Backtestshugorun-20260922-141813#15206

Issue #15206 Merged fix PR # Run 22 Sep 2026 · run-20260922-141813 gohugoio/hugo

ReadingTime and FuzzyWordCount have off-by-one errors in their ceiling calculations

Better

Our patch was judged better than the merged fix.

In this judgment, Patch A = our agent, Patch B = the merged fix.

Both patches make byte-identical source fixes to hugolib/page__content.go (fuzzyWordCount +99 and readingTime +211/+499), so on correctness they are tied — both correctly implement the ceil idiom (n+d-1)/d and both hold up at every boundary the issue names. Both add the same integration test TestReadingTimeAndFuzzyWordCountBoundaries with identical assertions that would fail before the fix (p100, p213, p500, p501) and pass after, so test value is also tied. The decisive difference is scope: Patch A touches only the source file and appends the one new test. Patch B carries substantial unrelated churn — reformatting roughly a dozen existing AssertFileContent/Assert call sites across the file (moving the first arg to its own line, adding trailing commas) that has nothing to do with the off-by-one bug. That incidental churn enlarges the diff, muddies blame, and risks review/merge conflicts for no functional gain. Under the scope criterion (change as small as a correct fix allows; drive-by reformatting counts against a patch), A is the cleaner, better-crafted fix.

The judge saw only the two patches, labelled A and B in random order, with issue links and version numbers removed. The reasoning above is shown verbatim.

Solve

1h 15m18 turns

Wall-clock time in the container

Full test suite

No new failures

3,692 passed · 99 skipped · 3,804 total

Contamination audit

Unknown

See the audit notes for this run

Patch size

80vs 128 merged

Changed lines, ours vs merged

Summary

Two related integer-arithmetic bugs in hugolib/page__content.go (in the plain-content calculation), causing behavior that deviates from documented intent at boundary values.

1. ReadingTime divides by 213/501 instead of 212/500

if isCJKLanguage {
    result.readingTime = (result.wordCount + 500) / 501
} else {
    result.readingTime = (result.wordCount + 212) / 213
}

The intent is a reading speed of 212 words per minute (500 for CJK), rounded up. But the correct ceiling-division idiom is (n + d - 1) / d. As written, the expression computes ceil(wordCount / 213) and ceil(wordCount / 501), so the effective reading speeds are 213 and 501 wpm.

wordCountcurrent: (wc+212)/213expected at 212 wpm: ceil(wc/212)
21211
213**1****2**
42422
425**2****3**

Proposed fix:

result.readingTime = (result.wordCount + 211) / 212   // non-CJK
result.readingTime = (result.wordCount + 499) / 500   // CJK

2. FuzzyWordCount overshoots on exact multiples of 100

result.fuzzyWordCount = (result.wordCount + 100) / 100 * 100

Documented as the word count "rounded up to the nearest multiple of 100", but adding 100 unconditionally bumps exact multiples to the next multiple:

wordCountcurrentexpected
99100100
100200100
101200200
0100**0**

Proposed fix:

result.fuzzyWordCount = (result.wordCount + 99) / 100 * 100

Both changes only affect output at exact boundary values, but they do change rendered output for existing sites (e.g., a 100-word page's FuzzyWordCount drops from 200 to 100), so this may warrant a changelog note.

Integration Test

<details>
<summary>hugolib/page_test.go</summary>

// See issue 15206.
func TestReadingTimeAndFuzzyWordCountBoundaries(t *testing.T) {
	t.Parallel()

	files := `
-- hugo.toml --
disableKinds = ["home", "section", "taxonomy", "term", "rss", "sitemap"]
-- content/p99.md --
---
title: p99
---
` + strings.Repeat("word ", 99) + `
-- content/p100.md --
---
title: p100
---
` + strings.Repeat("word ", 100) + `
-- content/p101.md --
---
title: p101
---
` + strings.Repeat("word ", 101) + `
-- content/p211.md --
---
title: p211
---
` + strings.Repeat("word ", 211) + `
-- content/p212.md --
---
title: p212
---
` + strings.Repeat("word ", 212) + `
-- content/p213.md --
---
title: p213
---
` + strings.Repeat("word ", 213) + `
-- content/p499.md --
---
title: p499
isCJKLanguage: true
---
` + strings.Repeat("你", 499) + `
-- content/p500.md --
---
title: p500
isCJKLanguage: true
---
` + strings.Repeat("你", 500) + `
-- content/p501.md --
---
title: p501
isCJKLanguage: true
---
` + strings.Repeat("你", 501) + `
-- layouts/page.html --
{{ .WordCount }}|{{ .FuzzyWordCount }}|{{ .ReadingTime }}
`

	b := Test(t, files)

	b.AssertFileContent("public/p99/index.html", "99|100|1")
	b.AssertFileContent("public/p100/index.html", "100|100|1") // fails
	b.AssertFileContent("public/p101/index.html", "101|200|1")

	b.AssertFileContent("public/p211/index.html", "211|300|1")
	b.AssertFileContent("public/p212/index.html", "212|300|1")
	b.AssertFileContent("public/p213/index.html", "213|300|2") // fails

	b.AssertFileContent("public/p499/index.html", "499|500|1")
	b.AssertFileContent("public/p500/index.html", "500|500|1") // fails
	b.AssertFileContent("public/p501/index.html", "501|600|2") // fails
}

</details>

The two patches

Our patch+77 −3
hugolib/page__content.goindex c89d6b614..725576f38 100644@@ -807,13 +807,13 @@ func (c *cachedContentScope) contentPlain(ctx context.Context) (contentPlainPlai  		// TODO(bep) is set in a test. Fix that. 		if result.fuzzyWordCount == 0 {-			result.fuzzyWordCount = (result.wordCount + 100) / 100 * 100+			result.fuzzyWordCount = (result.wordCount + 99) / 100 * 100 		}  		if isCJKLanguage {-			result.readingTime = (result.wordCount + 500) / 501+			result.readingTime = (result.wordCount + 499) / 500 		} else {-			result.readingTime = (result.wordCount + 212) / 213+			result.readingTime = (result.wordCount + 211) / 212 		}  		rs.Value = resulthugolib/page_test.goindex cad453d5b..ecd588900 100644@@ -1209,6 +1209,80 @@ func TestWordCount(t *testing.T) { 	testAllMarkdownEnginesForPages(t, assertFunc, nil, simplePageWithLongContent) } +// See issue 15206.+func TestReadingTimeAndFuzzyWordCountBoundaries(t *testing.T) {+	t.Parallel()++	files := `+-- hugo.toml --+disableKinds = ["home", "section", "taxonomy", "term", "rss", "sitemap"]+-- content/p99.md --+---+title: p99+---+` + strings.Repeat("word ", 99) + `+-- content/p100.md --+---+title: p100+---+` + strings.Repeat("word ", 100) + `+-- content/p101.md --+---+title: p101+---+` + strings.Repeat("word ", 101) + `+-- content/p211.md --+---+title: p211+---+` + strings.Repeat("word ", 211) + `+-- content/p212.md --+---+title: p212+---+` + strings.Repeat("word ", 212) + `+-- content/p213.md --+---+title: p213+---+` + strings.Repeat("word ", 213) + `+-- content/p499.md --+---+title: p499+isCJKLanguage: true+---+` + strings.Repeat("你", 499) + `+-- content/p500.md --+---+title: p500+isCJKLanguage: true+---+` + strings.Repeat("你", 500) + `+-- content/p501.md --+---+title: p501+isCJKLanguage: true+---+` + strings.Repeat("你", 501) + `+-- layouts/page.html --+{{ .WordCount }}|{{ .FuzzyWordCount }}|{{ .ReadingTime }}+`++	b := Test(t, files)++	b.AssertFileContent("public/p99/index.html", "99|100|1")+	b.AssertFileContent("public/p100/index.html", "100|100|1")+	b.AssertFileContent("public/p101/index.html", "101|200|1")++	b.AssertFileContent("public/p211/index.html", "211|300|1")+	b.AssertFileContent("public/p212/index.html", "212|300|1")+	b.AssertFileContent("public/p213/index.html", "213|300|2")++	b.AssertFileContent("public/p499/index.html", "499|500|1")+	b.AssertFileContent("public/p500/index.html", "500|500|1")+	b.AssertFileContent("public/p501/index.html", "501|600|2")+}+ func TestPagePaths(t *testing.T) { 	t.Parallel()  
Merged fix · PR #+109 −19
hugolib/page__content.goindex c89d6b614..725576f38 100644@@ -807,13 +807,13 @@ func (c *cachedContentScope) contentPlain(ctx context.Context) (contentPlainPlai  		// TODO(bep) is set in a test. Fix that. 		if result.fuzzyWordCount == 0 {-			result.fuzzyWordCount = (result.wordCount + 100) / 100 * 100+			result.fuzzyWordCount = (result.wordCount + 99) / 100 * 100 		}  		if isCJKLanguage {-			result.readingTime = (result.wordCount + 500) / 501+			result.readingTime = (result.wordCount + 499) / 500 		} else {-			result.readingTime = (result.wordCount + 212) / 213+			result.readingTime = (result.wordCount + 211) / 212 		}  		rs.Value = resulthugolib/page_test.goindex cad453d5b..635358590 100644@@ -412,7 +412,8 @@ baseURL = "http://example.com/" 	p := b.H.Sites[0].RegularPages()[0]  	b.Assert(p.Summary(context.Background()), qt.Equals, template.HTML(-		"<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup id=\"fnref:1\"><a href=\"#fn:1\" class=\"footnote-ref\" role=\"doc-noteref\">1</a></sup></p>"))+		"<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup id=\"fnref:1\"><a href=\"#fn:1\" class=\"footnote-ref\" role=\"doc-noteref\">1</a></sup></p>",+	))  	cnt := content(p) 	b.Assert(cnt, qt.Equals, "<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup id=\"fnref:1\"><a href=\"#fn:1\" class=\"footnote-ref\" role=\"doc-noteref\">1</a></sup></p>\n<div class=\"footnotes\" role=\"doc-endnotes\">\n<hr>\n<ol>\n<li id=\"fn:1\">\n<p>Many people say so.&#160;<a href=\"#fnref:1\" class=\"footnote-backref\" role=\"doc-backlink\">&#x21a9;&#xfe0e;</a></p>\n</li>\n</ol>\n</div>")@@ -730,7 +731,8 @@ This is **content**. Summary: {{ .Summary }}|Truncated: {{ .Truncated }}| Content: {{ .Content }}| -`).AssertFileContent("public/simple/index.html",+`).AssertFileContent(+		"public/simple/index.html", 		"Summary: <p>This is <strong>summary</strong>.</p>|", 		"Truncated: true|", 		"Content: <p>This is <strong>summary</strong>.</p>\n<p>This is <strong>content</strong>.</p>|",@@ -1286,13 +1288,15 @@ AllTranslations: {{ range .AllTranslations }}{{ .Language.Lang }}|{{ end }}|  ` 	b := Test(t, files)-	b.AssertFileContent("public/en/sect/p1/index.html",+	b.AssertFileContent(+		"public/en/sect/p1/index.html", 		"TranslationKey: adfasdf|", 		"AllTranslations: en|nn||", 		"Translations: nn||", 	) -	b.AssertFileContent("public/nn/sect/p1/index.html",+	b.AssertFileContent(+		"public/nn/sect/p1/index.html", 		"TranslationKey: adfasdf|", 		"Translations: en||", 		"AllTranslations: en|nn||",@@ -1378,12 +1382,14 @@ Resources: {{ range .Resources }}{{ .RelPermalink }}|{{ .Content }}|{{ end }}|  ` 	b := Test(t, files)-	b.AssertFileContent("public/en/sect/mybundle_en/index.html",+	b.AssertFileContent(+		"public/en/sect/mybundle_en/index.html", 		"TranslationKey: adfasdf|", 		"Resources: /en/sect/mybundle_en/f1.txt|f1.en|/en/sect/mybundle_en/f2.txt|f2.en||", 	) -	b.AssertFileContent("public/nn/sect/mybundle_nn/index.html",+	b.AssertFileContent(+		"public/nn/sect/mybundle_nn/index.html", 		"TranslationKey: adfasdf|", 		"Title: mybundle nn|TranslationKey: adfasdf|\nResources: /en/sect/mybundle_en/f1.txt|f1.en|/nn/sect/mybundle_nn/f2.nn.txt|f2.nn||", 	)@@ -1497,31 +1503,37 @@ CONTENT:{{ .Content }} ` 	b := Test(t, files) -	b.AssertFileContent("public/page-md-shortcode/index.html",+	b.AssertFileContent(+		"public/page-md-shortcode/index.html", 		"SUMMARY:<p>This is a a shortcode.</p>:END", 		"CONTENT:<p>This is a a shortcode.</p>\n\n<p>Content.</p>\n", 	) -	b.AssertFileContent("public/page-md-shortcode-same-line/index.html",+	b.AssertFileContent(+		"public/page-md-shortcode-same-line/index.html", 		"SUMMARY:<p>This is a a shortcode</p>:END", 		"CONTENT:<p>This is a a shortcode</p>\n\n<p>Same line.</p>\n", 	) -	b.AssertFileContent("public/page-md-shortcode-same-line-after/index.html",+	b.AssertFileContent(+		"public/page-md-shortcode-same-line-after/index.html", 		"SUMMARY:<p>Summary</p>:END", 		"CONTENT:<p>Summary</p>\n\na shortcode", 	) -	b.AssertFileContent("public/page-org-shortcode/index.html",+	b.AssertFileContent(+		"public/page-org-shortcode/index.html", 		"SUMMARY:<p>\nThis is a a shortcode.\n</p>:END", 		"CONTENT:<p>\nThis is a a shortcode.\n</p>\n<p>\nContent.\t\n</p>\n", 	)-	b.AssertFileContent("public/page-org-variant1/index.html",+	b.AssertFileContent(+		"public/page-org-variant1/index.html", 		"SUMMARY:<p>\nSummary.\n</p>:END", 		"CONTENT:<p>\nSummary.\n</p>\n<p>\nContent.\t\n</p>\n", 	) -	b.AssertFileContent("public/page-md-only-shortcode/index.html",+	b.AssertFileContent(+		"public/page-md-only-shortcode/index.html", 		"SUMMARY:a shortcode:END", 		"CONTENT:a shortcode\n\na shortcode\n", 	)@@ -1731,7 +1743,8 @@ c: {{ .Scratch.Get "c" }}  	b := Test(t, files) -	b.AssertFileContent("public/index.html",+	b.AssertFileContent(+		"public/index.html", 		".Scratch eq .Store: true", 		"a: b", 		"c: d",@@ -1928,7 +1941,8 @@ Site: {{ site.Store.Get "Site" }}|  	b := TestRunning(t, files) -	b.AssertFileContent("public/index.html",+	b.AssertFileContent(+		"public/index.html", 		` Shortcode: sh-Home| Page: p-Home|@@ -1939,7 +1953,8 @@ Hugo: h-Home|  	b.EditFileReplaceAll("content/_index.md", "Home", "Homer").Build() -	b.AssertFileContent("public/index.html",+	b.AssertFileContent(+		"public/index.html", 		` Shortcode: sh-Homer| Page: p-Homer|@@ -2149,7 +2164,8 @@ EF  	b := Test(t, files) -	b.AssertFileContent("public/s1/p1/index.html",+	b.AssertFileContent(+		"public/s1/p1/index.html", 		"ab|ab|ab", 		"cD|cD|cD", 		"EF|EF|EF",@@ -2368,3 +2384,77 @@ All. 	myothersection, _ := s.GetPage("myothersection") // backed by a content file. 	check(myothersection, true) }++// See issue 15206.+func TestReadingTimeAndFuzzyWordCountBoundaries(t *testing.T) {+	t.Parallel()++	files := `+-- hugo.toml --+disableKinds = ["home", "section", "taxonomy", "term", "rss", "sitemap"]+-- content/p99.md --+---+title: p99+---+` + strings.Repeat("word ", 99) + `+-- content/p100.md --+---+title: p100+---+` + strings.Repeat("word ", 100) + `+-- content/p101.md --+---+title: p101+---+` + strings.Repeat("word ", 101) + `+-- content/p211.md --+---+title: p211+---+` + strings.Repeat("word ", 211) + `+-- content/p212.md --+---+title: p212+---+` + strings.Repeat("word ", 212) + `+-- content/p213.md --+---+title: p213+---+` + strings.Repeat("word ", 213) + `+-- content/p499.md --+---+title: p499+isCJKLanguage: true+---+` + strings.Repeat("你", 499) + `+-- content/p500.md --+---+title: p500+isCJKLanguage: true+---+` + strings.Repeat("你", 500) + `+-- content/p501.md --+---+title: p501+isCJKLanguage: true+---+` + strings.Repeat("你", 501) + `+-- layouts/page.html --+{{ .WordCount }}|{{ .FuzzyWordCount }}|{{ .ReadingTime }}+`++	b := Test(t, files)++	b.AssertFileContent("public/p99/index.html", "99|100|1")+	b.AssertFileContent("public/p100/index.html", "100|100|1")+	b.AssertFileContent("public/p101/index.html", "101|200|1")++	b.AssertFileContent("public/p211/index.html", "211|300|1")+	b.AssertFileContent("public/p212/index.html", "212|300|1")+	b.AssertFileContent("public/p213/index.html", "213|300|2")++	b.AssertFileContent("public/p499/index.html", "499|500|1")+	b.AssertFileContent("public/p500/index.html", "500|500|1")+	b.AssertFileContent("public/p501/index.html", "501|600|2")+} 

What the agent reported

Review is clean. The fix is complete and verified.

Summary

What was wrong — Two off-by-one integer-arithmetic bugs in hugolib/page__content.go (in the plain-content calculation):

  1. ReadingTime used (wordCount + 212) / 213 (and +500)/501 for CJK), which computes ceiling at 213/501 wpm instead of the intended 212/500. Correct ceiling idiom is (n + d - 1) / d.
  2. FuzzyWordCount used (wordCount + 100) / 100 * 100, which overshoots exact multiples of 100 (e.g. 100 words → 200, 0 → 100).

What I changed

  • hugolib/page__content.go — three edits, nothing else:
  • fuzzyWordCount = (wordCount + 99) / 100 * 100
  • readingTime = (wordCount + 499) / 500 (CJK)
  • readingTime = (wordCount + 211) / 212 (non-CJK)
  • hugolib/page_test.go — added regression test TestReadingTimeAndFuzzyWordCountBoundaries (the exact integration test from the report, with the transient // fails annotations removed).

Test evidence

  • Before the fix, the new test failed at the reported boundary: public/p100/index.html rendered 100|200|1, expected 100|100|1.
  • After the fix, ./bt-test ./hugolib/ -run 'TestReadingTimeAndFuzzyWordCountBoundaries|TestWordCount' → the regression test passed; the hugolib package reported no failures.
  • The TestWordCount* tests are gated behind SkipSlowTestUnlessCI (skipped locally). Blast-radius check: the CJK TestWordCount* variants assert only WordCount, which my change does not touch; TestWordCount (483 words) also asserts FuzzyWordCount=500 and ReadingTime=3, both of which remain identical under the new divisors ((483+99)/100*100=500, (483+211)/212=3).
  • Reviewer confirmed the diff matches the report exactly with no unrelated changes.

Requirements disposition

  1. Fix ReadingTime divisors (212/500 instead of 213/501), non-CJK and CJK — implemented, verified by p213→...|2 and p501→...|2 assertions passing.
  2. Fix FuzzyWordCount overshoot on exact multiples of 100 (100→100, 0→0) — implemented, verified by p100→100|100|1 and p500→500|500|1 passing (previously 100|200|1).
  3. Regression test exercising the exact reported boundary scenarios — implemented, seen to fail before and pass after.
  4. Changelog note the report says the change "may warrant" — skipped: no CHANGELOG mechanism in this repo (releases are auto-generated via releaser: commits per git history), and it was offered as optional, not required; the behavior change is documented here instead.

Note: the pre-existing D AGENTS.md / D CLAUDE.md deletions are from the environment baseline and are unrelated to this fix. All changes are left uncommitted in the working tree.