resolves #47 stop requiring specific include file scheme (PR #302)

support conversion of article documents to a single-chapter file
resolves #205 support special chapters like bibliography
resolves #190 `basedir` now points to spine document directory when processing chapter files
resolves #178 fix image and listing numbers being reset in each chapter
resolves #166 fix xref resolving between sub-includes of chapter files
resolves #151 add support for contentless include files
resolves #136 drop nonstandard `<<chapter#>>` xref syntax and instead support vanilla `<<anchor>>` or `<<file#anchor>>` syntax
resolves #206 properly include bibliography generated by asciidoctor-bibtex
require Asciidoctor 1.5.6+

32 files changed, 1037 insertions, 1355 deletions

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index 6640cdd..10d6051 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -21,7 +21,7 @@ jobs:
       matrix:
         ruby: [jruby, 2.3, 2.4, 2.5, 2.6, 2.7]
         platform: [ubuntu-latest, macos-latest, windows-latest]
-        asciidoctor: [1.5.3, '']
+        asciidoctor: [1.5.6, '']
         exclude:
           # kindlegen fails to install on 2.3 on Windows. See https://github.com/asciidoctor/asciidoctor-epub3/pull/213
           - ruby: 2.3
diff --git a/CHANGELOG.adoc b/CHANGELOG.adoc
index 3b70374..864c390 100644
--- a/CHANGELOG.adoc
+++ b/CHANGELOG.adoc
@@ -7,6 +7,16 @@ For a detailed view of what has changed, refer to the {uri-repo}/commits/master[
 
 == Unreleased
 
+* support conversion of article documents to a single-chapter file
+* stop requiring specific include file scheme (#47)
+* support special chapters like bibliography (#205)
+* `basedir` now points to main document directory when processing chapter files (#190)
+* fix image and listing numbers being reset in each chapter (#178)
+* fix xref resolving between sub-includes of chapter files (#166)
+* add support for contentless include files (#151)
+* drop nonstandard `<<chapter#>>` xref syntax and instead support vanilla `<<anchor>>` or `<<file#anchor>>` syntax (#136)
+* properly include bibliography generated by asciidoctor-bibtex (#206)
+* require Asciidoctor 1.5.6+
 * do not include current date if `:reproducible:` attribute is set (#203)
 * respect `SOURCE_DATE_EPOCH` environment variable for reproducible builds
 * fix invalid markup produced for tables with footer (#295)
diff --git a/README.adoc b/README.adoc
index 5219b63..dfddc50 100644
--- a/README.adoc
+++ b/README.adoc
@@ -136,74 +136,22 @@ The EPUB3 publication, which can be thought of as the “digital master”, is a
 
 An EPUB3 archive is composed of multiple files. The content of each “chapter” is typically stored in a dedicated XHTML file.
 Therefore, the {project-name} converter “chunks” the AsciiDoc source document into multiple XHTML files to add to the EPUB3 archive.
-While other converters handle this chunking task by automatically slicing up the XHTML output at predetermined heading levels, this converter takes a different approach, outlined next.
-
-=== Composition Rules
-
-There are several rules you must follow to successfully convert your AsciiDoc source documents into an EPUB3 publication archive.
-These rules may be loosened in the future, but they are mandatory at the moment.
-
-Rule 1::
-Each chapter must begin in a dedicated file that resides at the root of the project.
-It must be included into the main document using the AsciiDoc include directive.
-The main document, which we call the “spine”, may only contain a document header and a sequence of include directives.
-See <<Declaring the Spine>> for details.
-+
-TIP: The include directive may be used in the chapter files to include other files (even files in other folders).
-
-Rule 2::
-Each chapter must start with a document title (i.e., level-0 heading).
-The document header must declare an ID that matches the chapter filename (minus its file extension).
-
-Rule 3::
-No sections in a chapter may have the same ID as the chapter ID.
-
-Rule 4::
-The cross reference (i.e., xref) from one chapter to another must be in the format `+xref:chapter-id#chapter-id[optional reftext]+` (or `+<<chapter-id#chapter-id,optional reftext>>+`).
-If you want to reference a section in a chapter, update the fragment (i.e., the value that follows the hash) to match the target section ID.
-The reftext is optional since the reftext of the target is used by default.
-+
-TIP: To assign reference text (i.e., reftext) for a chapter title, set the document attribute `docreftext` in the header for that chapter.
-+
-TIP: Rule 2 states that the chapter ID must match the basename of the chapter file.
-In truth, the chapter ID can be a _derivative of_ the chapter filename.
-In that case, an attribute reference can be used to prepend a prefix to an inter-document cross reference.
-For example, if you want each chapter file to begin with `chapter-`, you'd write the xref as `+xref:{chapter-prefix}chapter-id#chapter-id[]+`.
-You'd then assign an empty value to the `chapter-prefix` attribute when converting to an e-book and `chapter-` for all other formats.
-
-The next section goes into more detail about how to set up the spine document and include the chapter files.
-
-=== Declaring the Spine
-
-The spine (or master document) must be well-formed.
-Otherwise, {project-name} will not convert the document properly.
-If your AsciiDoc documents are not structured as explained in this section, you'll need to change them.
-
-Asciidoctor uses top-level include directives (i.e., include directives in the master document) to indicate where each chapter split should occur.
-The chapter files should begin with an AsciiDoc document header, which consists of an ID, a document title (i.e., level-0 heading), an author name, and a set of attribute entries.
-Only the document title is required.
-
-The document title in the chapter file is used as the chapter title and the label for the chapter in the TOC.
-The chapter ID, combined with the _.xhtml_ suffix, is used as the filename of the chapter inside the EPUB3 archive (though this could change in the future).
-We recommend that you base the filename of the chapter on the chapter ID.
-
-If you don't specify an ID for a chapter, one will be generated automatically from the document title.
-The rules for generating a chapter ID from the document title are as follows:
-
-* apply inline formatting, then remove XML elements
-* remove the `\&#8217;` character reference (so `John\&#8217;s` becomes `Johns`)
-* replace `\&amp;` with the word `and` (so `John \&amp; Jane` becomes `John and Jane`)
-* expand all other character references
-* lowercase all characters
-* replace illegal ID characters with the character defined by the `idseparator` attribute
-* prepend the value of the `idprefix` attribute
-* prepend an underscore if the ID begins with a number
-
-You can think of the master document as the spine of the book and the include directives the individual items being bound together.
-The target of each include directive in the master document is parsed and converted as a separate AsciiDoc document, with certain options and attributes passed down from the master to ensure consistent behavior.
-Each resulting XHTML document is then added to the EPUB3 archive as a chapter document and the master document becomes the navigation file (i.e, the table of contents).
-
-Here's an example showing the structure of a spine document:
+Like other converters, Asciidoctor EPUB3 handles this chunking task by automatically slicing up the XHTML output at predetermined heading levels.
+
+When `doctype` attribute is set to `book`, each top-level section will become a separate ebook "chapter" file.
+This includes preface, bibliography, appendix, etc.
+
+Otherwise, whole document is converted to a single ebook chapter.
+
+You may specify custom chapter filenames by assigning IDs to sections:
+
+[source,asciidoc]
+-----
+[#custom-chapter-id]
+= Chapter
+-----
+
+Here's an example showing the structure of a book:
 
 [source,asciidoc]
 ----
@@ -214,43 +162,21 @@ Author Name
 \ifndef::ebook-format[:leveloffset: 1]
 //...and so on
 
-\include::chapter-one.adoc[]
-
-\include::chapter-two.adoc[]
-
-\include::chapter-three.adoc[]
-----
-
-IMPORTANT: The spine document cannot contain any content other than include directives.
-
-Here's an example showing the structure of a chapter document:
-
-[source,asciidoc]
-----
-[#chapter-one]
-= Chapter One
-
-chapter content
-----
+== Chapter One
 
-CAUTION: Although an explicit ID over the chapter title is not required, it's recommended for stability.
+Some interesting text here.
 
-If your chapter files start with a level-1 section instead of a level-2 section, you need to make the opposite adjustment in the header of the spine document:
+== Chapter Two
 
-[source,asciidoc]
-----
-\ifndef::ebook-format[:leveloffset: -1]
+Even more exciting stuff.
 ----
 
-If the master document does not contain any include directives, then the converter treats the document as the sole chapter in the EPUB3 archive and automatically produces a navigation file that references it.
-(Currently broken. See issue {uri-issues}/47[#47]).
-
-NOTE: Eventually, we envision introducing a dedicated block macro to represent a spine item so that we don't overload the meaning of the include directive.
-However, for the time being, the include directive fills this role.
+In older Asciidoctor EPUB3 versions, there were strict rules on document organization: 'spine' master document with chapter includes.
+This is no longer the case. If you followed old rules, chances are your document will work with newer Asciidoctor EPUB3 either as-is or after minor adjustments.
 
 == Prerequisites
 
-All that's needed to use {project-name} is Ruby 2.3 or newer and a few Ruby gems (including at least Asciidoctor 1.5.3), which we'll explain how to install in the next section.
+All that's needed to use {project-name} is Ruby 2.3 or newer and a few Ruby gems (including at least Asciidoctor 1.5.6), which we'll explain how to install in the next section.
 
 To check if you have Ruby available, use the `ruby` command to query the installed version:
 
@@ -299,31 +225,6 @@ Assuming all the required gems install properly, verify you can run the `asciido
 If you see the version of {project-name} printed, you're ready to use {project-name}.
 Let's get an AsciiDoc document ready to convert to EPUB3.
 
-=== Prepare an AsciiDoc Document
-
-If you don't already have an AsciiDoc document, you can use the [file]_sample-book.adoc_ file and its chapters found in the [path]_data/samples_ directory of this project.
-
-.Master file named sample-book.adoc
-```asciidoc
-= Asciidoctor EPUB3: Sample Book
-Author Name
-v1.0, 2014-04-15
-:doctype: book
-:creator: {author}
-:producer: Asciidoctor
-:keywords: Asciidoctor, samples, e-book, EPUB3, KF8, MOBI, Asciidoctor.js
-:copyright: CC-BY-SA 3.0
-:imagesdir: images
-
-\include::asciidoctor-epub3-readme.adoc[]
-
-\include::sample-content.adoc[]
-
-\include::asciidoctor-js-introduction.adoc[]
-
-\include::asciidoctor-js-extension.adoc[]
-```
-
 === EPUB-related AsciiDoc Attributes
 
 The metadata in the generated EPUB3 file is populated from attributes in the AsciiDoc document.
@@ -362,7 +263,7 @@ The title is added to the metadata in plain text format.
 The authors in each chapter document are aggregated together with the authors in the master file.
 
 |username
-|Used to resolve an avatar for the author that is displayed in the header of a chapter when the `publication-type` is set to a value other than `book`.
+|Used to resolve an avatar for the author that is displayed in the header of a chapter when the `doctype` is set to a value other than `book`.
 The avatar image should be located at the path _\{imagesdir}/avatars/\{username}.jpg_, where `\{username}` is the value of this attribute.
 
 |producer
@@ -400,13 +301,9 @@ _Only applies to a chapter document._
 |The path to a directory that contains alternate epub3.css and epub3-css3-only.css files to customize the look and feel.
 
 |doctype
-|Effectively ignored.
-The master document is assumed to be a book and each chapter an article.
-
-|publication-type
 |Used to control the inclusion of special content in the generated HTML.
 If set to a value other than book, the byline information (author and avatar) is included below the chapter header and a typographic end mark is added at the end of the last paragraph.
-Suggested values include: book (default), anthology, magazine, journal, article.
+Suggested values include: book (default), article.
 |===
 
 When using the EPUB3 converter, the `ebook-format` attribute resolves to the name of the e-book format being generated (epub3 or kf8) and the corresponding attribute `ebook-format-<name>` is defined, where `<name>` is `epub3` or `kf8`.
@@ -618,9 +515,9 @@ mimetype
 
 Images referenced in your AsciiDoc document must be stored in the images catalog.
 The images catalog is defined by the `imagesdir` attribute.
-If set, the value of this attribute is resolved relative to the spine document and must be at or below (i.e., within) the directory of that document.
+If set, the value of this attribute is resolved relative to the document and must be at or below (i.e., within) the directory of that document.
 (In other words, it cannot point to a location outside the document directory).
-If this attribute is not set, the images catalog defaults to the directory of the spine document.
+If this attribute is not set, the images catalog defaults to the directory of the document.
 {project-name} will discover all local image references and insert those images into the EPUB3 archive at the same relative path.
 
 === Default Images
@@ -645,7 +542,7 @@ Once your image is ready, you can set the cover image by defining the `front-cov
 :front-cover-image: image:cover.png[Front Cover,1050,1600]
 ----
 
-The image is resolved relative to the directory specified in the `imagesdir` attribute, which defaults to the directory of the spine document.
+The image is resolved relative to the directory specified in the `imagesdir` attribute, which defaults to the directory of the3 document.
 The image can be in any format, though we recommend using PNG, JPG, or SVG as they are the most portable formats.
 
 IMPORTANT: *You should always specify the dimensions of the cover image.*
diff --git a/WORKLOG.adoc b/WORKLOG.adoc
index b82e35a..023b911 100644
--- a/WORKLOG.adoc
+++ b/WORKLOG.adoc
@@ -2,7 +2,6 @@
 
 == TODO
 
-* remove conum patch for listing block (fixed in Asciidoctor 1.5.6)
 * change has_role? to role?
 * use empty string in place of nil in interpolated strings (few more still to check)
 * add epub3width (or epubwidth) as alternative to scaledwidth for images
@@ -16,7 +15,6 @@
 * don't add image from front matter if already referenced by chapter
 * add document filename to warn messages (added for some)
 * look for front-matter.html relative to docdir (or base_dir?); also allow front-matter.xhtml
-* change references[:spine] to spine_document property
 * reorder sections in README; needs to flow a bit better...Getting Started is too large & after Structuring your Manuscript
 * add helper to resolve imagesdir
 * support inline SVG
@@ -50,7 +48,6 @@
 * CRITICAL: make generated file name strategy configurable (source file vs id)
 * HIGH: revisit the headshot image logic
 * HIGH: if running on a single article, then the document id has a conflicting purpose
-* HIGH: require spine option on include directive in order to create a chapter file (requires change in Asciidoctor)
 * HIGH: promote regexps to constants
 * HIGH: put shy marks after dashes to allow wrapping to occur in text justification
 * MEDIUM: add landmarks nav list to nav.xhtml; see https://github.com/IDPF/epub3-samples/blob/master/31/moby-dick-mo-xhtml/EPUB/toc.xhtml
diff --git a/asciidoctor-epub3.gemspec b/asciidoctor-epub3.gemspec
index 4ba454b..5b63da5 100644
--- a/asciidoctor-epub3.gemspec
+++ b/asciidoctor-epub3.gemspec
@@ -36,6 +36,6 @@ An extension for Asciidoctor that converts AsciiDoc documents to EPUB3 and KF8/M
   s.add_development_dependency 'rubocop', '~> 0.79.0'
   s.add_development_dependency 'rubocop-rspec', '~> 1.38.0'
 
-  s.add_runtime_dependency 'asciidoctor', '>= 1.5.3', '< 3.0.0'
+  s.add_runtime_dependency 'asciidoctor', '>= 1.5.6', '< 3.0.0'
   s.add_runtime_dependency 'gepub', '~> 1.0.0'
 end
diff --git a/data/samples/sample-book.adoc b/data/samples/sample-book.adoc
index 4b19c6f..2691064 100644
--- a/data/samples/sample-book.adoc
+++ b/data/samples/sample-book.adoc
@@ -1,29 +1,189 @@
-= Asciidoctor EPUB3: Sample Book
-Author Name
+= Asciidoctor Playground: Sample Content
+Sarah White <https://github.com/graphitefriction[@graphitefriction]>
 v1.0, 2014-04-15
 :doctype: book
 :producer: Asciidoctor
 :keywords: Asciidoctor, samples, e-book, EPUB3, KF8, MOBI, Asciidoctor.js
 :copyright: CC-BY-SA 3.0
-// NOTE anthology adds support for an author per chapter; use book for a single author
-:publication-type: anthology
-:idprefix:
-:idseparator: -
+:username: graphitefriction
+:description: This guide describes the Asciidoctor attributes, values, and layout options available for producing a customized and polished document.
+:keywords: AsciiDoc, Asciidoctor, syntax, reference
+:language: asciidoc
 :imagesdir: images
 :front-cover-image: images/default-cover.png
 
-ifeval::["{scripts}" == "multilingual"]
-include::i18n.adoc[]
-endif::[]
+[abstract]
+This sample guide describes several Asciidoctor attributes, values, and layout options available for producing a customized and polished document.
 
-include::../../README.adoc[]
+== Cross Reference
 
-include::sample-content.adoc[]
+This is a cross reference to <<block-titles>> and this is another one: xref:block-titles[].
 
-include::asciidoctor-js-introduction.adoc[]
+== Text size and justification
 
-include::asciidoctor-js-browser-extension.adoc[]
+* *http://guides.github.com/overviews/forking[Fork] the repository* and clone it locally.
+Connect your local to the original _upstream_ repository by adding it as a remote.
 
-include::../../NOTICE.adoc[]
+1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0
 
-include::../../LICENSE.adoc[]
+[[block-titles]]
+== Block titles
+
+You can assign a title to any paragraph, list, delimited block, or block macro.
+In most cases, the title is displayed immediately above the content.
+If the content is a figure or image, the title is displayed below the content.
+
+A block title is defined on a line above the element.
+The line must begin with a dot (`.`) and be followed immediately by the title text with no spaces inbetween.
+
+Here's an example of a list with a title:
+
+.A titled list
+====
+.TODO list
+- Learn the AsciiDoc syntax
+- Install Asciidoctor
+- Write my novel
+====
+
+== Metadata
+
+In addition to a title, lists and blocks can be assigned metadata, including:
+
+* Id (i.e., anchor)
+* Style (first unnamed block attribute)
+* Named block attributes
+* Options
+
+Here's an example of a quote block with metadata:
+
+[source]
+----
+.Gettysburg Address <1>
+[#gettysburg] <2>
+[quote, Abraham Lincoln, Soldiers' National Cemetery Dedication] <3> <4> <5>
+____
+Four score and seven years ago our fathers brought forth on this continent a new nation, conceived in Liberty, and dedicated to the proposition that all men are created equal.
+
+Now we are engaged in a great civil war,
+testing whether that nation,
+or any nation so conceived and so dedicated,
+can long endure.
+...
+
+Fourscoreandsevenyearsagoourfathersbroughtforthonthiscontinentanewnation...
+____
+----
+<1> Title: Gettysburg Address
+<2> Id: gettysburg
+<3> Style: quote
+<4> attribution: Abraham Lincoln (Named block attribute)
+<5> citetitle: Dedication of the Soldiers' National Cemetery (Named block attribute)
+
+TIP: A block can have multiple block attribute lines.
+The attributes will be aggregated.
+If there is a name conflict, the last attribute defined wins.
+
+Some metadata is used as supplementary content, such as the title, whereas other metadata controls how the block is rendered, such as the style.
+
+== Delimited blocks
+
+The AsciiDoc syntax provides a set of components for including non-paragraph text--such as block quotes, source code listings, sidebars and tables--in your document.
+These components are referred to as _delimited blocks_ because they are surrounded by delimiter lines.
+
+Within the boundaries of a delimited block, you can enter any content or blank lines.
+The block doesn't end until the ending delimiter is found.
+The delimiters around the block determine the type of block, how the content is processed and rendered, and what elements are used to wrap the content in the output.
+
+=== Optional delimiters
+
+If the content is contiguous (not interrupted by blank lines), you can forgo the use of the block delimiters and instead use the block style above a paragraph to repurpose it as one of the delimited block types.
+
+NOTE: This format is often used for single-line listings or quotes.
+
+.Single line listing block
+[source]
+----
+[listing]
+sudo yum install asciidoc
+----
+
+.Single line quote block
+[source]
+----
+[quote]
+Never do today what you can put off 'til tomorrow.
+----
+
+== Admonitions
+
+There are certain statements that you may want to draw attention to by taking them out of the content's flow and labeling them with a priority.
+These are called admonitions.
+Asciidoctor provides five built-in admonition style labels:
+
+Tip:: Used for giving the reader helpful advice.
+Note:: Used to communicate related or follow-up information.
+Important:: Used to remind the reader to do a critical step or task.
+Warning:: Used to inform the reader of potential risks.
+Caution:: Also used to inform the reader of potential risks, take your pick.
+
+When you want to call attention to a single paragraph, start the first line of the paragraph with the label you want to use.
+The label must be uppercase and followed by a colon (`:`).
+
+WARNING: Wolpertingers are known to nest in server racks.
+Enter at your own risk.
+
+When you want to apply an admonition label to complex content, set the label as a style attribute on a delimited block.
+Admonition labels are commonly set on example blocks.
+This behavior is referred to as _masquerading_.
+The label must be uppercase when set as an attribute on a block.
+
+[IMPORTANT]
+.Feeding the Werewolves
+====
+While werewolves are hardy community members, keep in mind the following dietary concerns:
+
+. They are allergic to cinnamon.
+. More than two glasses of orange juice in 24 hours makes them howl in harmony with alarms and sirens.
+. Celery makes them sad.
+====
+
+== Quotes, Prose Excerpts, and Verses
+
+Quotes, prose excerpts and verses share the same syntax structure, including:
+
+* style attribute, either `quote` or `verse`
+* name of who the content is attributed to
+* bibliographical information of the book, speech, play, etc., where the content was drawn from
+* excerpt text
+
+.Anatomy of a basic quote
+----
+[quote, attribution, citation title and information]
+Quote or excerpt text
+----
+
+For content that doesn't require the preservation of line breaks, set the +quote+ attribute in the first position of the attribute list.
+Next, set the attribution and relevant citation information.
+However, these positional attributes are optional.
+
+If the quote is a single line or paragraph, you can place the attribute list directly on top of the text.
+
+.after landing the cloaked Klingon bird of prey in Golden Gate park
+[quote, James Kirk, Star Trek IV: The Voyage Home]
+Everybody remember where we parked.
+
+For those of you who want a little bit of culture in this document, here's an excerpt of a poem by Carl Sandburg.
+
+[verse, Carl Sandburg, Fog]
+____
+The fog comes
+on little cat feet.
+
+It sits looking
+over harbor and city
+on silent haunches
+and then moves on.
+____
+
+That's all folks!
diff --git a/data/samples/sample-content.adoc b/data/samples/sample-content.adoc
deleted file mode 100644
index 757602c..0000000
--- a/data/samples/sample-content.adoc
+++ /dev/null
@@ -1,184 +0,0 @@
-[#sample-content]
-= Asciidoctor Playground: Sample Content
-Sarah White <https://github.com/graphitefriction[@graphitefriction]>
-:username: graphitefriction
-:description: This guide describes the Asciidoctor attributes, values, and layout options available for producing a customized and polished document.
-:keywords: AsciiDoc, Asciidoctor, syntax, reference
-:language: asciidoc
-
-[abstract]
-This sample guide describes several Asciidoctor attributes, values, and layout options available for producing a customized and polished document.
-
-== Cross Reference
-
-This is a cross reference to <<asciidoctor-js#usage>> and this is another one xref:asciidoctor-js#usage[].
-
-Referencing a document <<NOTICE#,NOTICE>>.
-
-== Text size and justification
-
-* *http://guides.github.com/overviews/forking[Fork] the repository* and clone it locally.
-Connect your local to the original _upstream_ repository by adding it as a remote.
-
-1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0
-
-== Block titles
-
-You can assign a title to any paragraph, list, delimited block, or block macro.
-In most cases, the title is displayed immediately above the content.
-If the content is a figure or image, the title is displayed below the content.
-
-A block title is defined on a line above the element.
-The line must begin with a dot (`.`) and be followed immediately by the title text with no spaces inbetween.
-
-Here's an example of a list with a title:
-
-.A titled list
-====
-.TODO list
-- Learn the AsciiDoc syntax
-- Install Asciidoctor
-- Write my novel
-====
-
-== Metadata
-
-In addition to a title, lists and blocks can be assigned metadata, including:
-
-* Id (i.e., anchor)
-* Style (first unnamed block attribute)
-* Named block attributes
-* Options
-
-Here's an example of a quote block with metadata:
-
-[source]
-----
-.Gettysburg Address <1>
-[#gettysburg] <2>
-[quote, Abraham Lincoln, Soldiers' National Cemetery Dedication] <3> <4> <5>
-____
-Four score and seven years ago our fathers brought forth on this continent a new nation, conceived in Liberty, and dedicated to the proposition that all men are created equal.
-
-Now we are engaged in a great civil war,
-testing whether that nation,
-or any nation so conceived and so dedicated,
-can long endure.
-...
-
-Fourscoreandsevenyearsagoourfathersbroughtforthonthiscontinentanewnation...
-____
-----
-<1> Title: Gettysburg Address
-<2> Id: gettysburg
-<3> Style: quote
-<4> attribution: Abraham Lincoln (Named block attribute)
-<5> citetitle: Dedication of the Soldiers' National Cemetery (Named block attribute)
-
-TIP: A block can have multiple block attribute lines.
-The attributes will be aggregated.
-If there is a name conflict, the last attribute defined wins.
-
-Some metadata is used as supplementary content, such as the title, whereas other metadata controls how the block is rendered, such as the style.
-
-== Delimited blocks
-
-The AsciiDoc syntax provides a set of components for including non-paragraph text--such as block quotes, source code listings, sidebars and tables--in your document.
-These components are referred to as _delimited blocks_ because they are surrounded by delimiter lines.
-
-Within the boundaries of a delimited block, you can enter any content or blank lines.
-The block doesn't end until the ending delimiter is found.
-The delimiters around the block determine the type of block, how the content is processed and rendered, and what elements are used to wrap the content in the output.
-
-=== Optional delimiters
-
-If the content is contiguous (not interrupted by blank lines), you can forgo the use of the block delimiters and instead use the block style above a paragraph to repurpose it as one of the delimited block types.
-
-NOTE: This format is often used for single-line listings or quotes.
-
-.Single line listing block
-[source]
-----
-[listing]
-sudo yum install asciidoc
-----
-
-.Single line quote block
-[source]
-----
-[quote]
-Never do today what you can put off 'til tomorrow.
-----
-
-== Admonitions
-
-There are certain statements that you may want to draw attention to by taking them out of the content's flow and labeling them with a priority.
-These are called admonitions.
-Asciidoctor provides five built-in admonition style labels:
-
-Tip:: Used for giving the reader helpful advice.
-Note:: Used to communicate related or follow-up information.
-Important:: Used to remind the reader to do a critical step or task.
-Warning:: Used to inform the reader of potential risks.
-Caution:: Also used to inform the reader of potential risks, take your pick.
-
-When you want to call attention to a single paragraph, start the first line of the paragraph with the label you want to use.
-The label must be uppercase and followed by a colon (`:`).
-
-WARNING: Wolpertingers are known to nest in server racks.
-Enter at your own risk.
-
-When you want to apply an admonition label to complex content, set the label as a style attribute on a delimited block.
-Admonition labels are commonly set on example blocks.
-This behavior is referred to as _masquerading_.
-The label must be uppercase when set as an attribute on a block.
-
-[IMPORTANT]
-.Feeding the Werewolves
-====
-While werewolves are hardy community members, keep in mind the following dietary concerns:
-
-. They are allergic to cinnamon.
-. More than two glasses of orange juice in 24 hours makes them howl in harmony with alarms and sirens.
-. Celery makes them sad.
-====
-
-== Quotes, Prose Excerpts, and Verses
-
-Quotes, prose excerpts and verses share the same syntax structure, including:
-
-* style attribute, either `quote` or `verse`
-* name of who the content is attributed to
-* bibliographical information of the book, speech, play, etc., where the content was drawn from
-* excerpt text
-
-.Anatomy of a basic quote
-----
-[quote, attribution, citation title and information]
-Quote or excerpt text
-----
-
-For content that doesn't require the preservation of line breaks, set the +quote+ attribute in the first position of the attribute list.
-Next, set the attribution and relevant citation information.
-However, these positional attributes are optional.
-
-If the quote is a single line or paragraph, you can place the attribute list directly on top of the text.
-
-.after landing the cloaked Klingon bird of prey in Golden Gate park
-[quote, James Kirk, Star Trek IV: The Voyage Home]
-Everybody remember where we parked.
-
-For those of you who want a little bit of culture in this document, here's an excerpt of a poem by Carl Sandburg.
-
-[verse, Carl Sandburg, Fog]
-____
-The fog comes
-on little cat feet.
-
-It sits looking
-over harbor and city
-on silent haunches
-and then moves on.
-____
-
-That's all folks!
diff --git a/lib/asciidoctor-epub3.rb b/lib/asciidoctor-epub3.rb
index 81017df..1ff6859 100644
--- a/lib/asciidoctor-epub3.rb
+++ b/lib/asciidoctor-epub3.rb
@@ -5,7 +5,6 @@ require 'asciidoctor/extensions'
 require 'gepub'
 require_relative 'asciidoctor-epub3/ext'
 require_relative 'asciidoctor-epub3/converter'
-require_relative 'asciidoctor-epub3/packager'
 
 # We need to be able to write files with unicode names. See https://github.com/asciidoctor/asciidoctor-epub3/issues/217
 ::Zip.unicode_names = true
diff --git a/lib/asciidoctor-epub3/converter.rb b/lib/asciidoctor-epub3/converter.rb
index 67fa444..271a6ce 100644
--- a/lib/asciidoctor-epub3/converter.rb
+++ b/lib/asciidoctor-epub3/converter.rb
@@ -1,6 +1,6 @@
 # frozen_string_literal: true
 
-require_relative 'spine_item_processor'
+require 'open3'
 require_relative 'font_icon_map'
 
 module Asciidoctor
@@ -14,56 +14,45 @@ module Asciidoctor
 
       register_for 'epub3'
 
-      def initialize backend, opts
-        super
-        basebackend 'html'
-        outfilesuffix '.epub' # dummy outfilesuffix since it may be .mobi
-        htmlsyntax 'xml'
-        @validate = false
-        @extract = false
-        @kindlegen_path = nil
-        @epubcheck_path = nil
-      end
-
-      def convert node, name = nil
-        if (name ||= node.node_name) == 'document'
-          @validate = node.attr? 'ebook-validate'
-          @extract = node.attr? 'ebook-extract'
-          @compress = node.attr 'ebook-compress'
-          @kindlegen_path = node.attr 'ebook-kindlegen-path'
-          @epubcheck_path = node.attr 'ebook-epubcheck-path'
-          spine_items = node.references[:spine_items]
-          if spine_items.nil?
-            logger.error %(#{::File.basename node.document.attr('docfile')}: failed to find spine items, produced file will be invalid)
-            spine_items = []
+      def write output, target
+        epub_file = @format == :kf8 ? %(#{::Asciidoctor::Helpers.rootname target}-kf8.epub) : target
+        output.generate_epub epub_file
+        logger.debug %(Wrote #{@format.upcase} to #{epub_file})
+        if @extract
+          extract_dir = epub_file.sub EpubExtensionRx, ''
+          ::FileUtils.remove_dir extract_dir if ::File.directory? extract_dir
+          ::Dir.mkdir extract_dir
+          ::Dir.chdir extract_dir do
+            ::Zip::File.open epub_file do |entries|
+              entries.each do |entry|
+                next unless entry.file?
+                unless (entry_dir = ::File.dirname entry.name) == '.' || (::File.directory? entry_dir)
+                  ::FileUtils.mkdir_p entry_dir
+                end
+                entry.extract
+              end
+            end
           end
-          Packager.new node, spine_items, node.attributes['ebook-format'].to_sym
-          # converting an element from the spine document, such as an inline node in the doctitle
-        elsif name.start_with? 'inline_'
-          (@content_converter ||= ::Asciidoctor::Converter::Factory.default.create 'epub3-xhtml5').convert node, name
-        else
-          raise ::ArgumentError, %(Encountered unexpected node in epub3 package converter: #{name})
+          logger.debug %(Extracted #{@format.upcase} to #{extract_dir})
         end
-      end
 
-      # FIXME: we have to package in write because we don't have access to target before this point
-      def write packager, target
-        packager.package validate: @validate, extract: @extract, compress: @compress, kindlegen_path: @kindlegen_path, epubcheck_path: @epubcheck_path, target: target
-        nil
+        if @format == :kf8
+          # QUESTION shouldn't we validate this epub file too?
+          distill_epub_to_mobi epub_file, target, @compress, @kindlegen_path
+        elsif @validate
+          validate_epub epub_file, @epubcheck_path
+        end
       end
-    end
 
-    # Public: The converter for the epub3 backend that converts the individual
-    # content documents in an EPUB3 publication.
-    class ContentConverter
-      include ::Asciidoctor::Converter
-      include ::Asciidoctor::Logging
+      CsvDelimiterRx = /\s*,\s*/
 
-      register_for 'epub3-xhtml5'
+      DATA_DIR = ::File.expand_path ::File.join(__dir__, '..', '..', 'data')
+      ImageMacroRx = /^image::?(.*?)\[(.*?)\]$/
+      ImgSrcScanRx = /<img src="(.+?)"/
+      SvgImgSniffRx = /<img src=".+?\.svg"/
 
       LF = ?\n
       NoBreakSpace = '&#xa0;'
-      ThinNoBreakSpace = '&#x202f;'
       RightAngleQuote = '&#x203a;'
       CalloutStartNum = %(\u2460)
 
@@ -87,13 +76,14 @@ module Asciidoctor
 
       ToHtmlSpecialCharsRx = /[#{ToHtmlSpecialCharsMap.keys.join}]/
 
-      def initialize backend, opts
+      EpubExtensionRx = /\.epub$/i
+      KindlegenCompression = ::Hash['0', '-c0', '1', '-c1', '2', '-c2', 'none', '-c0', 'standard', '-c1', 'huffdic', '-c2']
+
+      def initialize backend, opts = {}
         super
         basebackend 'html'
-        outfilesuffix '.xhtml'
+        outfilesuffix '.epub' # dummy outfilesuffix since it may be .mobi
         htmlsyntax 'xml'
-        @xrefs_seen = ::Set.new
-        @icon_names = []
       end
 
       def convert node, name = nil, _opts = {}
@@ -106,37 +96,209 @@ module Asciidoctor
       end
 
       def convert_document node
-        docid = node.id
-        pubtype = node.attr 'publication-type', 'book'
+        @format = node.attr('ebook-format').to_sym
+
+        @validate = node.attr? 'ebook-validate'
+        @extract = node.attr? 'ebook-extract'
+        @compress = node.attr 'ebook-compress'
+        @kindlegen_path = node.attr 'ebook-kindlegen-path'
+        @epubcheck_path = node.attr 'ebook-epubcheck-path'
+        @xrefs_seen = ::Set.new
+        @in_chapter = false
+        @icon_names = []
+        @images = []
+        @footnotes = []
+
+        @book = GEPUB::Book.new
+        @book.epub_backward_compat = @format != :kf8
+        @book.language node.attr('lang', 'en'), id: 'pub-language'
+
+        if node.attr? 'uuid'
+          @book.primary_identifier node.attr('uuid'), 'pub-identifier', 'uuid'
+        else
+          @book.primary_identifier node.id, 'pub-identifier', 'uuid'
+        end
+        # replace with next line once the attributes argument is supported
+        #unique_identifier doc.id, 'pub-id', 'uuid', 'scheme' => 'xsd:string'
+
+        # NOTE we must use :plain_text here since gepub reencodes
+        @book.add_title sanitize_doctitle_xml(node, :plain_text), id: 'pub-title'
+
+        # FIXME: this logic needs some work
+        if node.attr? 'publisher'
+          @book.publisher publisher_name = (node.attr 'publisher')
+          # marc role: Book producer (see http://www.loc.gov/marc/relators/relaterm.html)
+          @book.creator (node.attr 'producer', publisher_name), role: 'bkp'
+        elsif node.attr? 'producer'
+          # NOTE Use producer as both publisher and producer if publisher isn't specified
+          producer_name = node.attr 'producer'
+          @book.publisher producer_name
+          # marc role: Book producer (see http://www.loc.gov/marc/relators/relaterm.html)
+          @book.creator producer_name, role: 'bkp'
+        elsif node.attr? 'author'
+          # NOTE Use author as creator if both publisher or producer are absent
+          # marc role: Author (see http://www.loc.gov/marc/relators/relaterm.html)
+          @book.creator node.attr('author'), role: 'aut'
+        end
+
+        if node.attr? 'creator'
+          # marc role: Creator (see http://www.loc.gov/marc/relators/relaterm.html)
+          @book.creator node.attr('creator'), role: 'cre'
+        else
+          # marc role: Manufacturer (see http://www.loc.gov/marc/relators/relaterm.html)
+          # QUESTION should this be bkp?
+          @book.creator 'Asciidoctor', role: 'mfr'
+        end
+
+        if node.attr? 'reproducible'
+          # We need to set lastmodified to some fixed value. Otherwise, gepub will set it to current date.
+          @book.lastmodified = (::Time.at 0).utc
+          # Is it correct that we do not populate dc:date when 'reproducible' is set?
+        else
+          if node.attr? 'revdate'
+            begin
+              @book.date = node.attr 'revdate'
+            rescue ArgumentError => e
+              logger.error %(#{::File.basename node.attr('docfile')}: failed to parse revdate: #{e})
+              @book.date = node.attr 'docdatetime'
+            end
+          else
+            @book.date = node.attr 'docdatetime'
+          end
+          @book.lastmodified = node.attr 'localdatetime'
+        end
+
+        @book.description = node.attr 'description' if node.attr? 'description'
+        @book.source = node.attr 'source' if node.attr? 'source'
+        @book.rights = node.attr 'copyright' if node.attr? 'copyright'
+
+        (node.attr 'keywords', '').split(CsvDelimiterRx).each do |s|
+          @book.metadata.add_metadata 'subject', s
+        end
+
+        add_cover_image node
+        add_front_matter_page node
+
+        if node.doctype == 'book'
+          toc_items = []
+          node.sections.each do |item|
+            next unless item.parent == node
+            # Mark top-level sections as separate chapter files
+            item.set_attr 'ebook-chapter', item.id
+            toc_items << item
+          end
+          # TODO: this loses content between doc header and first chapter
+          node.content
+        else
+          toc_items = [node]
+          node.set_attr 'ebook-chapter', node.attr('docname')
+          add_chapter node
+        end
+
+        nav_xhtml = @book.add_item 'nav.xhtml', content: postprocess_xhtml(nav_doc(node, toc_items)), id: 'nav'
+        nav_xhtml.nav
+
+        # NOTE gepub doesn't support building a ncx TOC with depth > 1, so do it ourselves
+        toc_ncx = ncx_doc node, toc_items
+        @book.add_item 'toc.ncx', content: toc_ncx.to_ios, id: 'ncx'
+
+        docimagesdir = (node.attr 'imagesdir', '.').chomp '/'
+        docimagesdir = (docimagesdir == '.' ? nil : %(#{docimagesdir}/))
+
+        @images.each do |image|
+          if image[:name].start_with? %(#{docimagesdir}jacket/cover.)
+            logger.warn %(image path is reserved for cover artwork: #{image[:name]}; skipping image found in content)
+          elsif ::File.readable? image[:path]
+            @book.add_item image[:name], content: image[:path]
+          else
+            logger.error %(#{File.basename node.attr('docfile')}: image not found or not readable: #{image[:path]})
+          end
+        end
+
+        #add_metadata 'ibooks:specified-fonts', true
+
+        add_theme_assets node
+        if node.doctype != 'book'
+          usernames = [node].map {|item| item.attr 'username' }.compact.uniq
+          add_profile_images node, usernames
+        end
+
+        @book
+      end
 
-        if (doctitle = node.doctitle partition: true, use_fallback: true).subtitle?
-          title = %(#{doctitle.main} )
-          subtitle = doctitle.subtitle
+      # FIXME: move to Asciidoctor::Helpers
+      def sanitize_doctitle_xml doc, content_spec
+        doctitle = doc.doctitle use_fallback: true
+        sanitize_xml doctitle, content_spec
+      end
+
+      # FIXME: move to Asciidoctor::Helpers
+      def sanitize_xml content, content_spec
+        if content_spec != :pcdata && (content.include? '<')
+          if (content = (content.gsub XmlElementRx, '').strip).include? ' '
+            content = content.tr_s ' ', ' '
+          end
+        end
+
+        case content_spec
+        when :attribute_cdata
+          content = content.gsub '"', '&quot;' if content.include? '"'
+        when :cdata, :pcdata
+          # noop
+        when :plain_text
+          if content.include? ';'
+            content = content.gsub(CharEntityRx) { [$1.to_i].pack 'U*' } if content.include? '&#'
+            content = content.gsub FromHtmlSpecialCharsRx, FromHtmlSpecialCharsMap
+          end
+        else
+          raise ::ArgumentError, %(Unknown content spec: #{content_spec})
+        end
+        content
+      end
+
+      def add_chapter node
+        docid = node.attr 'ebook-chapter'
+
+        if node.context == :document
+          if (doctitle = node.doctitle partition: true, use_fallback: true).subtitle?
+            title = %(#{doctitle.main} )
+            subtitle = doctitle.subtitle
+          else
+            # HACK: until we get proper handling of title-only in CSS
+            title = ''
+            subtitle = doctitle.combined
+          end
         else
-          # HACK: until we get proper handling of title-only in CSS
           title = ''
-          subtitle = doctitle.combined
+          subtitle = node.title
         end
 
-        doctitle_sanitized = (node.doctitle sanitize: true, use_fallback: true).to_s
+        doctitle_sanitized = (node.document.doctitle sanitize: true, use_fallback: true).to_s
 
         # By default, Kindle does not allow the line height to be adjusted.
         # But if you float the elements, then the line height disappears and can be restored manually using margins.
         # See https://github.com/asciidoctor/asciidoctor-epub3/issues/123
         subtitle_formatted = subtitle.split.map {|w| %(<b>#{w}</b>) } * ' '
 
-        if pubtype == 'book'
+        if node.document.doctype == 'book'
           byline = ''
         else
           author = node.attr 'author'
           username = node.attr 'username', 'default'
-          imagesdir = (node.references[:spine].attr 'imagesdir', '.').chomp '/'
+          imagesdir = (node.document.attr 'imagesdir', '.').chomp '/'
           imagesdir = imagesdir == '.' ? '' : %(#{imagesdir}/)
           byline = %(<p class="byline"><img src="#{imagesdir}avatars/#{username}.jpg"/> <b class="author">#{author}</b></p>#{LF})
         end
 
-        mark_last_paragraph node unless pubtype == 'book'
-        content = node.content
+        mark_last_paragraph node unless node.document.doctype == 'book'
+
+        begin
+          @in_chapter = true
+          @xrefs_seen.clear
+          content = node.content
+        ensure
+          @in_chapter = false
+        end
 
         # NOTE must run after content is resolved
         # TODO perhaps create dynamic CSS file?
@@ -154,7 +316,7 @@ module Asciidoctor
 
         # NOTE kindlegen seems to mangle the <header> element, so we wrap its content in a div
         lines = [%(<!DOCTYPE html>
-<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="#{lang = node.attr 'lang', 'en'}" lang="#{lang}">
+<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="#{lang = node.document.attr 'lang', 'en'}" lang="#{lang}">
 <head>
 <meta charset="UTF-8"/>
 <title>#{doctitle_sanitized}</title>
@@ -179,12 +341,14 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
 </header>
 #{content})]
 
-        if node.footnotes?
+        unless (fns = node.document.footnotes - @footnotes).empty?
+          @footnotes += fns
+
           # NOTE kindlegen seems to mangle the <footer> element, so we wrap its content in a div
           lines << '<footer>
 <div class="chapter-footer">
 <div class="footnotes">'
-          node.footnotes.each do |footnote|
+          fns.each do |footnote|
             lines << %(<aside id="note-#{footnote.index}" epub:type="footnote">
 <p><sup class="noteref"><a href="#noteref-#{footnote.index}">#{footnote.index}</a></sup> #{footnote.text}</p>
 </aside>)
@@ -198,31 +362,39 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
 </body>
 </html>'
 
-        lines * LF
-      end
+        postprocessed_content = postprocess_xhtml lines * LF
+        chapter_item = @book.add_ordered_item %(#{docid}.xhtml), content: postprocessed_content
+        epub_properties = node.attr 'epub-properties'
+        chapter_item.add_property 'svg' if epub_properties&.include? 'svg'
 
-      # NOTE embedded is used for AsciiDoc table cell content
-      def convert_embedded node
-        node.content
+        # # QUESTION reenable?
+        # #linear 'yes' if i == 0
       end
 
       def convert_section node
-        hlevel = node.level + 1
-        epub_type_attr = node.special ? %( epub:type="#{node.sectname}") : ''
-        div_classes = [%(sect#{node.level}), node.role].compact
-        title = node.title
-        title_sanitized = xml_sanitize title
-        if node.document.header? || node.level != 1 || node != node.document.first_section
+        # TODO: single-document. Investigate if we can use Section.chapter?/Section.part?
+        # See https://github.com/asciidoctor/asciidoctor-pdf/blob/master/lib/asciidoctor/pdf/ext/asciidoctor/section.rb
+        # See https://asciidoctor.org/docs/user-manual/#book-parts-and-chapters
+        if @in_chapter
+          hlevel = node.level
+          epub_type_attr = node.special ? %( epub:type="#{node.sectname}") : ''
+          div_classes = [%(sect#{node.level}), node.role].compact
+          title = node.title
+          title_sanitized = xml_sanitize title
           %(<section class="#{div_classes * ' '}" title="#{title_sanitized}"#{epub_type_attr}>
 <h#{hlevel} id="#{node.id}">#{title}</h#{hlevel}>#{(content = node.content).empty? ? '' : %(
           #{content})}
 </section>)
         else
-          # document has no level-0 heading and this heading serves as the document title
-          node.content
+          add_chapter node
         end
       end
 
+      # NOTE embedded is used for AsciiDoc table cell content
+      def convert_embedded node
+        node.content
+      end
+
       # TODO: support use of quote block as abstract
       def convert_preamble node
         if (first_block = node.blocks[0]) && first_block.style == 'abstract'
@@ -330,10 +502,8 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
         pre_classes = node.style == 'source' ? ['source', %(language-#{node.attr 'language'})] : ['screen']
         title_div = node.title? ? %(<figcaption>#{node.captioned_title}</figcaption>
 ) : ''
-        # patches conums to fix extra or missing leading space
-        # TODO remove patch once upgrading to Asciidoctor 1.5.6
         %(<figure class="#{figure_classes * ' '}">
-#{title_div}<pre class="#{pre_classes * ' '}"><code>#{(node.content || '').gsub(/(?<! )<i class="conum"| +<i class="conum"/, ' <i class="conum"')}</code></pre>
+#{title_div}<pre class="#{pre_classes * ' '}"><code>#{node.content}</code></pre>
 </figure>)
       end
 
@@ -646,20 +816,21 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
       end
 
       def register_image node, target
+        if target.end_with? '.svg'
+          chapter = get_enclosing_chapter node
+          chapter.set_attr 'epub-properties', [] unless chapter.attr? 'epub-properties'
+          epub_properties = chapter.attr 'epub-properties'
+          epub_properties << 'svg' unless epub_properties.include? 'svg'
+        end
+
         out_dir = node.attr('outdir', nil, true) || doc_option(node.document, :to_dir)
         fs_path = (::File.join out_dir, target)
         unless ::File.exist? fs_path
-          # This is actually a hack. It would be more correct to set base_dir of chapter document to base_dir of spine document.
-          # That's how things would normally work if there was no separation between these documents, and instead chapters were normally included into spine document.
-          # However, setting chapter base_dir to spine base_dir breaks parser.rb because it resolves includes in chapter document relative to base_dir instead of actual location of chapter file.
-          # Choosing between two evils - a hack here or writing a full-blown include processor for chapter files, I chose the former.
-          # In the future, this all should be thrown away when we stop parsing chapters as a standalone documents.
-          # https://github.com/asciidoctor/asciidoctor-epub3/issues/47 is used to track that.
-          base_dir = root_document(node.document).references[:spine].base_dir
+          base_dir = root_document(node.document).base_dir
           fs_path = ::File.join base_dir, target
         end
         # We need *both* virtual and physical image paths. Unfortunately, references[:images] only has one of them.
-        (root_document(node.document).references[:epub_images] ||= []) << { name: target, path: fs_path } if doc_option node.document, :catalog_assets
+        @images << { name: target, path: fs_path }
       end
 
       def resolve_image_attrs node
@@ -679,13 +850,7 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
       def convert_image node
         target = node.image_uri node.attr 'target'
         register_image node, target
-        type = (::File.extname target)[1..-1]
         id_attr = node.id ? %( id="#{node.id}") : ''
-        if type == 'svg'
-          # TODO: make this a convenience method on document
-          epub_properties = (node.document.attributes['epub-properties'] ||= [])
-          epub_properties << 'svg' unless epub_properties.include? 'svg'
-        end
         img_attrs = resolve_image_attrs node
         %(<figure#{id_attr} class="image#{prepend_space node.role}">
 <div class="content">
@@ -695,84 +860,67 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
 </figure>)
       end
 
+      def get_enclosing_chapter node
+        loop do
+          return nil if node.nil?
+          return node if node.attr? 'ebook-chapter'
+          node = node.parent
+        end
+      end
+
       def convert_inline_anchor node
-        target = node.target
         case node.type
-        when :xref # TODO: would be helpful to know what type the target is (e.g., bibref)
-          doc, refid, text, path = node.document, ((node.attr 'refid') || target), node.text, (node.attr 'path')
-          # NOTE if path is non-nil, we have an inter-document xref
-          # QUESTION should we drop the id attribute for an inter-document xref?
-          if path
-            # ex. chapter-id#section-id
-            if node.attr 'fragment'
-              refdoc_id, refdoc_refid = refid.split '#', 2
-              if refdoc_id == refdoc_refid
-                target = target[0...(target.index '#')]
-                id_attr = %( id="xref--#{refdoc_id}")
+        when :xref
+          doc, refid, target, text = node.document, node.attr('refid'), node.target, node.text
+          id_attr = ''
+
+          if (path = node.attributes['path'])
+            # NOTE non-nil path indicates this is an inter-document xref that's not included in current document
+            text = node.text || path
+          elsif refid == '#'
+            logger.warn %(#{::File.basename doc.attr('docfile')}: <<chapter#>> xref syntax isn't supported anymore. Use either <<chapter>> or <<chapter#anchor>>)
+          elsif refid
+            ref = doc.references[:refs][refid]
+            our_chapter = get_enclosing_chapter node
+            ref_chapter = get_enclosing_chapter ref
+            if ref_chapter
+              ref_docname = ref_chapter.attr 'ebook-chapter'
+              if ref_chapter == our_chapter
+                # ref within same chapter file
+                id_attr = %( id="xref-#{refid}")
+                target = %(##{refid})
+              elsif refid == ref_docname
+                # ref to top section of other chapter file
+                id_attr = %( id="xref--#{refid}")
+                target = %(#{refid}.xhtml)
               else
-                id_attr = %( id="xref--#{refdoc_id}--#{refdoc_refid}")
+                # ref to section within other chapter file
+                id_attr = %( id="xref--#{ref_docname}--#{refid}")
+                target = %(#{ref_docname}.xhtml##{refid})
               end
-              # ex. chapter-id#
-            else
-              refdoc_id = refdoc_refid = refid
-              # inflate key to spine item root (e.g., transform chapter-id to chapter-id#chapter-id)
-              refid = %(#{refid}##{refid})
-              id_attr = %( id="xref--#{refdoc_id}")
-            end
-            id_attr = '' unless @xrefs_seen.add? refid
-            refdoc = doc.references[:spine_items].find {|it| refdoc_id == (it.id || (it.attr 'docname')) }
-            if refdoc
-              if (refs = refdoc.references[:refs])
-                ref = refs[refdoc_refid]
-                # If the reference is to a document, we have attached special reftext as 'docreftext'
-                if ::Asciidoctor::Document === ref
-                  xreftext = (ref.attr 'docreftext') || ref.doctitle
-                  # Otherwise use the usual xreftext
-                elsif ::Asciidoctor::AbstractNode === ref
-                  xreftext = (ref.xreftext node.attr('xrefstyle', nil, true)) || %([#{refdoc_refid}])
-                else
-                  warn %(asciidoctor: anchor in #{refdoc_id} chapter: #{refdoc_refid} no ref in #{refs.keys})
-                  xreftext = nil
-                end
-              else
-                # Fall back on looking up in 'ids' for backwards compatibility
-                xreftext = refdoc.references[:ids][refdoc_refid]
-              end
-
-              if xreftext
-                text ||= xreftext
-              else
-                logger.warn %(#{::File.basename doc.attr('docfile')}: invalid reference to unknown anchor in #{refdoc_id} chapter: #{refdoc_refid})
-              end
-            else
-              logger.warn %(#{::File.basename doc.attr('docfile')}: invalid reference to anchor in unknown chapter: #{refdoc_id})
-            end
-          else
-            id_attr = (@xrefs_seen.add? refid) ? %( id="xref-#{refid}") : ''
-            if (refs = doc.references[:refs])
-              if ::Asciidoctor::AbstractNode === (ref = refs[refid])
-                xreftext = (ref.xreftext node.attr('xrefstyle', nil, true)) || %([#{refid}])
-              else
-                xreftext = nil
-              end
-            else
-              xreftext = doc.references[:ids][refid]
-            end
 
-            if xreftext
-              text ||= xreftext
+              id_attr = '' unless @xrefs_seen.add? refid
+              text = (ref.xreftext node.attr('xrefstyle', nil, true))
             else
-              # FIXME: we get false negatives for reference to bibref when using Asciidoctor < 1.5.6
-              logger.warn %(#{::File.basename doc.attr('docfile')}: invalid reference to unknown local anchor (or valid bibref): #{refid})
+              logger.warn %(#{::File.basename doc.attr('docfile')}: invalid reference to unknown anchor: #{refid})
             end
           end
+
           %(<a#{id_attr} href="#{target}" class="xref">#{text || "[#{refid}]"}</a>)
         when :ref
-          %(<a id="#{target || node.id}"></a>)
+          # NOTE id is used instead of target starting in Asciidoctor 2.0.0
+          %(<a id="#{node.target || node.id}"></a>)
         when :link
-          %(<a href="#{target || node.id}" class="link">#{node.text}</a>)
+          %(<a href="#{node.target}" class="link">#{node.text}</a>)
         when :bibref
-          %(<a id="#{target || node.id}"></a>[#{node.reftext || target || node.id}])
+          # NOTE reftext is no longer enclosed in [] starting in Asciidoctor 2.0.0
+          # NOTE id is used instead of target starting in Asciidoctor 2.0.0
+          if (reftext = node.reftext)
+            reftext = %([#{reftext}]) unless reftext.start_with? '['
+          else
+            reftext = %([#{node.target || node.id}])
+          end
+          %(<a id="#{node.target || node.id}"></a>#{reftext})
         else
           logger.warn %(unknown anchor type: #{node.type.inspect})
           nil
@@ -815,12 +963,6 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
           target = node.image_uri node.target
           register_image node, target
 
-          if target.end_with? '.svg'
-            # TODO: make this a convenience method on document
-            epub_properties = (node.document.attributes['epub-properties'] ||= [])
-            epub_properties << 'svg' unless epub_properties.include? 'svg'
-          end
-
           img_attrs = resolve_image_attrs node
           img_attrs << %(class="inline#{prepend_space node.role}")
           %(<img src="#{target}"#{prepend_space img_attrs * ' '}/>)
@@ -908,6 +1050,425 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
       def prepend_space value
         value ? %( #{value}) : ''
       end
+
+      def add_theme_assets doc
+        format = @format
+        workdir = if doc.attr? 'epub3-stylesdir'
+                    stylesdir = doc.attr 'epub3-stylesdir'
+                    # FIXME: make this work for Windows paths!!
+                    if stylesdir.start_with? '/'
+                      stylesdir
+                    else
+                      docdir = doc.attr 'docdir', '.'
+                      docdir = '.' if docdir.empty?
+                      ::File.join docdir, stylesdir
+                    end
+                  else
+                    ::File.join DATA_DIR, 'styles'
+                  end
+
+        # TODO: improve design/UX of custom theme functionality, including custom fonts
+
+        if format == :kf8
+          # NOTE add layer of indirection so Kindle Direct Publishing (KDP) doesn't strip font-related CSS rules
+          @book.add_item 'styles/epub3.css', content: '@import url("epub3-proxied.css");'.to_ios
+          @book.add_item 'styles/epub3-css3-only.css', content: '@import url("epub3-css3-only-proxied.css");'.to_ios
+          @book.add_item 'styles/epub3-proxied.css', content: (postprocess_css_file ::File.join(workdir, 'epub3.css'), format)
+          @book.add_item 'styles/epub3-css3-only-proxied.css', content: (postprocess_css_file ::File.join(workdir, 'epub3-css3-only.css'), format)
+        else
+          @book.add_item 'styles/epub3.css', content: (postprocess_css_file ::File.join(workdir, 'epub3.css'), format)
+          @book.add_item 'styles/epub3-css3-only.css', content: (postprocess_css_file ::File.join(workdir, 'epub3-css3-only.css'), format)
+        end
+
+        font_files, font_css = select_fonts ::File.join(DATA_DIR, 'styles/epub3-fonts.css'), (doc.attr 'scripts', 'latin')
+        @book.add_item 'styles/epub3-fonts.css', content: font_css
+        unless font_files.empty?
+          # NOTE metadata property in oepbs package manifest doesn't work; must use proprietary iBooks file instead
+          #(@book.metadata.add_metadata 'meta', 'true')['property'] = 'ibooks:specified-fonts' unless format == :kf8
+          @book.add_optional_file 'META-INF/com.apple.ibooks.display-options.xml', '<?xml version="1.0" encoding="UTF-8"?>
+<display_options>
+<platform name="*">
+<option name="specified-fonts">true</option>
+</platform>
+</display_options>'.to_ios unless format == :kf8
+
+          # https://github.com/asciidoctor/asciidoctor-epub3/issues/120
+          #
+          # 'application/x-font-ttf' causes warnings in epubcheck 4.0.2,
+          # "non-standard font type". Discussion:
+          # https://www.mobileread.com/forums/showthread.php?t=231272
+          #
+          # 3.1 spec recommends 'application/font-sfnt', but epubcheck doesn't
+          # implement that yet (warnings). https://idpf.github.io/epub-cmt/v3/
+          #
+          # 3.0 spec recommends 'application/vnd.ms-opentype', this works without
+          # warnings.
+          # http://www.idpf.org/epub/30/spec/epub30-publications.html#sec-core-media-types
+          font_files.each do |font_file|
+            item = @book.add_item font_file, content: File.join(DATA_DIR, font_file)
+            item.set_media_type 'application/vnd.ms-opentype'
+          end
+        end
+        nil
+      end
+
+      def add_cover_image doc
+        return if (image_path = doc.attr 'front-cover-image').nil?
+
+        imagesdir = (doc.attr 'imagesdir', '.').chomp '/'
+        imagesdir = (imagesdir == '.' ? '' : %(#{imagesdir}/))
+
+        image_attrs = {}
+        if (image_path.include? ':') && image_path =~ ImageMacroRx
+          logger.warn %(deprecated block macro syntax detected in front-cover-image attribute) if image_path.start_with? 'image::'
+          image_path = %(#{imagesdir}#{$1})
+          (::Asciidoctor::AttributeList.new $2).parse_into image_attrs, %w(alt width height) unless $2.empty?
+        end
+
+        image_href = %(#{imagesdir}jacket/cover#{::File.extname image_path})
+
+        workdir = doc.attr 'docdir'
+        workdir = '.' if workdir.nil_or_empty?
+
+        unless ::File.readable? ::File.join(workdir, image_path)
+          logger.error %(#{::File.basename doc.attr('docfile')}: front cover image not found or readable: #{::File.expand_path image_path, workdir})
+          return
+        end
+
+        unless !image_attrs.empty? && (width = image_attrs['width']) && (height = image_attrs['height'])
+          width, height = 1050, 1600
+        end
+
+        @book.add_item(image_href, content: File.join(workdir, image_path)).cover_image
+
+        unless @format == :kf8
+          # NOTE SVG wrapper maintains aspect ratio and confines image to view box
+          content = %(<!DOCTYPE html>
+<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="en" lang="en">
+<head>
+<meta charset="UTF-8"/>
+<title>#{sanitize_doctitle_xml doc, :cdata}</title>
+<style type="text/css">
+@page {
+  margin: 0;
+}
+html {
+  margin: 0 !important;
+  padding: 0 !important;
+}
+body {
+  margin: 0;
+  padding: 0 !important;
+  text-align: center;
+}
+body > svg {
+  /* prevent bleed onto second page (removes descender space) */
+  display: block;
+}
+</style>
+</head>
+<body epub:type="cover"><svg version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"
+  width="100%" height="100%" viewBox="0 0 #{width} #{height}" preserveAspectRatio="xMidYMid meet">
+<image width="#{width}" height="#{height}" xlink:href="#{image_href}"/>
+</svg></body>
+</html>).to_ios
+
+          # Gitden expects a cover.xhtml, so add it to the spine
+          @book.add_ordered_item 'cover.xhtml', content: content, id: 'cover'
+        end
+        nil
+      end
+
+      def add_front_matter_page doc
+        workdir = doc.attr 'docdir'
+        workdir = '.' if workdir.nil_or_empty?
+
+        front_matter = File.join workdir, 'front-matter.html'
+        return unless ::File.file? front_matter
+        front_matter_content = ::File.read front_matter
+
+        item = @book.add_item 'front-matter.xhtml', content: (postprocess_xhtml front_matter_content)
+        item.add_property 'svg' if SvgImgSniffRx =~ front_matter_content
+
+        front_matter_content.scan ImgSrcScanRx do
+          @book.add_item $1
+        end
+        nil
+      end
+
+      def add_profile_images doc, usernames
+        imagesdir = (doc.attr 'imagesdir', '.').chomp '/'
+        imagesdir = (imagesdir == '.' ? nil : %(#{imagesdir}/))
+
+        @book.add_item %(#{imagesdir}avatars/default.jpg), content: ::File.join(DATA_DIR, 'images/default-avatar.jpg')
+        @book.add_item %(#{imagesdir}headshots/default.jpg), content: ::File.join(DATA_DIR, 'images/default-headshot.jpg')
+
+        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
+
+        usernames.each do |username|
+          avatar = %(#{imagesdir}avatars/#{username}.jpg)
+          if ::File.readable? (resolved_avatar = (::File.join workdir, avatar))
+            @book.add_item avatar, content: resolved_avatar
+          else
+            logger.error %(avatar for #{username} not found or readable: #{avatar}; falling back to default avatar)
+            @book.add_item avatar, content: ::File.join(DATA_DIR, 'images/default-avatar.jpg')
+          end
+
+          headshot = %(#{imagesdir}headshots/#{username}.jpg)
+          if ::File.readable? (resolved_headshot = (::File.join workdir, headshot))
+            @book.add_item headshot, content: resolved_headshot
+          elsif doc.attr? 'builder', 'editions'
+            logger.error %(headshot for #{username} not found or readable: #{headshot}; falling back to default headshot)
+            @book.add_item headshot, content: ::File.join(DATA_DIR, 'images/default-headshot.jpg')
+          end
+        end
+        nil
+      end
+
+      # TODO: aggregate authors of chapters into authors attribute(s) on main document
+      def nav_doc doc, items
+        lines = [%(<!DOCTYPE html>
+<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="#{lang = doc.attr 'lang', 'en'}" lang="#{lang}">
+<head>
+<meta charset="UTF-8"/>
+<title>#{sanitize_doctitle_xml doc, :cdata}</title>
+<link rel="stylesheet" type="text/css" href="styles/epub3.css"/>
+<link rel="stylesheet" type="text/css" href="styles/epub3-css3-only.css" media="(min-device-width: 0px)"/>
+</head>
+<body>
+<h1>#{sanitize_doctitle_xml doc, :pcdata}</h1>
+<nav epub:type="toc" id="toc">
+<h2>#{doc.attr 'toc-title'}</h2>)]
+        lines << (nav_level items, [(doc.attr 'toclevels', 1).to_i, 0].max)
+        lines << %(</nav>
+</body>
+</html>)
+        lines * LF
+      end
+
+      def nav_level items, depth, state = {}
+        lines = []
+        lines << '<ol>'
+        items.each do |item|
+          #index = (state[:index] = (state.fetch :index, 0) + 1)
+          if item.attr? 'ebook-chapter'
+            # NOTE we sanitize the chapter titles because we use formatting to control layout
+            if item.context == :document
+              item_label = sanitize_doctitle_xml item, :cdata
+            else
+              item_label = sanitize_xml item.title, :cdata
+            end
+            item_href = (state[:content_doc_href] = %(#{item.attr 'ebook-chapter'}.xhtml))
+          else
+            item_label = sanitize_xml item.title, :pcdata
+            item_href = %(#{state[:content_doc_href]}##{item.id})
+          end
+          lines << %(<li><a href="#{item_href}">#{item_label}</a>)
+          if depth == 0 || (child_sections = item.sections).empty?
+            lines[-1] = %(#{lines[-1]}</li>)
+          else
+            lines << (nav_level child_sections, depth - 1, state)
+            lines << '</li>'
+          end
+          state.delete :content_doc_href if item.attr? 'ebook-chapter'
+        end
+        lines << '</ol>'
+        lines * LF
+      end
+
+      def ncx_doc doc, items
+        # TODO: populate docAuthor element based on unique authors in work
+        lines = [%(<?xml version="1.0" encoding="utf-8"?>
+<ncx xmlns="http://www.daisy.org/z3986/2005/ncx/" version="2005-1" xml:lang="#{doc.attr 'lang', 'en'}">
+<head>
+<meta name="dtb:uid" content="#{@book.identifier}"/>
+%{depth}
+<meta name="dtb:totalPageCount" content="0"/>
+<meta name="dtb:maxPageNumber" content="0"/>
+</head>
+<docTitle><text>#{sanitize_doctitle_xml doc, :cdata}</text></docTitle>
+<navMap>)]
+        lines << (ncx_level items, [(doc.attr 'toclevels', 1).to_i, 0].max, state = {})
+        lines[0] = lines[0].sub '%{depth}', %(<meta name="dtb:depth" content="#{state[:max_depth]}"/>)
+        lines << %(</navMap>
+</ncx>)
+        lines * LF
+      end
+
+      def ncx_level items, depth, state = {}
+        lines = []
+        state[:max_depth] = (state.fetch :max_depth, 0) + 1
+        items.each do |item|
+          index = (state[:index] = (state.fetch :index, 0) + 1)
+          item_id = %(nav_#{index})
+          if item.attr? 'ebook-chapter'
+            if item.context == :document
+              item_label = sanitize_doctitle_xml item, :cdata
+            else
+              item_label = sanitize_xml item.title, :cdata
+            end
+            item_href = (state[:content_doc_href] = %(#{item.attr 'ebook-chapter'}.xhtml))
+          else
+            item_label = sanitize_xml item.title, :cdata
+            item_href = %(#{state[:content_doc_href]}##{item.id})
+          end
+          lines << %(<navPoint id="#{item_id}" playOrder="#{index}">)
+          lines << %(<navLabel><text>#{item_label}</text></navLabel>)
+          lines << %(<content src="#{item_href}"/>)
+          unless depth == 0 || (child_sections = item.sections).empty?
+            lines << (ncx_level child_sections, depth - 1, state)
+          end
+          lines << %(</navPoint>)
+          state.delete :content_doc_href if item.attr? 'ebook-chapter'
+        end
+        lines * LF
+      end
+
+      # Swap fonts in CSS based on the value of the document attribute 'scripts',
+      # then return the list of fonts as well as the font CSS.
+      def select_fonts filename, scripts = 'latin'
+        font_css = ::File.read filename
+        font_css = font_css.gsub(/(?<=-)latin(?=\.ttf\))/, scripts) unless scripts == 'latin'
+
+        # match CSS font urls in the forms of:
+        # src: url(../fonts/notoserif-regular-latin.ttf);
+        # src: url(../fonts/notoserif-regular-latin.ttf) format("truetype");
+        font_list = font_css.scan(/url\(\.\.\/([^)]+\.ttf)\)/).flatten
+
+        [font_list, font_css.to_ios]
+      end
+
+      def postprocess_css_file filename, format
+        return filename unless format == :kf8
+        postprocess_css ::File.read(filename), format
+      end
+
+      def postprocess_css content, format
+        return content.to_ios unless format == :kf8
+        # TODO: convert regular expressions to constants
+        content
+          .gsub(/^  -webkit-column-break-.*\n/, '')
+          .gsub(/^  max-width: .*\n/, '')
+          .to_ios
+      end
+
+      # NOTE Kindle requires that
+      #      <meta charset="utf-8"/>
+      #      be converted to
+      #      <meta http-equiv="Content-Type" content="application/xml+xhtml; charset=UTF-8"/>
+      def postprocess_xhtml content
+        return content.to_ios unless @format == :kf8
+        # TODO: convert regular expressions to constants
+        content
+          .gsub(/<meta charset="(.+?)"\/>/, '<meta http-equiv="Content-Type" content="application/xml+xhtml; charset=\1"/>')
+          .gsub(/<img([^>]+) style="width: (\d\d)%;"/, '<img\1 style="width: \2%; height: \2%;"')
+          .gsub(/<script type="text\/javascript">.*?<\/script>\n?/m, '')
+          .to_ios
+      end
+
+      def get_kindlegen_command kindlegen_path
+        unless kindlegen_path.nil?
+          logger.debug %(Using ebook-kindlegen-path attribute: #{kindlegen_path})
+          return [kindlegen_path]
+        end
+
+        unless (result = ENV['KINDLEGEN']).nil?
+          logger.debug %(Using KINDLEGEN env variable: #{result})
+          return [result]
+        end
+
+        begin
+          require 'kindlegen' unless defined? ::Kindlegen
+          result = ::Kindlegen.command.to_s
+          logger.debug %(Using KindleGen from gem: #{result})
+          [result]
+        rescue LoadError => e
+          logger.debug %(#{e}; Using KindleGen from PATH)
+          [%(kindlegen#{::Gem.win_platform? ? '.exe' : ''})]
+        end
+      end
+
+      def distill_epub_to_mobi epub_file, target, compress, kindlegen_path
+        mobi_file = ::File.basename target.sub(EpubExtensionRx, '.mobi')
+        compress_flag = KindlegenCompression[compress ? (compress.empty? ? '1' : compress.to_s) : '0']
+
+        argv = get_kindlegen_command(kindlegen_path) + ['-dont_append_source', compress_flag, '-o', mobi_file, epub_file].compact
+        begin
+          # This duplicates Kindlegen.run, but we want to override executable
+          out, err, res = Open3.capture3(*argv) do |r|
+            r.force_encoding 'UTF-8' if ::Gem.win_platform? && r.respond_to?(:force_encoding)
+          end
+        rescue Errno::ENOENT => e
+          raise 'Unable to run KindleGen. Either install the kindlegen gem or place `kindlegen` executable on PATH or set KINDLEGEN environment variable with path to it', cause: e
+        end
+
+        out.each_line do |line|
+          log_line line
+        end
+        err.each_line do |line|
+          log_line line
+        end
+
+        output_file = ::File.join ::File.dirname(epub_file), mobi_file
+        if res.success?
+          logger.debug %(Wrote MOBI to #{output_file})
+        else
+          logger.error %(KindleGen failed to write MOBI to #{output_file})
+        end
+      end
+
+      def get_epubcheck_command epubcheck_path
+        unless epubcheck_path.nil?
+          logger.debug %(Using ebook-epubcheck-path attribute: #{epubcheck_path})
+          return [epubcheck_path]
+        end
+
+        unless (result = ENV['EPUBCHECK']).nil?
+          logger.debug %(Using EPUBCHECK env variable: #{result})
+          return [result]
+        end
+
+        begin
+          result = ::Gem.bin_path 'epubcheck-ruby', 'epubcheck'
+          logger.debug %(Using EPUBCheck from gem: #{result})
+          [::Gem.ruby, result]
+        rescue ::Gem::Exception => e
+          logger.debug %(#{e}; Using EPUBCheck from PATH)
+          ['epubcheck']
+        end
+      end
+
+      def validate_epub epub_file, epubcheck_path
+        argv = get_epubcheck_command(epubcheck_path) + ['-w', epub_file]
+        begin
+          out, err, res = Open3.capture3(*argv)
+        rescue Errno::ENOENT => e
+          raise 'Unable to run EPUBCheck. Either install epubcheck-ruby gem or place `epubcheck` executable on PATH or set EPUBCHECK environment variable with path to it', cause: e
+        end
+
+        out.each_line do |line|
+          logger.info line
+        end
+        err.each_line do |line|
+          log_line line
+        end
+
+        logger.error %(EPUB validation failed: #{epub_file}) unless res.success?
+      end
+
+      def log_line line
+        line = line.strip
+
+        if line =~ /^fatal/i
+          logger.fatal line
+        elsif line =~ /^error/i
+          logger.error line
+        elsif line =~ /^warning/i
+          logger.warn line
+        else
+          logger.info line
+        end
+      end
     end
 
     class DocumentIdGenerator
@@ -968,11 +1529,8 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
       end
     end
 
-    require_relative 'packager'
-
     Extensions.register do
       if (document = @document).backend == 'epub3'
-        document.attributes['spine'] = ''
         document.set_attribute 'listing-caption', 'Listing'
         # pygments.rb hangs on JRuby for Windows, see https://github.com/asciidoctor/asciidoctor-epub3/issues/253
         if !(::RUBY_ENGINE == 'jruby' && Gem.win_platform?) && (Gem.try_activate 'pygments.rb')
@@ -991,8 +1549,6 @@ document.addEventListener('DOMContentLoaded', function(event, reader) {
           ebook_format = document.attributes['ebook-format'] = 'epub3'
         end
         document.attributes[%(ebook-format-#{ebook_format})] = ''
-        # Only fire SpineItemProcessor for top-level include directives
-        include_processor SpineItemProcessor.new(document)
         treeprocessor do
           process do |doc|
             doc.id = DocumentIdGenerator.generate_id doc, (doc.attr 'idprefix'), (doc.attr 'idseparator')
diff --git a/lib/asciidoctor-epub3/packager.rb b/lib/asciidoctor-epub3/packager.rb
deleted file mode 100644
index d048122..0000000
--- a/lib/asciidoctor-epub3/packager.rb
+++ /dev/null
@@ -1,735 +0,0 @@
-# frozen_string_literal: true
-
-autoload :FileUtils, 'fileutils'
-autoload :Open3, 'open3'
-
-module Asciidoctor
-  module Epub3
-    module GepubBuilderMixin
-      include ::Asciidoctor::Logging
-      DATA_DIR = ::File.expand_path ::File.join(__dir__, '..', '..', 'data')
-      SAMPLES_DIR = ::File.join DATA_DIR, 'samples'
-      LF = ?\n
-      CharEntityRx = ContentConverter::CharEntityRx
-      XmlElementRx = ContentConverter::XmlElementRx
-      FromHtmlSpecialCharsMap = ContentConverter::FromHtmlSpecialCharsMap
-      FromHtmlSpecialCharsRx = ContentConverter::FromHtmlSpecialCharsRx
-      CsvDelimiterRx = /\s*,\s*/
-      ImageMacroRx = /^image::?(.*?)\[(.*?)\]$/
-      ImgSrcScanRx = /<img src="(.+?)"/
-      SvgImgSniffRx = /<img src=".+?\.svg"/
-
-      attr_reader :book, :format, :spine
-
-      # FIXME: move to Asciidoctor::Helpers
-      def sanitize_doctitle_xml doc, content_spec
-        doctitle = doc.header? ? doc.doctitle : (doc.attr 'untitled-label')
-        sanitize_xml doctitle, content_spec
-      end
-
-      # FIXME: move to Asciidoctor::Helpers
-      def sanitize_xml content, content_spec
-        if content_spec != :pcdata && (content.include? '<')
-          if (content = (content.gsub XmlElementRx, '').strip).include? ' '
-            content = content.tr_s ' ', ' '
-          end
-        end
-
-        case content_spec
-        when :attribute_cdata
-          content = content.gsub '"', '&quot;' if content.include? '"'
-        when :cdata, :pcdata
-          # noop
-        when :plain_text
-          if content.include? ';'
-            content = content.gsub(CharEntityRx) { [$1.to_i].pack 'U*' } if content.include? '&#'
-            content = content.gsub FromHtmlSpecialCharsRx, FromHtmlSpecialCharsMap
-          end
-        else
-          raise ::ArgumentError, %(Unknown content spec: #{content_spec})
-        end
-        content
-      end
-
-      def add_theme_assets doc
-        builder = self
-        format = @format
-        workdir = if doc.attr? 'epub3-stylesdir'
-                    stylesdir = doc.attr 'epub3-stylesdir'
-                    # FIXME: make this work for Windows paths!!
-                    if stylesdir.start_with? '/'
-                      stylesdir
-                    else
-                      docdir = doc.attr 'docdir', '.'
-                      docdir = '.' if docdir.empty?
-                      ::File.join docdir, stylesdir
-                    end
-                  else
-                    ::File.join DATA_DIR, 'styles'
-                  end
-
-        # TODO: improve design/UX of custom theme functionality, including custom fonts
-        resources do
-          if format == :kf8
-            # NOTE add layer of indirection so Kindle Direct Publishing (KDP) doesn't strip font-related CSS rules
-            file 'styles/epub3.css' => '@import url("epub3-proxied.css");'.to_ios
-            file 'styles/epub3-css3-only.css' => '@import url("epub3-css3-only-proxied.css");'.to_ios
-            file 'styles/epub3-proxied.css' => (builder.postprocess_css_file ::File.join(workdir, 'epub3.css'), format)
-            file 'styles/epub3-css3-only-proxied.css' => (builder.postprocess_css_file ::File.join(workdir, 'epub3-css3-only.css'), format)
-          else
-            file 'styles/epub3.css' => (builder.postprocess_css_file ::File.join(workdir, 'epub3.css'), format)
-            file 'styles/epub3-css3-only.css' => (builder.postprocess_css_file ::File.join(workdir, 'epub3-css3-only.css'), format)
-          end
-        end
-
-        resources do
-          font_files, font_css = builder.select_fonts ::File.join(DATA_DIR, 'styles/epub3-fonts.css'), (doc.attr 'scripts', 'latin')
-          file 'styles/epub3-fonts.css' => font_css
-          unless font_files.empty?
-            # NOTE metadata property in oepbs package manifest doesn't work; must use proprietary iBooks file instead
-            #(@book.metadata.add_metadata 'meta', 'true')['property'] = 'ibooks:specified-fonts' unless format == :kf8
-            builder.optional_file 'META-INF/com.apple.ibooks.display-options.xml' => '<?xml version="1.0" encoding="UTF-8"?>
-<display_options>
-<platform name="*">
-<option name="specified-fonts">true</option>
-</platform>
-</display_options>'.to_ios unless format == :kf8
-
-            # https://github.com/asciidoctor/asciidoctor-epub3/issues/120
-            #
-            # 'application/x-font-ttf' causes warnings in epubcheck 4.0.2,
-            # "non-standard font type". Discussion:
-            # https://www.mobileread.com/forums/showthread.php?t=231272
-            #
-            # 3.1 spec recommends 'application/font-sfnt', but epubcheck doesn't
-            # implement that yet (warnings). https://idpf.github.io/epub-cmt/v3/
-            #
-            # 3.0 spec recommends 'application/vnd.ms-opentype', this works without
-            # warnings.
-            # http://www.idpf.org/epub/30/spec/epub30-publications.html#sec-core-media-types
-            with_media_type 'application/vnd.ms-opentype' do
-              font_files.each do |font_file|
-                file font_file => ::File.join(DATA_DIR, font_file)
-              end
-            end
-          end
-        end
-        nil
-      end
-
-      def add_cover_image doc
-        return if (image_path = doc.attr 'front-cover-image').nil?
-
-        imagesdir = (doc.attr 'imagesdir', '.').chomp '/'
-        imagesdir = (imagesdir == '.' ? '' : %(#{imagesdir}/))
-
-        image_attrs = {}
-        if (image_path.include? ':') && image_path =~ ImageMacroRx
-          logger.warn %(deprecated block macro syntax detected in front-cover-image attribute) if image_path.start_with? 'image::'
-          image_path = %(#{imagesdir}#{$1})
-          (::Asciidoctor::AttributeList.new $2).parse_into image_attrs, %w(alt width height) unless $2.empty?
-        end
-
-        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        unless ::File.readable? ::File.join(workdir, image_path)
-          logger.error %(#{::File.basename doc.attr('docfile')}: front cover image not found or readable: #{::File.expand_path image_path, workdir})
-          return
-        end
-
-        unless !image_attrs.empty? && (width = image_attrs['width']) && (height = image_attrs['height'])
-          width, height = 1050, 1600
-        end
-
-        resources do
-          cover_image %(#{imagesdir}jacket/cover#{::File.extname image_path}) => (::File.join workdir, image_path)
-          @last_defined_item.tap do |last_item|
-            last_item['width'] = width
-            last_item['height'] = height
-          end
-        end
-        nil
-      end
-
-      # NOTE must be called within the ordered block
-      def add_cover_page doc, spine_builder, manifest
-        return if (cover_item_attrs = manifest.items['item_cover'].instance_variable_get :@attributes).nil?
-
-        href = cover_item_attrs['href']
-        # NOTE we only store width and height temporarily to pass through the values
-        width = cover_item_attrs.delete 'width'
-        height = cover_item_attrs.delete 'height'
-
-        # NOTE SVG wrapper maintains aspect ratio and confines image to view box
-        content = %(<!DOCTYPE html>
-<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="en" lang="en">
-<head>
-<meta charset="UTF-8"/>
-<title>#{sanitize_doctitle_xml doc, :cdata}</title>
-<style type="text/css">
-@page {
-  margin: 0;
-}
-html {
-  margin: 0 !important;
-  padding: 0 !important;
-}
-body {
-  margin: 0;
-  padding: 0 !important;
-  text-align: center;
-}
-body > svg {
-  /* prevent bleed onto second page (removes descender space) */
-  display: block;
-}
-</style>
-</head>
-<body epub:type="cover"><svg version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"
-  width="100%" height="100%" viewBox="0 0 #{width} #{height}" preserveAspectRatio="xMidYMid meet">
-<image width="#{width}" height="#{height}" xlink:href="#{href}"/>
-</svg></body>
-</html>).to_ios
-        # Gitden expects a cover.xhtml, so add it to the spine
-        spine_builder.file 'cover.xhtml' => content
-        assigned_id = (spine_builder.instance_variable_get :@last_defined_item).item.id
-        spine_builder.id 'cover'
-        # clearly a deficiency of gepub that it does not match the id correctly
-        # FIXME can we move this hack elsewhere?
-        @book.spine.itemref_by_id[assigned_id].idref = 'cover'
-        nil
-      end
-
-      def add_images_from_front_matter doc
-        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        front_matter = File.join workdir, 'front-matter.html'
-        if ::File.file? front_matter
-          front_matter_content = ::File.read front_matter
-          front_matter_content.scan ImgSrcScanRx do
-            resources do
-              file $1 => File.join(workdir, $1)
-            end
-          end
-        end
-        nil
-      end
-
-      def add_front_matter_page doc, spine_builder
-        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        front_matter = File.join workdir, 'front-matter.html'
-        if ::File.file? front_matter
-          front_matter_content = ::File.read front_matter
-          spine_builder.file 'front-matter.xhtml' => (postprocess_xhtml front_matter_content, @format)
-          spine_builder.add_property 'svg' unless (spine_builder.property? 'svg') || SvgImgSniffRx !~ front_matter_content
-        end
-        nil
-      end
-
-      def add_content_images doc, images
-        docimagesdir = (doc.attr 'imagesdir', '.').chomp '/'
-        docimagesdir = (docimagesdir == '.' ? nil : %(#{docimagesdir}/))
-
-        self_logger = logger
-        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        resources workdir: workdir do
-          images.each do |image_name, image_data|
-            if image_name.start_with? %(#{docimagesdir}jacket/cover.)
-              self_logger.warn %(image path is reserved for cover artwork: #{image_name}; skipping image found in content)
-            elsif ::File.readable? image_data[:path]
-              file [image_name, image_data[:path]]
-            else
-              self_logger.error %(#{::File.basename image_data[:docfile]}: image not found or not readable: #{image_data[:path]})
-            end
-          end
-        end
-        nil
-      end
-
-      def add_profile_images doc, usernames
-        imagesdir = (doc.attr 'imagesdir', '.').chomp '/'
-        imagesdir = (imagesdir == '.' ? nil : %(#{imagesdir}/))
-
-        resources do
-          file %(#{imagesdir}avatars/default.jpg) => ::File.join(DATA_DIR, 'images/default-avatar.jpg')
-          file %(#{imagesdir}headshots/default.jpg) => ::File.join(DATA_DIR, 'images/default-headshot.jpg')
-        end
-
-        self_logger = logger
-        workdir = (workdir = doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        resources do
-          usernames.each do |username|
-            avatar = %(#{imagesdir}avatars/#{username}.jpg)
-            if ::File.readable? (resolved_avatar = (::File.join workdir, avatar))
-              file avatar => resolved_avatar
-            else
-              self_logger.error %(avatar for #{username} not found or readable: #{avatar}; falling back to default avatar)
-              file avatar => ::File.join(DATA_DIR, 'images/default-avatar.jpg')
-            end
-
-            headshot = %(#{imagesdir}headshots/#{username}.jpg)
-            if ::File.readable? (resolved_headshot = (::File.join workdir, headshot))
-              file headshot => resolved_headshot
-            elsif doc.attr? 'builder', 'editions'
-              self_logger.error %(headshot for #{username} not found or readable: #{headshot}; falling back to default headshot)
-              file headshot => ::File.join(DATA_DIR, 'images/default-headshot.jpg')
-            end
-          end
-        end
-        nil
-      end
-
-      def add_content doc
-        builder, spine, format, images = self, @spine, @format, {}
-        workdir = (doc.attr 'docdir').nil_or_empty? ? '.' : workdir
-        resources workdir: workdir do
-          extend GepubResourceBuilderMixin
-          builder.add_images_from_front_matter doc
-          builder.add_nav_doc doc, self, spine, format
-          builder.add_ncx_doc doc, self, spine
-          ordered do
-            builder.add_cover_page doc, self, @book.manifest unless format == :kf8
-            builder.add_front_matter_page doc, self
-            spine.each_with_index do |item, _i|
-              docfile = item.attr 'docfile'
-              file %(#{item.id || (item.attr 'docname')}.xhtml) => (builder.postprocess_xhtml item.convert, format)
-              add_property 'svg' if ((item.attr 'epub-properties') || []).include? 'svg'
-              # QUESTION should we pass the document itself?
-              next if (image_refs = item.references[:epub_images]).nil?
-              image_refs.each do |image|
-                images[image[:name]] ||= { docfile: docfile, path: image[:path] }
-              end
-              # QUESTION reenable?
-              #linear 'yes' if i == 0
-            end
-          end
-        end
-        add_content_images doc, images
-        nil
-      end
-
-      def add_nav_doc doc, spine_builder, spine, format
-        spine_builder.nav 'nav.xhtml' => (postprocess_xhtml nav_doc(doc, spine), format)
-        spine_builder.id 'nav'
-        nil
-      end
-
-      # TODO: aggregate authors of spine document into authors attribute(s) on main document
-      def nav_doc doc, spine
-        lines = [%(<!DOCTYPE html>
-<html xmlns="http://www.w3.org/1999/xhtml" xmlns:epub="http://www.idpf.org/2007/ops" xml:lang="#{lang = doc.attr 'lang', 'en'}" lang="#{lang}">
-<head>
-<meta charset="UTF-8"/>
-<title>#{sanitize_doctitle_xml doc, :cdata}</title>
-<link rel="stylesheet" type="text/css" href="styles/epub3.css"/>
-<link rel="stylesheet" type="text/css" href="styles/epub3-css3-only.css" media="(min-device-width: 0px)"/>
-</head>
-<body>
-<h1>#{sanitize_doctitle_xml doc, :pcdata}</h1>
-<nav epub:type="toc" id="toc">
-<h2>#{doc.attr 'toc-title'}</h2>)]
-        lines << (nav_level spine, [(doc.attr 'toclevels', 1).to_i, 0].max)
-        lines << %(</nav>
-</body>
-</html>)
-        lines * LF
-      end
-
-      def nav_level items, depth, state = {}
-        lines = []
-        lines << '<ol>'
-        items.each do |item|
-          #index = (state[:index] = (state.fetch :index, 0) + 1)
-          if item.context == :document
-            # NOTE we sanitize the chapter titles because we use formatting to control layout
-            item_label = sanitize_doctitle_xml item, :cdata
-            item_href = (state[:content_doc_href] = %(#{item.id || (item.attr 'docname')}.xhtml))
-          else
-            item_label = sanitize_xml item.title, :pcdata
-            item_href = %(#{state[:content_doc_href]}##{item.id})
-          end
-          lines << %(<li><a href="#{item_href}">#{item_label}</a>)
-          if depth == 0 || (child_sections = item.sections).empty?
-            lines[-1] = %(#{lines[-1]}</li>)
-          else
-            lines << (nav_level child_sections, depth - 1, state)
-            lines << '</li>'
-          end
-          state.delete :content_doc_href if item.context == :document
-        end
-        lines << '</ol>'
-        lines * LF
-      end
-
-      # NOTE gepub doesn't support building a ncx TOC with depth > 1, so do it ourselves
-      def add_ncx_doc doc, spine_builder, spine
-        spine_builder.file 'toc.ncx' => (ncx_doc doc, spine).to_ios
-        spine_builder.id 'ncx'
-        nil
-      end
-
-      def ncx_doc doc, spine
-        # TODO: populate docAuthor element based on unique authors in work
-        lines = [%(<?xml version="1.0" encoding="utf-8"?>
-<ncx xmlns="http://www.daisy.org/z3986/2005/ncx/" version="2005-1" xml:lang="#{doc.attr 'lang', 'en'}">
-<head>
-<meta name="dtb:uid" content="#{@book.identifier}"/>
-%{depth}
-<meta name="dtb:totalPageCount" content="0"/>
-<meta name="dtb:maxPageNumber" content="0"/>
-</head>
-<docTitle><text>#{sanitize_doctitle_xml doc, :cdata}</text></docTitle>
-<navMap>)]
-        lines << (ncx_level spine, [(doc.attr 'toclevels', 1).to_i, 0].max, state = {})
-        lines[0] = lines[0].sub '%{depth}', %(<meta name="dtb:depth" content="#{state[:max_depth]}"/>)
-        lines << %(</navMap>
-</ncx>)
-        lines * LF
-      end
-
-      def ncx_level items, depth, state = {}
-        lines = []
-        state[:max_depth] = (state.fetch :max_depth, 0) + 1
-        items.each do |item|
-          index = (state[:index] = (state.fetch :index, 0) + 1)
-          item_id = %(nav_#{index})
-          if item.context == :document
-            item_label = sanitize_doctitle_xml item, :cdata
-            item_href = (state[:content_doc_href] = %(#{item.id || (item.attr 'docname')}.xhtml))
-          else
-            item_label = sanitize_xml item.title, :cdata
-            item_href = %(#{state[:content_doc_href]}##{item.id})
-          end
-          lines << %(<navPoint id="#{item_id}" playOrder="#{index}">)
-          lines << %(<navLabel><text>#{item_label}</text></navLabel>)
-          lines << %(<content src="#{item_href}"/>)
-          unless depth == 0 || (child_sections = item.sections).empty?
-            lines << (ncx_level child_sections, depth - 1, state)
-          end
-          lines << %(</navPoint>)
-          state.delete :content_doc_href if item.context == :document
-        end
-        lines * LF
-      end
-
-      def collect_keywords doc, spine
-        ([doc] + spine).map {|item|
-          if item.attr? 'keywords'
-            (item.attr 'keywords').split CsvDelimiterRx
-          else
-            []
-          end
-        }.flatten.uniq
-      end
-
-      # Swap fonts in CSS based on the value of the document attribute 'scripts',
-      # then return the list of fonts as well as the font CSS.
-      def select_fonts filename, scripts = 'latin'
-        font_css = ::File.read filename
-        font_css = font_css.gsub(/(?<=-)latin(?=\.ttf\))/, scripts) unless scripts == 'latin'
-
-        # match CSS font urls in the forms of:
-        # src: url(../fonts/notoserif-regular-latin.ttf);
-        # src: url(../fonts/notoserif-regular-latin.ttf) format("truetype");
-        font_list = font_css.scan(/url\(\.\.\/([^)]+\.ttf)\)/).flatten
-
-        [font_list, font_css.to_ios]
-      end
-
-      def postprocess_css_file filename, format
-        return filename unless format == :kf8
-        postprocess_css ::File.read(filename), format
-      end
-
-      def postprocess_css content, format
-        return content.to_ios unless format == :kf8
-        # TODO: convert regular expressions to constants
-        content
-          .gsub(/^  -webkit-column-break-.*\n/, '')
-          .gsub(/^  max-width: .*\n/, '')
-          .to_ios
-      end
-
-      def postprocess_xhtml_file filename, format
-        return filename unless format == :kf8
-        postprocess_xhtml ::File.read(filename), format
-      end
-
-      # NOTE Kindle requires that
-      #      <meta charset="utf-8"/>
-      #      be converted to
-      #      <meta http-equiv="Content-Type" content="application/xml+xhtml; charset=UTF-8"/>
-      def postprocess_xhtml content, format
-        return content.to_ios unless format == :kf8
-        # TODO: convert regular expressions to constants
-        content
-          .gsub(/<meta charset="(.+?)"\/>/, '<meta http-equiv="Content-Type" content="application/xml+xhtml; charset=\1"/>')
-          .gsub(/<img([^>]+) style="width: (\d\d)%;"/, '<img\1 style="width: \2%; height: \2%;"')
-          .gsub(/<script type="text\/javascript">.*?<\/script>\n?/m, '')
-          .to_ios
-      end
-    end
-
-    module GepubResourceBuilderMixin
-      # Add missing method to builder to add a property to last defined item
-      def add_property property
-        @last_defined_item.add_property property
-      end
-
-      # Add helper method to builder to check if property is set on last defined item
-      def property? property
-        (@last_defined_item['properties'] || []).include? property
-      end
-    end
-
-    class Packager
-      include ::Asciidoctor::Logging
-
-      EpubExtensionRx = /\.epub$/i
-      KindlegenCompression = ::Hash['0', '-c0', '1', '-c1', '2', '-c2', 'none', '-c0', 'standard', '-c1', 'huffdic', '-c2']
-
-      def initialize spine_doc, spine, format = :epub3, _options = {}
-        @document = spine_doc
-        @spine = spine || []
-        @format = format
-      end
-
-      def package options = {}
-        doc = @document
-        spine = @spine
-        fmt = @format
-        target = options[:target]
-        dest = File.dirname target
-
-        # FIXME: authors should be aggregated already on parent document
-        if doc.attr? 'authors'
-          authors = (doc.attr 'authors').split(GepubBuilderMixin::CsvDelimiterRx).concat(spine.map {|item| item.attr 'author' }.compact).uniq
-        else
-          authors = []
-        end
-
-        builder = ::GEPUB::Builder.new do
-          extend GepubBuilderMixin
-          @document = doc
-          @spine = spine
-          @format = fmt
-          @book.epub_backward_compat = fmt != :kf8
-
-          language doc.attr('lang', 'en')
-          id 'pub-language'
-
-          if doc.attr? 'uuid'
-            unique_identifier doc.attr('uuid'), 'pub-identifier', 'uuid'
-          else
-            unique_identifier doc.id, 'pub-identifier', 'uuid'
-          end
-          # replace with next line once the attributes argument is supported
-          #unique_identifier doc.id, 'pub-id', 'uuid', 'scheme' => 'xsd:string'
-
-          # NOTE we must use :plain_text here since gepub reencodes
-          title sanitize_doctitle_xml(doc, :plain_text)
-          id 'pub-title'
-
-          # FIXME: this logic needs some work
-          if doc.attr? 'publisher'
-            publisher (publisher_name = (doc.attr 'publisher'))
-            # marc role: Book producer (see http://www.loc.gov/marc/relators/relaterm.html)
-            creator (doc.attr 'producer', publisher_name), 'bkp'
-          elsif doc.attr? 'producer'
-            # NOTE Use producer as both publisher and producer if publisher isn't specified
-            producer_name = doc.attr 'producer'
-            publisher producer_name
-            # marc role: Book producer (see http://www.loc.gov/marc/relators/relaterm.html)
-            creator producer_name, 'bkp'
-          elsif doc.attr? 'author'
-            # NOTE Use author as creator if both publisher or producer are absent
-            # marc role: Author (see http://www.loc.gov/marc/relators/relaterm.html)
-            creator doc.attr('author'), 'aut'
-          end
-
-          if doc.attr? 'creator'
-            # marc role: Creator (see http://www.loc.gov/marc/relators/relaterm.html)
-            creator doc.attr('creator'), 'cre'
-          else
-            # marc role: Manufacturer (see http://www.loc.gov/marc/relators/relaterm.html)
-            # QUESTION should this be bkp?
-            creator 'Asciidoctor', 'mfr'
-          end
-
-          # TODO: getting author list should be a method on Asciidoctor API
-          contributors(*authors)
-
-          if doc.attr? 'reproducible'
-            # We need to set lastmodified to some fixed value. Otherwise, gepub will set it to current date.
-            @book.lastmodified (::Time.at 0).utc
-            # Is it correct that we do not populate dc:date when 'reproducible' is set?
-          else
-            if doc.attr? 'revdate'
-              begin
-                date doc.attr('revdate')
-              rescue ArgumentError => e
-                logger.error %(#{::File.basename doc.attr('docfile')}: failed to parse revdate: #{e})
-                date doc.attr('docdatetime')
-              end
-            else
-              date doc.attr('docdatetime')
-            end
-            @book.lastmodified doc.attr('localdatetime')
-          end
-
-          description doc.attr('description') if doc.attr? 'description'
-
-          (collect_keywords doc, spine).each do |s|
-            subject s
-          end
-
-          source doc.attr('source') if doc.attr? 'source'
-
-          rights doc.attr('copyright') if doc.attr? 'copyright'
-
-          #add_metadata 'ibooks:specified-fonts', true
-
-          add_theme_assets doc
-          add_cover_image doc
-          if (doc.attr 'publication-type', 'book') != 'book'
-            usernames = spine.map {|item| item.attr 'username' }.compact.uniq
-            add_profile_images doc, usernames
-          end
-          add_content doc
-        end
-
-        ::FileUtils.mkdir_p dest unless ::File.directory? dest
-
-        epub_file = fmt == :kf8 ? %(#{::Asciidoctor::Helpers.rootname target}-kf8.epub) : target
-        builder.generate_epub epub_file
-        logger.debug %(Wrote #{fmt.upcase} to #{epub_file})
-        if options[:extract]
-          extract_dir = epub_file.sub EpubExtensionRx, ''
-          ::FileUtils.remove_dir extract_dir if ::File.directory? extract_dir
-          ::Dir.mkdir extract_dir
-          ::Dir.chdir extract_dir do
-            ::Zip::File.open epub_file do |entries|
-              entries.each do |entry|
-                next unless entry.file?
-                unless (entry_dir = ::File.dirname entry.name) == '.' || (::File.directory? entry_dir)
-                  ::FileUtils.mkdir_p entry_dir
-                end
-                entry.extract
-              end
-            end
-          end
-          logger.debug %(Extracted #{fmt.upcase} to #{extract_dir})
-        end
-
-        if fmt == :kf8
-          # QUESTION shouldn't we validate this epub file too?
-          distill_epub_to_mobi epub_file, target, options[:compress], options[:kindlegen_path]
-        elsif options[:validate]
-          validate_epub epub_file, options[:epubcheck_path]
-        end
-      end
-
-      def get_kindlegen_command kindlegen_path
-        unless kindlegen_path.nil?
-          logger.debug %(Using ebook-kindlegen-path attribute: #{kindlegen_path})
-          return [kindlegen_path]
-        end
-
-        unless (result = ENV['KINDLEGEN']).nil?
-          logger.debug %(Using KINDLEGEN env variable: #{result})
-          return [result]
-        end
-
-        begin
-          require 'kindlegen' unless defined? ::Kindlegen
-          result = ::Kindlegen.command.to_s
-          logger.debug %(Using KindleGen from gem: #{result})
-          [result]
-        rescue LoadError => e
-          logger.debug %(#{e}; Using KindleGen from PATH)
-          [%(kindlegen#{::Gem.win_platform? ? '.exe' : ''})]
-        end
-      end
-
-      def distill_epub_to_mobi epub_file, target, compress, kindlegen_path
-        mobi_file = ::File.basename target.sub(EpubExtensionRx, '.mobi')
-        compress_flag = KindlegenCompression[compress ? (compress.empty? ? '1' : compress.to_s) : '0']
-
-        argv = get_kindlegen_command(kindlegen_path) + ['-dont_append_source', compress_flag, '-o', mobi_file, epub_file].compact
-        begin
-          # This duplicates Kindlegen.run, but we want to override executable
-          out, err, res = Open3.capture3(*argv) do |r|
-            r.force_encoding 'UTF-8' if ::Gem.win_platform? && r.respond_to?(:force_encoding)
-          end
-        rescue Errno::ENOENT => e
-          raise 'Unable to run KindleGen. Either install the kindlegen gem or place `kindlegen` executable on PATH or set KINDLEGEN environment variable with path to it', cause: e
-        end
-
-        out.each_line do |line|
-          log_line line
-        end
-        err.each_line do |line|
-          log_line line
-        end
-
-        output_file = ::File.join ::File.dirname(epub_file), mobi_file
-        if res.success?
-          logger.debug %(Wrote MOBI to #{output_file})
-        else
-          logger.error %(KindleGen failed to write MOBI to #{output_file})
-        end
-      end
-
-      def get_epubcheck_command epubcheck_path
-        unless epubcheck_path.nil?
-          logger.debug %(Using ebook-epubcheck-path attribute: #{epubcheck_path})
-          return [epubcheck_path]
-        end
-
-        unless (result = ENV['EPUBCHECK']).nil?
-          logger.debug %(Using EPUBCHECK env variable: #{result})
-          return [result]
-        end
-
-        begin
-          result = ::Gem.bin_path 'epubcheck-ruby', 'epubcheck'
-          logger.debug %(Using EPUBCheck from gem: #{result})
-          [::Gem.ruby, result]
-        rescue ::Gem::Exception => e
-          logger.debug %(#{e}; Using EPUBCheck from PATH)
-          ['epubcheck']
-        end
-      end
-
-      def validate_epub epub_file, epubcheck_path
-        argv = get_epubcheck_command(epubcheck_path) + ['-w', epub_file]
-        begin
-          out, err, res = Open3.capture3(*argv)
-        rescue Errno::ENOENT => e
-          raise 'Unable to run EPUBCheck. Either install epubcheck-ruby gem or place `epubcheck` executable on PATH or set EPUBCHECK environment variable with path to it', cause: e
-        end
-
-        out.each_line do |line|
-          logger.info line
-        end
-        err.each_line do |line|
-          log_line line
-        end
-
-        logger.error %(EPUB validation failed: #{epub_file}) unless res.success?
-      end
-
-      def log_line line
-        line = line.strip
-
-        if line =~ /^fatal/i
-          logger.fatal line
-        elsif line =~ /^error/i
-          logger.error line
-        elsif line =~ /^warning/i
-          logger.warn line
-        else
-          logger.info line
-        end
-      end
-    end
-  end
-end
diff --git a/lib/asciidoctor-epub3/spine_item_processor.rb b/lib/asciidoctor-epub3/spine_item_processor.rb
deleted file mode 100644
index 1bf0aa8..0000000
--- a/lib/asciidoctor-epub3/spine_item_processor.rb
+++ /dev/null
@@ -1,93 +0,0 @@
-# frozen_string_literal: true
-
-module Asciidoctor
-  module Epub3
-    class SpineItemProcessor < Extensions::IncludeProcessor
-      include ::Asciidoctor::Logging
-
-      def initialize document
-        @document = document
-      end
-
-      # NOTE only fires for includes in spine document if registered directly on the instance of the spine document
-      def process doc, reader, target, _attributes
-        spine_doc = doc
-        # TODO: allow URI value
-        unless ::File.file? (include_file = (spine_doc.normalize_system_path target, reader.dir, nil, target_name: 'include file'))
-          logger.warn %(#{reader.line_info}: include file not found: #{include_file})
-          return
-        end
-        inherited_attrs = spine_doc.attributes.dup
-        # QUESTION should we keep backend-epub3 for convenience?
-        %w(backend-epub3 backend-epub3-doctype-book docdir docfile docname doctitle outfilesuffix spine).each {|key| inherited_attrs.delete key }
-        if (leveloffset = inherited_attrs['leveloffset'])
-          leveloffset = inherited_attrs['leveloffset'] = %(#{leveloffset}@) unless leveloffset.end_with? '@'
-        end
-
-        # parse header to get author information
-        spine_item_doc_meta = ::Asciidoctor.load_file include_file,
-            safe: spine_doc.safe,
-            backend: 'epub3-xhtml5',
-            doctype: :article,
-            parse_header_only: true,
-            attributes: leveloffset ? { 'leveloffset' => leveloffset } : nil
-
-        # blank out author information if present in sub-document
-        # FIXME this is a huge hack...we need a cleaner way to do this; perhaps an API method that retrieves all the author attribute names
-        if spine_item_doc_meta.attr? 'author'
-          %w(author firstname lastname email authorinitials authors authorcount).each {|key| inherited_attrs.delete key }
-          idx = 1
-          while inherited_attrs.key? %(author_#{idx})
-            %W[author_#{idx} firstname_#{idx} lastname_#{idx} email_#{idx} authorinitials_#{idx}].each {|key| inherited_attrs.delete key }
-            idx += 1
-          end
-        end
-
-        # REVIEW: reaching into converter to resolve document id feels like a hack; should happen in Asciidoctor parser
-        # also, strange that "id" doesn't work here
-        idprefix = (spine_doc.attr 'idprefix') || (spine_item_doc_meta.attr 'idprefix')
-        idseparator = (spine_doc.attr 'idseparator') || (spine_item_doc_meta.attr 'idseparator')
-        inherited_attrs['css-signature'] = DocumentIdGenerator.generate_id spine_item_doc_meta, idprefix, idseparator
-        inherited_attrs['docreldir'] = ::File.dirname target
-
-        # NOTE can't assign spine document as parent since there's too many assumptions in the Asciidoctor processor
-        spine_item_doc = ::Asciidoctor.load_file include_file,
-            # setting base_dir breaks if outdir is not a subdirectory of spine_doc.base_dir
-            #base_dir: spine_doc.base_dir,
-            # NOTE won't write to correct directory if safe mode is :secure
-            safe: spine_doc.safe,
-            backend: 'epub3-xhtml5',
-            doctype: :article,
-            header_footer: true,
-            catalog_assets: true,
-            to_dir: spine_doc.options[:to_dir],
-            attributes: inherited_attrs
-
-        # restore attributes to those defined in the document header
-        spine_item_doc.restore_attributes
-
-        # FIXME: core should register document ID if specified
-        unless (refs = spine_item_doc.references)[:ids].include? spine_item_doc.id
-          spine_item_doc.register :ids, [spine_item_doc.id, (spine_item_doc.attr 'docreftext') || spine_item_doc.doctitle]
-        end
-
-        refs[:spine] = spine_doc
-        refs[:spine_items] = ((spine_doc.references[:spine_items] ||= []) << spine_item_doc)
-        # NOTE if there are attribute assignments between the include directives,
-        # then this ordered list is not continguous, so bailing on the idea
-        #reader.replace_line %(. link:#{::File.basename(spine_item_doc.attr 'outfile')}[#{spine_item_doc.doctitle}])
-        nil
-      end
-
-      # handles? should get the attributes on include directive as the second argument
-      def handles? target
-        (@document.attr? 'spine') && (ASCIIDOC_EXTENSIONS.include? ::File.extname(target))
-      end
-
-      # FIXME: this method shouldn't be required
-      def update_config config
-        (@config ||= {}).update config
-      end
-    end
-  end
-end
diff --git a/spec/cli_spec.rb b/spec/cli_spec.rb
index 9c132b2..f3b39a2 100644
--- a/spec/cli_spec.rb
+++ b/spec/cli_spec.rb
@@ -21,11 +21,11 @@ describe 'asciidoctor-epub3' do
     _, err, res = run_command asciidoctor_epub3_bin,
         '--failure-level=ERROR',
         '-a', 'ebook-validate',
-        fixture_file('empty.adoc'),
-        '-o', temp_file('empty.epub')
+        fixture_file('invalid.adoc'),
+        '-o', temp_file('invalid.epub')
     expect(res.exitstatus).to eq(1)
     # Error from epubcheck
-    expect(err).to include 'ERROR(RSC-005)'
+    expect(err).to include 'ERROR(RSC-012)'
   end
 
   it 'converts sample book to epub and validates it' do
@@ -51,16 +51,16 @@ describe 'asciidoctor-epub3' do
   end
 
   it 'prints errors to stderr when converts invalid book to epub' do
-    _, err, res = to_epub fixture_file('empty.adoc'), temp_file('empty.epub')
+    _, err, res = to_epub fixture_file('invalid.adoc'), temp_file('invalid.epub')
     expect(res.exitstatus).to eq(0)
     # Error from epubcheck
-    expect(err).to include 'ERROR(RSC-005)'
-    # Error from packager.rb
+    expect(err).to include 'ERROR(RSC-012)'
+    # Error from converter.rb
     expect(err).to include 'EPUB validation failed'
   end
 
   it 'prints errors to stderr when converts invalid book to mobi' do
-    _, err, res = to_mobi fixture_file('empty.adoc'), temp_file('empty.mobi')
+    _, err, res = to_mobi fixture_file('invalid.adoc'), temp_file('invalid.mobi')
     expect(err).to include 'ERROR'
     expect(res.exitstatus).to eq(0)
   end
diff --git a/spec/converter_spec.rb b/spec/converter_spec.rb
index efdce6c..36c9678 100644
--- a/spec/converter_spec.rb
+++ b/spec/converter_spec.rb
@@ -32,6 +32,13 @@ describe Asciidoctor::Epub3::Converter do
       expect(chapter.content).to include '<figcaption>Listing 1. .gitattributes</figcaption>'
     end
 
+    it 'increments listing numbering across chapters' do
+      book, = to_epub 'listing-chapter/book.adoc'
+      chapter_b = book.item_by_href 'chapter-b.xhtml'
+      expect(chapter_b).not_to be_nil
+      expect(chapter_b.content).to include '<figcaption>Listing 2. .gitattributes</figcaption>'
+    end
+
     it 'populates ebook subject from keywords' do
       book, = to_epub 'keywords/book.adoc'
       keywords = book.subject_list.map(&:content)
diff --git a/spec/fixtures/bibliography-chapter/bibliography.adoc b/spec/fixtures/bibliography-chapter/bibliography.adoc
new file mode 100644
index 0000000..4eb6baa
--- /dev/null
+++ b/spec/fixtures/bibliography-chapter/bibliography.adoc
@@ -0,0 +1,7 @@
+[bibliography]
+= Bibliography
+
+- [[[pp]]] Andy Hunt & Dave Thomas. The Pragmatic Programmer:
+From Journeyman to Master. Addison-Wesley. 1999.
+- [[[gof,2]]] Erich Gamma, Richard Helm, Ralph Johnson & John Vlissides. Design Patterns:
+Elements of Reusable Object-Oriented Software. Addison-Wesley. 1994.
diff --git a/spec/fixtures/bibliography-chapter/book.adoc b/spec/fixtures/bibliography-chapter/book.adoc
new file mode 100644
index 0000000..e1f9a5b
--- /dev/null
+++ b/spec/fixtures/bibliography-chapter/book.adoc
@@ -0,0 +1,7 @@
+= Bibliography chapter book
+:doctype: book
+:idprefix:
+
+include::chapter.adoc[leveloffset=+1]
+
+include::bibliography.adoc[leveloffset=+1]
diff --git a/spec/fixtures/bibliography-chapter/chapter.adoc b/spec/fixtures/bibliography-chapter/chapter.adoc
new file mode 100644
index 0000000..8a43606
--- /dev/null
+++ b/spec/fixtures/bibliography-chapter/chapter.adoc
@@ -0,0 +1,4 @@
+= Chapter
+
+_The Pragmatic Programmer_ <<pp>> should be required reading for all developers.
+To learn all about design patterns, refer to the book by the "`Gang of Four`" <<gof>>.
diff --git a/spec/fixtures/diagram/book.adoc b/spec/fixtures/diagram/book.adoc
index a25e582..c64f816 100644
--- a/spec/fixtures/diagram/book.adoc
+++ b/spec/fixtures/diagram/book.adoc
@@ -1,7 +1,6 @@
 = Diagrams book
 :doctype: book
 
-// TODO: https://github.com/asciidoctor/asciidoctor-epub3/issues/190
-// :plantumlconfig: ./plantuml.cfg
+:plantumlconfig: ./plantuml.cfg
 
 include::subdir/chapter.adoc[leveloffset=+1]
diff --git a/spec/fixtures/inline-anchor-xref/chapter.adoc b/spec/fixtures/inline-anchor-xref/chapter.adoc
index c0352b9..475465b 100644
--- a/spec/fixtures/inline-anchor-xref/chapter.adoc
+++ b/spec/fixtures/inline-anchor-xref/chapter.adoc
@@ -1,6 +1,6 @@
 = Chapter
 
-* [[item1]]foo::bar
-* [[item2]]baz::qux
+[[item1]]foo::bar
+[[item2]]baz::qux
 
 Try to refer to <<item1>>.
diff --git a/spec/fixtures/inter-chapter-xref/chapter-a.adoc b/spec/fixtures/inter-chapter-xref/chapter-a.adoc
index 738dbd2..af4b89c 100644
--- a/spec/fixtures/inter-chapter-xref/chapter-a.adoc
+++ b/spec/fixtures/inter-chapter-xref/chapter-a.adoc
@@ -3,4 +3,4 @@
 This is chapter A.
 There's not much too it.
 
-Time to move on to <<chapter-b#>>.
+Time to move on to <<chapter-b>>.
diff --git a/spec/fixtures/inter-subchapter-xref/book.adoc b/spec/fixtures/inter-subchapter-xref/book.adoc
new file mode 100644
index 0000000..554c1d8
--- /dev/null
+++ b/spec/fixtures/inter-subchapter-xref/book.adoc
@@ -0,0 +1,8 @@
+= Book Title
+:doctype: book
+:idprefix:
+:idseparator: -
+
+include::chapter-a.adoc[leveloffset=+1]
+
+include::chapter-b.adoc[leveloffset=+1]
diff --git a/spec/fixtures/inter-subchapter-xref/chapter-a.adoc b/spec/fixtures/inter-subchapter-xref/chapter-a.adoc
new file mode 100644
index 0000000..05d3149
--- /dev/null
+++ b/spec/fixtures/inter-subchapter-xref/chapter-a.adoc
@@ -0,0 +1,3 @@
+= Chapter A
+
+include::subchapter-a.adoc[leveloffset=+1]
diff --git a/spec/fixtures/inter-subchapter-xref/chapter-b.adoc b/spec/fixtures/inter-subchapter-xref/chapter-b.adoc
new file mode 100644
index 0000000..c834622
--- /dev/null
+++ b/spec/fixtures/inter-subchapter-xref/chapter-b.adoc
@@ -0,0 +1,3 @@
+= Chapter B
+
+include::subchapter-b.adoc[leveloffset=+1]
diff --git a/spec/fixtures/inter-subchapter-xref/subchapter-a.adoc b/spec/fixtures/inter-subchapter-xref/subchapter-a.adoc
new file mode 100644
index 0000000..4953f23
--- /dev/null
+++ b/spec/fixtures/inter-subchapter-xref/subchapter-a.adoc
@@ -0,0 +1,3 @@
+= Subchapter A
+
+Look at <<subchapter-b#anchor>>.
diff --git a/spec/fixtures/inter-subchapter-xref/subchapter-b.adoc b/spec/fixtures/inter-subchapter-xref/subchapter-b.adoc
new file mode 100644
index 0000000..720b5d3
--- /dev/null
+++ b/spec/fixtures/inter-subchapter-xref/subchapter-b.adoc
@@ -0,0 +1,3 @@
+= Subchapter B
+
+[[anchor,label]]Anchors rock!
diff --git a/spec/fixtures/invalid.adoc b/spec/fixtures/invalid.adoc
new file mode 100644
index 0000000..0ffb786
--- /dev/null
+++ b/spec/fixtures/invalid.adoc
@@ -0,0 +1,3 @@
+= Title
+
+<<nonexistent,xref to nowhere>>
diff --git a/spec/fixtures/listing-chapter/book.adoc b/spec/fixtures/listing-chapter/book.adoc
new file mode 100644
index 0000000..342754a
--- /dev/null
+++ b/spec/fixtures/listing-chapter/book.adoc
@@ -0,0 +1,8 @@
+= Listing chapter book
+:doctype: book
+:idprefix:
+:idseparator: -
+
+include::chapter-a.adoc[leveloffset=+1]
+
+include::chapter-b.adoc[leveloffset=+1]
diff --git a/spec/fixtures/listing-chapter/chapter-a.adoc b/spec/fixtures/listing-chapter/chapter-a.adoc
new file mode 100644
index 0000000..2ae739d
--- /dev/null
+++ b/spec/fixtures/listing-chapter/chapter-a.adoc
@@ -0,0 +1,6 @@
+= Chapter A
+
+[title=.gitattributes]
+----
+*   -text
+----
diff --git a/spec/fixtures/listing-chapter/chapter-b.adoc b/spec/fixtures/listing-chapter/chapter-b.adoc
new file mode 100644
index 0000000..d5577b0
--- /dev/null
+++ b/spec/fixtures/listing-chapter/chapter-b.adoc
@@ -0,0 +1,6 @@
+= Chapter B
+
+[title=.gitattributes]
+----
+*   -text
+----
diff --git a/spec/fixtures/minimal/chapter.adoc b/spec/fixtures/minimal/chapter.adoc
index e69de29..31b7fa2 100644
--- a/spec/fixtures/minimal/chapter.adoc
+++ b/spec/fixtures/minimal/chapter.adoc
@@ -0,0 +1 @@
+= Chapter
diff --git a/spec/reproducible_spec.rb b/spec/reproducible_spec.rb
index 3a8ba79..d9d2771 100644
--- a/spec/reproducible_spec.rb
+++ b/spec/reproducible_spec.rb
@@ -23,8 +23,6 @@ describe Asciidoctor::Epub3::Converter do
   end
 
   it 'sets mod and creation dates to match SOURCE_DATE_EPOCH environment variable' do
-    skip %(Current Asciidoctor version doesn't support SOURCE_DATE_EPOCH) unless supports_source_date_epoch?
-
     old_source_date_epoch = ENV.delete 'SOURCE_DATE_EPOCH'
     begin
       ENV['SOURCE_DATE_EPOCH'] = '1234123412'
diff --git a/spec/spec_helper.rb b/spec/spec_helper.rb
index 7d9e98b..b93bcfe 100644
--- a/spec/spec_helper.rb
+++ b/spec/spec_helper.rb
@@ -61,11 +61,6 @@ RSpec.configure do |config|
     File.join examples_dir, path
   end
 
-  # Returns true if current version of Asciidoctor supports SOURCE_DATE_EPOCH environment variable
-  def supports_source_date_epoch?
-    Gem::Version.new(Asciidoctor::VERSION) >= Gem::Version.new('1.5.5')
-  end
-
   def has_logger?
     defined? Asciidoctor::LoggerManager
   end
diff --git a/spec/xref_spec.rb b/spec/xref_spec.rb
index 74ba392..f82e082 100644
--- a/spec/xref_spec.rb
+++ b/spec/xref_spec.rb
@@ -4,21 +4,28 @@ require_relative 'spec_helper'
 
 describe 'Asciidoctor::Epub3::Converter - Xref' do
   context 'inter-chapter' do
-    it 'should resolve xref to top of chapter' do
+    it 'resolves xref to top of chapter' do
       book, = to_epub 'inter-chapter-xref/book.adoc'
       chapter_a = book.item_by_href 'chapter-a.xhtml'
       expect(chapter_a).not_to be_nil
       expect(chapter_a.content).to include '<a id="xref--chapter-b" href="chapter-b.xhtml" class="xref">Chapter B</a>'
     end
 
-    it 'should resolve xref to section inside chapter' do
+    it 'resolves xref to section inside chapter' do
       book, = to_epub 'inter-chapter-xref-to-subsection/book.adoc'
       chapter_a = book.item_by_href 'chapter-a.xhtml'
       expect(chapter_a).not_to be_nil
       expect(chapter_a.content).to include '<a id="xref--chapter-b--getting-started" href="chapter-b.xhtml#getting-started" class="xref">Getting Started</a>'
     end
 
-    it 'should resolve xref to inline anchor' do
+    it 'resolves xref between subchapter include files' do
+      book, = to_epub 'inter-subchapter-xref/book.adoc'
+      chapter_a = book.item_by_href 'chapter-a.xhtml'
+      expect(chapter_a).not_to be_nil
+      expect(chapter_a.content).to include '<a id="xref--chapter-b--anchor" href="chapter-b.xhtml#anchor" class="xref">label</a>'
+    end
+
+    it 'resolves xref to inline anchor' do
       book, = to_epub 'inline-anchor-xref/book.adoc'
       chapter = book.item_by_href 'chapter.xhtml'
       expect(chapter).not_to be_nil
@@ -26,12 +33,19 @@ describe 'Asciidoctor::Epub3::Converter - Xref' do
       expect(chapter.content).to include '<a id="xref-item1" href="#item1" class="xref">[item1]</a>'
     end
 
-    it 'should resolve xref to bibliography anchor' do
+    it 'resolves xref to bibliography anchor' do
       book, = to_epub 'bibliography-xref/book.adoc'
       chapter = book.item_by_href 'chapter.xhtml'
       expect(chapter).not_to be_nil
       expect(chapter.content).to include '<a id="item1"></a>[item1] foo::bar'
       expect(chapter.content).to include '<a id="xref-item1" href="#item1" class="xref">[item1]</a>'
     end
+
+    it 'resolves xref to bibliography chapter' do
+      book, = to_epub 'bibliography-chapter/book.adoc'
+      chapter = book.item_by_href 'chapter.xhtml'
+      expect(chapter).not_to be_nil
+      expect(chapter.content).to include '<a id="xref--bibliography--pp" href="bibliography.xhtml#pp" class="xref">[pp]</a>'
+    end
   end
 end