hugolib: Fix broken manual summary handling

Fixes #5381

6 files changed, 146 insertions, 98 deletions

diff --git a/hugolib/page.go b/hugolib/page.go
index 578aed9ed..47083e9ef 100644
--- a/hugolib/page.go
+++ b/hugolib/page.go
@@ -730,56 +730,48 @@ func splitUserDefinedSummaryAndContent(markup string, c []byte) (sc *summaryCont
 		}
 	}()
 
-	c = bytes.TrimSpace(c)
-	startDivider := bytes.Index(c, internalSummaryDivider)
+	startDivider := bytes.Index(c, internalSummaryDividerBaseBytes)
 
 	if startDivider == -1 {
 		return
 	}
 
-	endDivider := startDivider + len(internalSummaryDivider)
-	endSummary := startDivider
-
-	var (
-		startMarkup []byte
-		endMarkup   []byte
-		addDiv      bool
-	)
-
+	startTag := "p"
 	switch markup {
-	default:
-		startMarkup = []byte("<p>")
-		endMarkup = []byte("</p>")
 	case "asciidoc":
-		startMarkup = []byte("<div class=\"paragraph\">")
-		endMarkup = []byte("</div>")
-	case "rst":
-		startMarkup = []byte("<p>")
-		endMarkup = []byte("</p>")
-		addDiv = true
+		startTag = "div"
+
+	}
+
+	// Walk back and forward to the surrounding tags.
+	start := bytes.LastIndex(c[:startDivider], []byte("<"+startTag))
+	end := bytes.Index(c[startDivider:], []byte("</"+startTag))
+
+	if start == -1 {
+		start = startDivider
+	} else {
+		start = startDivider - (startDivider - start)
 	}
 
-	// Find the closest end/start markup string to the divider
-	fromStart := -1
-	fromIdx := bytes.LastIndex(c[:startDivider], startMarkup)
-	if fromIdx != -1 {
-		fromStart = startDivider - fromIdx - len(startMarkup)
+	if end == -1 {
+		end = startDivider + len(internalSummaryDividerBase)
+	} else {
+		end = startDivider + end + len(startTag) + 3
 	}
-	fromEnd := bytes.Index(c[endDivider:], endMarkup)
 
-	if fromEnd != -1 && fromEnd <= fromStart {
-		endSummary = startDivider + fromEnd + len(endMarkup)
-	} else if fromStart != -1 && fromEnd != -1 {
-		endSummary = startDivider - fromStart - len(startMarkup)
+	var addDiv bool
+
+	switch markup {
+	case "rst":
+		addDiv = true
 	}
 
-	withoutDivider := bytes.TrimSpace(append(c[:startDivider], c[endDivider:]...))
-	var (
-		summary []byte
-	)
+	withoutDivider := append(c[:start], bytes.Trim(c[end:], "\n")...)
+
+	var summary []byte
 
 	if len(withoutDivider) > 0 {
-		summary = bytes.TrimSpace(withoutDivider[:endSummary])
+		summary = bytes.TrimSpace(withoutDivider[:start])
 	}
 
 	if addDiv {
@@ -793,7 +785,7 @@ func splitUserDefinedSummaryAndContent(markup string, c []byte) (sc *summaryCont
 
 	sc = &summaryContent{
 		summary: summary,
-		content: withoutDivider,
+		content: bytes.TrimSpace(withoutDivider),
 	}
 
 	return
diff --git a/hugolib/page_content.go b/hugolib/page_content.go
index dab46411e..5a8258279 100644
--- a/hugolib/page_content.go
+++ b/hugolib/page_content.go
@@ -27,7 +27,9 @@ import (
 )
 
 var (
-	internalSummaryDivider = []byte("HUGOMORE42")
+	internalSummaryDividerBase      = "HUGOMORE42"
+	internalSummaryDividerBaseBytes = []byte(internalSummaryDividerBase)
+	internalSummaryDividerPre       = []byte("\n\n" + internalSummaryDividerBase + "\n\n")
 )
 
 // The content related items on a Page.
@@ -111,7 +113,7 @@ Loop:
 			}
 
 		case it.Type == pageparser.TypeLeadSummaryDivider:
-			result.Write(internalSummaryDivider)
+			result.Write(internalSummaryDividerPre)
 			p.source.hasSummaryDivider = true
 			// Need to determine if the page is truncated.
 			f := func(item pageparser.Item) bool {
diff --git a/hugolib/page_test.go b/hugolib/page_test.go
index 4499a97e0..18be64cee 100644
--- a/hugolib/page_test.go
+++ b/hugolib/page_test.go
@@ -483,6 +483,7 @@ func checkPageContent(t *testing.T, page *Page, content string, msg ...interface
 	a := normalizeContent(content)
 	b := normalizeContent(string(page.content()))
 	if a != b {
+		t.Log(trace())
 		t.Fatalf("Page content is:\n%q\nExpected:\n%q (%q)", b, a, msg)
 	}
 }
@@ -553,6 +554,7 @@ func normalizeExpected(ext, str string) string {
 			}
 			expected += fmt.Sprintf("<div class=\"paragraph\">\n%s</p></div>\n", para)
 		}
+
 		return expected
 	case "rst":
 		return fmt.Sprintf("<div class=\"document\">\n\n\n%s</div>", str)
@@ -569,7 +571,6 @@ func testAllMarkdownEnginesForPages(t *testing.T,
 		{"md", func() bool { return true }},
 		{"mmark", func() bool { return true }},
 		{"ad", func() bool { return helpers.HasAsciidoc() }},
-		// TODO(bep) figure a way to include this without too much work.{"html", func() bool { return true }},
 		{"rst", func() bool { return helpers.HasRst() }},
 	}
 
@@ -642,57 +643,6 @@ func TestCreateNewPage(t *testing.T) {
 	testAllMarkdownEnginesForPages(t, assertFunc, settings, simplePage)
 }
 
-func TestSplitSummaryAndContent(t *testing.T) {
-	t.Parallel()
-	for i, this := range []struct {
-		markup          string
-		content         string
-		expectedSummary string
-		expectedContent string
-	}{
-		{"markdown", `<p>Summary Same LineHUGOMORE42</p>
-
-<p>Some more text</p>`, "<p>Summary Same Line</p>", "<p>Summary Same Line</p>\n\n<p>Some more text</p>"},
-		{"asciidoc", `<div class="paragraph"><p>sn</p></div><div class="paragraph"><p>HUGOMORE42Some more text</p></div>`,
-			"<div class=\"paragraph\"><p>sn</p></div>",
-			"<div class=\"paragraph\"><p>sn</p></div><div class=\"paragraph\"><p>Some more text</p></div>"},
-		{"rst",
-			"<div class=\"document\"><p>Summary Next Line</p><p>HUGOMORE42Some more text</p></div>",
-			"<div class=\"document\"><p>Summary Next Line</p></div>",
-			"<div class=\"document\"><p>Summary Next Line</p><p>Some more text</p></div>"},
-		{"markdown", "<p>a</p><p>b</p><p>HUGOMORE42c</p>", "<p>a</p><p>b</p>", "<p>a</p><p>b</p><p>c</p>"},
-		{"markdown", "<p>a</p><p>b</p><p>cHUGOMORE42</p>", "<p>a</p><p>b</p><p>c</p>", "<p>a</p><p>b</p><p>c</p>"},
-		{"markdown", "<p>a</p><p>bHUGOMORE42</p><p>c</p>", "<p>a</p><p>b</p>", "<p>a</p><p>b</p><p>c</p>"},
-		{"markdown", "<p>aHUGOMORE42</p><p>b</p><p>c</p>", "<p>a</p>", "<p>a</p><p>b</p><p>c</p>"},
-		{"markdown", "  HUGOMORE42 ", "", ""},
-		{"markdown", "HUGOMORE42", "", ""},
-		{"markdown", "<p>HUGOMORE42", "<p>", "<p>"},
-		{"markdown", "HUGOMORE42<p>", "", "<p>"},
-		{"markdown", "\n\n<p>HUGOMORE42</p>\n", "<p></p>", "<p></p>"},
-		// Issue #2586
-		// Note: Hugo will not split mid-sentence but will look for the closest
-		// paragraph end marker. This may be a change from Hugo 0.16, but it makes sense.
-		{"markdown", `<p>this is an example HUGOMORE42of the issue.</p>`,
-			"<p>this is an example of the issue.</p>",
-			"<p>this is an example of the issue.</p>"},
-		// Issue: #2538
-		{"markdown", fmt.Sprintf(` <p class="lead">%s</p>HUGOMORE42<p>%s</p>
-`,
-			strings.Repeat("A", 10), strings.Repeat("B", 31)),
-			fmt.Sprintf(`<p class="lead">%s</p>`, strings.Repeat("A", 10)),
-			fmt.Sprintf(`<p class="lead">%s</p><p>%s</p>`, strings.Repeat("A", 10), strings.Repeat("B", 31)),
-		},
-	} {
-
-		sc, err := splitUserDefinedSummaryAndContent(this.markup, []byte(this.content))
-
-		require.NoError(t, err)
-		require.NotNil(t, sc, fmt.Sprintf("[%d] Nil %s", i, this.markup))
-		require.Equal(t, this.expectedSummary, string(sc.summary), fmt.Sprintf("[%d] Summary markup %s", i, this.markup))
-		require.Equal(t, this.expectedContent, string(sc.content), fmt.Sprintf("[%d] Content markup %s", i, this.markup))
-	}
-}
-
 func TestPageWithDelimiter(t *testing.T) {
 	t.Parallel()
 	assertFunc := func(t *testing.T, ext string, pages Pages) {
@@ -720,11 +670,14 @@ func TestPageWithDelimiterForMarkdownThatCrossesBorder(t *testing.T) {
 
 	p := s.RegularPages[0]
 
-	if p.Summary() != template.HTML("<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup class=\"footnote-ref\" id=\"fnref:1\"><a href=\"#fn:1\">1</a></sup>\n</p>") {
+	if p.Summary() != template.HTML(
+		"<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup class=\"footnote-ref\" id=\"fnref:1\"><a href=\"#fn:1\">1</a></sup></p>") {
 		t.Fatalf("Got summary:\n%q", p.Summary())
 	}
 
-	if p.content() != template.HTML("<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup class=\"footnote-ref\" id=\"fnref:1\"><a href=\"#fn:1\">1</a></sup>\n</p>\n<div class=\"footnotes\">\n\n<hr />\n\n<ol>\n<li id=\"fn:1\">Many people say so.\n <a class=\"footnote-return\" href=\"#fnref:1\"><sup>[return]</sup></a></li>\n</ol>\n</div>") {
+	if p.content() != template.HTML(
+		"<p>The <a href=\"http://gohugo.io/\">best static site generator</a>.<sup class=\"footnote-ref\" id=\"fnref:1\"><a href=\"#fn:1\">1</a></sup></p>\n\n<div class=\"footnotes\">\n\n<hr />\n\n<ol>\n<li id=\"fn:1\">Many people say so.\n <a class=\"footnote-return\" href=\"#fnref:1\"><sup>[return]</sup></a></li>\n</ol>\n</div>") {
+
 		t.Fatalf("Got content:\n%q", p.content())
 	}
 }
@@ -1544,6 +1497,70 @@ func TestChompBOM(t *testing.T) {
 	checkPageTitle(t, p, "Simple")
 }
 
+// https://github.com/gohugoio/hugo/issues/5381
+func TestPageManualSummary(t *testing.T) {
+	b := newTestSitesBuilder(t)
+	b.WithSimpleConfigFile()
+
+	b.WithContent("page-md-shortcode.md", `---
+title: "Hugo"
+---
+This is a {{< sc >}}.
+<!--more--> 
+Content.
+`)
+
+	b.WithContent("page-md-shortcode-same-line.md", `---
+title: "Hugo"
+---
+This is a {{< sc >}}.<!--more-->Same line.
+`)
+
+	b.WithContent("page-org-shortcode.org", `#+TITLE: T1
+#+AUTHOR: A1
+#+DESCRIPTION: D1
+This is a {{< sc >}}.
+# more
+Content.	
+`)
+
+	b.WithContent("page-org-variant1.org", `#+TITLE: T1
+Summary.
+
+# more
+
+Content.	
+`)
+
+	b.WithTemplatesAdded("layouts/shortcodes/sc.html", "a shortcode")
+	b.WithTemplatesAdded("layouts/_default/single.html", `
+SUMMARY:{{ .Summary }}:END
+--------------------------
+CONTENT:{{ .Content }}
+`)
+
+	b.CreateSites().Build(BuildCfg{})
+
+	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",
+		"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-org-shortcode/index.html",
+		"SUMMARY:<p>This is a a shortcode.</p>:END",
+		"CONTENT:<p>This is a a shortcode.</p>\n\n<p>Content.\t</p>\n",
+	)
+	b.AssertFileContent("public/page-org-variant1/index.html",
+		"SUMMARY:<p>Summary.</p>:END",
+		"CONTENT:<p>Summary.</p>\n\n<p>Content.\t</p>\n",
+	)
+}
+
 // TODO(bep) this may be useful for other tests.
 func compareObjects(a interface{}, b interface{}) bool {
 	aStr := strings.Split(fmt.Sprintf("%v", a), "")
@@ -1705,6 +1722,8 @@ Len PlainWords: {{ len .PlainWords }}
 Truncated: {{ .Truncated }}
 Len Summary: {{ len .Summary }}
 Len Content: {{ len .Content }}
+
+SUMMARY:{{ .Summary }}:{{ len .Summary }}:END
 `}
 
 	b := newTestSitesBuilder(t)
@@ -1776,10 +1795,10 @@ Summary: In Chinese, 好 means good.
 
 	b.AssertFileContent("public/p1/index.html", "WordCount: 510\nFuzzyWordCount: 600\nReadingTime: 3\nLen Plain: 2550\nLen PlainWords: 510\nTruncated: false\nLen Summary: 2549\nLen Content: 2557")
 
-	b.AssertFileContent("public/p2/index.html", "WordCount: 314\nFuzzyWordCount: 400\nReadingTime: 2\nLen Plain: 1570\nLen PlainWords: 314\nTruncated: true\nLen Summary: 34\nLen Content: 1592")
+	b.AssertFileContent("public/p2/index.html", "WordCount: 314\nFuzzyWordCount: 400\nReadingTime: 2\nLen Plain: 1569\nLen PlainWords: 314\nTruncated: true\nLen Summary: 25\nLen Content: 1583")
 
-	b.AssertFileContent("public/p3/index.html", "WordCount: 206\nFuzzyWordCount: 300\nReadingTime: 1\nLen Plain: 639\nLen PlainWords: 7\nTruncated: true\nLen Summary: 52\nLen Content: 661")
-	b.AssertFileContent("public/p4/index.html", "WordCount: 7\nFuzzyWordCount: 100\nReadingTime: 1\nLen Plain: 639\nLen PlainWords: 7\nTruncated: true\nLen Summary: 52\nLen Content: 661")
+	b.AssertFileContent("public/p3/index.html", "WordCount: 206\nFuzzyWordCount: 300\nReadingTime: 1\nLen Plain: 638\nLen PlainWords: 7\nTruncated: true\nLen Summary: 43\nLen Content: 652")
+	b.AssertFileContent("public/p4/index.html", "WordCount: 7\nFuzzyWordCount: 100\nReadingTime: 1\nLen Plain: 638\nLen PlainWords: 7\nTruncated: true\nLen Summary: 43\nLen Content: 652")
 	b.AssertFileContent("public/p5/index.html", "WordCount: 206\nFuzzyWordCount: 300\nReadingTime: 1\nLen Plain: 638\nLen PlainWords: 7\nTruncated: true\nLen Summary: 229\nLen Content: 653")
 	b.AssertFileContent("public/p6/index.html", "WordCount: 7\nFuzzyWordCount: 100\nReadingTime: 1\nLen Plain: 638\nLen PlainWords: 7\nTruncated: false\nLen Summary: 637\nLen Content: 653")
 
diff --git a/hugolib/testhelpers_test.go b/hugolib/testhelpers_test.go
index 469cfeb4a..f80f40c48 100644
--- a/hugolib/testhelpers_test.go
+++ b/hugolib/testhelpers_test.go
@@ -4,7 +4,9 @@ import (
 	"io/ioutil"
 	"path/filepath"
 	"runtime"
+	"strconv"
 	"testing"
+	"unicode/utf8"
 
 	"bytes"
 	"fmt"
@@ -698,6 +700,26 @@ func dumpPages(pages ...*Page) {
 	}
 }
 
+func printStringIndexes(s string) {
+	lines := strings.Split(s, "\n")
+	i := 0
+
+	for _, line := range lines {
+
+		for _, r := range line {
+			fmt.Printf("%-3s", strconv.Itoa(i))
+			i += utf8.RuneLen(r)
+		}
+		i++
+		fmt.Println()
+		for _, r := range line {
+			fmt.Printf("%-3s", string(r))
+		}
+		fmt.Println()
+
+	}
+
+}
 func isCI() bool {
 	return os.Getenv("CI") != ""
 }
diff --git a/parser/pageparser/pagelexer.go b/parser/pageparser/pagelexer.go
index ddf109b3d..565be2994 100644
--- a/parser/pageparser/pagelexer.go
+++ b/parser/pageparser/pagelexer.go
@@ -194,6 +194,16 @@ func (l *pageLexer) consumeCRLF() bool {
 	return consumed
 }
 
+func (l *pageLexer) consumeSpace() {
+	for {
+		r := l.next()
+		if r == eof || !unicode.IsSpace(r) {
+			l.backup()
+			return
+		}
+	}
+}
+
 func lexMainSection(l *pageLexer) stateFunc {
 	// Fast forward as far as possible.
 	var l1, l2 int
@@ -234,6 +244,8 @@ func lexMainSection(l *pageLexer) stateFunc {
 				}
 				l.summaryDividerChecked = true
 				l.pos += len(l.summaryDivider)
+				// This makes it a little easier to reason about later.
+				l.consumeSpace()
 				l.emit(TypeLeadSummaryDivider)
 			}
 		}
diff --git a/parser/pageparser/pageparser_intro_test.go b/parser/pageparser/pageparser_intro_test.go
index ba4a2c84b..60c431c10 100644
--- a/parser/pageparser/pageparser_intro_test.go
+++ b/parser/pageparser/pageparser_intro_test.go
@@ -37,7 +37,7 @@ var (
 	tstFrontMatterYAMLCRLF = nti(TypeFrontMatterYAML, "foo: \"bar\"\r\n")
 	tstFrontMatterJSON     = nti(TypeFrontMatterJSON, tstJSON+"\r\n")
 	tstSomeText            = nti(tText, "\nSome text.\n")
-	tstSummaryDivider      = nti(TypeLeadSummaryDivider, "<!--more-->")
+	tstSummaryDivider      = nti(TypeLeadSummaryDivider, "<!--more-->\n")
 	tstHtmlStart           = nti(TypeHTMLStart, "<")
 
 	tstORG = `
@@ -65,8 +65,9 @@ var frontMatterTests = []lexerTest{
 	{"TOML front matter", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, tstEOF}},
 	{"JSON front matter", tstJSON + "\r\n\nSome text.\n", []Item{tstFrontMatterJSON, tstSomeText, tstEOF}},
 	{"ORG front matter", tstORG + "\nSome text.\n", []Item{tstFrontMatterORG, tstSomeText, tstEOF}},
-	{"Summary divider ORG", tstORG + "\nSome text.\n# more\nSome text.\n", []Item{tstFrontMatterORG, tstSomeText, nti(TypeLeadSummaryDivider, "# more"), tstSomeText, tstEOF}},
-	{"Summary divider", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n<!--more-->\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, tstSummaryDivider, tstSomeText, tstEOF}},
+	{"Summary divider ORG", tstORG + "\nSome text.\n# more\nSome text.\n", []Item{tstFrontMatterORG, tstSomeText, nti(TypeLeadSummaryDivider, "# more\n"), nti(tText, "Some text.\n"), tstEOF}},
+	{"Summary divider", "+++\nfoo = \"bar\"\n+++\n\nSome text.\n<!--more-->\nSome text.\n", []Item{tstFrontMatterTOML, tstSomeText, tstSummaryDivider, nti(tText, "Some text.\n"), tstEOF}},
+	{"Summary divider same line", "+++\nfoo = \"bar\"\n+++\n\nSome text.<!--more-->Some text.\n", []Item{tstFrontMatterTOML, nti(tText, "\nSome text."), nti(TypeLeadSummaryDivider, "<!--more-->"), nti(tText, "Some text.\n"), tstEOF}},
 }
 
 func TestFrontMatter(t *testing.T) {