Terminal semicolon warning

Author: ms609

NEWS.md

@@ -8,23 +8,21 @@
   at exact thresholds (e.g. a split in 2 of 3 trees with `p = 2/3`). The
   majority threshold `p = 0.5` is unchanged (a split must occur in more than
   half the trees).
+- `ReeadCharacters()` no longer warns on a `STATELABELS` block with a terminal
+  semicolon.
 
 ## Performance 
 
 - Guarantee preorder return from `root_on_node()` to simplify `Consensus()`
-  internal pre-processing
+  internal pre-processing.
 - `Consensus()` and `SplitFrequency()` defer materialising a split's bit pattern
   until it is needed, so splits that never reach the consensus threshold are no
-  longer built. Identical results; up to ~13× faster for large trees (greatest
-  gains for tall trees / many tips), with no change at small sizes.
+  longer built.
 - `RenumberTips()` relabels an unlabelled `multiPhylo` or `list` of trees in a
   single C++ pass instead of a per-tree R loop, with a no-op fast path for trees
-  already in the target order. Speeds up `Consensus()` and other callers when
-  combining many trees; results are unchanged.
+  already in the target order.
 - `Consensus()` no longer copies every input tree to strip branch lengths and
-  node labels (the consensus core ignores both); it now coerces in place,
-  trimming wrapper overhead (~25% faster on small forests of many short trees).
-  Results are unchanged.
+  node labels; it now coerces in place.
 
 ## Dependencies
 

R/parse_files.R

@@ -283,6 +283,13 @@ ReadCharacters <- function(filepath, character_num = NULL, encoding = "UTF8") {
       }
       stateStarts <- grep("^\\d+", stateLines)
       stateEnds <- grep("[,;]$", stateLines)
+      # When the closing ';' of the block also serves as the terminator of the
+      # last character entry (e.g. Wills 2012), stripping it above leaves the
+      # last entry without an explicit comma/semicolon. Treat end-of-block as
+      # an implicit terminator so the last entry is still parsed correctly.
+      if (length(stateStarts) == length(stateEnds) + 1L) {
+        stateEnds <- c(stateEnds, length(stateLines) + 1L)
+      }
       if (length(stateStarts) != length(stateEnds)) {
         warning("Could not parse character states; does each end with a ' or ;?.")
       } else {

inst/extdata/tests/statelabels-semicolon.nex

@@ -0,0 +1,29 @@
+#NEXUS
+BEGIN TAXA;
+  DIMENSIONS NTAX=2;
+  TAXLABELS
+    TaxonA TaxonB
+  ;
+END;
+
+BEGIN CHARACTERS;
+  DIMENSIONS NCHAR=2;
+  FORMAT DATATYPE=STANDARD MISSING=? GAP=-;
+  CHARLABELS
+    'First character'
+    'Second character'
+  ;
+  STATELABELS
+  1
+  "state zero"
+  "state one"
+  ,
+  2
+  "absent"
+  "present"
+  ;
+  MATRIX
+  TaxonA 01
+  TaxonB 10
+  ;
+END;

tests/testthat/test-parsers.R

@@ -321,6 +321,16 @@ test_that("ReadCharacters() reads CHARSTATELABELS", {
 
 })
 
+test_that("ReadCharacters() handles STATELABELS with terminal semicolon", {
+  # Last STATELABELS entry terminated by block ';' rather than a comma
+  labels <- expect_no_warning(
+    ReadCharacters(system.file("extdata/tests/statelabels-semicolon.nex",
+                               package = "TreeTools"))
+  )
+  expect_equal(attr(labels, "state.labels"),
+               list(c("state zero", "state one"), c("absent", "present")))
+})
+
 test_that("MorphoBankDecode() decodes", {
   expect_equal("' -- x  \n 1--2", MorphoBankDecode("'' - x^n 1-2"))
 })