Queue and progress treat the statically linked D3DX9 block and byte-matched library functions as library

harbor/queue.py

@@ -112,6 +112,18 @@ class FunctionQueue:
         return frozenset(int(m["address"], 16) for c in data.get("clusters", []) for m in c.get("members", [])
                          if m.get("best_readable") and m.get("status") != "readable")
 
+    def library_ranges(self):
+        """[(start, end)] of statically linked library blocks (the [reverse] library_ranges csv)."""
+        path = self.config.get("library_ranges")
+        if not path or not Path(path).exists():
+            return []
+        with open(path, encoding="utf-8") as f:
+            return [(int(r["start"], 16), int(r["end"], 16)) for r in csv.DictReader(f)
+                    if r.get("verdict", "").startswith("D3DX")]
+
+    def is_library(self, addr, library, ranges):
+        return addr in library or any(lo <= addr < hi for lo, hi in ranges)
+
     # ------------------------------------------------------------------ queue
     def score(self, addr, facts, named, rejected, similar_readable=frozenset()):
         d = facts[addr]
@@ -145,11 +157,12 @@ class FunctionQueue:
         statuses, deferred, rejected = self.statuses(), self.deferrals(), self.rejections()
         facts, named = self.r._facts(), self._named()
         library = self._results("library_results")
+        ranges = self.library_ranges()
         ranked = []
         for addr in facts:
             if not any(lo <= addr <= hi for lo, hi in bands) or any(lo <= addr <= hi for lo, hi in taken):
                 continue
-            if statuses.get(addr, ("raw",))[0] == "readable" or addr in deferred or addr in library:
+            if statuses.get(addr, ("raw",))[0] == "readable" or addr in deferred or self.is_library(addr, library, ranges):
                 continue
             if rejected.get(addr, 0) >= self.cap:
                 continue
@@ -181,12 +194,13 @@ class FunctionQueue:
         exact = {a for a, row in self._results("match_results").items() if str(row.get("status")) == "match"}
         verified = {a for a, row in self._results("diffrun_results").items() if str(row.get("verdict")) == "same"}
         library = set(self._results("library_results"))
+        ranges = self.library_ranges()
         tiers = {t: {"functions": 0, "bytes": 0} for t in TIER_ORDER + ("library", "engine")}
         for addr, d in facts.items():
             size = int(d.get("size") or 0)
             category = self._category(addr)
             readable = statuses.get(addr, ("raw",))[0] == "readable"
-            if addr in library or category in ("library", "platform"):
+            if self.is_library(addr, library, ranges) or category in ("library", "platform", "crt"):
                 tier = "library"
             elif category == "engine" and not readable:
                 tier = "engine"