docs(search): pin the stem-substitution limit wildcards inherit

Stemming substitutes as well as truncates ("copy" and "copies" both index as
"copi" while "copyright" keeps its literal y), so a stemmed pattern reaches a
word's inflections but no longer reaches compounds that keep the surface
spelling. That trade is accepted: the same substitution is what makes company*
and library* work, and no rule over one normalized string separates them. So
usage.md stops claiming a trailing star just works, and a test pins copy* to the
base word rather than the compound. Also adds a parity test tying
stem_pattern_text to paperless_text_analyzer's own output, so a filter added to
the index analyzer alone cannot silently diverge, and corrects the docstring
claim that a run can analyze to several tokens - the raw tokenizer emits one
token whatever the input, so only the remove_long zero-token case can fire.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

docs/usage.md

@@ -927,10 +927,12 @@ title:Invoice*
 ```
 
 Wildcards are matched against the _stemmed_ terms stored in the index, not
-against the words as they appear in the document. A trailing `*` therefore
-works as expected (`invoice*` finds "invoice", "invoices" and "invoiced"), but
-a pattern whose text continues past where stemming cuts a word off cannot
-match: `productname` is indexed as `productnam`, so `produ*name` finds nothing.
+against the words as they appear in the document. A trailing `*` matches a word
+and its inflections (`invoice*` finds "invoice", "invoices" and "invoiced"),
+but not every longer word that starts with the same letters: `copy*` finds
+"copy" and "copies", not "copyright". For the same reason, a pattern whose text
+continues past where stemming cuts a word off cannot match at all:
+`productname` is indexed as `productnam`, so `produ*name` finds nothing.
 
 Matching natural date keywords:
 

src/documents/search/_tokenizer.py

@@ -131,11 +131,14 @@ def stem_pattern_text(text: str, language: str | None) -> str:
     """Stem an already lowercased/ascii-folded run the way index terms are.
 
     Returns text unchanged when stemming is disabled for language, and also
-    when the stem step does not yield exactly one token: remove_long drops an
-    over-long run entirely, and there is no single stem to substitute for a run
-    that analyzes to several. Falling back to the text as typed is the safe
-    direction for a pattern prefix, since it can only be as narrow as it was
-    before stemming was considered.
+    when the stem step does not yield exactly one token: remove_long drops a run
+    past the length limit, leaving no stem to substitute. Falling back to the
+    text as typed is the safe direction for a pattern prefix, since it can only
+    be as narrow as it was before stemming was considered.
+
+    The raw tokenizer emits one token whatever the input and the stemmer is
+    1-to-1, so only the zero-token case can fire today; the guard covers both
+    counts so a tokenizer change cannot turn this into an IndexError.
     """
     analyzer = _pattern_stemmer(language)
     if analyzer is None:

src/documents/tests/search/test_pattern_stemming.py

@@ -15,6 +15,9 @@ import pytest
 
 from documents.models import Document
 from documents.search._registry import _make_pattern_normalizer
+from documents.search._tokenizer import ascii_fold
+from documents.search._tokenizer import paperless_text_analyzer
+from documents.search._tokenizer import stem_pattern_text
 
 if TYPE_CHECKING:
     from documents.search._backend import TantivyBackend
@@ -23,7 +26,7 @@ pytestmark = [pytest.mark.search, pytest.mark.django_db]
 
 CONTENT = (
     "invoice total due for electricity from both companies, "
-    "payments made to the university library"
+    "payments made to the university library, copies attached"
 )
 
 
@@ -94,6 +97,51 @@ class TestPrefixStemming:
         is deliberate, not accidental."""
         assert _matched_ids(backend, "produ*name") == set()
 
+    def test_stem_substitution_loses_compounds_accepted_trade(
+        self,
+        backend: TantivyBackend,
+        indexed_doc: Document,
+    ) -> None:
+        """English stemming substitutes as well as truncates: "copy" and
+        "copies" both index as "copi", while "copyright" keeps its literal "y".
+        So "copy*" reaches the base word and its inflections but no longer
+        reaches the compound, which it did before patterns were stemmed. That
+        trade is deliberate: the same substitution is what makes "company*" and
+        "library*" work at all, and no rule over one normalized string tells the
+        two apart. Matching both would need the pattern to be emitted as a
+        disjunction of the folded and stemmed forms, which belongs in the
+        emitter, not here.
+        """
+        compound = Document.objects.create(
+            title="Copyright notice",
+            content="copyright notice for the work",
+            checksum="pattern-stemming-2",
+            archive_serial_number=901,
+        )
+        backend.add_or_update(compound)
+
+        assert _matched_ids(backend, "copy*") == {indexed_doc.id}
+        assert _matched_ids(backend, "copyright*") == {compound.id}
+
+
+class TestStemsMatchTheIndexAnalyzer:
+    """stem_pattern_text rebuilds paperless_text_analyzer's stemming tail rather
+    than sharing it, so a filter added to the index analyzer alone would silently
+    stop patterns from reaching the terms it produces.
+    """
+
+    @pytest.mark.parametrize(
+        "language",
+        ["en", "de", "fr", "es", "sv", None, "klingon"],
+    )
+    @pytest.mark.parametrize(
+        "word",
+        ["Copies", "copyright", "Companies", "Invoices", "laufen", "casas", "Straße"],
+    )
+    def test_stem_equals_the_index_term(self, word: str, language: str | None) -> None:
+        indexed = paperless_text_analyzer(language).analyze(word)[0]
+        assert stem_pattern_text(ascii_fold(word.lower()), language) == indexed
+
 
 class TestPatternNormalizer:
     @pytest.mark.parametrize(