutdir GGIR time series output directory
+#' @param gpsdir Path to GPS directory
+#' @param outputdir Path to output directory
+#' @param dataset_name Name of dataset
+#' @param configfile Path to configuration file
+#' @return no object is returned, only a new file is created in the output directory
+#' @import hbGPS
+#' @export
+
+hbGPS_shiny = function(ggiroutdir = NULL, gpsdir = NULL, outputdir = NULL,
+                       dataset_name = "", configfile = c()) {
+  # create R script with the code to run the data analysis via a command line call
+  # in this way turning off or restarting the app will not kill the data analysis
+  fileConn <- file(paste0(outputdir, "/hbgps_cmdline.R"))
+  writeLines(c("#!/usr/bin/env Rscript",
+               "args = commandArgs(trailingOnly = TRUE)",
+               "if (length(args) < 4) {",
+               "stop(\"At least four arguments are expected\", call. = FALSE)",
+               "}",
+               "hbGPS::hbGPS(gps_file = args[1], ",
+                     "GGIRpath = args[2],",
+                     "outputDir = args[3],",
+                     "configFile = args[4])"
+               ),
+             fileConn)
+  close(fileConn)
+  if (.Platform$OS.type == "windows") {
+    logFile = paste0(outputdir, "/hbGPS.log")
+    fileConn = file(logFile)
+    writeLines(c("Hello, this Shiny app is primarily designed for Unix.",
+                 "In Windows OS live progress of the analysis can be followed in the RStudio console.",
+                 "In Unix-like systems the progress would be shown here inside this window in the Shiny app."), fileConn)
+    close(fileConn)
+
+    basecommand = paste0(outputdir, "/hbgps_cmdline.R ",
+                         gpsdir, " ",
+                         ggiroutdir, " ",
+                         outputdir, " ",
+                         configfile)
+    system2(command = "Rscript", args = basecommand,
+            stdout = "",
+            stderr = "", wait = TRUE)
+    write.table(x = paste0(""),
+                file = paste0(outputdir, "/hbGPS.log"))
+
+    fileConn = file(logFile)
+    writeLines(c(""), fileConn)
+    close(fileConn)
+
+  } else {
+    basecommand = paste0("cd ", outputdir, " ; nohup Rscript hbgps_cmdline.R ",
+                         gpsdir, " ",
+                         ggiroutdir, " ",
+                         outputdir, " ",
+                         configfile, " > ", outputdir, "/hbGPS.log 2>&1 &")
+    system2(command = "cd", args = gsub(pattern = "cd ", replacement = "", x = basecommand),
+            stdout = "", stderr = "", wait = TRUE)
+  }
+
+}

R/hbt_build_days.R

@@ -64,9 +64,10 @@ hbt_build_days <- function(data = NULL, verbose = TRUE,
     mutate_if(is.logical, as.integer)
 
   fields <- palmsplus_fields %>% filter(domain_field == TRUE) %>% pull(name)
-  data <- data %>%
+
+    data <- data %>%
     st_set_geometry(NULL) %>%
-    dplyr::select(identifier, datetime, domain_names, all_of(fields)) %>%
+    dplyr::select(identifier, datetime, any_of(domain_names), all_of(fields)) %>%
     mutate(duration = 1) %>%
     mutate_at(vars(-identifier,-datetime), ~ . * palms_epoch(data) / 60) %>%
     group_by(identifier, date = as.Date(datetime)) %>%

R/identify_tools.R

@@ -11,27 +11,36 @@
 # S4 class needs to be defined outside function
 setClass(Class = "toolio", slots = list(input = "character", output = "character", usecases = "character"))
 
-identify_tools = function(datatypes = c("AccRaw", "ACount", "GPS", "GIS", "PALMSpy_out"),
+identify_tools = function(datatypes = c("AccRaw", "ACount", "GPS", "GIS", 
+                                        "PALMSpy_out", "GGIR_out", "hbGPS_out"),
                           goals = c("PA", "QC", "Trips", "Environment"),
-                          available_tools = c("GGIR", "PALMSpy", "palmsplusr", "Counts")) {
+                          available_tools = c("GGIR", "PALMSpy", "palmsplusr", "CountConverter", "hbGPS")) {
   iotools = list(GGIR = new("toolio",
                             input = "AccRaw",
                             output = c("GGIR_out", "ACount"),
                             usecases = c("PA", "QC", "Trips", "Environment")), 
                  PALMSpy = new("toolio",
-                               input = c("ACount","GPS"),
+                               input = c("ACount", "GPS"),
                                output = c("PALMSpy_out"),
                                usecases = c("Trips", "QC", "Environment")),
-                 palmsplusr = new("toolio",
-                                  input = c("PALMSpy_out", "GIS"),
-                                  output = c("palmsplusr_out"),
-                                  usecases = c("Environment", "QC")),
-                 Counts = new("toolio",
-                              input = "AccRaw",
-                              output = c("Counts_out"),
-                              usecases = c("PA", "Trips", "QC", "Environment")))
+                 palmsplusr = new("toolio", # palmsplusr based on PALMSpy output
+                                   input = c("PALMSpy_out", "GIS"),
+                                   output = c("palmsplusr_out"),
+                                   usecases = c("Environment", "QC")),
+                 palmsplusr = new("toolio", # palmsplusr based on hbGPS output
+                                   input = c("hbGPS_out", "GIS"),
+                                   output = c("palmsplusr_out"),
+                                   usecases = c("Environment", "QC")),
+                 CountConverter = new("toolio",
+                                      input = "AccRaw",
+                                      output = c("Counts_out"),
+                                      usecases = c("PA", "Trips", "QC", "Environment")),
+                 hbGPS = new("toolio",
+                             input = c("GGIR_out","GPS"),
+                             output = c("hbGPS_out"),
+                             usecases = c("Trips", "QC", "Environment")))
   iotools = iotools[which(names(iotools) %in% available_tools)] # only look at available tools
-  allgoals = tools_needed = outputs = c()
+  tools_needed = outputs = c()
   # loop over tools and select the ones that generate the output users needs and is able to generate
   for (j in 1:length(available_tools)) { # assumption is that pipeline is never longer then number of tools
     for (i in 1:length(iotools)) {
@@ -46,10 +55,10 @@ identify_tools = function(datatypes = c("AccRaw", "ACount", "GPS", "GIS", "PALMS
   if ("AccRaw" %in% datatypes == FALSE & "GGIR" %in% tools_needed) {
     tools_needed = tools_needed[-which(tools_needed == "GGIR")]
   }
-  if ("Counts" %in% tools_needed) {
+  if ("CountConverter" %in% tools_needed) {
     if ("ACount" %in% datatypes == TRUE | # No need to estimate counts if they already exist
         ("AccRaw" %in% datatypes == TRUE & "GPS" %in% datatypes == FALSE)) { # No need to estimate counts if there is no GPS data
-      tools_needed = tools_needed[-which(tools_needed == "Counts")]
+      tools_needed = tools_needed[-which(tools_needed == "CountConverter")]
     }
   }
   invisible(list(tools_needed = tools_needed, iotools = iotools[which(names(iotools) %in% tools_needed)]))