rsyntaxtree 1.8.0 → 1.8.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 27e1c2eb4e8e276a63ab92e99b32bd3ef607451df95088f379ff093f6a206001
-  data.tar.gz: 5147922ca1269a629761d7b93f364f9ae9d22fc41d63ce28fdd4c75223cccc52
+  metadata.gz: 232efc991d608eb2af9ff19d6816d9f22f9bc4a1006506276507c1b8705f34ec
+  data.tar.gz: da8e1ab08b470d76d47809b89ae4fd5b7bf0ef7519847ef3bc7e2955814f5521
 SHA512:
-  metadata.gz: 36edfb6922b94b9e8c0821e72f9154b58ad58fc43bd7f55ae913807534f39acd9dbe8becd00e9f14c30c999bb90cc15b6d7ba9cf38e75a6e1feaeca9c090edb9
-  data.tar.gz: 58f60c25f6c8ab2566ff1755c74c353769ea80a970e020f3e2c656bb2b6ee7d8e75fe82831b5d10c3788f8539d601a6817a257c5f57006d492ca11526c30e77b
+  metadata.gz: 5af80d2af48070f3733c21f19cb543da6a2f443b4e4e2fb4fa2f5a0760584cedcfac5e3b86067623f5f24ba63cfe3267e69692f5ddbc44b3c9fdf35f1de34ace
+  data.tar.gz: 3d391279ae3b669f87f6c62cfa6b2f44cb607a6f5c9d0adbfd37e147cd9a9d2a8c4f998a3e508c423886a4adfe56610e5041c838ffcd40d47f958e96b45d9d49

data/CHANGELOG.md

@@ -1,5 +1,62 @@
 # Changelog
 
+## [1.8.2] - 2026-08
+
+### Changed
+- `tidy: high` now compresses as far as its name promises. The level-balance
+  floor added in 1.8.0, which keeps a pair of siblings from being tucked
+  narrower than the level below it, applied to `medium` and `high` alike and
+  bound first in nearly every tree: across the 75 gallery examples the two
+  modes produced identical figures 70 times. The floor now applies to `medium`
+  only, leaving `high` free to trade even branch angles for width — 29 of the
+  75 examples now differ, by up to 19%. `off`, `low` and `medium` are
+  unchanged; `high`, like `symmetric` at the other end of the scale, is for
+  figures that ask for it.
+- Mathematical alphanumerics (U+1D400–, such as the little *v* of *v*P) are
+  named in the family chains. Neither Noto Sans nor Noto Serif covers the
+  block, so the glyphs came from whatever the machine offered: Noto Sans Math
+  on Alpine, DejaVu Serif on Debian/Ubuntu, STIX Two Math on macOS. The serif
+  style now asks for a serif source first, so a serif tree no longer shows a
+  sans *v*. The Docker images install `font-dejavu` for it.
+
+### Fixed
+- The CLI printed parse errors to stdout and exited 0, so a script generating
+  figures in bulk could not tell a rejected input from a drawn one. Errors go
+  to stderr and the exit status is 1.
+- Dropped `Noto Sans Mono SemiCondensed` from the mono chain. It is a width
+  style of the variable Noto Sans Mono rather than a family of its own, and
+  resolved nowhere on macOS, Debian/Ubuntu or Alpine; the chain fell through
+  to `Noto Sans Mono`, which it now names directly.
+
+## [1.8.1] - 2026-08
+
+### Fixed
+- Arabic rendered as isolated, unjoined letterforms in some environments.
+  Scripts outside Latin and CJK were left to the system's generic font
+  fallback, which is not the same font on every machine: on Alpine the Arabic
+  block was claimed by Noto Sans Math, which has the glyphs but no joining
+  rules, while on Debian/Ubuntu the same text fell to DejaVu Sans. The family
+  chains now name the scripts the gallery covers — Arabic (`Noto Sans Arabic`,
+  with `Noto Naskh Arabic` for the serif style), Hebrew, Devanagari, Thai and
+  Khmer — so machines that have those fonts installed produce the same shapes.
+  Mathematical alphanumerics (U+1D400–) are not named yet and still vary by
+  environment. Gallery example 067 (Arabic) was affected and has been
+  regenerated.
+- Emoji were measured with one font and drawn with another where a colour
+  emoji font was installed: Pango selects `Noto Color Emoji`, but the
+  librsvg/Cairo pipeline does not rasterise its bitmap glyphs, so the drawing
+  fell back to whatever outline font happened to cover the codepoint. The
+  project's Docker images now install the monochrome Noto Emoji and leave the
+  colour build out.
+
+### Changed
+- The Docker images install the Noto packages for Arabic (including Naskh),
+  Hebrew, Devanagari, Thai and Khmer, and carry a fontconfig rule that removes
+  the Arabic ranges from Noto Sans Math while keeping its mathematical
+  alphanumerics (used for the little *v* of *v*P).
+- Documentation records how to override any of the named families with a
+  fontconfig alias, for users who prefer their own script fonts.
+
 ## [1.8.0] - 2026-08
 
 ### Added

data/CITATION.cff

@@ -11,7 +11,7 @@ repository-code: "https://github.com/yohasebe/rsyntaxtree"
 # Concept DOI: always resolves to the latest archived version on Zenodo
 doi: "10.5281/zenodo.21916150"
 license: MIT
-version: 1.8.0
+version: 1.8.2
 date-released: 2026-08-15
 keywords:
   - linguistics

data/Dockerfile

@@ -8,8 +8,27 @@ RUN apk update && \
     apk add --no-cache librsvg librsvg-dev pango-dev imagemagick imagemagick-dev xz-dev libbz2 && \
     apk add --no-cache gobject-introspection gobject-introspection-dev && \
     apk add --no-cache -t .build-packages --no-cache build-base curl-dev wget gcompat && \
-    apk add --no-cache font-noto font-noto-cjk font-noto-cjk-extra font-noto-emoji \
-        font-noto-arabic font-noto-devanagari font-noto-thai
+    apk add --no-cache font-noto font-noto-cjk font-noto-cjk-extra \
+        font-noto-arabic font-noto-naskh-arabic font-noto-hebrew \
+        font-noto-devanagari font-noto-thai font-noto-khmer \
+        font-dejavu
+
+# Noto Sans Math (pulled in by font-noto) claims the Arabic block but carries no
+# joining rules, and fontconfig ranks it above Noto Sans Arabic — Arabic then
+# renders as isolated letterforms. Keep the font, since trees use its
+# mathematical alphanumerics (the little v of vP), but drop Arabic from it.
+COPY dev/fontconfig/99-noto-math-no-arabic.conf /etc/fonts/conf.d/99-noto-math-no-arabic.conf
+
+# Emoji: the monochrome Noto Emoji, not Alpine's font-noto-emoji (which is the
+# colour NotoColorEmoji). Cairo does not rasterise its bitmap glyphs in this
+# pipeline, so emoji would silently fall back to whatever outline font happens
+# to cover them. Pinned to a google/fonts commit so the gallery stays stable.
+ARG NOTO_EMOJI_SHA=b979dba422e445492b0eb9951ac52ee0b4d648c3
+ARG NOTO_EMOJI_SHA256=de6c18832938afc99caf132b39d6a30a19bac7f2e812e28db2535b4608d27551
+RUN mkdir -p /usr/share/fonts/noto-emoji && \
+    wget -q -O /usr/share/fonts/noto-emoji/NotoEmoji-Regular.ttf \
+      "https://raw.githubusercontent.com/google/fonts/${NOTO_EMOJI_SHA}/ofl/notoemoji/NotoEmoji%5Bwght%5D.ttf" && \
+    echo "${NOTO_EMOJI_SHA256}  /usr/share/fonts/noto-emoji/NotoEmoji-Regular.ttf" | sha256sum -c -
 
 
 ADD Gemfile $WORKSPACE

data/README.md

@@ -112,13 +112,19 @@ See [RSyntaxTree Example Gallery](https://yohasebe.github.io/rsyntaxtree/example
 
 ## System Fonts
 
-RSyntaxTree resolves fonts by family name through fontconfig (measurement via Pango, rendering via librsvg), so the fonts must be installed on the system that generates the images. Recommended:
+RSyntaxTree resolves fonts by family name through fontconfig (measurement via Pango, rendering via librsvg), so the fonts must be installed on the machine that generates the images. As of v1.8.0 the gem no longer bundles font files; they were not used at runtime.
 
-- Debian/Ubuntu: `apt install fonts-noto-core fonts-noto-cjk fonts-noto-color-emoji`
-- Alpine: `apk add font-noto font-noto-cjk font-noto-cjk-extra font-noto-emoji` (the `-extra` package carries Noto Serif CJK)
-- macOS: install [Noto Sans/Serif](https://fonts.google.com/noto) (and Noto Sans/Serif JP for CJK); otherwise the system falls back to available fonts
+- Debian/Ubuntu: `apt install fonts-noto-core fonts-noto-cjk`
+- Alpine: `apk add font-noto font-noto-cjk font-noto-cjk-extra` (the `-extra` package carries Noto Serif CJK)
+- macOS: `brew install --cask font-noto-sans font-noto-serif font-noto-sans-mono font-noto-sans-jp font-noto-serif-jp font-noto-sans-cjk font-noto-serif-cjk font-noto-sans-mono-cjk-jp` (macOS itself ships Noto only for a few rare scripts)
 
-Any script supported by an installed font renders correctly — the family chains fall back from Noto Sans/Serif to the JP and CJK variants, so Japanese, Chinese (simplified and traditional) and Korean are covered by the Noto CJK package. As of v1.8.0 the gem no longer bundles font files (they were not used at runtime).
+Arabic, Hebrew, Devanagari, Thai and Khmer need their own packages, and emoji need the **monochrome** [Noto Emoji](https://fonts.google.com/noto/specimen/Noto+Emoji) — colour emoji fonts are not drawn by this pipeline:
+
+- Debian/Ubuntu: `fonts-noto-core` covers the scripts; on minimal images add `fonts-noto` for the full set
+- Alpine: `apk add font-noto-arabic font-noto-naskh-arabic font-noto-hebrew font-noto-devanagari font-noto-thai font-noto-khmer`
+- macOS: the same families are available as `font-noto-*` casks
+
+See [Fonts](https://yohasebe.github.io/rsyntaxtree/documentation#install-fonts-for-svg) in the documentation for which family is used for what, how to substitute your own, and the macOS caveats.
 
 ## Installation
 

data/bin/rsyntaxtree

@@ -173,8 +173,8 @@ data = RSyntaxTree::FormatConverter.to_bracket(data) if data
 begin
   RSyntaxTree::RSGenerator.check_data(data)
 rescue RSTError => e
-  puts e
-  exit
+  warn e
+  exit 1
 end
 
 begin
@@ -188,10 +188,10 @@ begin
   output = rsg.send(draw_method)
   File.binwrite(filepath, output)
 rescue RSTError => e
-  puts e
-  exit
+  warn e
+  exit 1
 rescue StandardError => e
-  p e
-  puts "Error: something unexpected occurred"
-  exit
+  warn "Error: something unexpected occurred"
+  warn "#{e.class}: #{e.message}"
+  exit 1
 end

data/lib/rsyntaxtree/base_graph.rb

@@ -376,7 +376,8 @@ module RSyntaxTree
     # being compressed narrower than the level right below it (an inversion
     # that reads as a needle-thin top over flat lower branches); branch
     # angles then stay comparable between adjacent levels. 1.0 = a pair is
-    # never tighter than the widest pair among its own children.
+    # never tighter than the widest pair among its own children. Applied in
+    # the :ordered nest mode only; see tidy_compress.
     TIDY_LEVEL_BALANCE = 1.0
 
     # Base multiplier for the minimum clearance between adjacent subtrees.
@@ -554,16 +555,21 @@ module RSyntaxTree
             end
           end
 
-          # Level-balance floor (nest mode only): keep the pair at least a
+          # Level-balance floor (:ordered only): keep the pair at least a
           # fraction of the widest child-level spread found inside its own
           # two subtrees. Nesting can tuck an upper pair far narrower than
           # the level right below it — an inversion that reads as a
-          # needle-thin top over flat lower branches. Without nesting the
-          # leaf-span guard already prevents such inversions, and applying
-          # the floor there would only forfeit legitimate compression.
+          # needle-thin top over flat lower branches.
+          #
+          # :none does not need it, since the leaf-span guard already
+          # prevents such inversions. :free deliberately goes without: it is
+          # the end of the scale where density outranks even branch angles,
+          # and with the floor in place it compressed no further than
+          # :ordered on all but a handful of trees.
+          #
           # Compression is a negative delta (the right subtree moves left),
           # so the floor bounds delta from below.
-          if @tidy_nest != :none && delta.negative?
+          if @tidy_nest == :ordered && delta.negative?
             dist = (right_child.horizontal_indent + right_child.content_width / 2.0) -
                    (left_child.horizontal_indent + left_child.content_width / 2.0)
             below = [left_child, right_child].map { |c| child_spread(c) }.compact.max

data/lib/rsyntaxtree/utils.rb

@@ -15,14 +15,50 @@ require 'pango'
 # helpers format it for SVG (quoted) or for Pango::FontDescription#family
 # (unquoted). Keeping a single source guarantees that measurement and
 # rendering resolve through the same fallback chain.
+#
+# Scripts beyond Latin and CJK are named explicitly rather than left to the
+# generic fallback, because the generic fallback is not the same font on every
+# machine: on Alpine the Arabic block was picked up by Noto Sans Math, which
+# has the glyphs but no joining rules and so rendered Arabic as isolated
+# letters, while on Debian the same text fell to DejaVu Sans. A family is
+# listed here when (i) the gallery has an example in that script, or (ii) a
+# concrete environment-dependent failure has been reported for it. Anything
+# else stays with the generic fallback — an unbounded list is unmaintainable.
+SCRIPT_FAMILIES_SANS = ["Noto Sans Arabic", "Noto Sans Hebrew", "Noto Sans Devanagari", "Noto Sans Thai", "Noto Sans Khmer"].freeze
+# Noto has no serif Arabic; Naskh is its serif-like counterpart, with Noto Sans
+# Arabic behind it for systems that ship one but not the other.
+SCRIPT_FAMILIES_SERIF = ["Noto Naskh Arabic", "Noto Sans Arabic", "Noto Serif Hebrew", "Noto Serif Devanagari", "Noto Serif Thai", "Noto Serif Khmer"].freeze
+
+# Monochrome emoji faces only. Colour emoji fonts carry their glyphs as bitmap
+# or COLR tables, which Pango happily measures but librsvg does not draw, so a
+# colour font in this list would give a figure whose emoji are blank or blobbed.
+#
+# This holds only where Pango resolves through fontconfig. On macOS it goes
+# through CoreText, which answers every emoji codepoint with Apple Color Emoji
+# whatever the chain asks for; emoji figures have to be generated elsewhere.
+EMOJI_FAMILIES = ["OpenMoji", "OpenMoji Color", "OpenMoji Black", "Noto Emoji"].freeze
+
+# Mathematical alphanumerics (U+1D400–) label heads such as the little v of vP.
+# Neither Noto Sans nor Noto Serif covers the block, so without these entries
+# the glyphs came from whatever the machine happened to offer — Noto Sans Math
+# on Alpine, DejaVu Serif on Debian/Ubuntu, STIX Two Math on macOS. Noto Sans
+# Math is the only Noto face that covers the block and it is a sans design, so
+# the serif style asks for a serif source first. Listed after the script
+# families: Noto Sans Math also claims the Arabic block without joining rules,
+# and must never be reached before Noto Sans Arabic.
+MATH_FAMILIES_SANS = ["Noto Sans Math"].freeze
+MATH_FAMILIES_SERIF = ["DejaVu Serif", "Noto Sans Math"].freeze
+
 FONT_FAMILIES = {
-  sans:  ["Noto Sans", "Noto Sans JP", "Noto Sans CJK JP", "OpenMoji", "OpenMoji Color", "OpenMoji Black", "Noto Emoji", "sans-serif"].freeze,
-  serif: ["Noto Serif", "Noto Serif JP", "Noto Serif CJK JP", "OpenMoji", "OpenMoji Color", "OpenMoji Black", "Noto Emoji", "serif"].freeze,
-  mono:  ["Noto Sans Mono SemiCondensed", "Noto Sans Mono", "Noto Sans JP", "Noto Sans Mono CJK JP", "OpenMoji", "OpenMoji Color", "OpenMoji Black", "Noto Emoji", "monospace"].freeze,
+  sans:  (["Noto Sans", "Noto Sans JP", "Noto Sans CJK JP"] + SCRIPT_FAMILIES_SANS + MATH_FAMILIES_SANS + EMOJI_FAMILIES + ["sans-serif"]).freeze,
+  serif: (["Noto Serif", "Noto Serif JP", "Noto Serif CJK JP"] + SCRIPT_FAMILIES_SERIF + MATH_FAMILIES_SERIF + EMOJI_FAMILIES + ["serif"]).freeze,
+  # Noto ships no monospaced faces for these scripts, so the mono style borrows
+  # the proportional ones rather than dropping to the generic fallback.
+  mono:  (["Noto Sans Mono", "Noto Sans JP", "Noto Sans Mono CJK JP"] + SCRIPT_FAMILIES_SANS + MATH_FAMILIES_SANS + EMOJI_FAMILIES + ["monospace"]).freeze,
   # The cjk style puts a full-coverage CJK family first, for text that mixes
   # Han, Hangul and kana. Latin falls back to the same Noto faces the sans
   # style uses. (Before 1.8.0 this was WQY Zen Hei.)
-  cjk:   ["Noto Sans CJK JP", "Noto Sans", "Noto Sans JP", "OpenMoji", "OpenMoji Color", "OpenMoji Black", "Noto Emoji", "sans-serif"].freeze
+  cjk:   (["Noto Sans CJK JP", "Noto Sans", "Noto Sans JP"] + SCRIPT_FAMILIES_SANS + MATH_FAMILIES_SANS + EMOJI_FAMILIES + ["sans-serif"]).freeze
 }.freeze
 
 module FontFamily

data/lib/rsyntaxtree/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RSyntaxTree
-  VERSION = "1.8.0"
+  VERSION = "1.8.2"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rsyntaxtree
 version: !ruby/object:Gem::Version
-  version: 1.8.0
+  version: 1.8.2
 platform: ruby
 authors:
 - Yoichiro Hasebe