Fix issue URLs

small_scripts/make_embedding/embedding.py

@@ -110,46 +110,60 @@ def _parse_issue_strict(issue_block: str, global_issue_index: int):
 
     return desc_text, option_lines
 
+BASE = "https://forum.nationstates.net/viewtopic.php?f=13&t=88"
 
-import re
-
-def format_issue_title_markdown(issue_block):
+def compute_start_from_anchor(anchor: int) -> int:
     """
-    Extracts anchor and visible title from the first line of the issue block,
-    and formats as markdown with a forum link.
+    Returns the 'start' offset for the forum URL given an integer anchor (issue number).
+    start increases by 25 every 500 anchors, beginning at 420.
     """
-    # Find the first non-empty line (should be the title line)
-    for line in issue_block.splitlines():
-        line = line.strip()
-        if not line:
-            continue
-
-        # Extract anchor (e.g., [anchor=1379])
-        anchor_match = re.search(r'\[anchor=(\d+)\]', line)
-        anchor = anchor_match.group(1) if anchor_match else None
-
-        # Extract visible title (after the closing [/anchor]:)
-        # This matches: [anchor=1379]#1379[/anchor]: <title>
-        title_match = re.search(r'\[anchor=(\d+)\]\#\d+\[\/anchor\]:\s*(.*)', line)
-        if title_match:
-            title_text = title_match.group(2).strip()
-        else:
-            # Fallback: try to find after the first colon
-            parts = line.split(':', 1)
-            title_text = parts[1].strip() if len(parts) > 1 else line
-
-        # Remove trailing BBCode tags from title (but keep chain/fancy formatting)
-        title_text = re.sub(r'\[\/?[^\]]+\]', '', title_text).strip()
-
-        # Compose markdown
-        if anchor:
-            return f"#{anchor}: [{title_text}](https://forum.nationstates.net/viewtopic.php?f=13&t=88#{anchor})"
-        else:
-            # Fallback: just return cleaned title
-            return title_text
+    # k is 0 for [0..419], 1 for [420..919], 2 for [920..1419], 3 for [1420..], etc.
+    anchor = int(anchor)
+    k = int(((anchor - 420) / 500) + 1)
+    if anchor < 420:
+        k = 0
+    if k < 0:
+        k = 0
+    return 25 * k
+
+def craft_issue_url(anchor: int) -> str:
+    start = compute_start_from_anchor(anchor)
+    if start == 0:
+        return f"{BASE}#{anchor}"
+    return f"{BASE}&start={start}#{anchor}"
+
+ANCHOR_RE = re.compile(r"\[anchor=(\d+)\]")
+
+def extract_anchor(issue_title_line: str):
+    """
+    From a title like:
+      [b][anchor=1379]#1379[/anchor]: [color=#CE532A][i]MADness:[/i][/color] A View to a Thrill ...
+    returns 1379 as int, or None if not found.
+    """
+    m = ANCHOR_RE.search(issue_title_line)
+    return m.group(1) if m else None
 
-    print(f"Could not find issue title in {issue_block}")
-    raise ValueError(f"Parse error in issue title")
+def format_issue_title_markdown(issue_block: str) -> tuple[str, str]:
+    """
+    Returns (display_markdown, url) such as:
+      "#1379: [MADness: A View to a Thrill](...#1379)"
+    Keeps chain/fancy formatting in the visible title (BBCode stripped),
+    and builds the correct paginated URL using the anchor.
+    """
+    # First non-empty line should be the title line
+    title_line = next((ln.strip() for ln in issue_block.splitlines() if ln.strip()), "")
+    anchor = extract_anchor(title_line)
+    # Extract visible title to the right of '[/anchor]:'
+    # Example matches "...[/anchor]: <title text>"
+    title_part = title_line.split('[/anchor]:', 1)[-1].strip() if '[/anchor]:' in title_line else title_line
+    # Strip BBCode for display text while preserving the chain wording itself
+    title_text = re.sub(r"\[/?[^\]]+\]", "", title_part).strip()
+    if anchor is None:
+        # Fallback: no anchor found; return plain title
+        return (title_text or "Untitled Issue", f"{BASE}")
+    url = craft_issue_url(anchor)
+    display = f"#{anchor}: [{title_text}]({url})"
+    return display
 
 def encode_issues_components_and_sparse():
     print("Initializing BGEM3FlagModel...")