leptris 1.9.193.3 → 1.9.197.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 49ef91f7ec6c9a0642a8a7663226bb98d5a1922adbb026354b9af783ab5f2821
-  data.tar.gz: 114f27006c38eb0949f7754b39b4428da84f6c9e5dcc67886d147ac9ac04ebfb
+  metadata.gz: 1fe0d9ad19af1ea0312a5a04daf2f3fededf7c630490ebcf337bd231c5e86f76
+  data.tar.gz: 69ba7d9ba6b929220068d310d98e06e42480764828f023ed981b4f6625adb193
 SHA512:
-  metadata.gz: 7ca23bf845f255e976988659785f79292fd131502b4ae04c0baaf95f813c06c22d1d57b76412ec52040e4a855cfffc3e122b82f3165a8244ae9c59e8cd1e527a
-  data.tar.gz: 914210f2e059b27479bad365a3f8beef6121109638dced1db909ad49869db27cceb45426630024301d1542841540c6b63db1fe703baaee56c0c7f79d78bdf7f5
+  metadata.gz: 4eb77398a80e630172e14fe9d4ae3cbed2dc0416d3b6f6ed5d2343b394eb24477e9b824d8cfa5d18cdcc61cc2788d812de9b9ea533bf15a34f7478c51882aee7
+  data.tar.gz: 2a10f88e73dceb1ae6b48bb3aeacd8f5f696d54cbb5906acece5d2dac19c81925742b575f42a51a5a5eab750981fa4e9691570359cfa7dafd97c617b5adefff8

data/CHANGELOG.md

@@ -5,6 +5,123 @@ All notable changes to Leptris will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [1.9.197.0] - 2026-09-18
+
+### Changed — libleptris 1.9.194 → 1.9.197 (lockstep)
+
+- **HTML: the html5lib corpus is at ZERO reds upstream** — #659's
+  closing state reached in v1.9.196 (tricky01:8 table scope +
+  marker prune was the last), with table-foreign-scope bursts,
+  after-body/after-after-body comment routing, plaintext
+  raw-to-EOF, nested-table-in-cell, and the document-epilog
+  comment rule. v1.9.197 adds a memchr text-run skip in the
+  tokenizer main loop. The binding's WHATWG battery grows two
+  pins from the zero-reds set: comments past `</html>` become
+  document-level epilog nodes, and a nested table inside a cell
+  nests instead of foster-parenting.
+- **HTML: WHATWG mode is feature-complete upstream (#659 closed)** —
+  the binding now pins the shipped behavior classes in
+  `spec/xml/html_whatwg_full_spec.rb`: active-formatting
+  reconstruction across paragraphs, the adoption agency (misnested
+  formatting reopens inside `<p>`), the Noah's Ark three-element
+  cap, template content outside the live table machinery,
+  RCDATA/rawtext (title/textarea/iframe/noembed/xmp/style),
+  in-table stray-text foster parenting, frameset-mode content
+  drop, foreign integration points, after-body restore, and the
+  script-data escape states. The `:html4` default lane is
+  unchanged.
+- **RNG: required attributes behind `<ref>` are enforced**
+  (#1164). Spec-pinned.
+- **RNG: stray character data under element-only content is
+  rejected (#1153)** — the anyName-wildcard laxness we filed from
+  the v188 pins. Spec-pinned at the verdict level; follow-up filed
+  upstream: the rejection does not yet record an error row
+  (report/legacy channels empty for this case).
+- Engine 1.9.195 also fixes public-API symbols kept through
+  ThinLTO (#1154) — the macOS dylib now exports all 323 declared
+  symbols.
+
+### Added
+
+- **gem-smoke workflow** (the teptris-ruby template, 9 legs): after
+  every release, install the PUBLISHED gem from rubygems on
+  ubuntu/arm64-ubuntu/macos-intel/macos-arm and Windows ucrt
+  (x64 3.3/3.4/4.0, arm64 3.4/4.0), then round-trip
+  parse/xpath/mutate/serialize/RelaxNG against the installed copy —
+  with an explicit guard that the NODE_* constants ship in the
+  Ruby layer (the sibling project's #318 class: a packaging skew
+  surfaced as NameErrors in host apps). `scripts/gem_smoke.rb`.
+- **Ruby-source parity gate in the publish job**: every platform
+  variant's `lib/**/*.rb` must be byte-identical before anything
+  is pushed — a skew (one cell packaging from a different source
+  state) aborts the release.
+
+### Changed
+
+- The armv7 platform legs (`arm-linux`, `arm-linux-musl`) are
+  **experimental** (continue-on-error) until upstream ports the
+  64-bit-pinned layouts: the engine's round-19 ABI asserts
+  correctly abort on 32-bit (`sizeof(leptris_element) == 72`
+  requires 8-byte pointers) — filed as leptris/leptris#1174 with
+  both logs. aarch64-linux (qemu bookworm, glibc floor 2.36)
+  stays strict. The 1.9.194.0 release run correctly refused to
+  publish on their failure; .1/.2 were the parallel hot-read
+  aliases.
+
+## [1.9.194.0] - 2026-09-18
+
+### Changed — libleptris 1.9.193 → 1.9.194 (lockstep)
+
+- **RELAX NG: define bodies are sibling lists** — a `<ref>` to a
+  multi-member define matches the WHOLE body (an implicit group);
+  previously only the body's head pattern was matched, so the
+  metanorma `DocumentBody` shape (optional preface + required
+  sections) never matched its tail. Spec-pinned.
+- **RELAX NG: choice starvation fix** — inside a CHOICE,
+  alternatives that consume a child win over the zero-width
+  `<text/>`/empty branch (inline TextElement models stopped
+  failing). Spec-pinned.
+- **RELAX NG diagnostics: Jing `PatternMatcher` semantics** — refs
+  splice body members into the walk, the invalid-attribute-value
+  message enumerates every reachable VALUE leaf through refs,
+  `incomplete` reports the leftmost required position, and missing
+  required attributes collapse to one message. The metanorma
+  standoc corpus is now byte-identical to Jing. Spec-pinned (the
+  attribution survives ref splicing).
+- Engine 1.9.193.x fixes ride along (no public-symbol changes:
+  audit 323/323). The 1.9.193.2/.3 binding patches shipped without
+  changelog entries.
+
+### Added
+
+- **32-bit ARM platform gems: `arm-linux` and `arm-linux-musl`** —
+  no GitHub-hosted armv7 runner exists, so the release builds them
+  under qemu binfmt emulation (ruby:3.3-bookworm / ruby:3.3-alpine,
+  `--platform linux/arm/v7`); `LEPTRIS_PGO=0` on the emulated legs
+  (the two-stage trainer under qemu costs 10-20x for the same
+  library). The ruby variant vendors both, and the ffi vendor map
+  resolves `armv7l-linux`-style platforms — zero-setup
+  TruffleRuby/JRuby on armv7. Completes platform parity with the
+  sibling project's table.
+
+### Fixed
+
+- **macOS floor of the vendored dylibs (was: macOS 26 only)**: the
+  published darwin gems' `libleptris.dylib` carried
+  `LC_BUILD_VERSION minos 26.0` — the CI runner's SDK default —
+  and refused to load on any older macOS. The compile task now
+  sets `MACOSX_DEPLOYMENT_TARGET=11.0` for cmake (a plain
+  `-DCMAKE_OSX_DEPLOYMENT_TARGET` lands UNINITIALIZED and is
+  ignored); verified `minos 11.0` locally. The extconf.rb pin had
+  only ever covered the native bundle.
+- **glibc floor of the linux gems (was: glibc ≥ 2.38)**: the
+  x86_64 gem was built on a 24.04 runner and leaked GLIBC_2.38
+  symbol requirements — Debian 12 / Ubuntu 22.04 hosts could not
+  load it (the aarch64 gem likewise via the 24.04-arm runner).
+  The x86_64 leg now builds natively on ubuntu-22.04 (floor
+  2.35) and the aarch64 leg builds in a qemu linux/arm64
+  bookworm container (floor 2.36).
+
 ## [1.9.193.0] - 2026-09-17
 
 ### Changed — libleptris 1.9.192 -> 1.9.193 (lockstep)

data/Rakefile

@@ -8,7 +8,7 @@ RSpec::Core::RakeTask.new(:spec)
 # Pin for `rake compile` and the platform-gem builds. Keep in lockstep
 # with .github/workflows/build.yml (which calls `rake compile`) and the
 # CHANGELOG when libleptris releases.
-LIBLEPTRIS_VERSION = "1.9.193"
+LIBLEPTRIS_VERSION = "1.9.197"
 # Vendored alongside libleptris for fn:normalize-unicode (TODO
 # .restructure/20): built per platform with a RELOCATABLE @rpath
 # install name, loaded by ffi.rb before libleptris so the
@@ -77,6 +77,14 @@ task :compile do
   fetch_tarball.call(
     "https://api.github.com/repos/leptris/leptris/tarball/v#{version}", build)
 
+  # NOTE: 32-bit ARM (armv7) is an ENGINE port task — the round-19
+  # ABI asserts pin layouts to 8-byte-pointer sizes and correctly
+  # abort the build on 32-bit. Filed upstream as leptris/leptris
+  # #1174; the release legs stay experimental (continue-on-error)
+  # and flip green automatically when the port ships. We patch
+  # nothing here — the emulated-leg smoke validates whatever the
+  # engine provides.
+
   # utf8proc: shared build only, @rpath install name, local prefix.
   u8_dir = File.expand_path("tmp/utf8proc-#{UTF8PROC_VERSION}", __dir__)
   u8_prefix = File.join(u8_dir, "prefix")
@@ -98,6 +106,16 @@ task :compile do
   # literal on Windows shells (they broke the MSVC build).
   # GCC-family only — MSVC's cl rejects the flag outright (D8021).
   cflags = Gem.win_platform? ? "" : "-Wno-error=incompatible-pointer-types"
+  # macOS floor for the VENDORED dylibs: without this, cmake uses
+  # the building machine's SDK default (a macos-latest runner
+  # shipped minos 26.0) and the dylib refuses to load on user
+  # machines older than the runner — the extconf.rb 11.0 pin only
+  # ever covered the native bundle. CMake initializes
+  # CMAKE_OSX_DEPLOYMENT_TARGET from this env var at first
+  # configure (a plain -D lands UNINITIALIZED and is ignored).
+  # 11.0 = the oldest macOS GitHub still hosts runners for;
+  # matches extconf.rb.
+  ENV["MACOSX_DEPLOYMENT_TARGET"] ||= "11.0" if RUBY_PLATFORM =~ /darwin/
   cmake_base =
     "#{CMAKE_FLAGS.join(' ').sub('-DLEPTRIS_ENABLE_UTF8PROC=OFF', '-DLEPTRIS_ENABLE_UTF8PROC=ON')} " \
     "-DCMAKE_PREFIX_PATH=#{u8_prefix} #{cflags.empty? ? '' : "-DCMAKE_C_FLAGS=#{cflags}"}"
@@ -328,6 +346,10 @@ platforms = [
   "aarch64-linux-musl",
   "x86_64-darwin",
   "arm64-darwin",
+  "arm-linux",
+  "arm-linux-musl",
+  "ppc64le-linux",
+  "s390x-linux-musl",
 ]
 
 platforms.each do |platform|

data/ext/leptris/native/native.c

@@ -1880,6 +1880,67 @@ static VALUE nf_snapshot_subtree(VALUE self, VALUE document, VALUE addr)
     return out;
 }
 
+/* ---- Lean plan rows (lutaml/moxml#249 plan layer) ----------------
+ * Elements only, one 4-tuple per element: [name, attrs_flat, text,
+ * depth] — no Hash per row, no text/comment rows, attr and element
+ * NAMES interned (shared frozen fstrings: names repeat across the
+ * document, so repeat reads allocate zero). Pre-order, depth 0 at
+ * the addressed node. */
+static void walk_rows_impl(void *node, int depth, VALUE out)
+{
+    if (f_node_type(node) == WS_NODE_ELEMENT) {
+        VALUE row = rb_ary_new_capa(4);
+        VALUE attrs;
+        const char *name = f_elem_name(node);
+        const char *an;
+        const char *av;
+        void *a;
+        void *c;
+
+        rb_ary_store(row, 0, name ? rb_enc_interned_str(name, strlen(name),
+                                                     rb_utf8_encoding())
+                                  : Qnil);
+        attrs = rb_ary_new();
+        for (a = f_attr_first(node); a; a = f_attr_next(a)) {
+            an = f_attr_name(a);
+            av = f_attr_value(node, a);
+            rb_ary_push(attrs, an ? rb_enc_interned_str(an, strlen(an),
+                                                         rb_utf8_encoding())
+                                  : Qnil);
+            rb_ary_push(attrs, av ? rb_utf8_str_new_cstr(av) : Qnil);
+        }
+        rb_ary_store(row, 1, attrs);
+        for (c = f_first_child(node); c; c = f_next_sibling(c)) {
+            if (f_node_type(c) == WS_NODE_TEXT) {
+                const char *t = f_text_content(c);
+                rb_ary_store(row, 2, t ? rb_utf8_str_new_cstr(t) : Qnil);
+                break;
+            }
+        }
+        rb_ary_store(row, 3, INT2NUM(depth));
+        rb_ary_push(out, row);
+
+        for (c = f_first_child(node); c; c = f_next_sibling(c)) {
+            walk_rows_impl(c, depth + 1, out);
+        }
+    } else if (f_node_type(node) == WS_NODE_DOCTYPE || f_node_type(node) == 9) {
+        void *dc = f_first_child(node);
+        while (dc) {
+            walk_rows_impl(dc, depth, out);
+            dc = f_next_sibling(dc);
+        }
+    }
+}
+
+static VALUE nf_snapshot_rows(VALUE self, VALUE document, VALUE addr)
+{
+    (void)self;
+    (void)document;
+    VALUE out = rb_ary_new();
+    walk_rows_impl((void *)(uintptr_t)NUM2ULL(addr), 0, out);
+    return out;
+}
+
 /* ---- inner_html in one C pass (TODO.perf/18) --------------------
  * Serializes the receiver's children into one growable buffer:
  * elements via leptris_element_serialize_into (the same opts the
@@ -2356,6 +2417,11 @@ void Init_native(void)
     rb_define_method(c_native_node, "content", nn_content, 0);
     rb_define_method(c_native_node, "attribute", nn_attribute, 1);
     rb_define_alias(c_native_node, "[]", "attribute");
+    /* Unshadowed aliases for consumers whose contract modules shadow
+     * attribute/content: a plain call resolves to these at normal
+     * method-cache cost instead of UnboundMethod#bind_call. */
+    rb_define_alias(c_native_node, "attr_read", "attribute");
+    rb_define_alias(c_native_node, "text_read", "content");
     rb_define_method(c_native_node, "children", nn_children, 0);
     rb_define_method(c_native_node, "element_children", nn_element_children, 0);
     rb_define_method(c_native_node, "next_sibling", nn_next_sibling, 0);
@@ -2422,6 +2488,8 @@ void Init_native(void)
                               nf_fast_inner_xml, 1);
     rb_define_module_function(m_native, "snapshot_subtree",
                               nf_snapshot_subtree, 2);
+    rb_define_module_function(m_native, "snapshot_rows",
+                              nf_snapshot_rows, 2);
     rb_define_module_function(m_native, "doc_handle_attach",
                               nf_doc_handle_attach, 1);
     rb_define_module_function(m_native, "doc_handle_release",

data/lib/leptris/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Leptris
-  VERSION = "1.9.193.3"
+  VERSION = "1.9.197.0"
 end

data/lib/leptris/xml/ffi.rb

@@ -21,6 +21,13 @@ module Leptris
         when /x86_64.*darwin/ then %w[x86_64-darwin]
         when /aarch64.*linux/ then %w[aarch64-linux aarch64-linux-musl]
         when /x86_64.*linux/ then %w[x86_64-linux x86_64-linux-musl]
+        # 32-bit ARM (armv[567l]: "arm-linux", "armv7l-linux",
+        # "armeb-linux", ...). Never matches aarch64 ("aarch" has
+        # no "arm" substring).
+        when /arm.*linux/ then %w[arm-linux arm-linux-musl]
+        # POWER8+ little-endian and IBM Z (qemu-built platform gems)
+        when /ppc64le.*linux/ then %w[ppc64le-linux]
+        when /s390x.*linux/ then %w[s390x-linux s390x-linux-musl]
         else []
         end
       end

data/scripts/gem_smoke.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+# Post-publish smoke (the teptris-ruby gem-smoke template): the
+# INSTALLED gem from rubygems — not the checkout — must parse,
+# query, mutate, serialize, and validate, with the native layer
+# when the platform ships it. Also guards the #318 class: the
+# NODE_* constants must exist in the shipped layer (a packaging
+# skew surfaces as a NameError on first use, not at require).
+require "leptris"
+
+abort "SMOKE FAIL: leptris did not define Leptris::VERSION" unless defined?(Leptris::VERSION)
+puts "leptris #{Leptris::VERSION} on #{RUBY_PLATFORM} (ruby #{RUBY_VERSION})"
+
+doc = Leptris::XML.parse("<r id='1'><a x='2'>t1</a><a>t2</a></r>")
+raise "parse: root" unless doc.root.name == "r"
+raise "xpath: children" unless doc.root.xpath("a").size == 2
+raise "attr read" unless doc.at_xpath("a")["x"] == "2"
+
+doc.at_xpath("a").content = "z"
+raise "mutate" unless doc.at_xpath("a").content == "z"
+raise "serialize" unless doc.to_xml.include?('<a x="2">z</a>')
+
+rng = Leptris::XML::RelaxNG.parse(
+  '<element name="r" xmlns="http://relaxng.org/ns/structure/1.0">' \
+  "<attribute name=\"id\"/></element>")
+rng_doc = Leptris::XML::Document.parse(%q{<r id='1'/>})
+raise "rng verdict" unless rng.valid?(rng_doc)
+raise "rng errors shape" unless rng.validate_errors(rng_doc).is_a?(Array)
+raise "rng report shape" unless rng.validate_report(rng_doc).is_a?(Array)
+
+# The #318 class: node-kind constants must live in the shipped
+# Ruby layer (plain constants cannot go missing with the lib).
+raise "NODE constants" unless Leptris::XML::FFI::NODE_ELEMENT
+
+layer = defined?(Leptris::XML::NATIVE_FAST) ? "native" : "ffi"
+puts "SMOKE OK (#{layer} layer)"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: leptris
 version: !ruby/object:Gem::Version
-  version: 1.9.193.3
+  version: 1.9.197.0
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-17 00:00:00.000000000 Z
+date: 2026-09-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ffi
@@ -146,6 +146,8 @@ files:
 - lib/leptris/vendor/aarch64-linux/libutf8proc.so.3
 - lib/leptris/vendor/arm64-darwin/libleptris.dylib
 - lib/leptris/vendor/arm64-darwin/libutf8proc.3.dylib
+- lib/leptris/vendor/ppc64le-linux/libleptris.so
+- lib/leptris/vendor/ppc64le-linux/libutf8proc.so.3
 - lib/leptris/vendor/x86_64-darwin/libleptris.dylib
 - lib/leptris/vendor/x86_64-darwin/libutf8proc.3.dylib
 - lib/leptris/vendor/x86_64-linux-musl/libleptris.so
@@ -193,6 +195,7 @@ files:
 - lib/leptris/xml/xpath.rb
 - lib/leptris/xml/xquery.rb
 - lib/leptris/xml/xslt.rb
+- scripts/gem_smoke.rb
 homepage: https://github.com/leptris/leptris-ruby
 licenses:
 - MIT