From 6a56330ddec37e510fa1acbd45ad6053d6c967ae Mon Sep 17 00:00:00 2001
From: Andy C <andy@oilshell.org>
Date: Tue, 5 Dec 2023 13:48:13 -0500
Subject: [PATCH] [doctools] Polish ref-check a bit more

---
 devtools/release.sh   |  2 +-
 doctools/ref_check.py | 27 ++++++++++++++-------------
 2 files changed, 15 insertions(+), 14 deletions(-)

diff --git a/devtools/release.sh b/devtools/release.sh
index d21007eaad..09c1e2c77d 100755
--- a/devtools/release.sh
+++ b/devtools/release.sh
@@ -28,7 +28,7 @@
 #
 #   Commit files to oilshell/benchmark-data repo and sync.
 #   benchmarks/report.sh all
-#   $0 deploy-tar  # needed to checksum
+#   $0 deploy-tar  # needed to publish tarball checksum in HTML
 #   build/doc.sh run-for-release
 #   $0 compress
 #   devtools/release-version.sh git-changelog-$VERSION
diff --git a/doctools/ref_check.py b/doctools/ref_check.py
index f19a3bdab0..304e3ccb05 100755
--- a/doctools/ref_check.py
+++ b/doctools/ref_check.py
@@ -76,13 +76,11 @@ def Check(all_toc_nodes, chap_tree):
 
   log('')
 
-  log('Topics in TOC: %d', len(all_topics))
-  log('Unique topics in TOC: %d', len(set(all_topics)))
-  log('Sections not implemented: %d', sections_not_impl)
-  log('Total topics not implemented: %d', topics_not_impl)
-  log('')
-
-  log('%d in link_from set: %s', len(link_from), sorted(link_from)[:10])
+  log('TOC stats:')
+  log('  Topics: %d', len(all_topics))
+  log('  Unique topics: %d', len(set(all_topics)))
+  log('  Topic Sections not implemented (X): %d', sections_not_impl)
+  log('  All topics not implemented: %d', topics_not_impl)
   log('')
 
   if 0:
@@ -99,7 +97,7 @@ def Check(all_toc_nodes, chap_tree):
   chap_topics = collections.defaultdict(list)  # topic_id -> list of chapters
   short_topics = []
 
-  min_words = 10  # arbitrary
+  min_words = 5  # arbitrary
 
   for chap in chap_tree.children:
 
@@ -125,19 +123,22 @@ def Check(all_toc_nodes, chap_tree):
         elif num_words > 1:
           short_topics.append((topic_id, topic.text))
 
-  log('%d in link_to set: %s', len(link_to), sorted(link_to)[:10])
-  log('')
-
   num_chapters = len(chap_tree.children)
 
-  log('Chapter stats')
+  log('Chapter stats:')
   log('  num chapters = %d', num_chapters)
   log('  num_sections = %d', num_sections)
   log('  num_topics = %d', num_topics)
-  log('  topics with more than %d words = %d', min_words, num_topics_written)
 
   chap_topic_set = set(chap_topics)
   log('  num unique topics = %d', len(chap_topic_set))
+  log('  topics with first draft (more than %d words) = %d', min_words,
+      num_topics_written)
+  log('')
+
+  log('%d in link_to set: %s', len(link_to), sorted(link_to)[:10])
+  log('')
+  log('%d in link_from set: %s', len(link_from), sorted(link_from)[:10])
   log('')
 
   index_topic_set = set(toc_topic_check)