leptris 1.9.239.0 → 1.9.240.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a9988358313da5babf4756d5a3393ff7a805eaabb0a6c04a240bf75173f8248b
-  data.tar.gz: 80ddb8c442105e7afad9a1cc40e010f670e115f61e295978157182889a6f2c6e
+  metadata.gz: 91e18b7955af279de38382142b10d20f972c5d8fb1ceb02216efdcae5b572e8d
+  data.tar.gz: f30f6e5564aac12de9a5a074be4ebd706bb6fba196911fbbd752229507763b0a
 SHA512:
-  metadata.gz: 194a8513e590353c118ff58cac0700789337ce254d8b138e5818a228134f75a5d40f2e4a54cbd8045900182bdb25ea8ca706efe5731e6ae0b3b086d16842f59a
-  data.tar.gz: 887dbf39367b2cf2397a493c59d2bb569212acefd4a11f8c1bcd5cbc167640d254e03f4099a71b11c3a9c5f19dcf7c1c75bc4831b6aee1b2533c3cf1f9c3a790
+  metadata.gz: 3cc5846626e6feac81081470224e77a47da5e6094c4d57c7444d2135837aea33cdb799d34437cd717f0a4f7ae28f9df2abd0f8a3772d204693b4c27871bc4d2f
+  data.tar.gz: fa24557f80e5182058a320a7cf87fae1644f324750a767c971aa0a836afe2fd611fc610b0f4e57d34a862b6eabe689b13c146b8db5a98fe25f37c68dfbbee882

data/lib/leptris/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Leptris
-  VERSION = "1.9.239.0"
+  VERSION = "1.9.240.0"
 end

data/lib/leptris/xml/ffi.rb

@@ -232,6 +232,32 @@ module Leptris
       SAX_EVENT_END_PREFIX     = 9
       SAX_EVENT_ERROR          = 10
 
+      # LeptrisSaxRecord (sax.h, #1298): one element or text record;
+      # tree as parent/next_sib record indices, strings (off,len)
+      # views into the records buffer. 10 x uint32 + uint8.
+      class SaxRecord < ::FFI::Struct
+        layout :kind, :uint32,
+               :parent, :uint32, :next_sib, :uint32,
+               :off, :uint32, :len, :uint32,
+               :line, :uint32, :start_tag_end, :uint32,
+               :elem_end, :uint32, :attr_first, :uint32,
+               :attr_count, :uint32,
+               :self_closing, :uint8
+      end
+
+      SAX_REC_ELEMENT = 0
+      SAX_REC_TEXT    = 1
+      SAX_REC_NO_INDEX = 0xFFFFFFFF
+
+      # LeptrisSaxAttr: flat attribute table entry — qname + raw
+      # value views, value_has_ws = literal \t/\n/\r present (the
+      # CONSUMER applies 3.3.3 normalization). 4 x uint32 + uint8.
+      class SaxAttr < ::FFI::Struct
+        layout :name_off, :uint32, :name_len, :uint32,
+               :value_off, :uint32, :value_len, :uint32,
+               :value_has_ws, :uint8
+      end
+
       ITERPARSE_TOP_LEVEL      = 0
       ITERPARSE_FULL_DOCUMENT  = 1
 
@@ -1049,7 +1075,10 @@ attach_function :leptris_parse_string,
       attach_function :leptris_sax_records_data, [:pointer], :pointer
       attach_function :leptris_sax_records_attrs,
         [:pointer, :pointer], :pointer
-      attach_function :leptris_sax_records_buffer, [:pointer], UTF8_STRING
+      # :pointer, NOT UTF8_STRING — the buffer carries embedded NULs
+      # between views; a C string read would truncate at the first.
+      # Views materialize via Pointer#get_string(off, len).
+      attach_function :leptris_sax_records_buffer, [:pointer], :pointer
       attach_function :leptris_sax_records_free, [:pointer], :void
       # XSLT 1.0 engine (libleptris 1.9.1): compile once, apply many.
       # XQuery 1.0 core (libleptris 1.9.64-1.9.66): orchestration
@@ -1284,6 +1313,7 @@ attach_function :leptris_parse_string,
       LEPTRIS_ERROR_NOT_FOUND = -6
       LEPTRIS_ERROR_IO = -7
       LEPTRIS_ERROR_NOT_IMPLEMENTED = -8
+      LEPTRIS_ERROR_NOT_SUPPORTED = -9
 
       XPATH_NODESET = 0
       XPATH_BOOLEAN = 1

data/lib/leptris/xml/sax/records.rb

@@ -0,0 +1,134 @@
+# frozen_string_literal: true
+
+# Bulk SAX records (leptris#1298, engine leptris_sax_records_parse,
+# 1.9.223+): ONE crossing drains the whole document into a flat
+# record table — elements and text runs in document order, the tree
+# as parent/next-sibling record indices, attributes an index range
+# into a flat table, every string an (off,len) view into one
+# NUL-padded buffer. The engine mints zero per-event strings; hosts
+# materialize only what they consume (name/value String = one
+# get_string each).
+#
+# Coverage: the interleaved lane's scannable subset — elements, text,
+# attributes; well-formed entity-free documents without comments/
+# PIs/CDATA/DOCTYPE/xmlns. Outside the subset .open answers nil
+# (LEPTRIS_ERROR_NOT_SUPPORTED, zero allocation) and callers fall
+# back to the callback Recorder; correctness is bit-identical.
+#
+# Field columns materialize once per drain via
+# Pointer#read_array_of_uint32 (no Ruby object per record); the
+# self_closing flag rides a separate uint8 column because it shares
+# its struct tail with padding.
+class Leptris::XML::SAX::Records
+  KIND_ELEMENT = Leptris::XML::FFI::SAX_REC_ELEMENT
+  KIND_TEXT = Leptris::XML::FFI::SAX_REC_TEXT
+
+  REC = Leptris::XML::FFI::SaxRecord
+  ATTR = Leptris::XML::FFI::SaxAttr
+  REC_U32 = REC.size / 4
+  ATTR_U32 = ATTR.size / 4
+
+  attr_reader :count, :attr_count
+
+  # Drains xml into a table. Returns nil unless the document drained
+  # cleanly — outside the scannable subset (NOT_SUPPORTED, zero
+  # allocation) OR malformed (the callback path reports parse errors
+  # with position) — callers fall back to the callback Recorder in
+  # every nil case; correctness is bit-identical.
+  def self.open(xml)
+    xml = xml.to_s
+    out = ::FFI::MemoryPointer.new(:pointer)
+    status = Leptris::XML::FFI.leptris_sax_records_parse(
+      xml, xml.bytesize, 0, out
+    )
+    return nil unless status == Leptris::XML::FFI::LEPTRIS_OK
+
+    new(out.read_pointer)
+  end
+
+  # Block form: the table dies at the end of the block — every view
+  # dies with it, so materialize inside.
+  def self.parse(xml)
+    table = open(xml)
+    return nil unless table
+
+    begin
+      yield table
+    ensure
+      table.free
+    end
+  end
+
+  def initialize(ptr)
+    @ptr = ptr
+    @count = Leptris::XML::FFI.leptris_sax_records_count(@ptr)
+    recs = Leptris::XML::FFI.leptris_sax_records_data(@ptr)
+    attr_out = ::FFI::MemoryPointer.new(:pointer)
+    attrs = Leptris::XML::FFI.leptris_sax_records_attrs(@ptr, attr_out)
+    @attr_count = attr_out.read_uint64
+    @buf = Leptris::XML::FFI.leptris_sax_records_buffer(@ptr)
+    # Whole-table single reads: no Ruby object per record. The u32
+    # columns stride by struct size; the flags (self_closing /
+    # value_has_ws) are u8 tails read per-byte off their layout
+    # offset — endian-safe.
+    @rec_u32 = recs.read_array_of_uint32(@count * REC_U32)
+    @rec_flag = recs.read_array_of_uint8(@count * REC.size)
+    @attr_u32 = attrs.read_array_of_uint32(@attr_count * ATTR_U32)
+    @attr_flag = attrs.read_array_of_uint8(@attr_count * ATTR.size)
+    @rec_flag_off = REC.offset_of(:self_closing)
+    @attr_flag_off = ATTR.offset_of(:value_has_ws)
+  end
+
+  def free
+    return if @ptr.nil?
+
+    Leptris::XML::FFI.leptris_sax_records_free(@ptr)
+    @ptr = nil
+  end
+
+  def kind(i)
+    @rec_u32[i * REC_U32]
+  end
+
+  # Parent record index (0xFFFFFFFF for document-level records).
+  def parent(i)
+    @rec_u32[i * REC_U32 + 1]
+  end
+
+  # Element name or text content — ONE String per call, UTF-8.
+  def view(i)
+    base = i * REC_U32
+    @buf.get_string(@rec_u32[base + 3], @rec_u32[base + 4])
+        .force_encoding(Encoding::UTF_8)
+  end
+
+  def self_closing?(i)
+    @rec_flag[i * REC.size + @rec_flag_off] != 0
+  end
+
+  def attr_first(i)
+    @rec_u32[i * REC_U32 + 8]
+  end
+
+  def attr_count_of(i)
+    @rec_u32[i * REC_U32 + 9]
+  end
+
+  def attr_name(k)
+    @buf.get_string(@attr_u32[k * ATTR_U32],
+                    @attr_u32[k * ATTR_U32 + 1])
+        .force_encoding(Encoding::UTF_8)
+  end
+
+  def attr_value(k)
+    @buf.get_string(@attr_u32[k * ATTR_U32 + 2],
+                    @attr_u32[k * ATTR_U32 + 3])
+        .force_encoding(Encoding::UTF_8)
+  end
+
+  # Literal \t/\n/\r in the raw value — the consumer applies
+  # XML 1.0 §3.3.3 (map each break byte to a space).
+  def attr_value_has_ws?(k)
+    @attr_flag[k * ATTR.size + @attr_flag_off] != 0
+  end
+end

data/lib/leptris/xml/sax.rb

@@ -9,6 +9,7 @@ module Leptris
       autoload :Parser, "leptris/xml/sax/parser"
       autoload :Recorder, "leptris/xml/sax/recorder"
       autoload :DomDispatch, "leptris/xml/sax/dom_dispatch"
+      autoload :Records, "leptris/xml/sax/records"
     end
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: leptris
 version: !ruby/object:Gem::Version
-  version: 1.9.239.0
+  version: 1.9.240.0
 platform: ruby
 authors:
 - Ribose Inc.
@@ -197,6 +197,7 @@ files:
 - lib/leptris/xml/sax/dom_dispatch.rb
 - lib/leptris/xml/sax/parser.rb
 - lib/leptris/xml/sax/recorder.rb
+- lib/leptris/xml/sax/records.rb
 - lib/leptris/xml/schematron.rb
 - lib/leptris/xml/searchable.rb
 - lib/leptris/xml/serialization.rb