From 75a5b5e4dec8695bbaac1c2ae8e1c2298b583c74 Mon Sep 17 00:00:00 2001
From: upstream <upstream@localhost>
Date: Wed, 14 Oct 2026 18:48:44 +0200
Subject: [PATCH] Upstream spencermountain/compromise state of the touched
 files before ff559ff7cd

---
 src/1-one/sweep/methods/tagger/index.js | 58 +++++++++++++++++++++++++
 tests/03-three/misc/sweep-tag.test.js   | 40 +++++++++++++++++
 2 files changed, 98 insertions(+)
 create mode 100644 src/1-one/sweep/methods/tagger/index.js
 create mode 100644 tests/03-three/misc/sweep-tag.test.js

diff --git a/src/1-one/sweep/methods/tagger/index.js b/src/1-one/sweep/methods/tagger/index.js
new file mode 100644
index 0000000..7f79c93
--- /dev/null
+++ b/src/1-one/sweep/methods/tagger/index.js
@@ -0,0 +1,58 @@
+import canBe from './canBe.js'
+
+const tagger = function (list, document, world) {
+  const { model, methods } = world
+  const { getDoc, setTag, unTag } = methods.one
+  const looksPlural = methods.two.looksPlural
+  if (list.length === 0) {
+    return list
+  }
+  return list.map(todo => {
+    if (!todo.tag && !todo.chunk && !todo.unTag) {
+      return
+    }
+    const reason = todo.reason || todo.match
+    const terms = getDoc([todo.pointer], document)[0]
+    let validated
+    // handle 'safe' tag
+    if (todo.safe === true) {
+      // check for conflicting tags
+      if (canBe(terms, todo.tag, model) === false) {
+        return
+      }
+      // dont tag half of a hyphenated word
+      if (terms[terms.length - 1].post === '-') {
+        return
+      }
+      // Only an exact single tag was checked by canBe; arrays and shorthand
+      // still need their individual checks in setTag.
+      if (typeof todo.tag === 'string' && Object.hasOwn(model.one.tagSet, todo.tag)) {
+        validated = model.one.tagSet[todo.tag]
+      }
+    }
+    if (todo.tag !== undefined) {
+      setTag(terms, todo.tag, world, todo.safe, `[post] '${reason}'`, validated)
+      // quick and dirty plural tagger 😕
+      if (todo.tag === 'Noun' && looksPlural) {
+        const term = terms[terms.length - 1]
+        if (looksPlural(term.text)) {
+          setTag([term], 'Plural', world, todo.safe, 'quick-plural')
+        } else {
+          setTag([term], 'Singular', world, todo.safe, 'quick-singular')
+        }
+      }
+      // allow freezing this match, too
+      if (todo.freeze === true) {
+        terms.forEach(term => (term.frozen = true))
+      }
+    }
+    if (todo.unTag !== undefined) {
+      unTag(terms, todo.unTag, model.one.tagSet, reason)
+    }
+    // allow setting chunks, too
+    if (todo.chunk) {
+      terms.forEach(t => (t.chunk = todo.chunk))
+    }
+  })
+}
+export default tagger
diff --git a/tests/03-three/misc/sweep-tag.test.js b/tests/03-three/misc/sweep-tag.test.js
new file mode 100644
index 0000000..b8235d8
--- /dev/null
+++ b/tests/03-three/misc/sweep-tag.test.js
@@ -0,0 +1,40 @@
+import test from 'tape'
+import nlp from '../../lib/three.js'
+const here = '[three/misc/sweep-tag] '
+
+test('cache-one:', function (t) {
+  const matches = [{ match: '(he|she|they|#Noun) (has|have) (a|an)' }]
+  const net = nlp.buildNet(matches)
+  const doc = nlp(`you have a appointment`)
+  t.equal(doc.has(net), true, here + 'Noun cache')
+  t.end()
+})
+
+test('sweep-root:', function (t) {
+  const matches = [{ match: '{appointment}' }]
+  const net = nlp.buildNet(matches)
+  const doc = nlp(`you have some appointments`)
+  doc.compute('root')
+  t.equal(doc.has(net), true, here + 'root cache')
+  t.end()
+})
+
+test('sweep-freeze:', function (t) {
+  const matches = [
+    { match: 'juicy fruit', tag: 'Singular', freeze: true },
+    { match: 'front steps', tag: 'Plural', freeze: true },
+    { match: 'juicy', tag: 'Adjective' },
+    { match: 'front', tag: 'Adjective' },
+  ]
+  const doc = nlp(`i ate juicy fruit on the front steps`)
+  const net = nlp.buildNet(matches)
+  doc.sweep(net)
+  let m = doc.match('juicy fruit')
+  t.equal(m.has('#Singular #Singular'), true, here + 'juicy fruit')
+  t.equal(m.has('#Adjective'), false, here + 'juicy fruit frozen')
+
+  m = doc.match('front steps')
+  t.equal(m.has('#Plural #Plural'), true, here + 'front steps')
+  t.equal(m.has('#Adjective'), false, here + 'front steps frozen')
+  t.end()
+})