@antora/assembler 1.0.0-alpha.9 → 1.0.0-beta.2

package/lib/{produce-aggregate-document.js → produce-assembly-file.js}

@@ -1,64 +1,63 @@
 'use strict'
 
-const File = require('vinyl')
+const createAsciiDocFile = require('./util/create-asciidoc-file')
 const path = require('node:path/posix')
 const sanitize = require('./util/sanitize')
 const unconvertInlineAsciiDoc = require('./util/unconvert-inline-asciidoc')
 
-function produceAggregateDocument (
+const AttributeEntryRx = /^:([^:-][^:]*):(?: .*)?$/
+const BuiltInNamedEntities = { amp: '&', apos: "'", gt: '>', lt: '<', nbsp: ' ', quot: '"' }
+const CharRefRx = /&(?:([a-z][a-z]+\d{0,2})|#(?:(\d{2,6})|x([a-z\d]{2,5})));/g
+const DiscardAttributes = 'doctype leveloffset assembly-style underscore'.split(' ')
+const ReservedIdNames = 'content header footnotes footer footer-text premable toc toctitle'.split(' ')
+
+function produceAssemblyFile (
   loadAsciiDoc,
   contentCatalog,
   componentVersion,
   outline,
-  doctype,
-  pages,
+  files,
   asciidocConfig,
   mutableAttributes,
-  sectionMergeStrategy = 'discrete'
+  assemblyModel
 ) {
-  const pagesInOutline = selectPagesInOutline(outline, pages)
+  const pagesByUrl = files.reduce((map, it) => (it.src.family === 'page' ? map.set(it.pub.url, it) : map), new Map())
+  if (outline.urlType === 'internal' && !pagesByUrl.get(outline.url) && !(outline.items || []).length) return
+  const pagesInOutline = selectPagesInOutline(outline, pagesByUrl, componentVersion)
   const navtitle = outline.content
-  const templateFile = contentCatalog.addFile({
-    src: {
-      component: componentVersion.name,
-      version: componentVersion.version,
-      module: 'ROOT',
-      family: 'page',
-      relative: generateSlug(navtitle),
-    },
-  })
-  const { dir: outDir, name: outName } = path.parse(templateFile.out.path)
-  const path_ = outName === templateFile.src.relative ? path.join(outDir, outName + '.adoc') : outDir + '.adoc'
-  contentCatalog.removeFile(templateFile)
-  const header = buildAsciiDocHeader(componentVersion, navtitle, doctype)
-  const body = aggregateAsciiDoc(
+  const buffer = mergeAsciiDoc(
     loadAsciiDoc,
     contentCatalog,
-    header,
+    buildAsciiDocHeader(componentVersion, navtitle, assemblyModel),
     componentVersion,
     outline,
+    files,
     pagesInOutline,
     asciidocConfig,
     mutableAttributes,
-    sectionMergeStrategy
+    assemblyModel
   )
-  return new File({
-    aggregate: true,
+  const rootLevel = assemblyModel.rootLevel
+  const stem = rootLevel === 0 ? 'index' : generateSlug(navtitle)
+  const downloadStem = [componentVersion.name, componentVersion.version, rootLevel === 0 ? '' : stem]
+    .filter((it) => it)
+    .join('-')
+  return createAsciiDocFile(contentCatalog, {
     asciidoc: asciidocConfig,
-    contents: Buffer.from([...header, ...body].join('\n') + '\n'),
-    mediaType: 'text/asciidoc',
-    path: path_,
+    assembler: { assembled: pagesInOutline.assembled, downloadStem, rootLevel },
+    contents: Buffer.from(buffer.join('\n') + '\n'),
     src: {
       component: componentVersion.name,
       version: componentVersion.version,
-      basename: path.basename(path_),
-      stem: path.basename(path_, '.adoc'),
-      extname: '.adoc',
+      module: 'ROOT',
+      family: 'export',
+      relative: stem + '.adoc',
     },
   })
 }
 
-function buildAsciiDocHeader (componentVersion, navtitle, doctype = 'book') {
+function buildAsciiDocHeader (componentVersion, navtitle, assemblyModel) {
+  const doctype = assemblyModel.doctype ?? 'book'
   const navtitlePlain = sanitize(navtitle)
   const navtitleAsciiDoc = unconvertInlineAsciiDoc(navtitle)
   let doctitle = navtitleAsciiDoc
@@ -67,7 +66,7 @@ function buildAsciiDocHeader (componentVersion, navtitle, doctype = 'book') {
   return [
     `= ${doctitle}`,
     ...(version ? [`:revnumber: ${version}`] : []),
-    ...(doctype === 'article' ? [] : [`:doctype: ${doctype}`]),
+    ...(doctype === 'article' ? [] : [`:doctype: ${doctype ?? 'book'}`]),
     ':underscore: _',
     // Q: should we pass these via the CLI so they cannot be modified?
     `:page-component-name: ${componentVersion.name}`,
@@ -78,46 +77,52 @@ function buildAsciiDocHeader (componentVersion, navtitle, doctype = 'book') {
   ]
 }
 
-function selectPagesInOutline (outlineEntry, pages) {
-  const page = outlineEntry.urlType === 'internal' ? pages.find((it) => it.pub.url === outlineEntry.url) : undefined
-  return (outlineEntry.items || []).reduce(
-    (accum, item) => new Map([...accum, ...selectPagesInOutline(item, pages)]),
-    new Map(
-      page && [
-        [`${page.src.module === 'ROOT' ? '' : page.src.module + ':'}${page.src.relative}`, page],
-        [page.pub.url, page],
-      ]
-    )
-  )
+function selectPagesInOutline (outlineEntry, pagesByUrl, componentVersion, accum) {
+  accum ??= Object.assign(new Map(), { assembled: { pages: new Map(), assets: new Set() } })
+  const page = outlineEntry.urlType === 'internal' ? pagesByUrl.get(outlineEntry.url) : undefined
+  if (page) {
+    if (page.src.component === componentVersion.name && page.src.version === componentVersion.version) {
+      accum.set(`${page.src.module === 'ROOT' ? '' : page.src.module + ':'}${page.src.relative}`, page)
+    }
+    accum.set(page.pub.url, page)
+  }
+  for (const item of outlineEntry.items || []) selectPagesInOutline(item, pagesByUrl, componentVersion, accum)
+  return accum
 }
 
-function aggregateAsciiDoc (
+function mergeAsciiDoc (
   loadAsciiDoc,
   contentCatalog,
-  header,
+  buffer,
   componentVersion,
   outlineEntry,
+  files,
   pagesInOutline,
   asciidocConfig,
   mutableAttributes,
-  sectionMergeStrategy,
+  assemblyModel,
   lastComponentVersion = componentVersion,
-  level = 0
+  level = 0,
+  supportsParts = false
 ) {
-  const buffer = []
   // TODO: we could try to be smart about it and make sure the page with fragment is included at least once
   if (outlineEntry.hash) return buffer
   const { content: navtitle, items = [], unresolved, urlType, url } = outlineEntry
+  const atDocumentRoot = !buffer.inBody
+  const atBookRoot = atDocumentRoot && !level && assemblyModel.doctype === 'book' && (supportsParts = true)
   const hasItems = items.length > 0
   const navtitlePlain = sanitize(navtitle)
   const navtitleAsciiDoc = unconvertInlineAsciiDoc(navtitle)
   const siteUrl = ((val) => {
     if (!val || val === '/') return ''
     return val.charAt(val.length - 1) === '/' ? val.slice(0, val.length - 1) : val
-  })(asciidocConfig.attributes['site-url'])
+  })(asciidocConfig.attributes['primary-site-url'] || asciidocConfig.attributes['site-url'])
+  const idSeparator = assemblyModel.xmlIds ? '-' : ':'
+  const idScopeSeparator = idSeparator.repeat(3)
+  const idCoordinateSeparator = idSeparator === '-' ? '----' : idSeparator
   // FIXME: ideally, resource ID would be stored in navigation so we can look up the page more efficiently
   let page = urlType === 'internal' && !unresolved ? pagesInOutline.get(url) : undefined
-  if (page && pagesInOutline.aggregated?.includes(page)) page = undefined
+  if (page && pagesInOutline.assembled.pages.has(page)) page = undefined
   if (page) {
     let contents = page.src.contents
     if (contents == null) return buffer
@@ -130,20 +135,51 @@ function aggregateAsciiDoc (
         .replace(/^(?:[ \t]*\r\n?|[ \t]*\n)+/, '')
         .trimRight()
     )
-    ;(pagesInOutline.aggregated ??= []).push(page)
-    page = new page.constructor(Object.assign({}, page, { contents, mediaType: 'text/asciidoc' }))
     const { component, version, module: module_, relative, origin } = page.src
-    const doc = loadAsciiDoc(page, contentCatalog, asciidocConfig)
-    const refs = doc.getCatalog().refs
+    const topicPrefix = ~relative.indexOf('/') ? path.dirname(relative) + '/' : ''
+    const pageAsAsciiDoc = new page.constructor(Object.assign({}, page, { contents, mediaType: page.src.mediaType }))
+    const doc = loadAsciiDoc(pageAsAsciiDoc, contentCatalog, asciidocConfig)
+    if (atDocumentRoot) {
+      const authors = doc.getAuthors()
+      if (authors.length) {
+        const authorLine = authors
+          .map((author) => {
+            const email = author.getEmail()
+            return email ? `${author.getName()} <${author.getEmail()}>` : author.getName()
+          })
+          .join('; ')
+        buffer.splice(1, 0, authorLine)
+      }
+    }
     // NOTE: in Antora, docname is relative src path from module without file extension
     const docname = doc.getAttribute('docname')
-    const docnameForId = docname.replace(/[/]/g, '::').replace(/[.]/g, '-')
-    const scopeId = component !== componentVersion.name
-    const idprefix =
-      (scopeId ? component + ':' : '') +
-      (module_ === 'ROOT' ? (scopeId ? ':' : '') : module_ + ':') +
-      docnameForId +
-      ':::'
+    const docnameForId = docname.replace(/[/.]/g, '-')
+    const qualifyId = component !== componentVersion.name
+    let idScope = docnameForId
+    let idPrefix
+    if (qualifyId) {
+      idScope = [component, module_ === 'ROOT' ? '' : module_, idScope].join(idCoordinateSeparator)
+    } else if (module_ !== 'ROOT') {
+      idScope = module_ + idCoordinateSeparator + idScope
+    } else if (ReservedIdNames.includes(docnameForId)) {
+      idScope = idPrefix = idScope + idScopeSeparator
+    }
+    idPrefix ??= idScope + idScopeSeparator
+    let pageFragment = ''
+    let pageRoles = ''
+    let pageStyle = doc.getAttribute('assembly-style', '')
+    let part
+    if ((part = pageStyle === 'part')) {
+      pageStyle = ''
+      if (!supportsParts) part = undefined
+    } else if ((part = pageStyle.endsWith('-part'))) {
+      pageStyle = pageStyle.slice(0, -5)
+      if (!supportsParts) part = undefined
+    }
+    let nextSectionLevel = 1
+    const lines = doc.getSourceLines()
+    const ignoreLines = []
+    buffer.inBody = true
     buffer.push('')
     buffer.push(`:docname: ${docname}`)
     if (component !== lastComponentVersion.name) {
@@ -168,25 +204,50 @@ function aggregateAsciiDoc (
     buffer.push(`:page-origin-refname: ${origin.branch || origin.tag}`)
     buffer.push(`:page-origin-reftype: ${origin.branch ? 'branch' : 'tag'}`)
     buffer.push(`:page-origin-refhash: ${origin.worktree ? '(worktree)' : origin.refhash}`)
-    let enclosed
-    // NOTE: if level is 0, doctitle has already been added and we're in the document header
-    if (level) {
-      if (level === 1 && navtitlePlain === componentVersion.title) {
-        level--
+    if (doc.hasHeader()) pageRoles = processDocumentHeader(doc, lines, buffer, ignoreLines)
+    let heading
+    if (pageStyle && (part && level === 1 ? (level = 0) : level) === 0) {
+      let htitleAsciiDoc = navtitleAsciiDoc
+      let htitlePlain = navtitlePlain
+      let htitleOverride
+      if (
+        (htitleOverride = pageStyle === 'preface' ? doc.getAttribute('preface-title') : undefined) ||
+        (atDocumentRoot && (htitleOverride = outlineEntry.navtitle))
+      ) {
+        htitleAsciiDoc = unconvertInlineAsciiDoc(htitleOverride)
+        htitlePlain = sanitize(htitleOverride)
+      }
+      if (atDocumentRoot && pageStyle === 'preface' && htitlePlain === componentVersion.title) {
+        assemblyModel = Object.assign({}, assemblyModel, { sectionMergeStrategy: 'discrete' })
       } else {
-        let hlevel = level + 1
-        if (hlevel > 6) {
-          hlevel = 6
-          buffer.push(`[discrete#${idprefix}]`)
+        if (atDocumentRoot) {
+          pageFragment = `#${idPrefix}${doc.getId() ?? pageStyle}`
+          buffer.unshift(`[#${idScope}]`)
         } else {
-          buffer.push(`[#${idprefix}]`)
+          pageFragment = `#${idScope}`
+        }
+        heading = { title: htitleAsciiDoc, level: part ? 1 : 2 }
+        nextSectionLevel++
+      }
+    } else if (level) {
+      if (atDocumentRoot && navtitlePlain === componentVersion.title) {
+        level--
+      } else {
+        pageFragment = `#${idScope}`
+        if (part && level === 1) level--
+        if ((heading = { title: navtitleAsciiDoc, level: level + 1 }).level > 6) {
+          Object.assign(heading, { level: 6, style: `discrete.h${heading.level}` })
         }
-        buffer.push(`${'='.repeat(hlevel)} ${navtitleAsciiDoc}`)
       }
-    } else {
-      header.unshift(`[#${idprefix}]`)
     }
-    if (sectionMergeStrategy === 'enclose' && hasItems && doc.hasSections()) {
+    if (heading) {
+      buffer.push(`[${heading.style ?? pageStyle}${pageFragment}${pageRoles}]`)
+      buffer.push(`${'='.repeat(heading.level)} ${heading.title}`)
+    } else if (atDocumentRoot) {
+      buffer.unshift(`[#${idScope}]`)
+    }
+    let enclosed
+    if (assemblyModel.sectionMergeStrategy === 'enclose' && hasItems && doc.hasSections()) {
       enclosed = true
       // TODO: make overview section title configurable
       //let overviewTitle = doc.getDocumentTitle()
@@ -205,23 +266,43 @@ function aggregateAsciiDoc (
       }
       let hlevel = level + 2
       if (hlevel > 6) {
+        const blockStyle = `discrete.h${hlevel}`
         hlevel = 6
-        buffer.push(syntheticId ? `[discrete#${syntheticId}]` : '[discrete]')
+        buffer.push(syntheticId ? `[${blockStyle}#${syntheticId}]` : `[${blockStyle}]`)
       } else if (syntheticId) {
         buffer.push(`[#${syntheticId}]`)
       }
       buffer.push(`${'='.repeat(hlevel)} ${overviewTitle}`)
       if (toggleSectids) buffer.push(':sectids:')
     }
-    const lines = doc.getSourceLines()
-    const ignoreLines = []
-    // TODO: think more about when multipart is allowed; perhaps configurable
-    if (doc.hasSections()) fixSectionLevels(doc.getSections(), level === 0)
+    pagesInOutline.assembled.pages.set(page, pageFragment)
+    if (doc.hasSections()) {
+      fixSectionLevels(doc.getSections(), atBookRoot && !pageStyle ? undefined : nextSectionLevel)
+    }
     const allBlocks = doc.findBy({ traverse_documents: true }, (it) =>
       it.getContext() === 'document'
         ? it.getDocument().isNested()
         : !(it.getContext() === 'table_cell' && it.getStyle() === 'asciidoc')
     )
+    if (doc.getDoctype() === 'manpage') {
+      const firstSectionIdx = doc.hasSections() ? doc.getSections()[0].getLineNumber() - 1 : lines.length
+      for (let idx = 0; idx < firstSectionIdx; idx++) {
+        if (~ignoreLines.indexOf(idx)) continue
+        const line = lines[idx]
+        if (line.startsWith('== ') && line.length > 3) {
+          allBlocks.unshift({
+            getContext: () => 'section',
+            getDocument: () => doc,
+            getId: () => doc.getAttribute('manname-id'),
+            getLineNumber: () => idx + 1,
+            getSectionName: () => undefined,
+            level: 1,
+          })
+          break
+        }
+      }
+    }
+    const refs = doc.getCatalog().refs
     allBlocks.forEach((block) => {
       const contentModel = block.content_model
       if (
@@ -249,7 +330,7 @@ function aggregateAsciiDoc (
       }
     })
     let skipping
-    for (let idx = 0, len = lines.length; idx < len; idx++) {
+    for (let idx = 0, lastIdx = lines.length - 1; idx <= lastIdx; idx++) {
       if (~ignoreLines.indexOf(idx)) continue
       let line = lines[idx]
       if (line.startsWith('//')) {
@@ -265,7 +346,11 @@ function aggregateAsciiDoc (
       } else if (skipping) {
         continue
       }
-      if (line.charAt() === ':' && /^:(?:leveloffset: .*|!leveloffset:|leveloffset!:)$/.test(line)) {
+      if (
+        line.charAt() === ':' &&
+        ~line.indexOf(':', 2) &&
+        (line.match(AttributeEntryRx) || ['', ''])[1].replace('!', '') === 'leveloffset'
+      ) {
         if (lines[idx - 1] === '') lines[idx - 1] = undefined
         lines[idx] = undefined
         continue
@@ -276,16 +361,16 @@ function aggregateAsciiDoc (
           if (!refs['$key?'](refid) && (~refid.indexOf(' ') || refid.toLowerCase() !== refid)) {
             if ((refid = doc.$resolve_id(refid))['$nil?']()) return m
           }
-          return `<<${idprefix}${refid}${text ? ',' + text : ''}>>`
+          return `<<${idPrefix}${refid}${text ? ',' + text : ''}>>`
         })
       }
       // NOTE: the next check takes care of inline and block anchors
       if (~line.indexOf('[[')) {
-        line = line.replace(/\[\[([\p{Alpha}_:][\p{Alpha}0-9_\-:.]*)(|, *.+?)\]\]/gu, `[[${idprefix}$1$2]]`)
+        line = line.replace(/\[\[([\p{Alpha}_:][\p{Alpha}0-9_\-:.]*)(|, *.+?)\]\]/gu, `[[${idPrefix}$1$2]]`)
       }
       if (~line.indexOf('xref:')) {
         // Q: should we allow : as first character of target?
-        line = line.replace(/(?<![\\+])xref:([\p{Alpha}0-9_/.{#].*?)\[(|.*?[^\\])\]/gu, (m, target, text) => {
+        line = line.replace(/(?<![\\+])xref:((?:\.\/)?[\p{Alpha}0-9_/.{#].*?)\[(|.*?[^\\])\]/gu, (m, target, text) => {
           let pagePart, fragment, targetPage
           const hashIdx = target.indexOf('#')
           if (~hashIdx) {
@@ -302,8 +387,8 @@ function aggregateAsciiDoc (
           if (!pagePart) {
             // Q: should we validate the internal ID here?
             return text && ~text.indexOf('=')
-              ? `xref:${idprefix}${fragment}[${text}]`
-              : `<<${idprefix}${fragment}${text ? ',' + text.replace(/\\]/g, ']') : ''}>>`
+              ? `xref:${idPrefix}${fragment}[${text}]`
+              : `<<${idPrefix}${fragment}${text ? ',' + text.replace(/\\]/g, ']') : ''}>>`
           }
           if (~pagePart.indexOf('@') || /:.*:/.test(pagePart)) {
             if (siteUrl && (targetPage = contentCatalog.resolvePage(pagePart, page.src)) && targetPage.out) {
@@ -312,24 +397,30 @@ function aggregateAsciiDoc (
             }
             // TODO: handle unresolved page better
             return m
-          } else if (pagePart.indexOf(':') < 0) {
-            if (module_ !== 'ROOT') pagePart = `${module_}:${pagePart}`
-          } else if (pagePart.startsWith('ROOT:')) {
-            pagePart = pagePart.slice(5)
           }
-          if (!(targetPage = pagesInOutline.get(pagePart))) {
-            if (siteUrl && (targetPage = contentCatalog.resolvePage(pagePart, page.src)) && targetPage.out) {
+          let targetModule
+          const colonIdx = pagePart.indexOf(':')
+          if (~colonIdx) {
+            targetModule = pagePart.slice(0, colonIdx)
+            pagePart = pagePart.slice(colonIdx + 1)
+          } else {
+            targetModule = module_
+          }
+          if (pagePart.startsWith('./')) pagePart = topicPrefix + pagePart.slice(2)
+          const pageResourceRef = targetModule === 'ROOT' ? pagePart : `${targetModule}:${pagePart}`
+          if (!(targetPage = pagesInOutline.get(pageResourceRef))) {
+            if (siteUrl && (targetPage = contentCatalog.resolvePage(pageResourceRef, page.src)) && targetPage.out) {
               text ||= targetPage.asciidoc?.xreftext || target
               return `${siteUrl}${targetPage.pub.url}${fragment && '#' + fragment}[${text}]`
             }
             // TODO: handle unresolved page better
             return m
           }
-          pagePart = pagePart
-            .replace(/[/]/g, '::')
-            .replace(/\.adoc$/, '')
-            .replace(/[.]/g, '-')
-          const refid = `${pagePart}:::${fragment}`
+          if (targetModule !== 'ROOT') pagePart = `${targetModule}${idCoordinateSeparator}${pagePart}`
+          pagePart = pagePart.replace(/\.adoc$/, '').replace(/[/.]/g, '-')
+          const refid = fragment
+            ? `${pagePart}${idScopeSeparator}${fragment}`
+            : pagePart + (ReservedIdNames.includes(pagePart) ? idScopeSeparator : '')
           return `<<${refid}${text && text !== targetPage.title ? ',' + text.replace(/\\]/g, ']') : ''}>>`
         })
       }
@@ -354,7 +445,7 @@ function aggregateAsciiDoc (
             const image = contentCatalog.resolveResource(target, page.src, 'image', ['image'])
             // TODO: handle (or report) unresolved image better
             if (image?.out) {
-              image.out.assembled = true
+              pagesInOutline.assembled.assets.add(image)
               return `image:${image.out.path.replace(/_/g, '{underscore}')}[${attrlist}]`
             }
           }
@@ -374,21 +465,18 @@ function aggregateAsciiDoc (
       const context = block.getContext()
       let idx = lineno - 1
       if (context === 'section' && !block.getDocument().isNested()) {
-        if (block.getSectionName() === 'header') {
-          lines[idx] = undefined
-          return
-        }
-        let blockStyle = sectionMergeStrategy === 'discrete' ? 'discrete' : undefined
+        if (block.getSectionName() === 'header') return
+        let blockStyle = (assemblyModel.sectionMergeStrategy || 'discrete') === 'discrete' ? 'discrete' : undefined
         lines[idx] = lines[idx].replace(/^=+ (.+)/, (_, rest) => {
           let targetMarkerLength = block.level + 1 + level + (enclosed ? 1 : 0)
           if (targetMarkerLength > 6) {
+            blockStyle = `discrete.h${targetMarkerLength}`
             targetMarkerLength = 6
-            blockStyle = 'discrete'
           }
           return '='.repeat(targetMarkerLength) + ' ' + rest
         })
         // NOTE: ID will be undefined if sectids are turned off
-        if (block.getId()) rewriteStyleAttribute(block, lines, idx, idprefix, blockStyle)
+        if (block.getId()) rewriteStyleAttribute(block, lines, idx, idPrefix, blockStyle)
       } else {
         if (context === 'image') {
           let line = lines[idx] || ''
@@ -414,7 +502,7 @@ function aggregateAsciiDoc (
               // FIXME: handle (or report) case when image is not resolved
               if (image?.out) {
                 const boxedAttrlist = line.slice(line.indexOf('['))
-                image.out.assembled = true
+                pagesInOutline.assembled.assets.add(image)
                 lines[idx] = `${prefix}image::${image.out.path}${boxedAttrlist}`
               }
             }
@@ -424,11 +512,14 @@ function aggregateAsciiDoc (
           // nested document
           idx = (block.getHeader().getLineNumber() || idx + 1) - 1
         }
-        if (block.getId()) rewriteStyleAttribute(block, lines, idx, idprefix)
+        if (block.getId()) rewriteStyleAttribute(block, lines, idx, idPrefix)
       }
     })
-    buffer.push(...lines.filter((it) => it !== undefined))
-    const attributeEntries = Object.entries(doc.attributes_defined_in_header || {})
+    safePush(
+      buffer,
+      lines.filter((it) => it !== undefined)
+    )
+    const attributeEntries = Object.entries(doc.source_header_attributes?.$$smap || {})
     if (attributeEntries.length) {
       const resolvedAttributeEntries = attributeEntries.reduce(
         (accum, [name, val]) => {
@@ -440,27 +531,20 @@ function aggregateAsciiDoc (
             } else if (val !== initialVal) {
               accum.push(`:${name}:${initialVal ? ' ' + initialVal : ''}`)
             }
-          } else if (
-            !(
-              val == null ||
-              doc.isAttributeLocked(name) ||
-              name === 'doctype' ||
-              name === 'leveloffset' ||
-              name === 'underscore'
-            )
-          ) {
+          } else if (!(val == null || doc.isAttributeLocked(name) || DiscardAttributes.includes(name))) {
             accum.push(`:!${name}:`)
           }
           return accum
         },
         ['']
       )
-      if (resolvedAttributeEntries.length > 1) buffer.push(...resolvedAttributeEntries)
+      if (resolvedAttributeEntries.length > 1) safePush(buffer, resolvedAttributeEntries)
     }
   } else if (level) {
-    if (level === 1 && navtitlePlain === componentVersion.title) {
+    if (atDocumentRoot && navtitlePlain === componentVersion.title) {
       level--
     } else {
+      buffer.inBody = true
       buffer.push('')
       // NOTE: try to toggle sectids; otherwise, fallback to globally unique synthetic ID
       // Q: should we unset docname, page-module, etc?
@@ -480,7 +564,7 @@ function aggregateAsciiDoc (
       if (urlType === 'external') {
         sectionTitle = `${url}[${navtitleAsciiDoc.replace(/\]/g, '\\]')}]`
       } else if (urlType === 'internal' && !unresolved && siteUrl) {
-        const resource = contentCatalog.getFiles().find((it) => it.out && it.pub.url === url)
+        const resource = files.find((it) => it.pub.url === url)
         if (resource) sectionTitle = `${siteUrl}${resource.pub.url}[${navtitleAsciiDoc.replace(/\]/g, '\\]')}]`
       }
       let hlevel = level + 1
@@ -495,50 +579,112 @@ function aggregateAsciiDoc (
     }
   }
 
-  const nextLevel = level + 1
   if (hasItems) {
+    const nextLevel = level + 1
     // NOTE: drop first child if same as parent; should we keep if content is different?
     ;(urlType === 'internal' && urlType === items[0].urlType && url === items[0].url && !items[0].items
       ? items.slice(1)
       : items
     ).forEach((item) => {
-      buffer.push(
-        ...aggregateAsciiDoc(
-          loadAsciiDoc,
-          contentCatalog,
-          header,
-          componentVersion,
-          item,
-          pagesInOutline,
-          asciidocConfig,
-          mutableAttributes,
-          sectionMergeStrategy,
-          lastComponentVersion,
-          nextLevel
-        )
+      mergeAsciiDoc(
+        loadAsciiDoc,
+        contentCatalog,
+        buffer,
+        componentVersion,
+        item,
+        files,
+        pagesInOutline,
+        asciidocConfig,
+        mutableAttributes,
+        assemblyModel,
+        lastComponentVersion,
+        nextLevel,
+        atBookRoot
       )
     })
   }
   return buffer
 }
 
+function processDocumentHeader (doc, lines, buffer, ignoreLines) {
+  const doctitleIdx = doc.getHeader().getLineNumber() - 1
+  const end = doc.getBlocks()[0]?.getLineNumber() ?? lines.length
+  let belowDoctitle
+  let open
+  const implicitLines = []
+  for (let idx = 0; idx < end; idx++) {
+    if (idx === doctitleIdx) {
+      lines[idx] = undefined
+      ignoreLines.push(idx)
+      belowDoctitle = true
+      continue
+    }
+    const line = lines[idx]
+    if (open === ':' || open === '-:') {
+      if (line) {
+        if (open === ':') buffer.push(line)
+        if (!line.endsWith(' \\')) open = undefined
+      } else {
+        open = undefined
+      }
+    } else if (line) {
+      const chr0 = line.charAt()
+      let attributeEntryMatch
+      if (chr0 === '/' && line.charAt(1) === '/') {
+        if (line.startsWith('////')) {
+          open = open ? (open === line ? undefined : open) : line
+        } else if (belowDoctitle && !open && line.charAt(2) === '/') {
+          break
+        }
+        buffer.push(line)
+      } else if (open) {
+        buffer.push(line)
+      } else if (chr0 === ':' && ~line.indexOf(':', 2) && (attributeEntryMatch = line.match(AttributeEntryRx))) {
+        const attributeName = attributeEntryMatch[1].replace('!', '')
+        if (DiscardAttributes.includes(attributeName)) {
+          if (line.endsWith(' \\')) open = '-:'
+        } else {
+          if (line.endsWith(' \\')) open = ':'
+          buffer.push(line)
+        }
+      } else if (belowDoctitle) {
+        if (implicitLines.length === 2 || !/[\p{Alpha}0-9]/u.test(chr0)) break
+        implicitLines.push(line)
+      }
+    } else if (belowDoctitle) {
+      break
+    }
+    lines[idx] = undefined
+    ignoreLines.push(idx)
+  }
+  return doc
+    .getRoles()
+    .map((role) => '.' + role)
+    .join('')
+}
+
 function generateSlug (title) {
   return title
     .toLowerCase()
-    .replace(/&.+?;|[^ \p{Alpha}0-9_\-.]/gu, '')
-    .replace(/[ _.]/g, '-')
-    .replace(/--+/g, '-')
+    .replace(/<[^>]+>/g, '')
+    .replace(CharRefRx, (_, name, dec, hex) => {
+      if (name) return BuiltInNamedEntities[name] ?? '?'
+      return String.fromCharCode(dec ? parseInt(dec, 10) : parseInt(hex, 16))
+    })
+    .replace(/[\x27\u2019]/g, '')
+    .replace(/[^\p{Alpha}0-9\-]/gu, '-')
+    .replace(/^-+|-+$|(-)-+/g, '$1')
 }
 
-function fixSectionLevels (sections, multipart) {
+function fixSectionLevels (sections, expectedLevel) {
   sections.forEach((sect) => {
-    const targetLevel = sect.getParent().getLevel() + 1
-    if (multipart ? sect.getLevel() > targetLevel : sect.getLevel() !== targetLevel) sect.level = targetLevel
-    if (sect.hasSections()) fixSectionLevels(sect.getSections(), multipart)
+    const forceLevel = expectedLevel ?? Math.min(1, sect.getLevel())
+    if (sect.getLevel() !== forceLevel) sect.level = forceLevel
+    if (sect.hasSections()) fixSectionLevels(sect.getSections(), forceLevel + 1)
   })
 }
 
-function rewriteStyleAttribute (block, lines, idx, idprefix, replacementStyle = '') {
+function rewriteStyleAttribute (block, lines, idx, idPrefix, replacementStyle = '') {
   let prevLine = lines[idx - 1]
   const char0 = prevLine?.charAt()
   if (char0) {
@@ -548,7 +694,7 @@ function rewriteStyleAttribute (block, lines, idx, idprefix, replacementStyle =
         prevLine.charAt(1) === '[' &&
         /^\[\[(?:|[\p{Alpha}_:][\p{Alpha}0-9_\-:.]*(?:, *.+)?)\]\]$/u.test(prevLine))
     ) {
-      return rewriteStyleAttribute(block, lines, idx - 1, idprefix, replacementStyle)
+      return rewriteStyleAttribute(block, lines, idx - 1, idPrefix, replacementStyle)
     }
   }
   let cellSpec
@@ -564,27 +710,27 @@ function rewriteStyleAttribute (block, lines, idx, idprefix, replacementStyle =
     let rawStyle
     const commaIdx = prevLine.indexOf(',')
     if (~commaIdx) {
-      rawStyle = prevLine.slice(1, commaIdx - 1)
+      rawStyle = prevLine.slice(1, commaIdx)
       if (~rawStyle.indexOf('=')) rawStyle = undefined
     } else if (!~prevLine.indexOf('=')) {
-      rawStyle = prevLine.slice(1, prevLine.length - 2)
+      rawStyle = prevLine.slice(1, prevLine.length - 1)
     }
     if (rawStyle) {
       if (~rawStyle.indexOf('#')) {
-        prevLine = prevLine.replace(/#[^.%,\]]+/, `#${idprefix}${block.getId()}`)
+        prevLine = prevLine.replace(/#[^.%,\]]+/, `#${idPrefix}${block.getId()}`)
         if (replacementStyle) prevLine = prevLine.replace(/^[^#.%,\]]+/, `[${replacementStyle}`)
       } else {
         prevLine = `[${
           replacementStyle ? rawStyle.replace(/^[^.%]*/, replacementStyle) : rawStyle
-        }#${idprefix}${block.getId()}${prevLine.slice(rawStyle.length + 1)}`
+        }#${idPrefix}${block.getId()}${prevLine.slice(rawStyle.length + 1)}`
       }
     } else {
-      prevLine = `[${replacementStyle}#${idprefix}${block.getId()}${rawStyle == null ? ',' : ''}${prevLine.slice(1)}`
+      prevLine = `[${replacementStyle}#${idPrefix}${block.getId()}${rawStyle == null ? ',' : ''}${prevLine.slice(1)}`
     }
     if (cellSpec) prevLine = `${cellSpec}|${prevLine}`
     lines[idx - 1] = prevLine
   } else {
-    lines.splice(idx, 0, `[${replacementStyle}#${idprefix}${block.getId()}]`)
+    lines.splice(idx, 0, `[${replacementStyle}#${idPrefix}${block.getId()}]`)
   }
 }
 
@@ -596,4 +742,14 @@ function getObjectId (obj) {
   return global.Opal.uid()
 }
 
-module.exports = produceAggregateDocument
+function safePush (onto, entries) {
+  try {
+    onto.push(...entries)
+  } catch (err) {
+    /* istanbul ignore if */
+    if (!(err instanceof RangeError)) throw err
+    for (const e of entries) onto.push(e)
+  }
+}
+
+module.exports = produceAssemblyFile