fixed bug when some node labels are blank in read.beast

[brj1]

Hi Yu,

I was trying to read in trees generated in BEAST using the MultiTypeTree package using the read.beast method and I noticed the data was not being read correctly. My trees had node labels for only some of the nodes (with node nodes having blank labels) and read.beast was getting confused about which data corresonds to which node. See the following example (the example uses read.beast.newick but this also applies for read.beast:

writeLines('((1[&x="1"]:1,2[&x="2"]:1)5[&x="3"]:1,(3[&x="4"]:1,4[&x="5"]:1)[&x="6"]:1)[&x="7"];', "temp")

tree <- read.beast.newick("temp")

tree@data

# A tibble: 7 x 2
x node

1 1 1
2 2 2
3 3 6
4 4 3
5 5 4
6 6 5
7 7 5

Note how there are two rows where the 'node' column is 5.

I fixed this by setting pseudo node labels to blank labels (like when the labels are completely blank). There will still be an issue if there are duplicate node labels or node labels are similar to the psuedo labels. With the example above my fix gives the correct:

writeLines('((1[&x="1"]:1,2[&x="2"]:1)5[&x="3"]:1,(3[&x="4"]:1,4[&x="5"]:1)[&x="6"]:1)[&x="7"];', "temp")

tree <- read.beast.newick("temp")

tree@data

# A tibble: 7 x 2
x node

1 1 1
2 2 2
3 3 6
4 4 3
5 5 4
6 6 7
7 7 5

I also fixed a dependency issue with filenamer::filename and purrr::is_numeric in the internal functions for read.beast.

[GuangchuangYu]

> treeio:::is_numeric
function (x) 
!anyNA(suppressWarnings(as.numeric(as.character(x))))
<bytecode: 0x558ba6879fe8>
<environment: namespace:treeio>
> treeio:::filename
function (file) 
{
    file_name <- ""
    if (is.character(file)) {
        file_name <- file
    }
    return(file_name)
}
<bytecode: 0x558ba687f4f8>
<environment: namespace:treeio>

The is_numeric and filename were provided by the treeio package and no need to import them from elsewhere.

[xiangpin]

I don't think this method can solve the problem. when the tree contained X1 of node label or the same node label. This method can't also solve the problem.

for example

> xx <- '((1[&x="1"]:1,2[&x="2"]:1)5[&x="3"]:1,(3[&x="4"]:1,4[&x="5"]:1)X1[&x="6"]:1)[&x="7"];'
> trda <- read.beast.newick(textConnection(xx))
> trda@data
# A tibble: 7 x 2
      x node
  <dbl> <chr>
1     1 1
2     2 2
3     3 6
4     4 3
5     5 4
6     6 5
7     7 5

I think all node.label can be replace with

nnode <- phylo$Nnode
phylo$node.label <- paste("X", 1:nnode, sep="")

Even node.label is not NULL sometimes. because the node.label does not effect the following calculation.

[GuangchuangYu]

@xiangpin pls confirm whether the bug was exists? If yes, pls try to fix it.

[brj1]

I added @xiangpin 's fix to my PR. It works as intended and preserves the original node labels in the output. I also removed the importation of is_numeric and filename from other packages.

[xiangpin]

Hello, @brj1
I think this request can not also work well, you can see the pull request. #64