disarm 0.15.0-x86_64-linux → 0.16.0-x86_64-linux

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: abc080c897b04a4bc4da77201766d150a429e26c7768ddca4c752b4e5b5c2c0c
-  data.tar.gz: 4b145b7f0184fbe3617c921f5a3ed222f712ba89a499c8352f8c404ca7279c3b
+  metadata.gz: 6721d5de6e559d2ca365c66853ced362a1df2e8b26c46b0e1b2c070441984081
+  data.tar.gz: 02e5b3546142c90f96d773ca4e859c06eb7187d562da242e64709d335d0faf1e
 SHA512:
-  metadata.gz: c3799a0aad40b3c3d5b26bf0017ea43ea4f05ec88c291d22c84674dfca84ec9941590d27155a0ec19bd9443c57a6c39ce821682cdca5142150b64b3db9e39fd9
-  data.tar.gz: 1a3f8dd7fc0c8c4b9a41c427ec854d91d46b4733626716d7e5e96e710a6b517202a061acec88c6f452180870445085c5b8be2570ebbf789919589dccf623de0d
+  metadata.gz: 4cd955deabb3aa23d54ff62e96e2c0fa7f7dfd480873d3f3de11b6eac5b11da472fd59da6ea1cd7933c27efd6e06a39e7d300eb6eda0ccc1915b87113126b42e
+  data.tar.gz: 480e07dbf124ea56c1d37bcdcc227317233544d7809e8b87fd647c05a3ad5cf59e6a0bca84dc4d2796a9d7bdde48b77e586c5199d181c3ce9b07c17bfe165f9a

data/lib/disarm/version.rb

@@ -2,5 +2,5 @@
 
 module Disarm
   # Kept in lockstep with the Rust crate / Python package version.
-  VERSION = "0.15.0"
+  VERSION = "0.16.0"
 end

data/lib/disarm.rb

@@ -116,11 +116,26 @@ module Disarm
       translate_errors { _demojize(text, strip_modifiers) }
     end
 
+    # Replace every emoji with +replacement+, verbatim.
+    #
+    # The counterpart to +demojize+, and a different question of a different table.
+    # +demojize+ asks what CLDR calls a character, so its domain is the name table, which
+    # is wider than the emoji: <tt>demojize("x\u2122y")</tt> is "x trade mark y". This
+    # asks whether the UCD calls it an emoji — Emoji_Presentation=Yes, an Emoji=Yes base
+    # carrying U+FE0F, and the ZWJ, modifier, keycap and flag sequences on those. Nothing
+    # else moves.
+    #
+    # +replacement+ is inserted exactly as given: "" closes an intra-word split and " "
+    # keeps two words apart, and no rule serves both.
+    def replace_emoji(text, replacement = "")
+      translate_errors { _replace_emoji(text, replacement) }
+    end
+
     # Canonicalize, but raise rather than silently normalize a structural difference
     # away — the half of the pair that lets a caller reject input instead of comparing
     # a value the sender never wrote.
-    def canonicalize_strict(text)
-      translate_errors { _canonicalize_strict(text) }
+    def canonicalize_strict(text, digit_policy: :numeric)
+      translate_errors { _canonicalize_strict(text, digit_policy.to_s) }
     end
 
     # Strip the non-interchange and invisible classes while KEEPING the script.
@@ -136,8 +151,8 @@ module Disarm
 
     # Remove obfuscation (zero-width, bidi, combining-mark abuse) while keeping
     # legible content.
-    def strip_obfuscation(text)
-      translate_errors { _strip_obfuscation(text) }
+    def strip_obfuscation(text, digit_policy: :numeric)
+      translate_errors { _strip_obfuscation(text, digit_policy.to_s) }
     end
 
     # Canonicalize text for security-sensitive comparison: strip obfuscation,
@@ -151,8 +166,8 @@ module Disarm
     # delimiter can leave here carrying one. #inspect_anomalies reports it as :confusable
     # WHEN the word also carries an ASCII letter, which is the gate that keeps ordinary
     # non-Latin text from firing; a delimiter-only string is not reported.
-    def canonicalize(text)
-      translate_errors { _canonicalize(text) }
+    def canonicalize(text, digit_policy: :numeric)
+      translate_errors { _canonicalize(text, digit_policy.to_s) }
     end
 
     # @deprecated Renamed to {#canonicalize} in 0.11 (the +_clean+ name
@@ -165,22 +180,60 @@ module Disarm
     # Case/accent/script-insensitive search lookup key. `lang:` applies a
     # language profile for transliteration (e.g. "ru", "uk"); nil means none.
     # Raises Disarm::InvalidArgument on an unknown lang.
-    def search_key(text, lang: nil)
-      translate_errors { _search_key(text, lang&.to_s) }
+    def search_key(text, lang: nil, digit_policy: :numeric)
+      translate_errors { _search_key(text, lang&.to_s, digit_policy.to_s) }
     end
 
     # Collation sort key (like #search_key, but keeps base accented characters
     # for correct ordering). `lang:` applies a language profile; nil means none.
     # Raises Disarm::InvalidArgument on an unknown lang.
-    def sort_key(text, lang: nil)
-      translate_errors { _sort_key(text, lang&.to_s) }
+    def sort_key(text, lang: nil, digit_policy: :numeric)
+      translate_errors { _sort_key(text, lang&.to_s, digit_policy.to_s) }
     end
 
     # Library catalog deduplication key (search_key plus confusable folding).
     # `lang:` applies a language profile; `strict_iso9:` selects the ISO 9:1995
     # Cyrillic scheme. Raises Disarm::InvalidArgument on an unknown lang.
-    def catalog_key(text, lang: nil, strict_iso9: false)
-      translate_errors { _catalog_key(text, lang&.to_s, strict_iso9) }
+    def catalog_key(text, lang: nil, strict_iso9: false, digit_policy: :numeric)
+      translate_errors { _catalog_key(text, lang&.to_s, strict_iso9, digit_policy.to_s) }
+    end
+
+    # The TR39 identifier skeleton plus the two prototype classes disarm keeps apart
+    # (#650). A spoof key: its only job is to make confusable identifiers collide, and its
+    # output is never for display. `digit_policy:` is `:numeric` (the letter half only),
+    # `:tr39` (adds `1 ≡ l` and `0 ≡ O`) or `:preserve` (a non-Latin numeral keeps its
+    # script).
+    def skeleton_key(text, digit_policy: :numeric)
+      translate_errors { _skeleton_key(text, digit_policy.to_s) }
+    end
+
+    # Levenshtein edit distance between `a` and `b`, in characters (#894). The one class
+    # of registry spoofing the confusable tables deliberately do not model — `paypa1`,
+    # `adm1n`. Canonicalize both sides first when composed and decomposed spellings
+    # should compare equal.
+    def edit_distance(left, right)
+      translate_errors { _edit_distance(left, right) }
+    end
+
+    # The candidate closest to `value`, as `{ value:, distance: }`, or nil beyond
+    # `max_distance:` (#894). Reports; it does not decide. An exact match is reported with
+    # distance 0, and ties go to the first candidate at the lowest distance.
+    def nearest_match(value, candidates, max_distance: 1)
+      hit = translate_errors { _nearest_match(value, candidates.map(&:to_s), max_distance) }
+      hit && { value: hit[0], distance: hit[1] }
+    end
+
+    # Whether `text` is already its own canonical form under `preset:` (#730).
+    #
+    # The verification-path counterpart to the presets: text in, normalized text out is
+    # the generation path, and this is the question a caller asks about bytes that arrive
+    # already bound. `has_anomalies?` is not this predicate — 142,760 assigned code points are
+    # reported clean by the detector and are not their own canonical form.
+    #
+    # `preset:` is any name in the preset registry or any profile. Raises
+    # Disarm::InvalidArgument on an unknown one.
+    def canonical?(text, preset: "canonicalize")
+      translate_errors { _is_canonical?(text, preset.to_s) }
     end
 
     # Strip diacritics ("café" → "cafe").
@@ -486,6 +539,22 @@ module Disarm
       translate_errors { _script_info(name.to_s) }
     end
 
+    # TR39 confusable sources whose prototype is in +script+, and how many of those
+    # disarm's bundled tables fold. Returns a Hash with +:script+, +:sources+ and
+    # +:folded+ keys.
+    #
+    # The denominator +unmapped_confusables+ does not have: that measures one bundled
+    # table against the whole 6,565-source population, so a script disarm ships no table
+    # for reports a number determined by that absence rather than by its coverage.
+    #
+    # +:folded+ counts sources any bundled table reaches, not sources folded *toward*
+    # this script — Greek is 71 of 159, because the Latin table folds Greek letters that
+    # look Latin. A script disarm knows that TR39 never uses as a prototype returns 0 of
+    # 0. Raises Disarm::InvalidArgument on an unknown script.
+    def confusable_coverage(script)
+      translate_errors { _confusable_coverage(script.to_s) }
+    end
+
     # The Unicode `confusables.txt` release the bundled confusable tables were folded
     # from, e.g. "17.0.0". Not a Unicode version for the library as a whole — the
     # case-folding and width tables track different releases (see docs/provenance.md).
@@ -574,7 +643,9 @@ module Disarm
     #   pipe.process("Café") # => "cafe"
     #   pipe.process("Köln") # reuse the same handle
     #
-    # Disarm::Pipeline#process is the Rust-defined instance method on the handle.
+    # Disarm::Pipeline#process is the Rust-defined instance method on the handle, and
+    # Disarm::Pipeline#with_digit_policy(policy) returns a copy whose confusable passes
+    # fold under `policy` (#646); a profile with no confusables step refuses one.
     def get_pipeline(profile)
       translate_errors { _get_pipeline(profile.to_s) }
     end
@@ -615,4 +686,17 @@ module Disarm
       raise Error, e.message, e.backtrace
     end
   end
+
+  # The reusable handle `Disarm.get_pipeline` returns. `#process` is Rust-defined; this
+  # reopens the class for the one method that can fail, so its error arrives as
+  # `Disarm::InvalidArgument` the way every module-level call's does.
+  class Pipeline
+    # A copy of this pipeline whose confusable passes fold under `policy` (#646):
+    # `:numeric`, `:tr39` or `:preserve`. Raises Disarm::InvalidArgument when the profile
+    # has no confusables step and the policy is not the default — a setting that would
+    # never run is refused rather than kept.
+    def with_digit_policy(policy)
+      Disarm.send(:translate_errors) { _with_digit_policy(policy.to_s) }
+    end
+  end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: disarm
 version: !ruby/object:Gem::Version
-  version: 0.15.0
+  version: 0.16.0
 platform: x86_64-linux
 authors:
 - Richard Quinn
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2026-09-01 00:00:00.000000000 Z
+date: 2026-09-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake