From 75a86f01e5401c085da1dac19dd456dded611136 Mon Sep 17 00:00:00 2001
From: Aleteoryx <alyx@aleteoryx.me>
Date: Mon, 8 Sep 2025 18:34:46 -0400
Subject: [PATCH] add linkage metrics

---
 gloss/modes/glossary.py | 49 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 47 insertions(+), 2 deletions(-)

diff --git a/gloss/modes/glossary.py b/gloss/modes/glossary.py
index aa29f832e2945d3c154256946e01e278c5c8fee4..340456d6395422093a469bcdcdf407680702dda8 100755
--- a/gloss/modes/glossary.py
+++ b/gloss/modes/glossary.py
@@ -43,6 +43,10 @@ class GlsFile:
 	names: Set[str]
 	blocks: List[Block]
 	see_also: Optional[List[str]]
+	links: Set['GlsFile']
+	
+	def __hash__(self):
+		return id(self)
 
 
 ### PARSING ###
@@ -113,7 +117,7 @@ def first_pass(slug, fp):
 	elif block_type is not None:
 		blocks.append(Block(block_type, block_text, block_meta))
 	
-	return GlsFile(slug, title, [*names], blocks, see_also)			
+	return GlsFile(slug, title, [*names], blocks, see_also, set())
 
 quote_pat = re.compile("((?:(?!@@)(?!//).)*)(?:@@((?:(?!//).)+))?(?://(.+))?")
 
@@ -169,10 +173,14 @@ def gen_inner_html(fmt, file, idx):
 				if m is None:
 					continue
 
+				other = idx.by_name[name]
+				blacklist.update(other.names)
+				file.links.add(other)
+
 				s, e = m.span()
 				block.text.pop(i)
 				block.text.insert(i, seg[e:])
-				block.text.insert(i, (f'<a href="{html.escape(idx.by_name[name].slug)}.html">{seg[s:e]}</a>',))
+				block.text.insert(i, (f'<a href="{html.escape(other.slug)}.html">{seg[s:e]}</a>',))
 				block.text.insert(i, seg[:s])
 				break
 	
@@ -212,11 +220,14 @@ def gen_inner_html(fmt, file, idx):
 				href = page.slug
 				name = page.title
 			else:
+				page = None
 				if '|' in what:
 					href, name = what.split('|', 1)
 				else:
 					href = name = what
 			see_also += f"\n\t<li><a href=\"{html.escape(href)}.html\">{html.escape(name)}</a></li>"
+			if page is not None:
+				file.links.add(page)
 		see_also += "\n</ul>"
 	else:
 		see_also = ""
@@ -224,6 +235,32 @@ def gen_inner_html(fmt, file, idx):
 	return content, see_also
 
 
+### METRICS ###
+
+def remove_island(root, nodes):
+	to_check = [root]
+	for node in to_check:
+		if node not in nodes:
+			continue
+		
+		to_check.extend(node.links)
+		nodes.remove(node)
+
+def count_orphans(root, nodes):
+	nodes = set(nodes)
+	
+	remove_island(root, nodes)
+	num_orphans = len(nodes)
+	orphans = [*nodes]
+	
+	num_islands = 0
+	while len(nodes) > 0:
+		remove_island(next(iter(nodes)), nodes)
+		num_islands += 1
+	
+	return num_orphans, num_islands, orphans
+	
+
 ### ENTRYPOINT ###
 
 def gloss(sfmt, srcdir, outdir):
@@ -270,5 +307,13 @@ def gloss(sfmt, srcdir, outdir):
 				'modtime': datetime.fromtimestamp(stat(f'{srcdir}/{file.slug}.gls').st_mtime)
 			}
 			fp.write(template.format(**ctx))
+
+	num_orphans, num_islands, orphans = count_orphans(indexes.by_slug['index'], files)
 	
 	print(sfmt.format('###', f'generated {len(files)} entries'))
+	if num_orphans != 0:
+		pl_entry  = 'entry'  if num_orphans == 1 else 'entries'
+		pl_island = 'island' if num_islands == 1 else 'islands'
+		pl_orphan = 'orphan' if num_orphans == 1 else 'orphans'
+		print(sfmt.format('!!!', f'{num_orphans} orphan {pl_entry} across {num_islands} {pl_island}'))
+		print(sfmt.format('!!!', f'{pl_orphan}: {", ".join(x.title for x in orphans)}'))