parsanol 1.3.120 → 1.3.121

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e61dcbc90916d0c5bc6d82ecfa1f4c1f76987df549b49460a20e01622bb0a159
-  data.tar.gz: 3a4214dcc35b10e44819910f78cc1905dce036c1789ab7162c8819f92ffc15a6
+  metadata.gz: 6b2308bfe0b4b5787cbfac5725e10e2afd01ade41f3ae373dd09df250dbaebd3
+  data.tar.gz: b49dc9c6b767be5c8ec34492b3e1a63683aa8bba12f8a4d194ddd107169b51da
 SHA512:
-  metadata.gz: b9be0ac6018e475fbaba69dc3f95a0ca55567f8664049aaab3b7ffa20a5dafc7c2508da7d3aa66768010373a6b926426761e134ce39b490e0663b0394ea0ffb7
-  data.tar.gz: 0373d7ad8a5e0c5dad37702b59c22e12fd75646a2de59fbe1dbaa7d02dcd9a297b099c8a32759a3a048acf0c63e9d7b7a8f212a616e4f7c8caf41fa8be29776c
+  metadata.gz: 280e47f635e0319aa01a2b5b4936ddc3aa62c9eae8e6f3c13f07600a5904bafb2e7bbd7521f984e5689f5e550eb964c09ac23c8ca36ab7913d4c776f83fafea4
+  data.tar.gz: bab0183da6fcf3b5c837ba9f46e675269b5b567e77a3333ea8b5c41c5c85f5390a3f82cc8d9778b63cd57308fa7b0053d67da12d055f9a70c18e0c7c397b1c7f

data/Cargo.lock

@@ -307,9 +307,9 @@ checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"
 
 [[package]]
 name = "parsanol"
-version = "0.19.17"
+version = "0.19.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "761b572291fd7bad4514b55492cc402d297dd1586e48d1daad6ca25a927968b1"
+checksum = "7ec17007676a476ef23db64b69c3bf5b36d741763ac11ced3089b90a96c9a716"
 dependencies = [
  "ahash",
  "getrandom 0.3.4",
@@ -326,9 +326,9 @@ dependencies = [
 
 [[package]]
 name = "parsanol-derive"
-version = "0.19.17"
+version = "0.19.19"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3831668982431ef76e72800bd4cb399c38f0b51554541d4e4263a09fe461f460"
+checksum = "fc342af122d8854b8916eaacc6a6c2ad911419c605ea47b74b575c2ca24ab23d"
 dependencies = [
  "proc-macro2",
  "quote",

data/lib/parsanol/native/parser.rb

@@ -178,13 +178,27 @@ module Parsanol
 
         def append_atom_structure(atom, visited = {}.compare_by_identity, output = +"")
           if visited[atom]
+            # An Entity's self-reference keeps its name in the
+            # fingerprint, mirroring the non-visited branch (ruby#404).
+            if atom.is_a?(Parsanol::Atoms::Entity)
+              return output << "[:entity, " << atom.rule_name.to_s.inspect << ", [:cycle, " << atom.class.name.inspect << "]]"
+            end
+
             return output << "[:cycle, " << atom.class.name.inspect << "]"
           end
 
           visited[atom] = true
           case atom
           when Parsanol::Atoms::Entity
-            append_atom_structure(atom.parslet, visited, output)
+            # The rule name belongs in the fingerprint (rs#404-class
+            # collision): Entity is transparent in the serialized graph's
+            # STRUCTURE only when the target is spelled out, but the JSON
+            # keeps the Entity node itself — two webs whose Entities point
+            # at different rules share a transparent fingerprint while
+            # serializing differently, and the serializer cache then
+            # serves one grammar's JSON for the other.
+            output << "[:entity, " << atom.rule_name.to_s.inspect << ", "
+            append_atom_structure(atom.parslet, visited, output) << "]"
           when Parsanol::Atoms::Str
             output << "[:str, " << atom.str.inspect << "]"
           when Parsanol::Atoms::Re

data/lib/parsanol/parg/compiler.rb

@@ -261,6 +261,10 @@ module Parsanol
         # the declaration means the tree never sees this subtree named,
         # inside or out.
         return atom if glue_rule?(rule_name)
+        # A capture spans the rules it references (ruby#400): their
+        # bodies compile unwrapped so the capture's value equals its
+        # edition-1 value.
+        return atom if captured_reachable?(rule_name)
 
         wrap_uncaptured(atom)
       end
@@ -273,6 +277,57 @@ module Parsanol
         @document.glue_rules.include?(name.to_s)
       end
 
+      # The rules a text capture transitively spans (ruby#400): a
+      # capture on a reference captures the referenced rule's whole
+      # match, so every rule under a capture atom — the direct
+      # reference and everything its body transitively references —
+      # compiles unwrapped. Edition 2's Named leaves then attach only
+      # to non-captured structure, and a capture's value equals its
+      # edition-1 value (`{num: "123"}`, never a leaf array).
+      def captured_reachable?(name)
+        @capture_reachable ||= begin
+          seeds = []
+          @document.rules.each_value do |node|
+            collect_refs_under_captures(node, seeds)
+          end
+          reachable = []
+          until seeds.empty?
+            rule = seeds.pop
+            next if reachable.include?(rule) || !@document.rules.key?(rule)
+
+            reachable << rule
+            collect_all_refs(@document.rules[rule], seeds)
+          end
+          reachable
+        end
+        @capture_reachable.include?(name.to_s)
+      end
+
+      def collect_refs_under_captures(node, out)
+        walk = lambda do |n, in_capture|
+          case n.kind
+          when :ref then out << n.a if in_capture
+          when :cap then walk.call(n.b, true)
+          when :pred then walk.call(n.b, in_capture)
+          when :alt, :seq then n.a.each { |child| walk.call(child, in_capture) }
+          when :rep, :opt then walk.call(n.a, in_capture)
+          end
+        end
+        walk.call(node, false)
+      end
+
+      def collect_all_refs(node, out)
+        walk = lambda do |n|
+          case n.kind
+          when :ref then out << n.a
+          when :alt, :seq then n.a.each { |child| walk.call(child) }
+          when :cap, :pred then walk.call(n.b)
+          when :rep, :opt then walk.call(n.a)
+          end
+        end
+        walk.call(node)
+      end
+
       def wrap_uncaptured(atom)
         case atom
         when Atoms::Named
@@ -289,6 +344,7 @@ module Parsanol
           # agree by design.
           if trivia_only?(atom.rule_name.to_s, { atom.rule_name.to_s => true }) ||
               glue_rule?(atom.rule_name.to_s) ||
+              captured_reachable?(atom.rule_name.to_s) ||
               produces_capture?(atom.parslet)
             atom
           else
@@ -476,9 +532,12 @@ module Parsanol
         when Atoms::Named then nil
         when Atoms::Entity
           # Mirror wrap_uncaptured's verdicts: trivia glue, declared
-          # glue, and capture-producing refs all stay bare.
+          # glue, capture-reachable refs (ruby#400 — a capture spans
+          # them, so modern mode never names them), and
+          # capture-producing refs all stay bare.
           unless trivia_only?(atom.rule_name.to_s, { atom.rule_name.to_s => true }) ||
               glue_rule?(atom.rule_name.to_s) ||
+              captured_reachable?(atom.rule_name.to_s) ||
               produces_capture?(atom.parslet)
             lines << "rule #{site}: reference to #{atom.rule_name} produces no capture -- " \
                      "modern mode would name it #{atom.rule_name}"

data/lib/parsanol/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Parsanol
-  VERSION = "1.3.120"
+  VERSION = "1.3.121"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: parsanol
 version: !ruby/object:Gem::Version
-  version: 1.3.120
+  version: 1.3.121
 platform: ruby
 authors:
 - Ribose Inc.