all: support document attribute "leveloffset"

The ":leveloffset:" on document attribute allow increment
or decrement the heading level on included files.

Reference: https://docs.asciidoctor.org/asciidoc/latest/directives/include-with-leveloffset/

8 files changed, 404 insertions, 230 deletions

diff --git a/README.md b/README.md
index f104633..95d8bdf 100644
--- a/README.md
+++ b/README.md
@@ -126,6 +126,8 @@ Supported document attribute references,
 * `idseparator`
 * `lastname(_x)`
 * `last-update-label`
+* [`leveloffset`](https://docs.asciidoctor.org/asciidoc/latest/directives/include-with-leveloffset/).
+Only on document attributes, not on include directive.
 * `middlename(_x)`
 * `nofooter`
 * `noheader`
@@ -216,7 +218,6 @@ List of features which may be implemented,
 * Cross References
   * Inter-document Cross References
 * Include Directive
-  * Partitioning large documents and using leveloffset
   * AsciiDoc vs non-AsciiDoc files
   * Normalize Block Indentation
   * Include a File Multiple Times in the Same Document
diff --git a/document_attribute.go b/document_attribute.go
index 357a0e0..7d40a50 100644
--- a/document_attribute.go
+++ b/document_attribute.go
@@ -3,7 +3,11 @@
 
 package asciidoctor
 
-import "strings"
+import (
+	"fmt"
+	"strconv"
+	"strings"
+)
 
 // List of document attribute.
 const (
@@ -22,6 +26,7 @@ const (
 	docAttrLastName        = `lastname`
 	docAttrLastUpdateLabel = `last-update-label`
 	docAttrLastUpdateValue = `last-update-value`
+	docAttrLevelOffset     = `leveloffset`
 	docAttrMiddleName      = `middlename`
 	docAttrNoFooter        = `nofooter`
 	docAttrNoHeader        = `noheader`
@@ -57,7 +62,8 @@ const (
 // DocumentAttribute contains the mapping of global attribute keys in the
 // headers with its value.
 type DocumentAttribute struct {
-	Entry map[string]string
+	Entry       map[string]string
+	LevelOffset int
 }
 
 func newDocumentAttribute() DocumentAttribute {
@@ -74,18 +80,33 @@ func newDocumentAttribute() DocumentAttribute {
 	}
 }
 
-func (docAttr *DocumentAttribute) apply(key, val string) {
+func (docAttr *DocumentAttribute) apply(key, val string) (err error) {
 	if key[0] == '!' {
 		key = strings.TrimSpace(key[1:])
 		delete(docAttr.Entry, key)
-		return
+		return nil
 	}
 	var n = len(key)
 	if key[n-1] == '!' {
 		key = strings.TrimSpace(key[:n-1])
 		delete(docAttr.Entry, key)
-		return
+		return nil
+	}
+
+	if key == docAttrLevelOffset {
+		var offset int64
+		offset, err = strconv.ParseInt(val, 10, 32)
+		if err != nil {
+			return fmt.Errorf(`DocumentAttribute: %s: invalid value %q`, key, val)
+		}
+		if val[0] == '+' || val[0] == '-' {
+			docAttr.LevelOffset += int(offset)
+			goto valid
+		}
+		docAttr.LevelOffset = int(offset)
 	}
 
+valid:
 	docAttr.Entry[key] = val
+	return nil
 }
diff --git a/document_attribute_test.go b/document_attribute_test.go
new file mode 100644
index 0000000..cec898e
--- /dev/null
+++ b/document_attribute_test.go
@@ -0,0 +1,100 @@
+// SPDX-FileCopyrightText: 2024 M. Shulhan <ms@kilabit.info>
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+package asciidoctor
+
+import (
+	"testing"
+
+	"git.sr.ht/~shulhan/pakakeh.go/lib/test"
+)
+
+func TestDocumentAttributeApply(t *testing.T) {
+	type testCase struct {
+		desc     string
+		key      string
+		val      string
+		expError string
+		exp      DocumentAttribute
+	}
+
+	var docAttr = DocumentAttribute{
+		Entry: map[string]string{
+			`key1`: ``,
+			`key2`: ``,
+		},
+	}
+
+	var listCase = []testCase{{
+		key: `key3`,
+		exp: docAttr,
+	}, {
+		desc: `prefix negation`,
+		key:  `!key1`,
+		exp: DocumentAttribute{
+			Entry: map[string]string{
+				`key2`: ``,
+				`key3`: ``,
+			},
+		},
+	}, {
+		desc: `suffix negation`,
+		key:  `key2!`,
+		exp: DocumentAttribute{
+			Entry: map[string]string{
+				`key3`: ``,
+			},
+		},
+	}, {
+		desc: `leveloffset +`,
+		key:  docAttrLevelOffset,
+		val:  `+2`,
+		exp: DocumentAttribute{
+			Entry: map[string]string{
+				`key3`:        ``,
+				`leveloffset`: `+2`,
+			},
+			LevelOffset: 2,
+		},
+	}, {
+		desc: `leveloffset -`,
+		key:  docAttrLevelOffset,
+		val:  `-2`,
+		exp: DocumentAttribute{
+			Entry: map[string]string{
+				`key3`:        ``,
+				`leveloffset`: `-2`,
+			},
+			LevelOffset: 0,
+		},
+	}, {
+		desc: `leveloffset`,
+		key:  docAttrLevelOffset,
+		val:  `1`,
+		exp: DocumentAttribute{
+			Entry: map[string]string{
+				`key3`:        ``,
+				`leveloffset`: `1`,
+			},
+			LevelOffset: 1,
+		},
+	}, {
+		desc:     `leveloffset: invalid`,
+		key:      docAttrLevelOffset,
+		val:      `*1`,
+		expError: `DocumentAttribute: leveloffset: invalid value "*1"`,
+	}}
+
+	var (
+		tc  testCase
+		err error
+	)
+	for _, tc = range listCase {
+		err = docAttr.apply(tc.key, tc.val)
+		if err != nil {
+			test.Assert(t, `apply: `+tc.desc, tc.expError, err.Error())
+			continue
+		}
+		test.Assert(t, `apply: `+tc.desc, tc.exp, docAttr)
+	}
+}
diff --git a/document_parser.go b/document_parser.go
index e1acb4b..96c37db 100644
--- a/document_parser.go
+++ b/document_parser.go
@@ -141,19 +141,20 @@ func (docp *documentParser) hasPreamble() bool {
 
 		notEmtpy int
 		line     []byte
-		kind     int
 	)
 	for ; start < len(docp.lines); start++ {
 		line = docp.lines[start]
 		if len(line) == 0 {
 			continue
 		}
-		kind, _, _ = whatKindOfLine(line)
-		if kind == elKindSectionL1 || kind == elKindSectionL2 ||
-			kind == elKindSectionL3 || kind == elKindSectionL4 ||
-			kind == elKindSectionL5 ||
-			kind == lineKindID ||
-			kind == lineKindIDShort {
+		_, _ = docp.whatKindOfLine(line)
+		if docp.kind == elKindSectionL1 ||
+			docp.kind == elKindSectionL2 ||
+			docp.kind == elKindSectionL3 ||
+			docp.kind == elKindSectionL4 ||
+			docp.kind == elKindSectionL5 ||
+			docp.kind == lineKindID ||
+			docp.kind == lineKindIDShort {
 			return notEmtpy > 0
 		}
 		notEmtpy++
@@ -202,7 +203,7 @@ func (docp *documentParser) line(logp string) (spaces, line []byte, ok bool) {
 	}
 	docp.lineNum++
 
-	docp.kind, spaces, line = whatKindOfLine(line)
+	spaces, line = docp.whatKindOfLine(line)
 	return spaces, line, true
 }
 
@@ -397,7 +398,7 @@ func (docp *documentParser) parseBlock(parent *element, term int) {
 					}
 					el.Attrs[key] = value
 				} else {
-					docp.doc.Attributes.apply(key, value)
+					_ = docp.doc.Attributes.apply(key, value)
 					parent.addChild(&element{
 						kind:  docp.kind,
 						key:   key,
@@ -749,7 +750,7 @@ func (docp *documentParser) parseHeader() {
 			var key, value string
 			key, value, ok = docp.parseAttribute(line, false)
 			if ok {
-				docp.doc.Attributes.apply(key, value)
+				_ = docp.doc.Attributes.apply(key, value)
 			}
 			line = nil
 			continue
@@ -1599,3 +1600,247 @@ func (docp *documentParser) skipCommentAndEmptyLine() (line []byte, ok bool) {
 	}
 	return line, true
 }
+
+// whatKindOfLine return the kind of line.
+// It will return lineKindText if the line does not match with known syntax.
+func (docp *documentParser) whatKindOfLine(line []byte) (spaces, got []byte) {
+	docp.kind = lineKindText
+
+	line = bytes.TrimRight(line, " \f\n\r\t\v")
+
+	// All of the comparison MUST be in order.
+
+	if len(line) == 0 {
+		docp.kind = lineKindEmpty
+		return nil, line
+	}
+	if bytes.HasPrefix(line, []byte(`////`)) {
+		// Check for comment block first, since we use HasPrefix to
+		// check for single line comment.
+		docp.kind = lineKindBlockComment
+		return spaces, line
+	}
+	if bytes.HasPrefix(line, []byte(`//`)) {
+		// Use HasPrefix to allow single line comment without space,
+		// for example "//comment".
+		docp.kind = lineKindComment
+		return spaces, line
+	}
+
+	var strline = string(line)
+
+	switch strline {
+	case `'''`, `---`, `- - -`, `***`, `* * *`:
+		docp.kind = lineKindHorizontalRule
+		return spaces, line
+	case `<<<`:
+		docp.kind = lineKindPageBreak
+		return spaces, line
+	case `--`:
+		docp.kind = elKindBlockOpen
+		return spaces, line
+	case `____`:
+		docp.kind = elKindBlockExcerpts
+		return spaces, line
+	case `....`:
+		docp.kind = elKindBlockLiteral
+		return nil, line
+	case `++++`:
+		docp.kind = elKindBlockPassthrough
+		return spaces, line
+	case `****`:
+		docp.kind = elKindBlockSidebar
+		return nil, line
+	case `====`:
+		docp.kind = elKindBlockExample
+		return spaces, line
+	case `[listing]`:
+		docp.kind = elKindBlockListingNamed
+		return nil, line
+	case `[literal]`:
+		docp.kind = elKindBlockLiteralNamed
+		return nil, line
+	case `toc::[]`:
+		docp.kind = elKindMacroTOC
+		return spaces, line
+	}
+
+	if bytes.HasPrefix(line, []byte(`|===`)) {
+		docp.kind = elKindTable
+		return nil, line
+	}
+	if bytes.HasPrefix(line, []byte(`image::`)) {
+		docp.kind = elKindBlockImage
+		return spaces, line
+	}
+	if bytes.HasPrefix(line, []byte(`include::`)) {
+		docp.kind = lineKindInclude
+		return nil, line
+	}
+	if bytes.HasPrefix(line, []byte(`video::`)) {
+		docp.kind = elKindBlockVideo
+		return nil, line
+	}
+	if bytes.HasPrefix(line, []byte(`audio::`)) {
+		docp.kind = elKindBlockAudio
+		return nil, line
+	}
+	if isAdmonition(line) {
+		docp.kind = lineKindAdmonition
+		return nil, line
+	}
+
+	var (
+		x        int
+		r        byte
+		hasSpace bool
+	)
+	for x, r = range line {
+		if r == ' ' || r == '\t' {
+			hasSpace = true
+			continue
+		}
+		break
+	}
+	if hasSpace {
+		spaces = line[:x]
+		line = line[x:]
+
+		// A line indented with space only allowed on list item,
+		// otherwise it would be set as literal paragraph.
+
+		if isLineDescriptionItem(line) {
+			docp.kind = elKindListDescriptionItem
+			return spaces, line
+		}
+
+		if line[0] != '*' && line[0] != '-' && line[0] != '.' {
+			docp.kind = elKindLiteralParagraph
+			return spaces, line
+		}
+	}
+
+	switch line[0] {
+	case ':':
+		docp.kind = lineKindAttribute
+	case '[':
+		var (
+			newline = bytes.TrimRight(line, " \t")
+			l       = len(newline)
+		)
+
+		if newline[l-1] != ']' {
+			return nil, line
+		}
+		if l >= 5 {
+			// [[x]]
+			if newline[1] == '[' && newline[l-2] == ']' {
+				docp.kind = lineKindID
+				return nil, line
+			}
+		}
+		if l >= 4 {
+			// [#x]
+			if line[1] == '#' {
+				docp.kind = lineKindIDShort
+				return nil, line
+			}
+			// [.x]
+			if line[1] == '.' {
+				docp.kind = lineKindStyleClass
+				return nil, line
+			}
+		}
+		docp.kind = lineKindAttributeElement
+		return spaces, line
+	case '=', '#':
+		var subs = bytes.Fields(line)
+
+		switch string(subs[0]) {
+		case `=`, `#`:
+			docp.kind = elKindSectionL0
+		case `==`, `##`:
+			docp.kind = elKindSectionL1
+		case `===`, `###`:
+			docp.kind = elKindSectionL2
+		case `====`, `####`:
+			docp.kind = elKindSectionL3
+		case `=====`, `#####`:
+			docp.kind = elKindSectionL4
+		case `======`, `######`:
+			docp.kind = elKindSectionL5
+		default:
+			return spaces, line
+		}
+		docp.kind += docp.doc.Attributes.LevelOffset
+		if docp.kind < elKindSectionL0 || docp.kind > elKindSectionL5 {
+			docp.kind = elKindText
+		}
+		return spaces, line
+
+	case '.':
+		switch {
+		case len(line) <= 1:
+			docp.kind = lineKindText
+		case ascii.IsAlnum(line[1]):
+			docp.kind = lineKindBlockTitle
+		default:
+			x = 0
+			for ; x < len(line); x++ {
+				if line[x] == '.' {
+					continue
+				}
+				if line[x] == ' ' || line[x] == '\t' {
+					docp.kind = elKindListOrderedItem
+					return spaces, line
+				}
+			}
+		}
+	case '*', '-':
+		if len(line) <= 1 {
+			return spaces, line
+		}
+
+		var (
+			listItemChar = line[0]
+			count        = 0
+		)
+		x = 0
+		for ; x < len(line); x++ {
+			if line[x] == listItemChar {
+				count++
+				continue
+			}
+			if line[x] == ' ' || line[x] == '\t' {
+				docp.kind = elKindListUnorderedItem
+				return spaces, line
+			}
+			// Break on the first non-space, so from above
+			// condition we have,
+			// - item
+			// -- item
+			// --- item
+			// ---- // block listing
+			// --unknown // break here
+			break
+		}
+		if listItemChar == '-' && count == 4 && x == len(line) {
+			docp.kind = elKindBlockListing
+		} else {
+			docp.kind = lineKindText
+		}
+		return spaces, line
+	default:
+		switch string(line) {
+		case `+`:
+			docp.kind = lineKindListContinue
+		case `----`:
+			docp.kind = elKindBlockListing
+		default:
+			if isLineDescriptionItem(line) {
+				docp.kind = elKindListDescriptionItem
+			}
+		}
+	}
+	return spaces, line
+}
diff --git a/element.go b/element.go
index aaa05da..bdda706 100644
--- a/element.go
+++ b/element.go
@@ -804,7 +804,7 @@ func (el *element) setStyleAdmonition(admName string) {
 func (el *element) toHTML(doc *Document, w io.Writer) {
 	switch el.kind {
 	case lineKindAttribute:
-		doc.Attributes.apply(el.key, el.value)
+		_ = doc.Attributes.apply(el.key, el.value)
 
 	case elKindCrossReference:
 		var (
diff --git a/parser.go b/parser.go
index 67f43b4..979811e 100644
--- a/parser.go
+++ b/parser.go
@@ -654,216 +654,3 @@ func parseStyle(styleName string) (styleKind int64) {
 
 	return 0
 }
-
-// whatKindOfLine return the kind of line.
-// It will return lineKindText if the line does not match with known syntax.
-func whatKindOfLine(line []byte) (kind int, spaces, got []byte) {
-	kind = lineKindText
-
-	line = bytes.TrimRight(line, " \f\n\r\t\v")
-
-	// All of the comparison MUST be in order.
-
-	if len(line) == 0 {
-		return lineKindEmpty, nil, line
-	}
-	if bytes.HasPrefix(line, []byte(`////`)) {
-		// Check for comment block first, since we use HasPrefix to
-		// check for single line comment.
-		return lineKindBlockComment, spaces, line
-	}
-	if bytes.HasPrefix(line, []byte(`//`)) {
-		// Use HasPrefix to allow single line comment without space,
-		// for example "//comment".
-		return lineKindComment, spaces, line
-	}
-
-	var strline = string(line)
-
-	switch strline {
-	case `'''`, `---`, `- - -`, `***`, `* * *`:
-		return lineKindHorizontalRule, spaces, line
-	case `<<<`:
-		return lineKindPageBreak, spaces, line
-	case `--`:
-		return elKindBlockOpen, spaces, line
-	case `____`:
-		return elKindBlockExcerpts, spaces, line
-	case `....`:
-		return elKindBlockLiteral, nil, line
-	case `++++`:
-		return elKindBlockPassthrough, spaces, line
-	case `****`:
-		return elKindBlockSidebar, nil, line
-	case `====`:
-		return elKindBlockExample, spaces, line
-	case `[listing]`:
-		return elKindBlockListingNamed, nil, line
-	case `[literal]`:
-		return elKindBlockLiteralNamed, nil, line
-	case `toc::[]`:
-		return elKindMacroTOC, spaces, line
-	}
-
-	if bytes.HasPrefix(line, []byte(`|===`)) {
-		return elKindTable, nil, line
-	}
-	if bytes.HasPrefix(line, []byte(`image::`)) {
-		return elKindBlockImage, spaces, line
-	}
-	if bytes.HasPrefix(line, []byte(`include::`)) {
-		return lineKindInclude, nil, line
-	}
-	if bytes.HasPrefix(line, []byte(`video::`)) {
-		return elKindBlockVideo, nil, line
-	}
-	if bytes.HasPrefix(line, []byte(`audio::`)) {
-		return elKindBlockAudio, nil, line
-	}
-	if isAdmonition(line) {
-		return lineKindAdmonition, nil, line
-	}
-
-	var (
-		x        int
-		r        byte
-		hasSpace bool
-	)
-	for x, r = range line {
-		if r == ' ' || r == '\t' {
-			hasSpace = true
-			continue
-		}
-		break
-	}
-	if hasSpace {
-		spaces = line[:x]
-		line = line[x:]
-
-		// A line indented with space only allowed on list item,
-		// otherwise it would be set as literal paragraph.
-
-		if isLineDescriptionItem(line) {
-			return elKindListDescriptionItem, spaces, line
-		}
-
-		if line[0] != '*' && line[0] != '-' && line[0] != '.' {
-			return elKindLiteralParagraph, spaces, line
-		}
-	}
-
-	switch line[0] {
-	case ':':
-		kind = lineKindAttribute
-	case '[':
-		var (
-			newline = bytes.TrimRight(line, " \t")
-			l       = len(newline)
-		)
-
-		if newline[l-1] != ']' {
-			return lineKindText, nil, line
-		}
-		if l >= 5 {
-			// [[x]]
-			if newline[1] == '[' && newline[l-2] == ']' {
-				return lineKindID, nil, line
-			}
-		}
-		if l >= 4 {
-			// [#x]
-			if line[1] == '#' {
-				return lineKindIDShort, nil, line
-			}
-			// [.x]
-			if line[1] == '.' {
-				return lineKindStyleClass, nil, line
-			}
-		}
-		return lineKindAttributeElement, spaces, line
-	case '=':
-		var subs = bytes.Fields(line)
-
-		switch string(subs[0]) {
-		case `=`:
-			kind = elKindSectionL0
-		case `==`:
-			kind = elKindSectionL1
-		case `===`:
-			kind = elKindSectionL2
-		case `====`:
-			kind = elKindSectionL3
-		case `=====`:
-			kind = elKindSectionL4
-		case `======`:
-			kind = elKindSectionL5
-		}
-		return kind, spaces, line
-
-	case '.':
-		switch {
-		case len(line) <= 1:
-			kind = lineKindText
-		case ascii.IsAlnum(line[1]):
-			kind = lineKindBlockTitle
-		default:
-			x = 0
-			for ; x < len(line); x++ {
-				if line[x] == '.' {
-					continue
-				}
-				if line[x] == ' ' || line[x] == '\t' {
-					kind = elKindListOrderedItem
-					return kind, spaces, line
-				}
-			}
-		}
-	case '*', '-':
-		if len(line) <= 1 {
-			kind = lineKindText
-			return kind, spaces, line
-		}
-
-		var (
-			listItemChar = line[0]
-			count        = 0
-		)
-		x = 0
-		for ; x < len(line); x++ {
-			if line[x] == listItemChar {
-				count++
-				continue
-			}
-			if line[x] == ' ' || line[x] == '\t' {
-				kind = elKindListUnorderedItem
-				return kind, spaces, line
-			}
-			// Break on the first non-space, so from above
-			// condition we have,
-			// - item
-			// -- item
-			// --- item
-			// ---- // block listing
-			// --unknown // break here
-			break
-		}
-		if listItemChar == '-' && count == 4 && x == len(line) {
-			kind = elKindBlockListing
-		} else {
-			kind = lineKindText
-		}
-		return kind, spaces, line
-	default:
-		switch string(line) {
-		case `+`:
-			kind = lineKindListContinue
-		case `----`:
-			kind = elKindBlockListing
-		default:
-			if isLineDescriptionItem(line) {
-				kind = elKindListDescriptionItem
-			}
-		}
-	}
-	return kind, spaces, line
-}
diff --git a/testdata/_includes/section.adoc b/testdata/_includes/section.adoc
new file mode 100644
index 0000000..aa66f8c
--- /dev/null
+++ b/testdata/_includes/section.adoc
@@ -0,0 +1,3 @@
+= Section 1
+
+This is included with leveloffset +1.
diff --git a/testdata/leveloffset_test.txt b/testdata/leveloffset_test.txt
new file mode 100644
index 0000000..61f5b95
--- /dev/null
+++ b/testdata/leveloffset_test.txt
@@ -0,0 +1,17 @@
+Test the ":leveloffset:" document attribute.
+
+>>> leveloffset
+:leveloffset: +1
+
+include::testdata/_includes/section.adoc[]
+
+<<< leveloffset
+
+<div class="sect1">
+<h2 id="section_1">Section 1</h2>
+<div class="sectionbody">
+<div class="paragraph">
+<p>This is included with leveloffset +1.</p>
+</div>
+</div>
+</div>