parsanol 1.3.124 → 1.3.125

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fd3371b993b77e992ef44c6a12c46460595e88be97b77dadbcef91d891f2edd0
-  data.tar.gz: 90780bf0b7e04e9afcec6910327df787d97bb32658e9a4963e9d70c7dc79890a
+  metadata.gz: 9300ad7831ee208e1424f8a65c035db1db73f335bedfa2e3cb77027e7ae048ba
+  data.tar.gz: ec7e0cc9a73e967e957eba039f52b66e80c63cf53ca4048e69c6303b2d09f23b
 SHA512:
-  metadata.gz: a85f8d30e83a3ae06d7161e7a3193a4f476f46c32c84e7b4083d6308c486ee21cc69cca41c125b030e6a4330702e0396a0e6563f51643757f366a892fd8e24f8
-  data.tar.gz: d55182b8bf7ae973aeaa5ccb20b22f0bd7c4e7c878272d6cf42acc88ff0e0c5e87d4b2426bac4ab8e963f10642a69c2f07b052ca88c1bb34343c5d50be9ef50c
+  metadata.gz: c57b6923afc7be098b06844fd17fc42957c84ca7829363b8ce90fc04947c64ff7d30d16acef7aa8352f15c2b0b6f406095b6c97acd4aed4806057d4e96c31bf7
+  data.tar.gz: 200c2f64e60befe15181fbb92735d65428f02e3c01245fc66732da1e38083dfdc5143968d5b86f8d3957d4572c42a74af114eea7e0ce9917abe9379885a43aa9

data/lib/parsanol/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Parsanol
-  VERSION = "1.3.124"
+  VERSION = "1.3.125"
 end

data/lib/parsanol/vm.rb

@@ -990,8 +990,12 @@ module Parsanol
         @memoize = memoize
       end
 
-      # The RE match's [char width, byte advance]: single-class
-      # patterns match one character; multi-character patterns (%i
+      # The RE match's [span length, byte advance], both in BYTES: the
+      # interpreter's packed value spans are byte-based (STR packs the
+      # literal's byte length; ANY and the RUN ops pack the char's
+      # UTF-8 byte width; materialize byteslices the span), so the RE
+      # span must count bytes too. Single-class patterns match one
+      # character (its byte width); multi-character patterns (%i
       # literals reach RE with no byte table) consume their full
       # length (ruby#357: advancing one character skipped the
       # pattern's remainder and let the parse re-match inside it).
@@ -1002,12 +1006,17 @@ module Parsanol
           [w, w]
         else
           len = scanner.skip(pattern)
-          [input.byteslice(pos, len).length, len]
+          [len, len]
         end
       end
 
-      # An RE instruction's match at pos: [matched?, char width, byte
-      # advance]. Single-class patterns match one character;
+      # An RE instruction's match at pos: [matched?, span length, byte
+      # advance], both lengths in BYTES (see #re_match_width: the
+      # interpreter's spans are byte-based; a character count here
+      # truncated every slice containing a multibyte character —
+      # ruby#407: "Class[éin]" lost é's second continuation byte
+      # because the span stored the run's char count and materialize
+      # bytesliced it). Single-class patterns match one character;
       # multi-character patterns (%i literals reach RE with no byte
       # table) consume their FULL length (ruby#357: advancing one
       # character skipped the pattern's remainder and let the parse
@@ -1030,7 +1039,7 @@ module Parsanol
           [true, w, w]
         else
           len = scanner.skip(pattern)
-          [true, input.byteslice(pos, len).length, len]
+          [true, len, len]
         end
       end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: parsanol
 version: !ruby/object:Gem::Version
-  version: 1.3.124
+  version: 1.3.125
 platform: ruby
 authors:
 - Ribose Inc.