twinbasic · KubaO · May 17, 2026 · May 15, 2026 · May 15, 2026 · May 15, 2026
diff --git a/.gitignore b/.gitignore
@@ -1,3 +1,4 @@
 .claude/
 CLAUDE.md
 /.htmltest.yml
+node_modules/
diff --git a/BOOKPLAN.md b/BOOKPLAN.md
diff --git a/WIP.md b/WIP.md
@@ -420,16 +420,18 @@ WIP.md itself (and other files outside `docs/`) is not part of the Jekyll site a
 
 ## Scripts and tooling
 
-Any new helper script (content conversion, link checks beyond `check.bat`, etc.) should be written in **Python**. Do not add new Ruby code to this repo. The only Ruby allowed is the existing Jekyll/`just-the-docs` build chain (`Gemfile`, `Gemfile.lock`, `_plugins/`) — that stays as-is. The one carve-out is `_plugins/offlinify.rb`, the link rewriter that powers the offline build (see [Build / preview](#build--preview)); future build-time concerns that are tightly coupled to Jekyll's internal model may go there too, but anything that can stand alone should still be Python.
+**Anything that participates in rendering the online site, the offline site, or the PDF book is handled by Jekyll** — Liquid templates, includes, layouts, data files (`_data/*.yml`), and Ruby plugins under `_plugins/`. Build-time concerns tightly coupled to Jekyll's internal model (URL knowledge, page model, render order, `site.data` injection) belong in a `_plugins/*.rb` plugin. Existing examples: `_plugins/offlinify.rb` (the offline-site link rewriter) and `_plugins/build-info.rb` (the git commit-hash capture that stamps the PDF title page). Adding a Python pre-build step that writes a YAML file into `_data/` and then invokes Jekyll is **not** the way — `bundle exec jekyll build` and `book.bat` must remain self-contained.
+
+Python scripts are reserved for non-render concerns: one-off content conversion (e.g. `scripts/convert_em_dash_separators.py`), repo audits, dev tooling, link checks beyond `check.bat`, anything that runs *outside* a Jekyll build. They should never be a prerequisite for the render pipeline.
 
 ## Build / preview
 
 From `docs/`:
 
-- `bundle exec jekyll build` (or `build.bat`) — builds the online copy to `_site/` **and** a `file://`-browsable copy to `_site-offline/` in a single Jekyll run. The offline pass adds ~3-5s on top of the normal ~13s build; activated by `also_build_offline: true` in `_config.yml`. After Jekyll's WRITE phase, `_plugins/offlinify.rb` walks `_site/`, copies binary assets verbatim into `_site-offline/`, and for each HTML and CSS file rewrites every root-absolute `href` / `src` / `url()` to a page-relative path with the resolved file extension (`/FAQ` → `../../FAQ.html`, `/Tutorials/CEF/` → `../../Tutorials/CEF/index.html`). It also patches the offline copy of `assets/js/just-the-docs.js` in two places — `navLink()` to match the active nav entry by resolved DOM `link.href` rather than `document.location.pathname` (the upstream pathname-vs-attribute compare returns no match under `file://`, leaving the sidebar with no `.active` class so the nav appears collapsed on every navigation), and `initSearch()` to read the lunr index from `window.SEARCH_DATA` rather than fetching `search-data.json` over `XMLHttpRequest` (XHR to `file://` resources is blocked by browsers; classic `<script src=>` is not). To support that, the plugin (a) generates `_site-offline/assets/js/search-data.js` once per build by wrapping the rendered `search-data.json` in `window.SEARCH_DATA = {...};`, and (b) injects two `<script>` tags per page right before `just-the-docs.js`: one that sets `window.OFFLINE_SITE_ROOT` to the per-page relative prefix to the offline site root, and one that loads `search-data.js`. The patched `initSearch()` rewrites every `doc.url` from a root-absolute permalink (`/tB/Core/Const`) to a page-relative path (`<OFFLINE_SITE_ROOT>tB/Core/Const.html`) so search-result clicks land on the actual file regardless of which page the user is on.
+- `bundle exec jekyll build` (or `build.bat`) — builds three trees in a single Jekyll run: the online copy at `_site/`, a `file://`-browsable copy at `_site-offline/`, and the sparse pagedjs source at `_site-pdf/`. The offline pass (`_plugins/offlinify.rb`, activated by `also_build_offline: true` in `_config.yml`) adds ~3-5s and the PDF pass (`_plugins/pdfify.rb`, activated by `also_build_pdf: true`) adds <1s on top of the normal ~13s build. The PDF plugin copies `_site/book.html` (the concatenated chapter document rendered via `_layouts/book-combined.html`) verbatim into `_site-pdf/`, along with `assets/css/print.css`, `assets/css/rouge.css`, and every relative `<img src=>` target -- just what pagedjs needs to render the book PDF. After the copy, the plugin deletes `_site/book.html`: the concatenated document is a build artifact for the PDF render path alone, not a public page on the online site. The companion `offline_exclude: [..., book.html]` entry in `_config.yml` keeps `offlinify.rb` from copying it into `_site-offline/`. The two safeguards are independent -- the exclude pattern works regardless of whether offlinify walks `_site/` before or after pdfify's delete, and pdfify's delete works regardless of whether offlinify is enabled. After Jekyll's WRITE phase, the offline plugin walks `_site/`, copies binary assets verbatim into `_site-offline/`, and for each HTML and CSS file rewrites every root-absolute `href` / `src` / `url()` to a page-relative path with the resolved file extension (`/FAQ` → `../../FAQ.html`, `/Tutorials/CEF/` → `../../Tutorials/CEF/index.html`). It also patches the offline copy of `assets/js/just-the-docs.js` in two places — `navLink()` to match the active nav entry by resolved DOM `link.href` rather than `document.location.pathname` (the upstream pathname-vs-attribute compare returns no match under `file://`, leaving the sidebar with no `.active` class so the nav appears collapsed on every navigation), and `initSearch()` to read the lunr index from `window.SEARCH_DATA` rather than fetching `search-data.json` over `XMLHttpRequest` (XHR to `file://` resources is blocked by browsers; classic `<script src=>` is not). To support that, the plugin (a) generates `_site-offline/assets/js/search-data.js` once per build by wrapping the rendered `search-data.json` in `window.SEARCH_DATA = {...};`, and (b) injects two `<script>` tags per page right before `just-the-docs.js`: one that sets `window.OFFLINE_SITE_ROOT` to the per-page relative prefix to the offline site root, and one that loads `search-data.js`. The patched `initSearch()` rewrites every `doc.url` from a root-absolute permalink (`/tB/Core/Const`) to a page-relative path (`<OFFLINE_SITE_ROOT>tB/Core/Const.html`) so search-result clicks land on the actual file regardless of which page the user is on.
 - `bundle exec jekyll serve` (or `serve.bat`) — local server at `localhost:4000`. Note that `_site-offline/` is also produced on the initial build, but live-reload only updates `_site/`; manual rebuild needed for offline updates.
 - `check.bat` — link check (offline Lychee against `_site/`).
-- `build-offline.bat` — produce **only** the offline copy, writing directly to `_site-offline/` (no `_site/` is generated). Layers `_config_offline.yml` over `_config.yml` to set `offline_build: true` (activates the plugin's standalone in-place mode) and override `also_build_offline: false`. The output tree is byte-equivalent to what the combined build writes to `_site-offline/` — same URL rewriting, same JS patches, same offline-search wiring. Faster than the combined build when only the offline copy is wanted (no `_site/` rendered, no per-file copy step). Useful for shipping just the offline copy as a downloadable bundle.
+- `book.bat` — renders the PDF from `_site-pdf/book.html` via `pagedjs-cli` into `_pdf/book.pdf`. Run `build.bat` first to populate `_site-pdf/`.
 
 ## Site integrity check
 

diff --git a/docs/.gitignore b/docs/.gitignore
@@ -1,5 +1,7 @@
-_site
-_site-offline
+_site/
+_site-offline/
+_site-pdf/
+_pdf/
 .sass-cache
 .jekyll-cache
 .jekyll-metadata

diff --git a/docs/Miscellaneous/Documentation Development.md b/docs/Miscellaneous/Documentation Development.md
@@ -183,17 +183,15 @@ Also ensure that Jekyll is in the PATH. To adjust the path on Windows, press <kb
 
 ### Building
 
-To build the documentation, i.e. render it from `.md` files to the `_site` folder:
+To build the documentation, i.e. render it from `.md` files into the `_site/` (online), `_site-offline/` (offline mirror), and `_site-pdf/` (sparse PDF source) folders:
 
     bundle exec jekyll build
 
 or, on Windows only:
 
     build.bat
 
-To produce **only** the offline-browsable copy (no `_site/`), writing directly to `_site-offline/`:
-
-    build-offline.bat
+A single Jekyll run produces all three trees; toggle `also_build_offline` / `also_build_pdf` in `_config.yml` to skip a sibling output if you only want `_site/`.
 
 ### Checking Link Integrity
 

diff --git a/docs/_config.yml b/docs/_config.yml
@@ -134,11 +134,22 @@ gh_offline_link_url: "https://github.com/twinbasic/documentation/releases/latest
 # the online _site/ output is unaffected.
 also_build_offline: true
 
+# When true, the Pdfify plugin (_plugins/pdfify.rb) runs at the end of
+# the build and emits a sparse _site-pdf/ tree containing only the
+# files pagedjs-cli needs: book.html, the two stylesheets the book
+# layout links (print.css, rouge.css), and every image referenced
+# from book.html. book.bat then renders the PDF from _site-pdf/book.html.
+# Adds <1s to the build (it's a few file copies); the online _site/
+# output is unaffected.
+also_build_pdf: true
+
 # Patterns for files Jekyll produces in _site/ that have no purpose
 # in the offline tree -- Pages / crawler metadata, jekyll-redirect-
 # from output, Windows batch scripts Jekyll picks up from the source
-# directory. The online _site/ keeps them; offlinify strips them
-# from _site-offline/.
+# directory, and the concatenated `book.html` that exists only to
+# feed `_plugins/pdfify.rb` (which copies it to _site-pdf/ and deletes
+# the _site/ copy). The online _site/ keeps the metadata files;
+# offlinify strips them from _site-offline/.
 #
 # Patterns are File.fnmatch-style with FNM_PATHNAME, matched against
 # each file's path relative to the site root. `*` does NOT cross
@@ -148,6 +159,7 @@ offline_exclude:
   - CNAME
   - robots.txt
   - sitemap.xml
+  - book.html
 
 # Excludes for both the build (Jekyll won't try to process these as
 # source) and the watcher (`jekyll serve` won't trigger a rebuild
@@ -157,5 +169,6 @@ offline_exclude:
 # rebuild loop. Keep this in sync with also_build_offline above.
 exclude:
   - _site-offline
+  - _site-pdf
   - redirects.json
   - "*.bat"
diff --git a/docs/_config_offline.yml b/docs/_config_offline.yml
diff --git a/docs/_data/book.yml b/docs/_data/book.yml
@@ -0,0 +1,181 @@
+# Chapter manifest for the twinBASIC documentation PDF.
+#
+# `front_matter` is an ordered list of sections that emit between the
+# title page and the first numbered Part. No divider page, no part
+# number; each entry just contributes its chapter(s) inline. Each
+# entry has:
+#   title    -- the conceptual section name (used by future TOC and as
+#               the anchor seed when the URL would otherwise produce an
+#               empty anchor, e.g. the root `/` page).
+#   page     -- single absolute URL (alternative to `prefixes`); use
+#               when the section is exactly one page (the root index).
+#   prefixes -- URL-prefix list (alternative to `page`); same shape as
+#               on a numbered part.
+#
+# `parts` is an ordered list of numbered book parts. Each part has:
+#   title         -- shown on the part divider page and in the global TOC.
+#   subtitle      -- optional second-line descriptor on the divider.
+#   prefixes      -- one or more URL prefixes; book.html gathers every
+#                    page in `site.pages` whose URL contains a prefix,
+#                    sorts by URL, and emits each as a chapter. Used by
+#                    "flat" parts that map directly to a folder of pages.
+#   page          -- single absolute URL alternative to `prefixes`, used
+#                    for one-chapter parts (e.g. the FAQ). Match is
+#                    exact, not contains-based, so a one-page section
+#                    doesn't accidentally sweep in siblings.
+#   foreword_page -- single absolute URL emitted as a `<article
+#                    class="part-foreword">` right after the part
+#                    divider, before any chapter dividers. No running
+#                    header on these pages (CSS suppresses chrome via a
+#                    named page). Used by the Packages part to surface
+#                    the `/tB/Packages/` landing as introductory
+#                    material for the whole section.
+#   chapters      -- ordered list of per-chapter entries (1.9). Each
+#                    chapter has its own divider page and its own
+#                    chapter-level content; the part becomes a
+#                    container for several distinct subjects rather
+#                    than a flat folder. Each entry has:
+#                      title         -- the chapter divider title.
+#                      subtitle      -- optional descriptor.
+#                      landing_page  -- the chapter's intro page; always
+#                                       emitted first within the
+#                                       chapter, and the plugin strips
+#                                       its source H1 so the chapter
+#                                       divider's H2 is the sole
+#                                       outline entry for the chapter.
+#                      prefixes      -- additional URL prefixes for
+#                                       content pages (members, etc.);
+#                                       sorted by URL after the
+#                                       landing. A page that matches
+#                                       a prefix AND is the landing is
+#                                       emitted only once (as the
+#                                       landing).
+#
+# Index pages (URL ending in `/`) sort before sibling content pages
+# under ASCII lexicographic order, so each part naturally opens with
+# its package landing page followed by the individual symbols.
+
+front_matter:
+  - title: Introduction
+    page: /
+
+parts:
+  - title: Features
+    subtitle: Language additions, attributes, tooling, and packaging
+    prefixes:
+      - /Features/
+
+  - title: Frequently Asked Questions
+    subtitle: Status, scope, and how to get help
+    page: /FAQ
+
+  - title: Tutorials
+    subtitle: Worked examples for CEF, WebView2, and CustomControls
+    prefixes:
+      - /Tutorials/
+
+  - title: The Core Language
+    subtitle: Statements, operators, and built-in keywords
+    prefixes:
+      - /tB/Core/
+
+  - title: Reference Section
+    subtitle: Alphabetical indexes of statements, operators, and compiler constants, plus the project glossary
+    # The live site has a top-level "Reference Section" nav item that groups
+    # these alphabetical-index pages. The five Reference/* pages plus the
+    # Controls table and the Glossary live at three different URL prefixes;
+    # `contains`-match catches all of them. (Attributes also belongs here in
+    # the live nav but its permalink is /tB/Core/Attributes, which the Core
+    # part above already sweeps up; we leave it there rather than introduce
+    # an excludes schema for one page.)
+    prefixes:
+      - /Reference
+      - /tB/Controls
+      - /tB/Gloss
+
+  - title: Packages
+    subtitle: The runtime and library packages shipped with twinBASIC
+    # The /tB/Packages/ landing page is the part's foreword; its content
+    # (an overview of all built-in packages) sits between the part divider
+    # and the first chapter divider, with no running header. Each package
+    # below becomes a chapter with a dedicated chapter-divider title
+    # page; the chapter's own source landing page emits next (with its
+    # source H1 stripped by the plugin to avoid the redundancy with the
+    # chapter divider's H2), followed by the package's class / module /
+    # control pages in URL order.
+    foreword_page: /tB/Packages/
+    chapters:
+      - title: VBA Package
+        subtitle: Standard runtime modules --- Strings, Math, FileSystem, and the rest
+        # The VBA landing lives at /tB/Packages/VBA (added when Packages
+        # was promoted to a top-level nav item), while individual module
+        # members keep their legacy /tB/Modules/... URLs.
+        landing_page: /tB/Packages/VBA
+        prefixes:
+          - /tB/Modules/
+
+      - title: VBRUN Package
+        subtitle: Runtime types for controls, errors, and the property bag
+        landing_page: /tB/Packages/VBRUN/
+        prefixes:
+          - /tB/Packages/VBRUN/
+
+      - title: VB Package
+        subtitle: Classic VB6 forms and intrinsic controls
+        landing_page: /tB/Packages/VB/
+        prefixes:
+          - /tB/Packages/VB/
+
+      - title: WebView2 Package
+        subtitle: Chromium-based browser control via Microsoft Edge WebView2
+        landing_page: /tB/Packages/WebView2/
+        prefixes:
+          - /tB/Packages/WebView2/
+
+      - title: Assert Package
+        subtitle: Test assertions with Exact, Strict, and Permissive comparison
+        landing_page: /tB/Packages/Assert/
+        prefixes:
+          - /tB/Packages/Assert/
+
+      - title: CustomControls Package
+        subtitle: The WaynesControls suite and its supporting framework
+        landing_page: /tB/Packages/CustomControls/
+        prefixes:
+          - /tB/Packages/CustomControls/
+
+      - title: CEF Package
+        subtitle: Chromium Embedded Framework browser control
+        landing_page: /tB/Packages/CEF/
+        prefixes:
+          - /tB/Packages/CEF/
+
+      - title: WinEventLogLib Package
+        subtitle: Windows Event Log integration
+        landing_page: /tB/Packages/WinEventLogLib/
+        prefixes:
+          - /tB/Packages/WinEventLogLib/
+
+      - title: WinNamedPipesLib Package
+        subtitle: Asynchronous named-pipe framework over IOCP
+        landing_page: /tB/Packages/WinNamedPipesLib/
+        prefixes:
+          - /tB/Packages/WinNamedPipesLib/
+
+      - title: WinServicesLib Package
+        subtitle: Windows Services hosting
+        landing_page: /tB/Packages/WinServicesLib/
+        prefixes:
+          - /tB/Packages/WinServicesLib/
+
+      - title: tbIDE Package
+        subtitle: IDE Extensibility --- the addin SDK
+        landing_page: /tB/Packages/tbIDE/
+        prefixes:
+          - /tB/Packages/tbIDE/
+
+      - title: WinNativeCommonCtls Package
+        subtitle: VB6-compatible Common Controls replacement on top of Win32 ComCtl32
+        landing_page: /tB/Packages/WinNativeCommonCtls/
+        prefixes:
+          - /tB/Packages/WinNativeCommonCtls/