relaton-cli 3.0.0 → 3.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 38a1d2c4a0c1ae1bd81c6fd588a3b1b83f98828e11620fe6b981e17223f093cf
-  data.tar.gz: 7cb684d5bf4d320606c643d53b27b63ba5354c89e73019b4455dd2942bd42c52
+  metadata.gz: 9e09dab74faf731f638fb7b31271e8dedd483dec7752b232ffe1d1ed82230bb4
+  data.tar.gz: d35d03f375a776575bfc86d58cf188b3da4f9120907562ceb9120b72083dfda3
 SHA512:
-  metadata.gz: 52517b0e9ccaa115aea67a37ff21af8163b0657a51710217b8f02e5a7f39cf51609b2582c90bfe0d5b02e040752a97672c8892d1a26c84ad95e083a8c9085bec
-  data.tar.gz: cb9bf490e6c3dc59c6ca6c8b096867dc51be43b91eaeee8759ccb5684cb9be02715477f211e4a554adda9c5661ca26ff7d2b64ef54e00ecd0f59394c031c5a63
+  metadata.gz: f53553f7d991fb26fe8679ce9c5f898d8d33fa836cf7224e12ffaa1b831d0da27693916f9d99d3aab365f4698900e29c7b356c73bf7c497ffe3c453534ea000c
+  data.tar.gz: 5f9ddb547964442309a65aea8ba3b28959693ea4c373bb1594cdeafcae9bcc0c28957b68516613dbff917bd0aa7b5208adfacb6def16597e340352c536753fec

data/Gemfile

@@ -6,6 +6,17 @@ gemspec
 # The single combined `relaton` gem lives at the repo root.
 gem "relaton", path: "../.."
 
+# parsanol 1.3.111 is the trivia-exempt release (parsanol#355): the
+# emission-2 artifacts (pubid#614) parse parameter-shaped under it, so
+# the pin rides it; the root gemspec floor carries the same story.
+
+# The pubid 2.0.0 FINAL gem still ships emission-1 artifacts - under
+# 1.3.111 its parses break exactly as parsanol#352 describes. The
+# emission-2 set ships from 2.0.0.pre.alpha.64, so the pin names it;
+# without it the resolver picks the final and every structured-corpus
+# docid with a part+date fails ("Unknown parameter type: colon").
+gem "pubid", "2.0.0.pre.alpha.64"
+
 # https://github.com/rails/rails/issues/55886
 # add openssl as an explicit gem to fix SSL verification issues in GHA.
 # gem "openssl"
@@ -33,3 +44,4 @@ gem "webmock"
 # (leptris#1742), pinned to match the relaton root bundle — the battery
 # re-run accompanies this bump.
 gem "leptris", "1.9.370.1"
+

data/lib/relaton/cli/index_site_generator.rb

@@ -127,7 +127,14 @@ module Relaton
         # per row against 0.44 KB for the derived values alone — 543 MB vs 76 MB
         # on a 177k-row corpus. Anything else derived from pubid must therefore
         # be computed in `add` too; by write time the object is gone.
-        Row = Struct.new(:rendered, :file, :key, :id_hash)
+        #
+        # `derived` marks an alias row (this document's other spellings, or a
+        # committed fetcher alias) as opposed to the document's own primary
+        # id — `indexed_rows` prunes a derived row whose render a document
+        # owns as its primary, so an alias can never repoint a spelling at
+        # the wrong file (`Type#add_or_update` keys on the rendered string
+        # and overwrites).
+        Row = Struct.new(:rendered, :file, :key, :id_hash, :derived)
 
         attr_reader :rows
 
@@ -144,7 +151,24 @@ module Relaton
           @rows = []
         end
 
-        def add(rendered, file)
+        # Committed rows an id-sharing file already covers: the fetcher's
+        # aliases (a record-less registry's id pointing at the record-bearing
+        # document's file -- relaton-iana#8, #9). Carried as rows of their
+        # own so either spelling resolves to the one document.
+        def add_alias_rows(rows)
+          rows.each do |id_hash, file|
+            pubid = from_hash(id_hash)
+            next unless pubid
+
+            row = Row.new(pubid.to_s, file)
+            row.id_hash = id_hash
+            row.key = key_string(pubid)
+            row.derived = true
+            @rows << row
+          end
+        end
+
+        def add(rendered, file, alias_ids = nil)
           row = Row.new(rendered, file)
           # Precompute the expensive derived values (root.number walk,
           # lutaml to_hash) once during the streaming pass — computing
@@ -163,6 +187,43 @@ module Relaton
             row.key = key_string(pubid)
           end
           @rows << row
+          add_rendered_alias_rows(pubid, hash, rendered, file, alias_ids)
+        end
+
+        # The document's other spellings as rows of their own, so every id
+        # the dataset prints resolves: the parse's canonical render of the
+        # primary ("ISO/IEC DIR 1" from the file's own
+        # "ISO/IEC Directives Part 1") and each further docidentifier's
+        # render. All point at the same file. A docid that does not parse
+        # is skipped, and one that renders as an already-added spelling of
+        # this document adds no second row. Cross-document collisions are
+        # not decided here — `indexed_rows` prunes a derived row whose
+        # spelling a document owns as its primary.
+        def add_rendered_alias_rows(pubid, primary_hash, rendered, file, alias_ids)
+          seen = { rendered => true }
+          canon = pubid&.to_s
+          if canon && !canon.empty? && !seen.key?(canon)
+            seen[canon] = true
+            push_derived_row(canon, file, primary_hash, key_string(pubid))
+          end
+          Array(alias_ids).each do |content|
+            alias_pubid = parse(content.to_s)
+            next unless alias_pubid
+
+            render = alias_pubid.to_s
+            next if render.empty? || seen.key?(render)
+
+            seen[render] = true
+            push_derived_row(render, file, alias_pubid.to_hash, key_string(alias_pubid))
+          end
+        end
+
+        def push_derived_row(rendered, file, id_hash, key)
+          row = Row.new(rendered, file)
+          row.id_hash = id_hash
+          row.key = key
+          row.derived = true
+          @rows << row
         end
 
         def count
@@ -223,12 +284,36 @@ module Relaton
         # `Relaton::Ieee::DataFetcher#build_index` already skips the same rows
         # for the same reason and logs the loss; this matches it.
         #
+        # A derived (alias) row is dropped when a document owns the spelling
+        # as its own primary id, or when an earlier alias row claimed it:
+        # `Type#add_or_update` keys id_lookup on the rendered string, so
+        # letting both publish would silently repoint the spelling at
+        # whichever file sorted last. The document's primary id — and, when
+        # two derived rows survive, the earlier document — wins.
+        #
         # Memoized, and the single place the sort happens — `each_shard` and
         # `write_monolith` previously sorted the corpus independently, so both
         # ran on every build.
         def indexed_rows
-          @indexed_rows ||= @rows.reject { |row| row.id_hash.nil? }
-                                 .sort_by { |row| [row.key, row.rendered] }
+          @indexed_rows ||= begin
+            owned = {}
+            @rows.each { |row| owned[row.rendered] = true unless row.derived }
+            claimed = {}
+            @rows.reject { |row| row.id_hash.nil? }
+                 .reject do |row|
+                   next false unless row.derived
+
+                   next true if owned.key?(row.rendered)
+
+                   if claimed.key?(row.rendered)
+                     true
+                   else
+                     claimed[row.rendered] = true
+                     false
+                   end
+                 end
+                 .sort_by { |row| [row.key, row.rendered] }
+          end
         end
 
         # Rows dropped by `indexed_rows` — reported so the loss is never silent.
@@ -504,12 +589,34 @@ module Relaton
       # the same name carries plain strings, and one of those written into a
       # row would break both the monolith (`yaml_nested` walks a Hash) and the
       # consumer (`FileIO#deserialize_id` calls `from_hash`).
+      #
+      # The first id per file becomes the map; ids beyond the first are the
+      # fetcher's aliases onto that file (see .committed_alias_rows).
       def read_committed_index(path)
         rows = YAML.safe_load(File.read(path), permitted_classes: [Symbol])
-        Array(rows).each_with_object({}) do |row, acc|
-          next unless row.is_a?(Hash) && row[:id].is_a?(Hash)
+        @committed_rows = Array(rows).select do |row|
+          row.is_a?(Hash) && row[:id].is_a?(Hash) && row[:file]
+        end
+        @committed_rows.each_with_object({}) do |row, acc|
+          acc[row[:file]] ||= row[:id]
+        end
+      end
 
-          acc[row[:file]] = row[:id]
+      # `{ id hash => file }` for committed rows beyond each file's first --
+      # the fetcher's alias ids, which no data file of their own carries.
+      # Without this pass the machine index drops them (the file-keyed
+      # fallback resolves one id per file), and a spelling the corpus
+      # aliased stops resolving.
+      def committed_alias_rows
+        committed_index # force the load and @committed_rows
+        seen = {}
+        (@committed_rows || []).filter_map do |row|
+          if seen[row[:file]]
+            [row[:id], row[:file]]
+          else
+            seen[row[:file]] = true
+            nil
+          end
         end
       end
 
@@ -567,7 +674,8 @@ module Relaton
       # The machine-index input: rendered primary docid + repo-relative
       # path (clients prefix their own baseurl onto file).
       def machine_record(item)
-        [item["id"], item["yaml_path"]]
+        [item["id"], item["yaml_path"],
+         Array(item["docids"]).filter_map { |d| d["id"] if d.is_a?(Hash) }]
       end
 
       def write_machine_index_manifest(total)
@@ -659,6 +767,7 @@ module Relaton
           detail << detail_record(doc) if emit_detail?
           machine&.add(*machine_record(doc))
         end
+        machine&.add_alias_rows(committed_alias_rows)
         summary.flush
         detail.flush
         @machine_index = machine

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: relaton-cli
 version: !ruby/object:Gem::Version
-  version: 3.0.0
+  version: 3.0.1
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-10-27 00:00:00.000000000 Z
+date: 2026-10-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: liquid
@@ -44,14 +44,14 @@ dependencies:
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 3.0.0
+        version: 3.0.1
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: 3.0.0
+        version: 3.0.1
 - !ruby/object:Gem::Dependency
   name: relaton-render
   requirement: !ruby/object:Gem::Requirement