retronews

commit 506b77a0a07446f9ec17888e08d91048cffbc7ca
parent 275c4429be1f5b1e080f631c3354c256737a4451
Author: luke8086 <55237178+luke8086@users.noreply.github.com>
Date:   Tue,  2 Aug 2022 22:13:28 +0000

Improve rendering of link references

Diffstat:
M
retronews.py
 | 
4
++++
M
tests.py
 | 
21
+++++++++++++++++++++

2 files changed, 25 insertions(+), 0 deletions(-)
diff --git a/retronews.py b/retronews.py
@@ -88,6 +88,7 @@ Press any key to continue...
 
 REQUEST_TIMEOUT = 10
 QUOTE_REX = re.compile(r"^(> ?)+")
+REFERENCE_REX = re.compile(r"^\[\d+\][ :-]*https?://[^ ]*$")
 
 T = TypeVar("T")
 
@@ -294,6 +295,9 @@ def wrap_paragraph(text: str) -> list[str]:
         # Preserve code indentation
         return [text]
 
+    if REFERENCE_REX.match(text):
+        return [text]
+
     indent = ""
 
     if (match := QUOTE_REX.match(text)) is not None:
diff --git a/tests.py b/tests.py
@@ -44,10 +44,31 @@ class TestHtmlParser(unittest.TestCase):
         self.assertLines(html, lines)
 
     def test_expanding_links(self):
+        # Ensure links shortened with ellipsis are rendered in full
         html = '<a href="https://example.com/foo/bar">https://example.com/foo...</a>'
         lines = ["https://example.com/foo/bar"]
         self.assertLines(html, lines)
 
+    def test_link_references(self):
+        # Ensure links in numbered references are not shifted to the next line
+        html = (
+            "<p>Lorem ipsum dolor sit amet, pro eu soleat civibus. Mel quas sensibus</p>"
+            + "<p>[0]: <a>https://long.long.long.long.long.long.long.long.long.long.long.example.com</a></p>"
+            + "<p>[1] - <a>https://long.long.long.long.long.long.long.long.long.long.long.example.com</a></p>"
+            + "<p>[2] <a>https://long.long.long.long.long.long.long.long.long.long.long.example.com</a></p>"
+        )
+        lines = [
+            "Lorem ipsum dolor sit amet, pro eu soleat civibus. Mel quas sensibus",
+            "",
+            "[0]: https://long.long.long.long.long.long.long.long.long.long.long.example.com",
+            "",
+            "[1] - https://long.long.long.long.long.long.long.long.long.long.long.example.com",
+            "",
+            "[2] https://long.long.long.long.long.long.long.long.long.long.long.example.com",
+        ]
+
+        self.assertLines(html, lines)
+
     def test_code_blocks(self):
         # This HTML is not particularly correct, but it's how HN renders code blocks
         html = (