From 1edc232afc5b622e2a1905a20e5a5b2bec4f2047 Mon Sep 17 00:00:00 2001
From: Sebastian Gibb <mail@sebastiangibb.de>
Date: Fri, 4 Aug 2023 22:31:32 +0200
Subject: [PATCH] refactor: countElements returns NA if an element is invalid
 instead of dropping it

---
 R/adducts.R                       |  5 +++-
 R/chemFormula.R                   | 50 +++++++++++++++++--------------
 tests/testthat/test_chemFormula.R |  9 ++++++
 3 files changed, 41 insertions(+), 23 deletions(-)

diff --git a/R/adducts.R b/R/adducts.R
index 65e4694..fc0c1b7 100644
--- a/R/adducts.R
+++ b/R/adducts.R
@@ -233,6 +233,9 @@ adductFormula <- function(formulas, adduct = "[M+H]+", standardize = TRUE) {
     ionMatrix <- lapply(formulas, function(formula, adduct) {
         formulaAdduct <- apply(adduct, 1, function(x) {
             current_f <- formula
+
+            if (is.null(current_f)) return(NULL)
+
             multiplicity <- round(as.numeric(x["mass_multi"]) *
                                       as.numeric(x["charge"]))
             if (multiplicity != 1) {
@@ -264,7 +267,7 @@ adductFormula <- function(formulas, adduct = "[M+H]+", standardize = TRUE) {
         return(formulaAdduct)
     }, adduct = adduct)
     ionMatrix <- do.call(rbind, ionMatrix)
-    rownames(ionMatrix) <- formulas
+    rownames(ionMatrix) <- formulas[as.logical(lengths(formulas))]
     colnames(ionMatrix) <- rownames(adduct)
     return(ionMatrix)
 }
diff --git a/R/chemFormula.R b/R/chemFormula.R
index 948f859..8936fba 100644
--- a/R/chemFormula.R
+++ b/R/chemFormula.R
@@ -25,25 +25,25 @@ countElements <- function(x) {
         "(?<Element>",
             paste0("[0-9]*",
                 c(
-                    "[A][cglmrstu]|",
-                    "[B][aehikr]?|",
-                    "[C][adeflmnorsu]?|",
-                    "[D][bsy]|",
-                    "[E][rsu]|",
-                    "[F][elmr]?|",
-                    "[G][ade]|",
-                    "[H][efgos]?|",
-                    "[I][nr]?|",
-                    "[K][r]?|",
-                    "[L][airuv]|",
-                    "[M][cdgnot]|",
-                    "[N][abdehiop]?|",
-                    "[O][gs]?|",
-                    "[P][abdmortu]?|",
-                    "[R][abefghnu]|",
-                    "[S][bcegimnr]?|",
-                    "[T][abcehilms]|",
-                    "[U]|[V]|[W]|[X][e]|[Y][b]?|[Z][nr]"
+                    "A[cglmrstu]|",
+                    "B[aehikr]?|",
+                    "C[adeflmnorsu]?|",
+                    "D[bsy]|",
+                    "E[rsu]|",
+                    "F[elmr]?|",
+                    "G[ade]|",
+                    "H[efgos]?|",
+                    "I[nr]?|",
+                    "K[r]?|",
+                    "L[airuv]|",
+                    "M[cdgnot]|",
+                    "N[abdehiop]?|",
+                    "O[gs]?|",
+                    "P[abdmortu]?|",
+                    "R[abefghnu]|",
+                    "S[bcegimnr]?|",
+                    "T[abcehilms]|",
+                    "U|V|W|Xe|Yb?|Z[nr]"
                 ),
                 collapse = ""
             ),
@@ -58,6 +58,10 @@ countElements <- function(x) {
 
         if (is.na(xx))
             return(NA_integer_)
+        if (sum(attr(rr, "match.length")) != nchar(gsub("\\[|\\]", "", xx))) {
+            warning("The given formula '", xx, "' contains invalid symbols.")
+            return(NA_integer_)
+        }
 
         start <- attr(rr, "capture.start")
         end <- start + attr(rr, "capture.length") - 1L
@@ -175,10 +179,10 @@ standardizeFormula <- function(x) {
 #' @examples
 #' .sum_elements(c(H = 6, C = 3, O = 6, C = 3, H = 6))
 .sum_elements <- function(x) {
-    if (!is.character(names(x)))
-        stop("element names missing")
     if (anyNA(x))
         return(NA_integer_)
+    if (!is.character(names(x)))
+        stop("element names missing")
     unlist(lapply(split(x, names(x)), sum))
 }
 
@@ -336,7 +340,9 @@ calculateMass <- function(x) {
         stop("x must be either a character or a list with element counts.")
     vapply(x, function(z) {
         isotopes <- names(z)
-        if (!length(z) || !all(isotopes %in% names(.ISOTOPES))) {
+        if (!length(z) ||
+            is.null(isotopes) ||
+            !all(isotopes %in% names(.ISOTOPES))) {
             message("not for all isotopes a mass is found")
             return(NA_real_)
         }
diff --git a/tests/testthat/test_chemFormula.R b/tests/testthat/test_chemFormula.R
index c6abbb0..f6349e3 100644
--- a/tests/testthat/test_chemFormula.R
+++ b/tests/testthat/test_chemFormula.R
@@ -53,6 +53,15 @@ test_that("countElements", {
             ), names = c("C6H12O6", NA, "H2O")
         )
     )
+    expect_warning(countElements("Foo"), "invalid")
+    expect_identical(
+        suppressWarnings(countElements(c("C6H12O6", "Foo", "H2O"))),
+        list(
+            "C6H12O6" = c(C = 6L, H = 12L, O = 6L),
+            Foo = NA_integer_,
+            "H2O" = c(H = 2L, O = 1L)
+        )
+    )
 
     ## heavy isotopes
     expect_identical(