@antora/assembler 1.0.0-alpha.8 → 1.0.0-alpha.9

package/lib/assemble-content.js

@@ -5,7 +5,7 @@ const produceAggregateDocuments = require('./produce-aggregate-documents')
 const PromiseQueue = require('./util/promise-queue')
 const runCommand = require('./util/run-command')
 
-async function assembleContent (playbook, contentCatalog, converter, { siteCatalog, configSource }) {
+async function assembleContent (playbook, contentCatalog, convertDocument, { siteCatalog, configSource }) {
   // Q: could we get ContentCatalog#getComponentVersionStartPage() in Antora core?
   if (typeof contentCatalog.getComponentVersionStartPage !== 'function') {
     contentCatalog.getComponentVersionStartPage = function (component, version) {
@@ -17,14 +17,14 @@ async function assembleContent (playbook, contentCatalog, converter, { siteCatal
   const generatorFunctions = this ? this.getFunctions() : {}
   const { loadAsciiDoc = require('@antora/asciidoc-loader') } = generatorFunctions
   const aggregateDocuments = produceAggregateDocuments(loadAsciiDoc, contentCatalog, assemblerConfig)
-  if (!converter) return aggregateDocuments
+  if (!convertDocument) return aggregateDocuments
   const { publishSite: publishFiles = require('@antora/site-publisher') } = generatorFunctions
   const buildConfig = assemblerConfig.build
   await prepareWorkspace(publishFiles, aggregateDocuments, contentCatalog, buildConfig)
-  // TODO: pass more information to converter so it doesn't have to compute internal stuff
+  // TODO: pass more information to convertDocument so it doesn't have to compute internal stuff
   // Q: don't we need to pass in the combined/resolved AsciiDoc attributes per file or component version?
   return new PromiseQueue({ concurrency: buildConfig.processLimit })
-    .add(aggregateDocuments.map((doc) => () => converter.call(this, doc, buildConfig, runCommand)))
+    .add(aggregateDocuments.map((doc) => () => convertDocument.call(this, doc, buildConfig, runCommand)))
     .toPromise()
     .then((files) => {
       if (buildConfig.publish && siteCatalog) siteCatalog.addFiles(files)

package/lib/produce-aggregate-document.js

@@ -3,6 +3,7 @@
 const File = require('vinyl')
 const path = require('node:path/posix')
 const sanitize = require('./util/sanitize')
+const unconvertInlineAsciiDoc = require('./util/unconvert-inline-asciidoc')
 
 function produceAggregateDocument (
   loadAsciiDoc,
@@ -26,7 +27,8 @@ function produceAggregateDocument (
       relative: generateSlug(navtitle),
     },
   })
-  const path_ = `${templateFile.out.path}.adoc`
+  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(
@@ -57,7 +59,8 @@ function produceAggregateDocument (
 }
 
 function buildAsciiDocHeader (componentVersion, navtitle, doctype = 'book') {
-  const [navtitlePlain, navtitleAsciiDoc] = sanitize(navtitle)
+  const navtitlePlain = sanitize(navtitle)
+  const navtitleAsciiDoc = unconvertInlineAsciiDoc(navtitle)
   let doctitle = navtitleAsciiDoc
   if (navtitlePlain !== componentVersion.title) doctitle = `${componentVersion.title}: ${doctitle}`
   const version = componentVersion.version === 'master' ? '' : componentVersion.version
@@ -104,8 +107,10 @@ function aggregateAsciiDoc (
   const buffer = []
   // TODO: we could try to be smart about it and make sure the page with fragment is included at least once
   if (outlineEntry.hash) return buffer
-  const { content: navtitle, items, unresolved, urlType, url } = outlineEntry
-  const [navtitlePlain, navtitleAsciiDoc, navtitlePass] = sanitize(navtitle)
+  const { content: navtitle, items = [], unresolved, urlType, url } = outlineEntry
+  const hasItems = items.length > 0
+  const navtitlePlain = sanitize(navtitle)
+  const navtitleAsciiDoc = unconvertInlineAsciiDoc(navtitle)
   const siteUrl = ((val) => {
     if (!val || val === '/') return ''
     return val.charAt(val.length - 1) === '/' ? val.slice(0, val.length - 1) : val
@@ -181,7 +186,7 @@ function aggregateAsciiDoc (
     } else {
       header.unshift(`[#${idprefix}]`)
     }
-    if (sectionMergeStrategy === 'enclose' && items && doc.hasSections()) {
+    if (sectionMergeStrategy === 'enclose' && hasItems && doc.hasSections()) {
       enclosed = true
       // TODO: make overview section title configurable
       //let overviewTitle = doc.getDocumentTitle()
@@ -340,9 +345,7 @@ function aggregateAsciiDoc (
               relative,
             })
           // TODO: handle unresolved attachment page
-          return attachment && attachment.out
-            ? `${siteUrl}${attachment.pub.url.replace(/_/g, '{underscore}')}[${text}]`
-            : m
+          return attachment?.out ? `${siteUrl}${attachment.pub.url.replace(/_/g, '{underscore}')}[${text}]` : m
         })
       }
       if (~line.indexOf('image:') && !line.startsWith('image::')) {
@@ -350,7 +353,7 @@ function aggregateAsciiDoc (
           if (isResourceSpec(target)) {
             const image = contentCatalog.resolveResource(target, page.src, 'image', ['image'])
             // TODO: handle (or report) unresolved image better
-            if (image && image.out) {
+            if (image?.out) {
               image.out.assembled = true
               return `image:${image.out.path.replace(/_/g, '{underscore}')}[${attrlist}]`
             }
@@ -409,7 +412,7 @@ function aggregateAsciiDoc (
             if (isResourceSpec(target)) {
               const image = contentCatalog.resolveResource(target, page.src, 'image', ['image'])
               // FIXME: handle (or report) case when image is not resolved
-              if (image && image.out) {
+              if (image?.out) {
                 const boxedAttrlist = line.slice(line.indexOf('['))
                 image.out.assembled = true
                 lines[idx] = `${prefix}image::${image.out.path}${boxedAttrlist}`
@@ -475,13 +478,10 @@ function aggregateAsciiDoc (
       }
       let sectionTitle = navtitleAsciiDoc
       if (urlType === 'external') {
-        sectionTitle = `${url}[${navtitlePass ? navtitleAsciiDoc : navtitleAsciiDoc.replace(/\]/g, '\\]')}]`
+        sectionTitle = `${url}[${navtitleAsciiDoc.replace(/\]/g, '\\]')}]`
       } else if (urlType === 'internal' && !unresolved && siteUrl) {
-        const resource = contentCatalog.getFiles().find((it) => it.pub.url === url)
-        if (resource?.out) {
-          sectionTitle = navtitlePass ? navtitleAsciiDoc : navtitleAsciiDoc.replace(/\]/g, '\\]')
-          sectionTitle = `${siteUrl}${resource.pub.url}[${sectionTitle}]`
-        }
+        const resource = contentCatalog.getFiles().find((it) => it.out && it.pub.url === url)
+        if (resource) sectionTitle = `${siteUrl}${resource.pub.url}[${navtitleAsciiDoc.replace(/\]/g, '\\]')}]`
       }
       let hlevel = level + 1
       if (hlevel > 6) {
@@ -496,7 +496,7 @@ function aggregateAsciiDoc (
   }
 
   const nextLevel = level + 1
-  if (items) {
+  if (hasItems) {
     // 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)

package/lib/produce-aggregate-documents.js

@@ -130,10 +130,13 @@ function isResourceRef (target) {
 // in this case, if there's only a single navigation menu with no title, promote each top-level item to a menu
 function prepareOutlines (navigation, rootEntry, rootLevel) {
   if (rootLevel === 0 || navigation.length === 1) {
-    const navBranch =
-      navigation.length === 1
-        ? navigation[0]
-        : { items: navigation.reduce((navTree, it) => navTree.concat(it.content ? it : it.items), []) }
+    let navBranch
+    if (navigation.length === 1) {
+      navBranch = navigation[0]
+    } else {
+      const items = navigation.reduce((accum, it) => accum.concat(it.content ? it : it.items), [])
+      navBranch = items.length ? { items } : {}
+    }
     return rootLevel === 0 || navBranch.content ? [Object.assign(rootEntry, navBranch)] : navBranch.items
   }
   return navigation.reduce((navTree, it) => navTree.concat(it.content ? it : it.items), [rootEntry])

package/lib/util/sanitize.js

@@ -5,11 +5,9 @@ const XML_SPECIAL_CHARS = { '&lt;': '<', '&gt;': '>', '&amp;': '&' }
 const XML_SPECIAL_CHARS_RX = /&(?:[lg]t|amp);/g
 
 function sanitize (str) {
-  return ~str.indexOf('<')
-    ? [str.replace(XML_TAG_RX, '').replace(XML_SPECIAL_CHARS_RX, (m) => XML_SPECIAL_CHARS[m]), `+++${str}+++`, true]
-    : ~str.indexOf('&')
-        ? Array(2).fill(str.replace(XML_SPECIAL_CHARS_RX, (m) => XML_SPECIAL_CHARS[m]))
-        : [str, str]
+  if (~str.indexOf('<')) str = str.replace(XML_TAG_RX, '')
+  if (~str.indexOf('&')) str = str.replace(XML_SPECIAL_CHARS_RX, (m) => XML_SPECIAL_CHARS[m])
+  return str
 }
 
 module.exports = sanitize

package/lib/util/unconvert-inline-asciidoc.js

@@ -0,0 +1,95 @@
+'use strict'
+
+const ATTRIBUTE_REFERENCE_RX = /\{[a-z0-9_][a-z0-9_-]*\}/g
+const STRICT_WORD_CHAR_RX = /[\p{L}\d_]/u
+const WORD_CHAR_RX = /[\p{L}\d_;}:<>]/u
+
+const MARK_FOR_TAG = { code: '`', em: '_', mark: '#', span: '#', strong: '*' }
+const SKIP_SPAN = { icon: '<i ', image: '<img ' }
+
+module.exports = (str) => {
+  if (!str) return str
+  let matchIndex = str.indexOf('<')
+  if (!~matchIndex) return ~str.indexOf('{') ? str.replace(ATTRIBUTE_REFERENCE_RX, '\\$&') : str
+  let current = { contents: '' }
+  const stack = [current]
+  let lastIndex = 0
+  do {
+    if (matchIndex > lastIndex) {
+      const matched = str.slice(lastIndex, matchIndex)
+      current.contents += ~matched.indexOf('{') ? matched.replace(ATTRIBUTE_REFERENCE_RX, '\\$&') : matched
+    }
+    const isCloseTag = str[++matchIndex] === '/' ? ++matchIndex : false
+    let tagName = str.slice(matchIndex, (lastIndex = str.indexOf('>', matchIndex) + 1) - 1)
+    if (isCloseTag) {
+      const parent = current // TODO expect tagName to equal current.tagName
+      stack.pop()
+      current = stack[stack.length - 1]
+      if (parent.mark) {
+        let { contents, mark, id, role } = parent
+        const attrlist = (id ? '#' + id : '') + (role ? '.' + role.replace(/ /g, '.') : '')
+        if (
+          current.mark === mark ||
+          current.mark === '_' ||
+          isWordChar(str.charAt(lastIndex), true) ||
+          isWordChar(current.contents[current.contents.length - 1] || current.mark)
+        ) {
+          mark = mark.repeat(2)
+        }
+        current.contents += (attrlist ? '[' + attrlist + ']' : '') + mark + contents + mark
+      } else {
+        current.contents += parent.contents
+      }
+    } else {
+      let attrs, attrlistIndex, role
+      if (~(attrlistIndex = tagName.indexOf(' '))) {
+        role = (attrs = parseAttrlist(tagName.slice(attrlistIndex))).class
+        tagName = tagName.slice(0, attrlistIndex)
+      }
+      if (tagName === 'img') {
+        current.contents += 'image:' + attrs.src + '[' + attrs.alt + ']'
+      } else if (tagName === 'i' && current.tagName === 'span' && current.role === 'icon') {
+        current.contents += 'icon:' + role.slice(6) + '[]'
+        lastIndex += 4
+      } else {
+        let check, mark
+        const id = attrs?.id
+        if (tagName !== 'span' || id || (role && !((check = SKIP_SPAN[role]) && str.startsWith(check, lastIndex)))) {
+          mark = MARK_FOR_TAG[tagName]
+        }
+        stack.push((current = { tagName, role, id, mark, contents: '' }))
+      }
+    }
+  } while (~(matchIndex = str.indexOf('<', lastIndex)))
+  const rest = str.slice(lastIndex)
+  if (rest) current.contents += ~rest.indexOf('{') ? rest.replace(ATTRIBUTE_REFERENCE_RX, '\\$&') : rest
+  return current.contents
+}
+
+function parseAttrlist (str) {
+  let lastIndex = 0
+  const attrs = {}
+  while (str.charAt(lastIndex++) === ' ') {
+    const spaceIndex = str.indexOf(' ', lastIndex)
+    const equalsIndex = str.indexOf('=', lastIndex)
+    if (~spaceIndex && spaceIndex < equalsIndex) {
+      attrs[str.slice(lastIndex, (lastIndex = spaceIndex))] = true
+    } else if (~equalsIndex) {
+      const name = str.slice(lastIndex, equalsIndex)
+      const valueIndex = equalsIndex + 1
+      attrs[name] =
+        str.charAt(valueIndex) === '"'
+          ? str.slice(valueIndex + 1, (lastIndex = str.indexOf('"', valueIndex + 1) + 1) - 1)
+          : str.slice(valueIndex, (lastIndex = ~spaceIndex ? spaceIndex : str.length))
+    } else {
+      attrs[str.slice(lastIndex)] = true
+      break
+    }
+  }
+  return attrs
+}
+
+function isWordChar (str, strict) {
+  if (!str) return false
+  return (strict ? STRICT_WORD_CHAR_RX : WORD_CHAR_RX).test(str)
+}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@antora/assembler",
-  "version": "1.0.0-alpha.8",
+  "version": "1.0.0-alpha.9",
   "description": "An extension library for Antora that assembles content from multiple pages into a single AsciiDoc file to converted and publish.",
   "license": "MPL-2.0",
   "author": "OpenDevise Inc. (https://opendevise.com)",
@@ -28,6 +28,9 @@
     "./produce-aggregate-documents": "./lib/produce-aggregate-documents.js",
     "./select-mutable-attributes": "./lib/select-mutable-attributes.js"
   },
+  "imports": {
+    "#unconvert-inline-asciidoc": "./lib/util/unconvert-inline-asciidoc.js"
+  },
   "dependencies": {
     "@antora/expand-path-helper": "~2.0",
     "braces": "~3.0",