Replace watchlist match reasons with MatchReason enum (#60)

Author: henry0816191

src/paperscout/models.py

@@ -159,6 +159,13 @@ class Tier(str, Enum):
     COLD = "cold"
 
 
+class MatchReason(str, Enum):
+    """Why a watchlist entry matched a paper or probe hit."""
+
+    AUTHOR = "author"
+    PAPER = "paper"
+
+
 @dataclass(slots=True)
 class ProbeHit:
     """Successful HEAD to an unpublished draft URL plus optional excerpt text."""
@@ -211,7 +218,7 @@ def __post_init__(self) -> None:
 
 @dataclass
 class PerUserMatches:
-    """One user's watchlist hits: ``(paper|hit, 'author'|'paper')`` tuples."""
+    """One user's watchlist hits: ``(paper|hit, MatchReason)`` tuples."""
 
-    papers: list[tuple[Paper, str]] = field(default_factory=list)
-    probe_hits: list[tuple[ProbeHit, str]] = field(default_factory=list)
+    papers: list[tuple[Paper, MatchReason]] = field(default_factory=list)
+    probe_hits: list[tuple[ProbeHit, MatchReason]] = field(default_factory=list)

src/paperscout/scout.py

@@ -521,15 +521,15 @@ def notify_users(app: App, result: PollResult, mq: MessageQueue) -> None:
             lines.append("*:rotating_light: Papers matching your watchlist:*")
             for paper, reason in matches.papers:
                 p_link = _paper_link(paper)
-                tag = f"[{reason} match]"
+                tag = f"[{reason.value} match]"
                 lines.append(f"• {p_link} — {paper.title} (by *{paper.author}*) {tag}")
 
         if matches.probe_hits:
             lines.append("*:rotating_light: New drafts matching your watchlist:*")
             for hit, reason in matches.probe_hits:
                 h_link = _hit_label(hit.url, hit.prefix, hit.number, hit.revision, hit.extension)
                 lm = _fmt_lm(hit.last_modified)
-                tag = f"[{reason} match]"
+                tag = f"[{reason.value} match]"
                 lines.append(f"• {h_link} — {lm} {tag}")
 
         if not lines:

src/paperscout/storage.py

@@ -10,7 +10,7 @@
 from contextlib import contextmanager
 from typing import TYPE_CHECKING, Any, cast
 
-from .models import PerUserMatches
+from .models import MatchReason, PerUserMatches
 
 if TYPE_CHECKING:
     from psycopg2.pool import ThreadedConnectionPool
@@ -371,29 +371,29 @@ def matches_for_users(
             authors = user_authors.get(uid, [])
             paper_nums = user_papers.get(uid, set())
 
-            matched_papers: list[tuple[Paper, str]] = []
+            matched_papers: list[tuple[Paper, MatchReason]] = []
             for paper in new_papers:
                 # Author match
                 if authors and paper.author:
                     author_lower = paper.author.lower()
                     if any(a in author_lower for a in authors):
-                        matched_papers.append((paper, "author"))
+                        matched_papers.append((paper, MatchReason.AUTHOR))
                         continue
                 # Paper-number match
                 if paper_nums and paper.number is not None and paper.number in paper_nums:
-                    matched_papers.append((paper, "paper"))
+                    matched_papers.append((paper, MatchReason.PAPER))
 
-            matched_hits: list[tuple[ProbeHit, str]] = []
+            matched_hits: list[tuple[ProbeHit, MatchReason]] = []
             for hit in probe_hits:
                 # Author match via front_text
                 if authors and hit.front_text:
                     text_lower = hit.front_text.lower()
                     if any(a in text_lower for a in authors):
-                        matched_hits.append((hit, "author"))
+                        matched_hits.append((hit, MatchReason.AUTHOR))
                         continue
                 # Paper-number match via probe hit number
                 if paper_nums and hit.number in paper_nums:
-                    matched_hits.append((hit, "paper"))
+                    matched_hits.append((hit, MatchReason.PAPER))
 
             if matched_papers or matched_hits:
                 result[uid] = PerUserMatches(

tests/test_monitor.py

@@ -12,7 +12,7 @@
 import pytest
 
 from paperscout.errors import ConfigurationError
-from paperscout.models import CycleResult, CycleStatus, Paper, PerUserMatches, ProbeHit
+from paperscout.models import CycleResult, CycleStatus, MatchReason, Paper, PerUserMatches, ProbeHit
 from paperscout.monitor import (
     DiffResult,
     PollResult,
@@ -189,7 +189,7 @@ def test_explicit_dp_transitions(self):
     def test_explicit_per_user_matches(self):
         diff = DiffResult(new_papers=[], updated_papers=[])
         paper = Paper(id="P2300R11")
-        pum = PerUserMatches(papers=[(paper, "author")], probe_hits=[])
+        pum = PerUserMatches(papers=[(paper, MatchReason.AUTHOR)], probe_hits=[])
         result = PollResult(diff=diff, probe_hits=[], per_user_matches={"U1": pum})
         assert "U1" in result.per_user_matches
 
@@ -358,7 +358,7 @@ async def test_poll_once_populates_per_user_matches(self, fake_pool):
         prober.run_cycle = AsyncMock(return_value=_empty_cycle())
 
         user_watchlist.matches_for_users.return_value = {
-            "U123": PerUserMatches(papers=[(new_paper, "author")], probe_hits=[])
+            "U123": PerUserMatches(papers=[(new_paper, MatchReason.AUTHOR)], probe_hits=[])
         }
         result = await scheduler.poll_once()
         assert "U123" in result.per_user_matches
@@ -373,7 +373,7 @@ async def test_poll_once_per_user_probe_hit(self, fake_pool):
         index.papers = {}
 
         user_watchlist.matches_for_users.return_value = {
-            "U123": PerUserMatches(papers=[], probe_hits=[(hit, "author")])
+            "U123": PerUserMatches(papers=[], probe_hits=[(hit, MatchReason.AUTHOR)])
         }
         result = await scheduler.poll_once()
         assert "U123" in result.per_user_matches
@@ -403,7 +403,7 @@ async def test_restart_with_prior_poll_notifies_seed_hits(self, fake_pool):
         hit = _recent_hit()
         prober.run_cycle = AsyncMock(return_value=_success_cycle([hit]))
         user_watchlist.matches_for_users.return_value = {
-            "U123": PerUserMatches(papers=[], probe_hits=[(hit, "author")])
+            "U123": PerUserMatches(papers=[], probe_hits=[(hit, MatchReason.AUTHOR)])
         }
         result = await scheduler.poll_once()
         assert len(notified) == 1
@@ -418,7 +418,7 @@ async def test_restart_with_discovered_urls_notifies(self, fake_pool):
         hit = _recent_hit()
         prober.run_cycle = AsyncMock(return_value=_success_cycle([hit]))
         user_watchlist.matches_for_users.return_value = {
-            "U123": PerUserMatches(papers=[], probe_hits=[(hit, "author")])
+            "U123": PerUserMatches(papers=[], probe_hits=[(hit, MatchReason.AUTHOR)])
         }
         result = await scheduler.poll_once()
         assert len(notified) == 1

tests/test_scout.py

@@ -5,7 +5,7 @@
 from datetime import datetime, timedelta, timezone
 from unittest.mock import MagicMock, patch
 
-from paperscout.models import Paper, PerUserMatches, ProbeHit
+from paperscout.models import MatchReason, Paper, PerUserMatches, ProbeHit
 from paperscout.monitor import DiffResult, DPTransition, PollResult
 from paperscout.scout import (
     _batch_lines,
@@ -278,7 +278,7 @@ def test_author_match_sends_dm(self):
         paper = Paper(
             id="P2300R11", title="Senders", author="Eric Niebler", url="https://wg21.link/P2300R11"
         )
-        pum = PerUserMatches(papers=[(paper, "author")], probe_hits=[])
+        pum = PerUserMatches(papers=[(paper, MatchReason.AUTHOR)], probe_hits=[])
         result = _make_result(per_user_matches={"U123": pum})
         notify_users(app, result, mq)
         mq.enqueue.assert_called_once()
@@ -291,7 +291,7 @@ def test_paper_match_sends_dm(self):
         app = MagicMock()
         mq = MagicMock()
         paper = Paper(id="P2300R11", title="X", author="Someone", url="https://wg21.link/P2300R11")
-        pum = PerUserMatches(papers=[(paper, "paper")], probe_hits=[])
+        pum = PerUserMatches(papers=[(paper, MatchReason.PAPER)], probe_hits=[])
         result = _make_result(per_user_matches={"U456": pum})
         notify_users(app, result, mq)
         channel, text = mq.enqueue.call_args[0]
@@ -302,7 +302,7 @@ def test_probe_hit_match_sends_dm(self):
         app = MagicMock()
         mq = MagicMock()
         hit = _recent_hit()
-        pum = PerUserMatches(papers=[], probe_hits=[(hit, "author")])
+        pum = PerUserMatches(papers=[], probe_hits=[(hit, MatchReason.AUTHOR)])
         result = _make_result(per_user_matches={"U789": pum})
         notify_users(app, result, mq)
         mq.enqueue.assert_called_once()
@@ -313,7 +313,7 @@ def test_multiple_users_get_separate_dms(self):
         app = MagicMock()
         mq = MagicMock()
         paper = Paper(id="P2300R11", title="X", author="Niebler")
-        pum = PerUserMatches(papers=[(paper, "author")], probe_hits=[])
+        pum = PerUserMatches(papers=[(paper, MatchReason.AUTHOR)], probe_hits=[])
         result = _make_result(per_user_matches={"U1": pum, "U2": pum})
         notify_users(app, result, mq)
         assert mq.enqueue.call_count == 2

tests/test_storage.py

@@ -8,7 +8,7 @@
 
 import pytest
 
-from paperscout.models import Paper
+from paperscout.models import MatchReason, Paper
 from paperscout.storage import (
     PaperCache,
     ProbeState,
@@ -309,6 +309,8 @@ def test_matches_for_users_author_match(self, fake_pool):
         assert "U1" in result
         matched_papers = [p for p, _ in result["U1"].papers]
         assert paper in matched_papers
+        _, reason = result["U1"].papers[0]
+        assert reason is MatchReason.AUTHOR
 
     def test_matches_for_users_paper_match(self, fake_pool):
         wl = UserWatchlist(fake_pool)
@@ -406,7 +408,7 @@ def test_matches_skips_bad_paper_row_author_match_still_works(self, fake_pool):
         result = wl.matches_for_users([paper], [])
         assert "U1" in result
         reasons = [r for _, r in result["U1"].papers]
-        assert "author" in reasons
+        assert MatchReason.AUTHOR in reasons
 
     def test_matches_paper_with_none_number_never_paper_matched(self, fake_pool):
         wl = UserWatchlist(fake_pool)