Fixes #381

DOI-USGS · Oct 25, 2024 · de4ea92 · de4ea92
1 parent 2ecb8ee
commit de4ea92
Show file tree

Hide file tree

Showing 4 changed files with 65 additions and 49 deletions.
diff --git a/R/compressData.r b/R/compressData.r
@@ -51,24 +51,28 @@ compressData <- function(data, verbose = TRUE){
     i <- i + 1
   }
 
-  names(data) <- c('dateTime', 'code', 'value')
-  returnDataFrame <- as.data.frame(matrix(ncol=3,nrow=nrow(data)))
-  names(returnDataFrame) <- c('dateTime', 'ConcLow', 'ConcHigh')
+  names(data)[1:3] <- c('dateTime', 'code', 'value')
 
   data$dateTime <- as.character(data$dateTime)
   if(dateFormatCheck(data$dateTime)){
-    returnDataFrame$dateTime <- as.Date(data$dateTime)  
+    data$dateTime <- as.Date(data$dateTime)  
   } else {
     data$dateTime <- as.Date(data$dateTime,format="%m/%d/%Y")
-    returnDataFrame$dateTime <- as.Date(data$dateTime,format="%m/%d/%Y")
   }
-  returnDataFrame$ConcLow <- as.numeric(lowConcentration)
-  returnDataFrame$ConcHigh <- as.numeric(highConcentration)
-  Uncen1<-ifelse(returnDataFrame$ConcLow==returnDataFrame$ConcHigh,1,0)
-  returnDataFrame$Uncen<-ifelse(is.na(returnDataFrame$ConcLow)|is.na(returnDataFrame$ConcHigh),0,Uncen1)
+
+
+  data$ConcLow <- as.numeric(lowConcentration)
+  data$ConcHigh <- as.numeric(highConcentration)
+  Uncen1<-ifelse(data$ConcLow==data$ConcHigh,1,0)
+  data$Uncen<-ifelse(is.na(data$ConcLow)|is.na(data$ConcHigh),0,Uncen1)
 
-  flaggedData1 <- returnDataFrame[(returnDataFrame$ConcLow == 0 & returnDataFrame$ConcHigh == 0),]
-  returnDataFrame <- returnDataFrame[!(returnDataFrame$ConcLow == 0 & returnDataFrame$ConcHigh == 0),]
+  return(data)
+}
+
+
+remove_zeros <- function(data, verbose){
+  flaggedData1 <- data[(data$ConcLow == 0 & data$ConcHigh == 0),]
+  data <- data[!(data$ConcLow == 0 & data$ConcHigh == 0),]
 
   if (nrow(flaggedData1) > 0){
     WarningMessage <- paste("Deleted", nrow(flaggedData1), "rows of data because concentration was reported as 0.0, the program is unable to interpret that result and is therefore deleting it.")    
@@ -78,6 +82,5 @@ compressData <- function(data, verbose = TRUE){
       print(flaggedData1)
     }
   }
-
-  return(returnDataFrame)
+  return(data)
 }
diff --git a/R/populateSampleColumns.r b/R/populateSampleColumns.r
@@ -14,8 +14,8 @@
 #' dataInput <- data.frame(dateTime, ConcLow, ConcHigh, Uncen, stringsAsFactors=FALSE)
 #' Sample <- populateSampleColumns(dataInput)
 populateSampleColumns <- function(rawData){  # rawData is a dataframe with dateTime, ConcLow, ConcHigh, Uncen
-  Sample <- as.data.frame(matrix(ncol=3,nrow=length(rawData$dateTime)))
-  colnames(Sample) <- c('Date', 'ConcLow','ConcHigh')
+  Sample <- rawData
+
   Sample$Date <- rawData$dateTime
   Sample$ConcLow <- rawData$ConcLow
   Sample$ConcHigh <- rawData$ConcHigh
@@ -29,6 +29,7 @@ populateSampleColumns <- function(rawData){  # rawData is a dataframe with dateT
 
   Sample$SinDY <- sin(2*pi*Sample$DecYear)
   Sample$CosDY <- cos(2*pi*Sample$DecYear)
-  # Sample2 <- subset(Sample, (!is.na(Sample$ConcHigh)))  # Was just ConcHigh.....
+
+  Sample <- Sample[,names(Sample)[!names(Sample) %in% c("code", "value")]]
   return (Sample)  
 }
diff --git a/R/readUserSample.r b/R/readUserSample.r
@@ -54,6 +54,12 @@ readUserSample <- function (filePath, fileName,
                            format = format)
 
   compressedData <- compressData(data, verbose = verbose)
+  compressedData <- remove_zeros(compressedData, verbose = verbose)
   Sample <- populateSampleColumns(compressedData)
+  orig_Sample <- c("Date", "ConcLow", "ConcHigh", "Uncen", "ConcAve",
+                   "Julian", "Month", "Day", "DecYear", "waterYear", "MonthSeq",
+                   "SinDY", "CosDY")
+  Sample <- Sample[, c(orig_Sample, names(Sample)[!names(Sample) %in% orig_Sample])]
+  Sample <- Sample[order(Sample$Date), ]
   return(Sample)
 }
diff --git a/R/readWQPSample.R b/R/readWQPSample.R
@@ -44,7 +44,14 @@ readWQPSample <- function(siteNumber,
                           endDate = "",
                           verbose = TRUE){
 
-
+  extra_cols <- c("ActivityStartDateTime",
+                  "USGSPCode",
+                  "ActivityMediaSubdivisionName",
+                  "ActivityMediaName",
+                  "ResultSampleFractionText",
+                  "ResultStatusIdentifier",
+                  "ResultValueTypeName")
+
   if(packageVersion("dataRetrieval") >= "2.7.17"){
     data <- suppressMessages(dataRetrieval::readWQPqw(siteNumbers = siteNumber,
                                      parameterCd = characteristicName,
@@ -54,29 +61,23 @@ readWQPSample <- function(siteNumber,
                                      legacy = FALSE)  )
 
     conversion_names <- data.frame(legacy_names = c("ResultDetectionConditionText",
-                      "ResultMeasureValue",
-                      "DetectionQuantitationLimitMeasure.MeasureValue",
-                      "CharacteristicName",
-                      "ActivityStartDate",
-                      "ActivityStartDateTime",
-                      "USGSPCode",
-                      "ActivityMediaSubdivisionName",
-                      "ActivityMediaName",
-                      "ResultSampleFractionText",
-                      "ResultStatusIdentifier",
-                      "ResultValueTypeName"),
-    new_names = c("Result_ResultDetectionCondition",
-                   "Result_Measure",
-                   "DetectionLimit_TypeA",
-                   "Result_Characteristic",
-                   "Activity_StartDate",
-                   "Activity_StartDateTime",
-                   "USGSpcode",
-                   "Activity_MediaSubdivisionName",
-                   "Activity_Media",
-                   "Result_SampleFraction",
-                   "Result_MeasureStatusIdentifier",
-                   "Result_MeasureType"))
+                                                    "ResultMeasureValue",
+                                                    "DetectionQuantitationLimitMeasure.MeasureValue",
+                                                    "CharacteristicName",
+                                                    "ActivityStartDate",
+                                                    extra_cols),
+                                   new_names = c("Result_ResultDetectionCondition",
+                                                 "Result_Measure",
+                                                 "DetectionLimit_MeasureA",
+                                                 "Result_Characteristic",
+                                                 "Activity_StartDate",
+                                                 "Activity_StartDateTime",
+                                                 "USGSpcode",
+                                                 "Activity_MediaSubdivisionName",
+                                                 "Activity_Media",
+                                                 "Result_SampleFraction",
+                                                 "Result_MeasureStatusIdentifier",
+                                                 "Result_MeasureType"))
 
     for(i in seq_len(nrow(conversion_names))){
       names(data)[which(names(data) == conversion_names$new_names[i])] <- conversion_names$legacy_names[i]
@@ -97,15 +98,20 @@ readWQPSample <- function(siteNumber,
   if(nrow(data) > 0){
 
     data <- processQWData(data)
+    first_three <- c("dateTime",
+                     "qualifier",
+                     "value")
+    compressedData <- compressData(data[, first_three],
+                                   verbose = verbose)
+    combined_data <- cbind(compressedData, data[, names(data)[!names(data) %in% first_three]])
+    combined_data <- remove_zeros(combined_data, verbose = verbose)
 
-    compressedData <- compressData(data[, c("dateTime",
-                                            "qualifier",
-                                            "value")],
-                                   verbose=verbose)
-    Sample <- populateSampleColumns(compressedData)
-    Sample <- cbind(Sample, data[, names(data)[!names(data) %in% c("dateTime",
-                                                                  "qualifier",
-                                                                  "value")]])
+    Sample <- populateSampleColumns(combined_data)
+    orig_Sample <- c("Date", "ConcLow", "ConcHigh", "Uncen", "ConcAve",
+                     "Julian", "Month", "Day", "DecYear", "waterYear", "MonthSeq",
+                     "SinDY", "CosDY")
+    Sample <- Sample[, c(orig_Sample, names(Sample)[!names(Sample) %in% orig_Sample])]
+    Sample <- Sample[order(Sample$Date), ]
   } else {
     Sample <- data.frame(Date=as.Date(character()),
                          ConcLow=numeric(), 
@@ -121,6 +127,6 @@ readWQPSample <- function(siteNumber,
                          CosDY=numeric(),
                          stringsAsFactors=FALSE)
   }
-  Sample <- Sample[order(Sample$Date), ]
+
   return(Sample)
 }