Merge pull request #335 from USEPA/162-update-tcplLoadData-to-support-chem-queries

162 update tcplLoadData to support chemical queries

Author: cthunes

Diff for: R/chemQueries.R

@@ -2,6 +2,18 @@
 # .ChemQ: Create tcplLoadChem query string 
 #-------------------------------------------------------------------------------
 
+#' @title Function to support queries by chemical
+#' @note This function is not exported and not intended to be used by the user.
+#' 
+#' @param field Character, the field to query on
+#' @param val Vector of values to subset on
+#' @param exact Logical, should chemical names be considered exact?
+#' 
+#' @description
+#' \code{.ChemQ} creates tcplLoadChem query string
+#' 
+#' @seealso \code{\link{tcplLoadData}}
+#' @seealso \code{\link{tcplLoadChem}}
 .ChemQ <- function(field, val, exact) {
 
   qformat <- 
@@ -29,11 +41,11 @@
       if (exact) {
         qformat <- paste(qformat, "chnm IN (%s);")
         val <- paste0("\"", val, "\"", collapse = ",")
-        qstring <- sprintf(qformat, val, val)
+        qstring <- sprintf(qformat, val)
       } else {
         qformat <- paste(qformat, "chnm RLIKE %s;")
         val <- paste0("\"", paste(val, collapse = "|"), "\"")
-        qstring <- sprintf(qformat, val, val)
+        qstring <- sprintf(qformat, val)
       }
     } else if (nfld == 'chem.only') {
       qstring <- "
@@ -64,6 +76,16 @@
 # .ChemListQ: Create tcplLoadChemList query string 
 #-------------------------------------------------------------------------------
 
+#' @title Function to support queries by chemical
+#' @note This function is not exported and not intended to be used by the user.
+#' 
+#' @param field Character, the field to query on
+#' @param val Vector of values to subset on
+#' 
+#' @description
+#' \code{.ChemListQ} creates tcplLoadChemList query string
+#' 
+#' @seealso \code{\link{tcplLoadChemList}}
 .ChemListQ <- function(field, val) {
 
   qformat <- "SELECT * FROM chemical_lists"
@@ -90,5 +112,4 @@
 
 #-------------------------------------------------------------------------------
 # END .ChemListQ
-#-------------------------------------------------------------------------------
-
+#-------------------------------------------------------------------------------

Diff for: R/tcplLoadChem.R

@@ -49,6 +49,8 @@
 tcplLoadChem <- function(field = NULL, val = NULL, exact = TRUE,
                          include.spid = TRUE) {
 
+  if (length(field) > 1) stop("'field' must be of length 1.")
+  
   if (getOption("TCPL_DRVR") == "API") {
     if (is.null(field) || tolower(field) != "spid") stop("When drvr option is set to 'API', only 'spid' is a valid 'field' value.")
     if (!exact) exact <- TRUE

Diff for: R/tcplLoadData.R

@@ -15,6 +15,8 @@
 #' the same order as 'fld'.
 #' @param add.fld Boolean if true we want to return 
 #' the additional parameters fit with tcplfit2
+#' @param exact Logical, passed to tcplLoadChem -- should chemical names be 
+#' considered exact?
 #' 
 #' @details
 #' The data type can be either 'mc' for mutliple concentration data, or 'sc'
@@ -71,6 +73,12 @@
 #' ## Load level 0 data where the well type is "t" and the concentration
 #' ## index is 3 or 4
 #' tcplLoadData(lvl = 1, fld = c("wllt", "cndx"), val = list("t", c(3:4)))
+#'
+#' ## Load level 4 data using a chemical name
+#' tcplLoadData(lvl = 4, fld = "chnm", val = "Bisphenol A")
+#'
+#' ## Load level 3 data using a partial chemical name
+#' tcplLoadData(lvl = 3, fld = "chnm", val = "phenol", exact = FALSE)
 #' }
 #' @return A data.table containing data for the given fields.
 #'
@@ -82,7 +90,7 @@
 #' @importFrom rlang exec sym
 #' @export
 
-tcplLoadData <- function(lvl, fld = NULL, val = NULL, type = "mc", add.fld = TRUE) {
+tcplLoadData <- function(lvl, fld = NULL, val = NULL, type = "mc", add.fld = TRUE, exact = TRUE) {
   #variable binding
   model <- model_param <- model_val <- NULL
   hit_param <- hit_val <- sc_vignette <- mc_vignette <- NULL
@@ -230,6 +238,12 @@ tcplLoadData <- function(lvl, fld = NULL, val = NULL, type = "mc", add.fld = TRU
     # add.fld is not possible if invitrodb version less than 4
     if (!check_tcpl_db_schema()) add.fld <- FALSE
 
+    if (any(fld %in% c("chid", "casn", "chnm", "dsstox_substance_id", "code"))) {
+      chem <- tcplLoadChem(field = fld, val = val, exact = exact)
+      fld <- "spid"
+      val <- chem$spid
+    }
+    
     table <- paste0(type, lvl)
     tbls_joins <- case_when(
       table == "sc0" ~ list(tbls = "sc0",