updated pc_synonyms cts_convert, and cir_query to use match i…

…nstead of `choices` ropensci#263

NEWS.md

@@ -11,6 +11,7 @@
 * The `"type"` argument in `ci_query()` and `aw_query()` has been changed to `"from"` for consistency with other functions
 * `fn_percept()` and `cts_compinfo()` now have `"query"` and `"from"` arguments for consistency with other functions
 * Possible values for `"from"` have been made more consistent across functions
+* `pc_synonyms()` and `cir_query()` have been changed to use the `match` argument instead of `choices` for consistency with other functions
 
 # webchem 1.0.0
 

R/cir.R

@@ -12,13 +12,14 @@
 #' @param resolver character; what resolver should be used? If NULL (default)
 #'  the identifier type is detected and the different resolvers are used in turn.
 #'  See details for possible resolvers.
-#' @param first deprecated, use choices = 1 to return only the first result
-#' @param choices if \code{choices = 1}, returns only the first result. To get a
-#' number of results to choose from in an interactive menu, provide the number
-#' of choices you want or "all" to choose from all synonyms.
+#' @param match character; How should multiple hits be handled? \code{"all"}
+#' returns all matches, \code{"first"} returns only the first result,
+#' \code{"ask"} enters an interactive mode and the user is asked for input,
+#' \code{"na"} returns \code{NA} if multiple hits are found.
+#' @param choices deprecated.  Use the \code{match} argument instead.
 #' @param verbose logical; should a verbose output be printed on the console?
 #' @param ... currently not used.
-#' @return A list of character vectors. If first = TRUE a vector.
+#' @return A list of character vectors.
 #' @details
 #'  CIR can resolve can be of the following \code{identifier}: Chemical Names,
 #'  IUPAC names,
@@ -61,14 +62,14 @@
 #'      \item \code{'protonable_group_count'} (Number of protonable groups).
 #'  }
 #'
-#'  CIR first tries to determine the indetifier type submitted and then
+#'  CIR first tries to determine the identifier type submitted and then
 #'  uses 'resolvers' to look up the data.
 #'  If no \code{resolver} is supplied, CIR tries different resolvers in
 #'  turn till a hit is found.
 #'  E.g. for names CIR tries first to look up in OPSIN and if this fails
 #'  the local name index of CIR.
 #'  However, it can be also specified which resolvers to use
-#'  (if you know e.g. know your indentifier type)
+#'  (if you know e.g. know your identifier type)
 #'  Possible \code{resolvers} are:
 #'  \itemize{
 #'    \item \code{'name_by_cir'} (Lookup in name index of CIR),
@@ -100,24 +101,29 @@
 #' @examples
 #' \donttest{
 #' # might fail if API is not available
-#' cir_query('Triclosan', 'cas')
-#' cir_query("3380-34-5", 'cas', first = TRUE)
-#' cir_query("3380-34-5", 'cas', resolver = 'cas_number')
-#' cir_query("3380-34-5", 'smiles')
-#' cir_query('Triclosan', 'mw')
+#' cir_query("Triclosan", "cas")
+#' cir_query("3380-34-5", "cas", match = "first")
+#' cir_query("3380-34-5", "cas", resolver = "cas_number")
+#' cir_query("3380-34-5", "smiles")
+#' cir_query("Triclosan", "mw")
 #'
 #' # multiple inputs
-#' comp <- c('Triclosan', 'Aspirin')
-#' cir_query(comp, 'cas', first = TRUE)
+#' comp <- c("Triclosan", "Aspirin")
+#' cir_query(comp, "cas", match = "first")
 #'
 #'}
 #' @export
-cir_query <- function(identifier, representation = 'smiles', resolver = NULL,
-                      first = FALSE, choices = NULL, verbose = TRUE, ...){
-  if (first == TRUE) {
-    message("`first` is deprecated.  Using `choices = 1` instead.")
-    choices = 1
+cir_query <- function(identifier, representation = "smiles",
+                      resolver = NULL,
+                      first = FALSE,
+                      match = c("all", "first", "ask", "na"),
+                      verbose = TRUE,
+                      choices = NULL,
+                      ...){
+  if (!missing("choices")) {
+    stop("`choices` is deprecated.  Use `match` instead.")
   }
+  match <- match.arg(match)
   foo <- function(identifier, representation, resolver, first, verbose) {
     if (is.na(identifier)) {
       return(NA)
@@ -144,9 +150,7 @@ cir_query <- function(identifier, representation = 'smiles', resolver = NULL,
         message('No representation found... Returning NA.')
         return(NA)
       }
-      # if (first)
-      #   out <- out[1]
-      out <- chooser(out, choices)
+      out <- matcher(out, query = identifier, match = match, verbose = verbose)
       # convert to numeric
       if (representation %in% c('mw', 'monoisotopic_mass', 'h_bond_donor_count',
                                 'h_bond_acceptor_count', 'h_bond_center_count',
@@ -162,8 +166,5 @@ cir_query <- function(identifier, representation = 'smiles', resolver = NULL,
   out <- lapply(identifier, foo, representation = representation,
                 resolver = resolver, first = first, verbose = verbose)
   out <- setNames(out, identifier)
-  # if (first)
-  if(!is.null(choices))
-    out <- unlist(out)
   return(out)
 }

R/cts.R

@@ -71,8 +71,11 @@ cts_compinfo <- function(query, from = "inchikey", verbose = TRUE, inchikey){
 #' @param from character; type of query ID, e.g. \code{'Chemical Name'} , \code{'InChIKey'},
 #'  \code{'PubChem CID'}, \code{'ChemSpider'}, \code{'CAS'}.
 #' @param to character; type to convert to.
-#' @param first deprecated.  Use choices = 1 instead.
-#' @param choices to return only the first result, use 'choices = 1'.  To choose a result from an interactive menu, provide a number of choices to choose from or "all".
+#' @param match character; How should multiple hits be handled? \code{"all"}
+#' returns all matches, \code{"first"} returns only the first result,
+#' \code{"ask"} enters an interactive mode and the user is asked for input,
+#' \code{"na"} returns \code{NA} if multiple hits are found.
+#' @param choices deprecated.  Use the \code{match} argument instead.
 #' @param verbose logical; should a verbose output be printed on the console?
 #' @param ... currently not used.
 #' @return a list of character vectors or if \code{choices} is used, then a single named vector.
@@ -95,15 +98,22 @@ cts_compinfo <- function(query, from = "inchikey", verbose = TRUE, inchikey){
 #' comp <- c("triclosan", "hexane")
 #' cts_convert(comp, "Chemical Name", "cas")
 #' }
-cts_convert <- function(query, from, to, first = FALSE, choices = NULL, verbose = TRUE, ...){
-  if(!missing("first"))
-    stop('"first" is deprecated.  Use "choices = 1" instead.')
+cts_convert <- function(query,
+                        from,
+                        to,
+                        match = c("all", "first", "ask", "na"),
+                        verbose = TRUE,
+                        choices = NULL,
+                        ...){
+  if(!missing("choices"))
+    stop('"choices" is deprecated.  Use "match" instead.')
   if (length(from) > 1 | length(to) > 1) {
     stop('Cannot handle multiple input or output types.  Please provide only one argument for `from` and `to`.')
   }
 
   from <-  match.arg(tolower(from), c(cts_from(), "name"))
   to <-  match.arg(tolower(to), c(cts_to(), "name"))
+  match <- match.arg(match)
 
   if (from == "name") {
     from <- "chemical name"
@@ -113,7 +123,7 @@ cts_convert <- function(query, from, to, first = FALSE, choices = NULL, verbose
     to <- "chemical name"
   }
 
-  foo <- function(query, from, to , first, verbose){
+  foo <- function(query, from, to, first, verbose){
     if (is.na(query)) return(NA)
     baseurl <- "http://cts.fiehnlab.ucdavis.edu/service/convert"
     qurl <- paste0(baseurl, '/', from, '/', to, '/', query)
@@ -131,13 +141,12 @@ cts_convert <- function(query, from, to, first = FALSE, choices = NULL, verbose
         return(NA)
     }
     out <- out$result[[1]]
-    out <- chooser(out, choices)
+    out <- matcher(out, match = match, query = query, verbose = verbose)
     return(out)
   }
   out <- lapply(query, foo, from = from, to = to, first = first, verbose = verbose)
   out <- setNames(out, query)
-  if(!is.null(choices))
-    out <- unlist(out)
+
   return(out)
 }
 

R/pubchem.R

@@ -375,15 +375,16 @@ pc_prop <- function(cid, properties = NULL, verbose = TRUE, ...) {
 #' @param query character; search term.
 #' @param from character; type of input, can be one of "name" (default), "cid",
 #'     "sid", "aid", "smiles", "inchi", "inchikey"
-#' @param interactive deprecated.  Use the \code{choices} argument instead
-#' @param choices to get only the first synonym, use \code{choices = 1}, to get
-#' a number of synonyms to choose from in an interactive menu, provide the
-#' number of choices you want or "all" to choose from all synonyms.
+#' @param match character; How should multiple hits be handled? \code{"all"}
+#' returns all matches, \code{"first"} returns only the first result,
+#' \code{"ask"} enters an interactive mode and the user is asked for input,
+#' \code{"na"} returns \code{NA} if multiple hits are found.
+#' @param choices deprecated.  Use the \code{match} argument instead.
 #' @param verbose logical; should a verbose output be printed on the console?
-#' @param arg character; optinal arguments like "name_type=word" to match
+#' @param arg character; optional arguments like "name_type=word" to match
 #' individual words.
-#' @param ... optional arguments
-#' @return a list of character vectors (one per query). If \code{choices} is used, a single named vector is returned instead.
+#' @param ... currently unused
+#' @return a named list.
 #'
 #' @references Wang, Y., J. Xiao, T. O. Suzek, et al. 2009 PubChem: A Public
 #' Information System for
@@ -411,16 +412,20 @@ pc_prop <- function(cid, properties = NULL, verbose = TRUE, ...) {
 #' pc_synonyms("Aspirin")
 #' pc_synonyms(c("Aspirin", "Triclosan"))
 #' pc_synonyms(5564, from = "cid")
-#' pc_synonyms(c("Aspirin", "Triclosan"), choices = 10)
+#' pc_synonyms(c("Aspirin", "Triclosan"), match = "ask")
 #' }
-pc_synonyms <- function(query, from = c("name", "cid", "sid", "aid", "smiles", "inchi", "inchikey"), choices = NULL, verbose = TRUE,
-                        arg = NULL, interactive = 0, ...) {
+pc_synonyms <- function(query,
+                        from = c("name", "cid", "sid", "aid", "smiles", "inchi", "inchikey"),
+                        match = c("all", "first", "ask", "na"),
+                        verbose = TRUE,
+                        arg = NULL, choices = NULL, ...) {
   # from can be cid | name | smiles | inchi | sdf | inchikey | formula
   # query <- c("Aspirin")
   # from = "name"
   from <- match.arg(from)
-  if (!missing("interactive"))
-    stop("'interactive' is deprecated. Use 'choices' instead.")
+  match <- match.arg(match)
+  if (!missing("choices"))
+    stop("'choices' is deprecated. Use 'match' instead.")
   foo <- function(query, from, verbose, ...) {
     if (is.na(query)) return(NA)
     prolog <- "https://pubchem.ncbi.nlm.nih.gov/rest/pug"
@@ -444,10 +449,10 @@ pc_synonyms <- function(query, from = c("name", "cid", "sid", "aid", "smiles", "
       warning(cont$Fault$Details, ". Returning NA.")
       return(NA)
     }
-    out <- unlist(cont)
+    out <- unlist(cont)[-1] #first result is always an ID number
     names(out) <- NULL
 
-    out <- chooser(out, choices)
+    out <- matcher(out, query = query, match = match, verbose = verbose)
 
   }
   out <- lapply(query, foo, from = from, verbose = verbose)

R/utils.R

@@ -436,7 +436,7 @@ chooser <- function(x, choices){
 #' @param x a vector
 #' @param query what the query was, only used if match = "best"
 #' @param result what the result of the query was, only used if match = "best
-#' @param match haracter; How should multiple hits be handeled? "all" returns
+#' @param match character; How should multiple hits be handled? "all" returns
 #' all matched IDs, "first" only the first match, "best" the best matching (by
 #' name) ID, "ask" is a interactive mode and the user is asked for input, "na"
 #' @param verbose print messages?

tests/testthat/test-cir.R

@@ -8,12 +8,12 @@ test_that("cir_query()", {
   expect_equal(cir_query("3380-34-5", 'stdinchikey', resolver = 'cas_number', verbose = FALSE)[[1]],
             "InChIKey=XEFQLINVKFYRCS-UHFFFAOYSA-N")
   expect_true(length(cir_query('Triclosan', 'cas', verbose = FALSE)[[1]]) > 1)
-  expect_message(cir_query("acetic acid", "mw", first = TRUE))
-  expect_length(cir_query('Triclosan', 'cas', choices = 1, verbose = FALSE)[[1]], 1)
+  expect_message(cir_query("acetic acid", "mw", match = "first"))
+  expect_length(cir_query('Triclosan', 'cas', match = "first", verbose = FALSE)[[1]], 1)
   expect_length(cir_query(c('Triclosan', 'Aspirin'), 'cas', verbose = FALSE), 2)
 
-  skip("I have no clue why this one fails on R CMD check.  It works when run in the console!")
-  expect_equivalent(cir_query('acetic acid', 'mw', choices = 1), c(`acetic acid` = 60.0524))
+  # skip("I have no clue why this one fails on R CMD check.  It works when run in the console!")
+  expect_equivalent(cir_query('acetic acid', 'mw', match = "first"), c(`acetic acid` = 60.0524))
 
 })
 

tests/testthat/test-cts.R

@@ -25,25 +25,14 @@ test_that("cts_convert()", {
   expect_error(cts_convert(comp, c('Chemical Name', 'CAS'), 'CAS'))
   expect_error(cts_convert('Triclosan', 'CAS'))
   expect_true(is.na(suppressWarnings(cts_convert('xxxx', 'Chemical Name', 'inchikey'))[[1]]))
-  o1 <- cts_convert(comp, 'Chemical Name', 'inchikey', choices = 1, verbose = FALSE)
+  o1 <- cts_convert(comp, 'Chemical Name', 'inchikey', match = "first", verbose = FALSE)
   expect_length(o1, 2)
 
   expect_equal(o1[[1]], 'XEFQLINVKFYRCS-UHFFFAOYSA-N')
-  # cts_convert('acetic acid', 'Chemical Name', 'CAS', choices = 1)
   expect_equivalent(cts_convert(NA, from = "Chemical Name", to = "inchikey"), NA)
 })
 
 
-# # integration tests
-# test_that("cts_compinfo(cir_query())", {
-#   chk_cts()
-#   chk_cir()
-#   inchikey <- cir_query('Triclosan', representation = 'stdinchikey', verbose = FALSE)
-#   inchikey <- gsub('InChIKey=', '', inchikey)
-#   expect_equal(round(cts_compinfo(inchikey, verbose = FALSE)[[1]][["molweight"]], 3), 289.542)
-# })
-
-
 test_that("fromto", {
   skip_on_cran()
   skip_if_not(up, "CTS service down")

tests/testthat/test-pubchem.R

@@ -87,10 +87,10 @@ test_that("pc_synonyms", {
   skip_on_cran()
   skip_if_not(up, "PubChem service is down")
   expect_equivalent(pc_synonyms(NA), NA)
-  expect_equal(pc_synonyms("Triclosan")[[1]][1], "5564")
+  expect_equal(pc_synonyms("Acetyl Salicylic Acid")[[1]][1], "aspirin")
   expect_equal(length(pc_synonyms(c("Triclosan", "Aspirin"))), 2)
   expect_equal(pc_synonyms("BPGDAMSIGCZZLK-UHFFFAOYSA-N",
-                           from = "inchikey")[[1]][1], "12345")
+                           from = "inchikey")[[1]][1], "Methylene diacetate")
   expect_true(is.na(suppressWarnings(pc_synonyms("xxxx"))[[1]]))
 })