leptris 1.9.396.0-ppc64le-linux → 1.9.398.0-ppc64le-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 948d64de60780abbd35c5eb75ecbb69609396b681b219047ef0ac7775b9383c3
-  data.tar.gz: 69236732f06743bd08714bbf46697cadd63febc486e324ffeee46798c5182fce
+  metadata.gz: ac182cc0317771a9f1487906920c6d50bf545c80315b811654d84b59679c71bc
+  data.tar.gz: 0ddcd84f4438034af6a96526542174c15060ae75e6ea6df798c38a38f25a9b15
 SHA512:
-  metadata.gz: d3a2ee8d49e508f8d27bbf36610c2d9cd1975c5ec8fdf6f8f752d81c1154f22c630fc5883ab2fd5af5afcafa920eebb8342a7e8777af3dd82fef88d6acf7957b
-  data.tar.gz: 955dccf5c781bd9e1906bade2ae4182fa50d28c15dc899950f0edb340cb77c8deacc307a9132f9e7ba23348398ddd71e911d1082105c07940576f8fb9600956c
+  metadata.gz: 4b1207872fb11a74902eff09dd71d529c2407be2c77962928953a990973c37f329bc976996394b27f2d9414e7823647f8fae9e4ddbb3e70a80bb55638e3541fe
+  data.tar.gz: e6452c37d0afffcd4e98f222d4f6fa0854cb1b06e39182149e1c38bee4e160aed816ca5c6322588fc5c2571774f1ed4e6aed4db1027c82e4b1c6cf1d4f0e42b6

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 = "1.9.396"
+LIBLEPTRIS_VERSION = "1.9.398"
 # 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 = "1.9.396.0"
+  VERSION = "1.9.398.0"
 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 v1.9.396
+- libleptris v1.9.398
 - utf8proc v2.11.0
 
 Rebuild by hand:

data/vendor-src/libleptris/CHANGELOG.md

@@ -1,5 +1,42 @@
 ## [Unreleased]
 
+## [1.9.398] - 2026-10-31
+
+### Fixed
+
+- **gMonth accepts both XSD 1.0 spellings (leptris#1936, PR
+  #1937).** The first edition defined the lexical as `--MM--` (the
+  W3C suite's gMonth002/004 carry `--03--`); the errata added
+  `--MM`. Both validate now, each with its optional timezone.
+- **Absent timezone must end the lexical.** `tz_optional`'s absent
+  path returned success without reaching end of string, so the
+  g-regional/date family accepted trailing garbage (`--03garbage`,
+  `2026-10-08x`). Pinned by a 15-row red-first spec; the XSD
+  compile suite stays 138/138.
+- **The automaton size cap is inert at instance validation
+  (moxml#514, PR #1938).** The #1894 cap is a resource guard, not a
+  validity signal: a document whose content model hits the cap
+  validates on its merits instead of failing.
+
+
+
+## [1.9.397] - 2026-10-31
+
+### Performance
+
+- **Attribute-value tail fused into one pass (leptris#1842, PR
+  #1934).** With the UTF-8 gate on the SIMD validator, the line
+  profile moved to `dp_parse_attrs`: values of ~50 bytes just miss
+  the 48-byte inline probe, so every attribute paid the quote
+  memchr AND a SIMD-contains CALL for `&` AND a second full byte
+  loop for literal whitespace — three scans of every value. One
+  fused early-exit loop computes both flags: **parse +28%** on the
+  2.2MB dense-UTF-8 fixture (4.11 -> 3.20 ms, 536 -> 686 MB/s with
+  frame pointers). Flag semantics unchanged; amp/whitespace hits
+  from the inline probe still short-circuit the tail.
+
+
+
 ## [1.9.396] - 2026-10-31
 
 ### Fixed

data/vendor-src/libleptris/CMakeLists.txt

@@ -3,7 +3,7 @@
 cmake_minimum_required(VERSION 3.20)
 
 project(leptris
-    VERSION 1.9.396
+    VERSION 1.9.398
     DESCRIPTION "Fast XML parser and XPath evaluator in pure C"
     LANGUAGES C CXX
 )

data/vendor-src/libleptris/src/leptris/flat/direct_parse.c

@@ -1114,17 +1114,17 @@ static int dp_parse_attrs(DParser* p, LeptrisElement elem) {
             if (q >= p->end) return -1; /* unterminated */
             val_end = (char*)memchr(q, quote, p->end - q);
             if (!val_end) return -1;
-            if (q < val_end) {
-                if (!has_amp &&
-                    leptris_text_contains(q, (size_t)(val_end - q), '&'))
-                    has_amp = 1;
-                if (!has_ws) {
-                    for (const char* w = q; w < val_end; w++) {
-                        if (*w == '\t' || *w == '\n' || *w == '\r') {
-                            has_ws = 1;
-                            break;
-                        }
-                    }
+            if (q < val_end && (!has_amp || !has_ws)) {
+                /* One fused pass for '&' and literal \t/\n/\r: the old
+                 * shape paid a SIMD-contains CALL plus a second byte
+                 * loop per attribute — 3 scans of every value on the
+                 * #1842 profile (50-byte values just miss the 48-byte
+                 * inline probe, so this tail ran for all of them). */
+                for (const char* w = q; w < val_end; w++) {
+                    unsigned char c = (unsigned char)*w;
+                    has_amp |= (c == '&');
+                    has_ws |= (c == '\t' || c == '\n' || c == '\r');
+                    if (has_amp && has_ws) break;
                 }
             }
         }

data/vendor-src/libleptris/src/leptris/xsd/datatypes.c

@@ -146,7 +146,9 @@ static int two_digits(const char** p) {
     return 1;
 }
 static int tz_optional(const char** p) {
-    if (**p != 'Z' && **p != '+' && **p != '-') return 1; /* absent */
+    /* #1936: the absent timezone is only absent at END of string —
+     * "2000-01-01garbage" used to pass the g-regional family. */
+    if (**p != 'Z' && **p != '+' && **p != '-') return **p == 0;
     if (**p == 'Z') { (*p)++; return **p == 0; }
     (*p)++;
     return two_digits(p) && *(*p) == ':' && (*p)++ && two_digits(p) &&
@@ -393,7 +395,15 @@ int xsd_builtin_valid(const char* type, const char* v) {
         const char* p = v;
         if (*p++ != '-') return 0;
         if (*p++ != '-') return 0;
-        return two_digits_bounded(&p, 1, 12) && tz_optional(&p);
+        if (!two_digits_bounded(&p, 1, 12)) return 0;
+        /* #1936: XSD 1.0 first edition spelled the lexical --MM--
+         * (the errata added --MM) — both are valid 1.0 forms, each
+         * carrying the optional timezone. */
+        if (p[0] == '-' && p[1] == '-') {
+            const char* q = p + 2;
+            return tz_optional(&q);
+        }
+        return tz_optional(&p);
     }
     if (strcmp(type, "xs:dateTime") == 0) {
         const char* p = v;

data/vendor-src/libleptris/src/leptris/xsd/validate.c

@@ -392,14 +392,13 @@ static void validate_children(struct xsd_validator* v,
     int ok = xsd_content_valid(s, model, xsd_target_ns(s), names, uris,
                                n);
     if (ok == -2) {
-        /* #1894: pathological occurrence product — the automaton
-         * exceeds the compile-time size cap. Loud, not a hang. */
-        const char* l2 = NULL;
-        leptris_element_expanded_name(elem, &l2, NULL, NULL);
-        verrf_atk(v, elem, "content-model", l2 ? l2 : "", "element",
-                  l2 ? l2 : "",
-                  "content model exceeds the automaton size cap "
-                  "(pathological occurrence product)");
+        /* #1894/#514: pathological occurrence product — the automaton
+         * exceeds the compile-time size cap. The cap is a resource
+         * guard, not a validity verdict: the content model is
+         * engine-INERT for this element (no instance finding — the
+         * consuming validator decides), the same posture as the
+         * compile-time guard. The element's content check is skipped
+         * and the walk continues; under-report, never false-report. */
         free(names);
         free(uris);
         return;

data/vendor-src/libleptris/test/xsd/test_xsd_compile.cpp

@@ -460,6 +460,35 @@ TEST(XsdDatatypes, BuiltinsLexicalMatrix) {
     }
 }
 
+/* leptris#1936: gMonth accepts BOTH XSD 1.0 spellings — the
+ * errata's --MM and the first edition's --MM-- — each with the
+ * optional timezone; and the g-regional/date family rejects
+ * trailing garbage after an absent timezone. */
+TEST(XsdDatatypes, GMonthSpellingsAndStrictTail) {
+    const BuiltinCase rows[] = {
+        {"xs:gMonth", "--03", 1},
+        {"xs:gMonth", "--03--", 1},
+        {"xs:gMonth", "--03Z", 1},
+        {"xs:gMonth", "--03--Z", 1},
+        {"xs:gMonth", "--03--+05:00", 1},
+        {"xs:gMonth", "--03-05:00", 1},
+        {"xs:gMonth", "--12", 1},
+        {"xs:gMonth", "--13", 0},
+        {"xs:gMonth", "--00", 0},
+        {"xs:gMonth", "--03garbage", 0},
+        {"xs:gMonth", "--03-x", 0},
+        {"xs:gMonth", "--03--05:00", 0},
+        {"xs:date", "2026-10-08garbage", 0},
+        {"xs:gDay", "---05garbage", 0},
+        {"xs:gYear", "2026x", 0},
+    };
+    for (const BuiltinCase& c : rows) {
+        SCOPED_TRACE(std::string(c.type) + " : " + c.value);
+        EXPECT_EQ(leptris_xsd_builtin_valid(c.type, c.value), c.expect)
+            << c.type << " : " + std::string(c.value);
+    }
+}
+
 /* leptris#1883: the builtin lexical table converges on Datatypes
  * Part 2 — collapse semantics, unbounded integers, +INF spelling,
  * 64-bit bounds, calendar/duration/binary grammars, xmlns prefix. */
@@ -3657,6 +3686,33 @@ TEST(XsdContent, PathologicalProductFailsLoudlyNotOom) {
     leptris_xsd_free(s);
 }
 
+/* lutaml/moxml#514: the cap is a resource guard, not a validity
+ * verdict. A content model whose NFA crosses the cap is engine-INERT
+ * at INSTANCE validation — no structured row flips a suite-valid
+ * document to invalid (the compile-time loud face above is the one
+ * place the guard reports). */
+TEST(XsdContent, Issue514CapIsInertAtInstanceValidation) {
+    std::string xsd =
+        "<xs:schema xmlns:xs=\"http://www.w3.org/2001/XMLSchema\">"
+        "<xs:element name=\"r\"><xs:complexType><xs:sequence"
+        " minOccurs=\"0\" maxOccurs=\"2000\"><xs:element name=\"e\""
+        " minOccurs=\"0\" maxOccurs=\"2000\"/></xs:sequence>"
+        "</xs:complexType></xs:element></xs:schema>";
+    LeptrisStatus st = LEPTRIS_OK;
+    LeptrisXsdSchema s =
+        leptris_xsd_compile(xsd.data(), xsd.size(), &st);
+    ASSERT_EQ(st, LEPTRIS_OK);
+    ASSERT_NE(s, (LeptrisXsdSchema)0);
+    LeptrisDocument d = leptris_parse_string("<r><e/><e/></r>", 15, &st);
+    ASSERT_NE(d, nullptr);
+    /* The instance is suite-valid and stays valid: the capped model
+     * is skipped, not failed. */
+    EXPECT_EQ(leptris_xsd_validate(s, d), 1);
+    EXPECT_EQ(leptris_xsd_error_count(s), 0u);
+    leptris_document_free(d);
+    leptris_xsd_free(s);
+}
+
 TEST(XsdContent, Issue1894ReproStaysUnderCapAndCorrect) {
     /* The issue's exact shape — seq maxOccurs=100 x element
      * unbounded — lowers to a loop region (not copies) and must stay

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": "1.9.396",
+  "version": "1.9.398",
   "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: 1.9.396.0
+  version: 1.9.398.0
 platform: ppc64le-linux
 authors:
 - Ribose Inc.