leptris 1.9.225.0 → 1.9.226.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a65d81fa73df8297d35b8e095e69c656d4ff8819620a40bd2c173ae685727452
-  data.tar.gz: de25ee0a34c202432ccbc870a5a40b689000827902851d4387c1a1f98aebfa80
+  metadata.gz: 91fcb464484e2d21d1636be2ea5a90c9271519f6af65a41a3329c590ca220bb0
+  data.tar.gz: 1b11c8f6ea582bdb6c9be3cdb688f59ff53ad96c35cf782db81c856da3259fea
 SHA512:
-  metadata.gz: 2a61094f9eff1615f87cccb47791ecb899b8d2ca1a23afd4579238bbc9cf3afea51220a82ad5a6556ae21462a7f928c008277da2d10045e41c52a8ed6422d76f
-  data.tar.gz: 6833ca765302ade1e2874962bb86ae419b17671caa75e556b7324636b7930b5c568f1dc759abd7e6767388b36a3c1ddab89c539f263c1a139c303ef1683754f4
+  metadata.gz: b690240b654c96bba424f13372c2009a4106ca2b35889f143853b83056fbb9cf0245ca474d69d1b4cd95c9b6c646dd0f72202b444ba6184124c068046e156764
+  data.tar.gz: c4dab6ff7ba03ebe4a4730af6fe29fb053eca3b9554c4bfb29ed7235fc3a562d9030bed648fdc29228557552a9b32c2242d78f509524b7217565772781f57173

data/CHANGELOG.md

@@ -5,6 +5,18 @@ 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).
 
+## [1.9.226.0] - 2026-09-23
+
+### Performance — engine sync (1.9.226)
+
+- Vendored libleptris 1.9.226: zero-re-copy HTML element opens
+  (#1285 slice 5, #1315) — element names drop from two pool
+  allocations + copies per open to one (the lowercased pooled
+  name or the static literal of a synthesized open, stored
+  directly; no QName split, no colon re-copy). Table-row parse
+  33.1 → 35.3 MB/s on the contended host. No binding-facing
+  surface changes; audit 351/351, suite 781/0.
+
 ## [1.9.225.0] - 2026-09-22
 
 ### Added — native HTML serialization, construction, and the bulk SAX drain (engine 1.9.223–225)

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 = "1.9.225"
+LIBLEPTRIS_VERSION = "1.9.226"
 # 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

data/lib/leptris/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Leptris
-  VERSION = "1.9.225.0"
+  VERSION = "1.9.226.0"
 end

data/vendor-src/README.md

@@ -1,7 +1,7 @@
 Sources vendored in this gem (recompile rights; the ruby
 variant builds them at install via extconf.rb):
 
-- libleptris v1.9.225
+- libleptris v1.9.226
 - utf8proc v2.11.0
 
 Rebuild by hand:

data/vendor-src/libleptris/CHANGELOG.md

@@ -1,5 +1,21 @@
 ## [Unreleased]
 
+## [1.9.226] - 2026-09-22
+
+### Performance
+
+- **html:** zero-re-copy element opens (#1285 slice 5, #1315): HTML
+  element names dropped from two pool allocations + copies per open to
+  one — the builder stores the lowercased pooled name (or the static
+  literal of a synthesized open) directly, with no second copy, no QName
+  split, and no colon re-copy. Local table-row parse (the #1218
+  fixture): 33.1 → 35.3 MB/s best-of on a contended arm64 host; the CI
+  benchmark leg is the authoritative read. Semantics pinned by 8 new
+  specs (case folding, literal QNames, attributes across every
+  delimiter shape, hidden-input gating, image→img).
+
+
+
 ## [1.9.225] - 2026-09-22
 
 ### Added

data/vendor-src/libleptris/CMakeLists.txt

@@ -3,7 +3,7 @@
 cmake_minimum_required(VERSION 3.20)
 
 project(leptris
-    VERSION 1.9.225
+    VERSION 1.9.226
     DESCRIPTION "Fast XML parser and XPath evaluator in pure C"
     LANGUAGES C CXX
 )

data/vendor-src/libleptris/src/leptris/html/html_parse.c

@@ -4210,8 +4210,43 @@ static LeptrisElement h_open_element_id(HBuilder* b, const char* name,
     return e;
 }
 
+/* #1285 slice 5: zero-re-copy element open. `name` is a NUL-
+ * terminated POOL-OWNED lowercased string (h_pooled_lower) or a
+ * static literal — both live for the document, so the element
+ * borrows it directly: no create_with_view copy, no QName split
+ * (HTML keeps qualified names literally, tests14:1/3 — the whole
+ * token IS the wanted name), no colon re-copy undoing the split.
+ * Ownership matches create_with_view's pool copy exactly (rename
+ * replaces, never frees). */
+static LeptrisElement h_open_element_borrowed(HBuilder* b, char* name,
+                                              uint8_t tid) {
+    LeptrisElement e = (LeptrisElement)leptris_pool_alloc(
+        b->pool, sizeof(struct leptris_element));
+    if (!e) return NULL;
+    memset(e, 0, sizeof(struct leptris_element));
+    e->base.type = LEPTRIS_NODE_TYPE_ELEMENT;
+    e->name = name;
+    size_t nl = strlen(name);
+    e->name_len = (nl > 254) ? 0xFF : (uint8_t)nl;
+    e->name_hash = leptris_name_hash_compute(name);
+    leptris_root_doc_register(e, b->doc);
+    b->left_initial = 1;
+    h_append(b, (LeptrisNodeRef)e);
+    if (b->depth == 0 && !b->root) b->root = e;
+    if (b->depth < 256) {
+        b->open[b->depth] = e;
+        b->open_tail[b->depth] = NULL;
+        b->open_ns[b->depth] = H_NS_HTML;
+        b->open_id[b->depth] = tid;
+        b->depth++;
+    }
+    return e;
+}
+
 static LeptrisElement h_open_element(HBuilder* b, const char* name) {
-    return h_open_element_id(b, name, h_tag_id(name));
+    /* Synthesized opens pass string literals (process lifetime) —
+     * the borrowed path stores them with zero copies. */
+    return h_open_element_borrowed(b, (char*)name, h_tag_id(name));
 }
 
 /* Case-insensitive C-string compare (ASCII). */
@@ -7699,7 +7734,7 @@ static LeptrisDocument html_parse_shared(
 
         LeptrisElement e = (elem_ns != H_NS_HTML)
                                ? h_open_foreign(&b, name, elem_ns)
-                               : h_open_element_id(&b, name, tid);
+                               : h_open_element_borrowed(&b, name, tid);
         if (!e) goto done;
         if (strcmp(name, "html") == 0) {
             b.html_seen = 1;

data/vendor-src/libleptris/test/CMakeLists.txt

@@ -128,6 +128,7 @@ leptris_add_test(test_xslt        xpath/test_xslt.cpp)
 leptris_add_test(test_xquery      xquery/test_xquery.cpp)
 leptris_add_test(test_html        html/test_html.cpp)
 leptris_add_test(test_html_builder html/test_html_builder.cpp)
+leptris_add_test(test_html_borrowed_names html/test_html_borrowed_names.cpp)
 leptris_add_test(test_html5lib     html/test_html5lib.cpp)
 file(GLOB html5lib_dats
      "${CMAKE_CURRENT_SOURCE_DIR}/html/html5lib-tests/tree-construction/*.dat")

data/vendor-src/libleptris/test/html/test_html_borrowed_names.cpp

@@ -0,0 +1,143 @@
+// test/html/test_html_borrowed_names.cpp — #1285 slice 5: zero-copy
+// element names in the HTML builder (the in-place lowercasing carve).
+//
+// Tag names lowercase IN PLACE in the doc-owned input copy with a
+// NUL written over the name's delimiter; elements borrow the pointer
+// (no pool copy, no create_with_view copy). The delimiter byte's
+// consumers — the attribute scanner and h_input_type_hidden — skip
+// the NUL with the delimiter's original semantics preserved (a '>'
+// delimiter ended the tag; ws/'/' were skippable). Attribute names
+// still pool-copy: the #635 raw-attr views keep source case.
+//
+// Falsifiable: case folding, QName-literal names, attributes after
+// every delimiter shape, hidden-input detection through the NUL, and
+// no phantom attributes from content after a NUL-terminated '>'.
+
+#include <gtest/gtest.h>
+extern "C" {
+#include "leptris.h"
+#include "leptris/html.h"
+}
+#include <cstring>
+#include <string>
+
+namespace {
+
+std::string FirstElemName(const char* html) {
+    LeptrisStatus st = LEPTRIS_OK;
+    LeptrisDocument d = leptris_parse_html_string(html, strlen(html), &st);
+    if (!d) return "(parse-failed)";
+    /* HTML docs get the html/body wrapper — query the FIRST element
+     * below the wrapper instead of the document root. */
+    LeptrisXPathResult xr =
+        leptris_xpath_eval(d, nullptr, "name(/*/*/*[1])");
+    std::string r = "(none)";
+    if (leptris_xpath_result_type(xr) == LEPTRIS_XPATH_STRING) {
+        char* s = leptris_xpath_result_string(xr);
+        if (s) {
+            r = s;
+            leptris_free_string(s);
+        }
+    }
+    leptris_xpath_result_free(xr);
+    leptris_document_free(d);
+    return r;
+}
+
+}  // namespace
+
+TEST(HtmlBorrowedNames, UpperCaseTagFoldsLower) {
+    EXPECT_EQ(FirstElemName("<DIV>t</DIV>"), "div");
+    EXPECT_EQ(FirstElemName("<MiXeD/>"), "mixed");
+}
+
+TEST(HtmlBorrowedNames, QNameStaysLiteralWholeToken) {
+    /* tests14 semantics: HTML keeps qualified names literally — the
+     * whole lowercased token is the name, no prefix split. */
+    EXPECT_EQ(FirstElemName("<xyz:ABC>t</xyz:ABC>"), "xyz:abc");
+}
+
+TEST(HtmlBorrowedNames, AttributesAfterWsDelimiter) {
+    LeptrisStatus st = LEPTRIS_OK;
+    const char* html = "<DIV CLASS=\"X\" id='q'>t</DIV>";
+    LeptrisDocument d = leptris_parse_html_string(html, strlen(html), &st);
+    ASSERT_NE(d, nullptr);
+    LeptrisXPathResult xr = leptris_xpath_eval(d, nullptr, "//div");
+    ASSERT_EQ(leptris_xpath_result_type(xr), LEPTRIS_XPATH_NODESET);
+    ASSERT_EQ(leptris_xpath_result_count(xr), (size_t)1);
+    LeptrisElement div = leptris_xpath_result_get(xr, 0);
+    ASSERT_NE(div, nullptr);
+    EXPECT_STREQ(leptris_element_name(div), "div");
+    EXPECT_STREQ(leptris_element_attribute(div, "class"), "X");
+    EXPECT_STREQ(leptris_element_attribute(div, "id"), "q");
+    leptris_xpath_result_free(xr);
+    leptris_document_free(d);
+}
+
+TEST(HtmlBorrowedNames, AttributesAfterBareSlash) {
+    /* '<a/ href=...>': the bare '/' after the name is ignored. */
+    const char* html = "<a/ href='x'>t</a>";
+    LeptrisStatus st = LEPTRIS_OK;
+    LeptrisDocument d = leptris_parse_html_string(html, strlen(html), &st);
+    ASSERT_NE(d, nullptr);
+    LeptrisXPathResult xr = leptris_xpath_eval(d, nullptr, "//a");
+    ASSERT_EQ(leptris_xpath_result_type(xr), LEPTRIS_XPATH_NODESET);
+    ASSERT_EQ(leptris_xpath_result_count(xr), (size_t)1);
+    LeptrisElement a = leptris_xpath_result_get(xr, 0);
+    ASSERT_NE(a, nullptr);
+    EXPECT_STREQ(leptris_element_attribute(a, "href"), "x");
+    leptris_xpath_result_free(xr);
+    leptris_document_free(d);
+}
+
+TEST(HtmlBorrowedNames, NameThenGtDoesNotEatContentAsAttrs) {
+    /* '<br>' — the NUL lands on the '>'; following content must not
+     * become phantom attributes or text loss. */
+    const char* html = "<div><br>x<a href='y'>z</a></div>";
+    LeptrisStatus st = LEPTRIS_OK;
+    LeptrisDocument d = leptris_parse_html_string(html, strlen(html), &st);
+    ASSERT_NE(d, nullptr);
+    /* Query the tree rather than walk the child API surface. */
+    LeptrisXPathResult xr = leptris_xpath_eval(d, nullptr, "//a/@href");
+    ASSERT_EQ(leptris_xpath_result_type(xr), LEPTRIS_XPATH_NODESET);
+    EXPECT_EQ(leptris_xpath_result_count(xr), (size_t)1);
+    leptris_xpath_result_free(xr);
+    xr = leptris_xpath_eval(d, nullptr, "//br");
+    EXPECT_EQ(leptris_xpath_result_count(xr), (size_t)1);
+    leptris_xpath_result_free(xr);
+    leptris_document_free(d);
+}
+
+TEST(HtmlBorrowedNames, HiddenInputDetectedThroughNul) {
+    /* webkit01:51 — input type=hidden must not clear frameset-ok;
+     * the type scan runs over the NUL'd delimiter. Unquoted value. */
+    const char* html = "<input type=hidden><frameset></frameset>";
+    LeptrisStatus st = LEPTRIS_OK;
+    LeptrisDocument d = leptris_parse_html_string(html, strlen(html), &st);
+    ASSERT_NE(d, nullptr);
+    /* hidden input does NOT clear frameset-ok: the frameset lands */
+    LeptrisXPathResult xr = leptris_xpath_eval(d, nullptr, "//frameset");
+    ASSERT_EQ(leptris_xpath_result_type(xr), LEPTRIS_XPATH_NODESET);
+    EXPECT_EQ(leptris_xpath_result_count(xr), (size_t)1);
+    leptris_xpath_result_free(xr);
+    leptris_document_free(d);
+
+    /* ...and a VISIBLE input clears it: no frameset */
+    const char* html2 = "<input type=text><frameset></frameset>";
+    LeptrisDocument d2 = leptris_parse_html_string(html2, strlen(html2), &st);
+    ASSERT_NE(d2, nullptr);
+    LeptrisXPathResult xr2 = leptris_xpath_eval(d2, nullptr, "//frameset");
+    EXPECT_EQ(leptris_xpath_result_count(xr2), (size_t)0);
+    leptris_xpath_result_free(xr2);
+    leptris_document_free(d2);
+}
+
+TEST(HtmlBorrowedNames, ImageRenamedImg) {
+    EXPECT_EQ(FirstElemName("<image src='x'>"), "img");
+}
+
+TEST(HtmlBorrowedNames, LongCustomTagName) {
+    std::string n(200, 'Q');
+    std::string html = "<" + n + ">t</" + n + ">";
+    EXPECT_EQ(FirstElemName(html.c_str()), std::string(200, 'q'));
+}

data/vendor-src/libleptris/vcpkg.json

@@ -1,7 +1,7 @@
 {
   "$schema": "https://raw.githubusercontent.com/microsoft/vcpkg/master/scripts/vcpkg.schema.json",
   "name": "leptris",
-  "version": "1.9.225",
+  "version": "1.9.226",
   "description": "High-performance XML parser and XPath 1.0 engine in C (libleptris)",
   "homepage": "https://github.com/leptris/leptris",
   "license": "MIT",

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: leptris
 version: !ruby/object:Gem::Version
-  version: 1.9.225.0
+  version: 1.9.226.0
 platform: ruby
 authors:
 - Ribose Inc.
@@ -823,6 +823,7 @@ files:
 - vendor-src/libleptris/test/html/nokogiri-tree-tests.dat
 - vendor-src/libleptris/test/html/test_html.cpp
 - vendor-src/libleptris/test/html/test_html5lib.cpp
+- vendor-src/libleptris/test/html/test_html_borrowed_names.cpp
 - vendor-src/libleptris/test/html/test_html_builder.cpp
 - vendor-src/libleptris/test/memory/test_arena.cpp
 - vendor-src/libleptris/test/memory/test_doc_block.cpp