test: lock markdown round-trip identity across boundary cases

Added a round-trip test for the markdown boundary cases the coverage audit flagged: multiple links on a line, parens inside a URL, inline code whose contents look like markdown, non-ASCII prose, and unclosed / trailing fenced code. All round-trip to identity, so a fetch with no edit cannot change the remote on push. Two cases (paren-URL and markdown-looking inline code) round-trip safely despite an imperfect intermediate Org form, noted in the test and filed as a display-fidelity follow-up. 376 tests green.

1 files changed, 14 insertions, 0 deletions

diff --git a/tests/test-pearl-convert.el b/tests/test-pearl-convert.el
index 4125e22..1337439 100644
--- a/tests/test-pearl-convert.el
+++ b/tests/test-pearl-convert.el
@@ -168,5 +168,19 @@ the conversion-tier docstring) and are excluded here."
                 "| a | b |\n|---|---|"))
     (should (string= md (pearl--org-to-md (pearl--md-to-org md))))))
 
+(ert-deftest test-pearl-convert-roundtrip-boundary-cases ()
+  "Round-trip identity holds across markdown boundary cases — the property that
+keeps a fetch + no-edit push from changing the remote.  Two of these round-trip
+safely despite an imperfect intermediate Org form: a URL containing parens, and
+inline code whose contents look like markdown (rendered converted in the Org
+view, but reversed exactly on push)."
+  (dolist (md '("[a](u1) and [b](u2)"                    ; multiple links on a line
+                "see [docs](http://x.com/(p))"           ; parens inside the URL
+                "`**b** and [l](u) and _i_`"             ; markdown-looking inline code
+                "café — naïve 日本語 ✓"                  ; non-ASCII prose
+                "before\n```\nunclosed code\nmore code"  ; unclosed fence, no swallow
+                "text\n```python\nx = 1\n```\nafter"))   ; fence with trailing text
+    (should (string= md (pearl--org-to-md (pearl--md-to-org md))))))
+
 (provide 'test-pearl-convert)
 ;;; test-pearl-convert.el ends here