fix: Do not split on Unicode characters with Line_Break=Glue property

Specifically, this allows for using non-breaking space ("\u00A0", " ")
in text which is preserved when using SVG output.

Connects to:
- <wilkelab#35>,
- <wilkelab/ggtext#95>.

R/process-tags.R

@@ -1,5 +1,12 @@
 process_text <- function(node, drawing_context) {
-  tokens <- stringr::str_split(stringr::str_squish(node), "[[:space:]]+")[[1]]
+  # regex: [:space:] set-minus \p{Line_Break=Glue}
+  re_whitespace_non_glue <- "[[:space:]-[\\p{Line_Break=Glue}]]+"
+  re_whitespace_non_glue_trim <- paste0("^", re_whitespace_non_glue,
+                                        "|",
+                                        re_whitespace_non_glue, "$")
+  node_squish_internal <- stringr::str_replace_all(node, re_whitespace_non_glue, " ")
+  node_squish_trim <- stringr::str_replace_all(node_squish_internal, re_whitespace_non_glue_trim, "")
+  tokens <- stringr::str_split(node_squish_trim, re_whitespace_non_glue)[[1]]
 
   # make interior boxes
   boxes <- lapply(tokens,

tests/testthat/test-textbox-grob.R

@@ -276,3 +276,39 @@ test_that("visual tests", {
   expect_doppelganger("Rotation around fixed point", draw_rotated_fixedpoint())
 
 })
+
+describe("text breaking behavior", {
+  get_labels <- function(x) {
+    g <- textbox_grob( x, width = unit(50, "pt"))
+    labels <- sapply(grid.force(g)$children, function(x) x$label)
+    return(labels)
+  }
+
+  it("does not break no-space text",
+    expect_contains(
+      get_labels(" ThisIsALongWordThatShouldNotBreak "),
+      c("ThisIsALongWordThatShouldNotBreak"))
+  )
+
+  it("does break spaced text",
+    expect_contains(
+      get_labels( "This Is Text That Should Break" ),
+      c("This", "Is", "Text", "That", "Should", "Break"))
+  )
+
+  it("does not break at non-breaking space but still at regular space",
+    expect_contains(
+      get_labels("  This&nbsp;Is&nbsp;Text That&nbsp;Should\n\nNot&nbsp;Break  "),
+      c("This\u00A0Is\u00A0Text",
+        "That\u00A0Should",
+        "Not\u00A0Break"))
+  )
+
+  it("preserves non-breaking space at start and end",
+    expect_contains(
+      get_labels("   &nbsp;Beginning Should Not Trim Nor End&nbsp;   "),
+      c("\u00A0Beginning",
+        "Should", "Not", "Trim", "Nor",
+        "End\u00A0"))
+  )
+})