Fix gen-release-notes.py script

Something changed in the RST->md conversion since the script was written,
so fixed the script by trimming the releases based on the RST file instead,
which we control.

Author: nicoddemus

scripts/gen-release-notes.py

@@ -1,36 +1,47 @@
 """
 Generates the release notes for the latest release, in Markdown.
 
-Convert CHANGELOG.rst to Markdown, and extracts just the latest release.
-
-Writes to ``scripts/latest-release-notes.md``, which can be
-used with https://github.com/softprops/action-gh-release.
+1. Extracts the latest release from the CHANGELOG.rst file.
+2. Converts it to Markdown using pypandoc.
+3. Writes to ``scripts/latest-release-notes.md``, which can be
+   used with https://github.com/softprops/action-gh-release.
 """
 from pathlib import Path
 
 import pypandoc
 
 this_dir = Path(__file__).parent
 rst_text = (this_dir.parent / "CHANGELOG.rst").read_text(encoding="UTF-8")
-md_text = pypandoc.convert_text(
-    rst_text, "md", format="rst", extra_args=["--wrap=preserve"]
-)
 
 output_lines = []
-first_heading_found = False
-for line in md_text.splitlines():
-    if line.startswith("# "):
-        # Skip the first section (normally # Releases).
-        pass
-    elif line.startswith("## "):
-        # First second-level section, note it and skip the text,
-        # as we are only interested in the individual release items.
-        if first_heading_found:
+capture = False
+for line in rst_text.splitlines():
+    # Only start capturing after the latest release section.
+    if line.startswith("-------"):
+        capture = not capture
+        if not capture:
+            # We only need to capture the latest release, so stop.
             break
-        first_heading_found = True
-    else:
+        continue
+
+    if capture:
         output_lines.append(line)
 
+# Drop last line, as it contains the previous release section title.
+del output_lines[-1]
+
+trimmed_rst = "\n".join(output_lines).strip()
+print(">>Trimmed RST follows:")
+print(trimmed_rst)
+print(">>Trimmed RST ends")
+
+md_text = pypandoc.convert_text(
+    trimmed_rst, "md", format="rst", extra_args=["--wrap=preserve"]
+)
+print(">>Converted Markdown follows:")
+print(md_text)
+print(">>Converted Markdown ends")
+
 output_fn = this_dir / "latest-release-notes.md"
-output_fn.write_text("\n".join(output_lines), encoding="UTF-8")
+output_fn.write_text(md_text, encoding="UTF-8")
 print(output_fn, "generated.")