fix(verify): use source chipset year for late SoC placeholders

app/verify/common.py

@@ -10,6 +10,7 @@
 
 import hashlib
 import json
+import re
 import sys
 from collections.abc import Iterable
 from pathlib import Path
@@ -30,6 +31,7 @@
 )
 
 VERIFY_DIR = DATA_DIR / "_verify"
+_RAW_CHIPSET_YEAR = re.compile(r"^[^,]+,\s*((?:19|20)\d{2})\s*,")
 LEDGER_PATH = VERIFY_DIR / "ledger.jsonl"  # git-tracked: promotion decisions only
 STATE_DIR = VERIFY_DIR / "state"  # gitignored caches
 SCORES_PATH = STATE_DIR / "scores.jsonl"  # full Tier 0 results (cheap to recompute)
@@ -95,14 +97,22 @@ def foreign_key_sets(
     """Build FK lookups the way ``app.validate`` does, plus a SoC release-date map.
 
     Returns ``(brand_slugs, soc_slugs, soc_release_date)`` where ``soc_release_date``
-    maps a SoC slug to its ISO release date (used for "chip can't postdate device").
+    maps a SoC slug to its best available ISO release date (used for "chip can't
+    postdate device"). A structured year in ``raw_chipset`` takes precedence
+    when the normalized January 1 placeholder is later. The raw import often
+    records the chip's year while the normalized date is several years late.
     """
     brand_slugs = {r.slug for r in records.get("brand", []) if r.slug}
     soc_slugs = {r.slug for r in records.get("soc", []) if r.slug}
     soc_release: dict[str, str] = {}
     for r in records.get("soc", []):
         rd = r.data.get("release_date")
         if r.slug and isinstance(rd, str):
+            raw = r.data.get("raw_chipset")
+            if rd.endswith("-01-01") and isinstance(raw, str):
+                match = _RAW_CHIPSET_YEAR.search(raw)
+                if match and match.group(1) < rd[:4]:
+                    rd = f"{match.group(1)}-01-01"
             soc_release[r.slug] = rd
     return brand_slugs, soc_slugs, soc_release
 

tests/verify/test_common_soc_dates.py

@@ -0,0 +1,44 @@
+"""Use a source chip year when a normalized SoC date is a later placeholder."""
+
+from app.verify import offline
+from app.verify.common import Record, foreign_key_sets
+
+
+def _release_for(release_date: str, raw_chipset: str | None) -> str:
+    soc = Record("soc", "soc/chip-x.json", {
+        "slug": "chip-x", "release_date": release_date, "raw_chipset": raw_chipset,
+    })
+    return foreign_key_sets({"soc": [soc]})[2]["chip-x"]
+
+
+def test_earlier_structured_chip_year_corrects_late_placeholder():
+    assert _release_for(
+        "2025-01-01", "Qualcomm Snapdragon 680 SM6225, 2021, 64 bit, octa-core"
+    ) == "2021-01-01"
+
+    phone = Record("smartphone", "smartphone/phone-x.json", {
+        "slug": "phone-x", "soc": "chip-x", "release_date": "2022-08-01",
+        "source_urls": ["https://www.qualcomm.com/products/mobile/snapdragon"],
+    })
+    before = offline.score_record(phone, 2026, {"chip-x": "2025-01-01"})
+    after = offline.score_record(phone, 2026, {"chip-x": "2021-01-01"})
+    assert "soc_not_after_device" in before.flags
+    assert "soc_not_after_device" not in after.flags
+
+
+def test_precise_or_unstructured_dates_keep_the_normalized_date():
+    assert _release_for("2024-05-07", "Chip X, 2021, 8 cores") == "2024-05-07"
+    assert _release_for("2025-01-01", "Chip X with 2021 revision") == "2025-01-01"
+    assert _release_for("2025-01-01", None) == "2025-01-01"
+    assert _release_for("2025-01-01", "Chip X, 2026, 8 cores") == "2025-01-01"
+
+
+def test_raw_chip_year_still_flags_a_genuine_era_mismatch():
+    release = _release_for(
+        "2024-01-01", "Qualcomm Snapdragon 8s Gen 3 SM8635, 2024, 8 cores"
+    )
+    phone = Record("smartphone", "smartphone/phone-x.json", {
+        "slug": "phone-x", "soc": "chip-x", "release_date": "2015-01-09",
+    })
+    score = offline.score_record(phone, 2026, {"chip-x": release})
+    assert "soc_not_after_device" in score.flags