kind fixes

app/bot/handlers/admin.py

@@ -53,7 +53,7 @@ HELP = (
     "<b>Smart upload (default):</b> just send or FORWARD files. The bot finds the course, type, "
     "title and skips duplicates. If unsure it asks you (tap a button).\n"
     "/recent : last 15 files · /review : files waiting for you\n"
-    "/move &lt;id&gt; &lt;course_id&gt; [kind] · /rename &lt;id&gt; &lt;title&gt; · /del &lt;id&gt;\n"
+    "/move &lt;id&gt; &lt;course_id&gt; [kind] · /kind &lt;id&gt; &lt;kind&gt; · /rename &lt;id&gt; &lt;title&gt; · /del &lt;id&gt;\n"
     "/mergecourses : merge courses shared by both streams (run once, then /seedcourses)\n\n"
     "<b>Force a course:</b> <code>/upload 3 exam</code>, then send files, /done when finished.\n"
     "<b>Or</b> send a file with caption\n"
@@ -398,6 +398,18 @@ async def move(message: Message, command: CommandObject, session: AsyncSession)
     await message.answer(f"✅ #{material.id} → {course.name} · {material.kind}")
 
 
+@router.message(Command("kind"))
+async def set_kind(message: Message, command: CommandObject, session: AsyncSession) -> None:
+    parts = (command.args or "").split()
+    kind = KIND_ALIASES.get(parts[1].lower()) if len(parts) == 2 else None
+    material = await repo.get_material(session, int(parts[0])) if parts and parts[0].isdigit() else None
+    if not material or not kind:
+        await message.answer("Usage: <code>/kind &lt;file_id&gt; &lt;slide|exam|note|book&gt;</code>")
+        return
+    material.kind = kind
+    await message.answer(f"✅ #{material.id} is now <b>{kind}</b>")
+
+
 @router.message(Command("rename"))
 async def rename(message: Message, command: CommandObject, session: AsyncSession) -> None:
     parts = (command.args or "").split(maxsplit=1)

app/services/classifier.py

@@ -133,7 +133,8 @@ def decide(ranked: Sequence[tuple[float, object]]) -> tuple[Optional[object], bo
 
 
 # ---------------------------------------------------------------- kind
-_EXAM = ("exam", "midterm", "mid exam", "mid-exam", "final", "quiz", "test", "question", "past paper", "previous", "ፈተና")
+_EXAM_STRONG = ("exam", "midterm", "mid exam", "quiz", "past paper", "question paper", "ፈተና")
+_EXAM_WEAK = ("final", "test", "question", "previous")  # "Geography Module Final" is NOT an exam
 _SLIDE = ("slide", "lecture", "ppt", "presentation", "chapter")
 _NOTE = ("note", "handout", "summary", "worksheet", "assignment", "tutorial", "module", "ማስታወሻ")
 _BOOK = ("textbook", "book", "edition", "isbn", "volume")
@@ -144,19 +145,21 @@ def detect_kind(name: str, caption: str, text: str, pages: Optional[int], size:
     label = f" {(name or '').lower()} {(caption or '').lower()} ".replace("_", " ").replace("-", " ")
     if ext in (".ppt", ".pptx"):
         return "slide", "presentation file"
-    if any(k in label for k in _EXAM):
+    if any(k in label for k in _EXAM_STRONG):
         return "exam", "name says exam"
     if any(k in label for k in _SLIDE):
         return "slide", "name says slides/chapter"
     if any(k in label for k in _BOOK):
         return "book", "name says book"
+    if any(k in label for k in _NOTE):
+        return "note", "name says module/notes"  # module/notes beat the ambiguous word "final"
+    if any(k in label for k in _EXAM_WEAK) or re.search(r"\bmid\b", label):
+        return "exam", "name hints at an exam"
     if sum(1 for k in _EXAM_TEXT if k in (text or "").lower()) >= 2:
         return "exam", "content looks like an exam"
-    if any(k in label for k in _NOTE):
-        return "note" if not pages or pages < 150 else "book", "name says module/notes"
-    if pages and pages >= 200:
+    if pages and pages >= 350:  # national modules are 150-300 pages: those stay "note"
         return "book", f"{pages} pages"
-    if size and size > 30 * 1024 * 1024:
+    if size and size > 40 * 1024 * 1024:
         return "book", "very large file"
     return "note", "default"
 

tests/test_logic.py

@@ -193,7 +193,12 @@ def test_kind_detection():
     assert C.detect_kind("Phys final exam.pdf", "", "", None, None, ".pdf")[0] == "exam"
     assert C.detect_kind("lecture1.pptx", "", "", None, None, ".pptx")[0] == "slide"
     assert C.detect_kind("Math module.pdf", "", "", 120, None, ".pdf")[0] == "note"
-    assert C.detect_kind("Math module.pdf", "", "", 260, None, ".pdf")[0] == "book"
+    assert C.detect_kind("Math module.pdf", "", "", 260, None, ".pdf")[0] == "note"
+    assert C.detect_kind("calculus.pdf", "", "", 400, None, ".pdf")[0] == "book"
+    assert C.detect_kind("Mathematics for Social Sciences.pdf", "", "", 280, None, ".pdf")[0] == "note"  # was #6
+    assert C.detect_kind("Geography Module Final.pdf", "", "", 150, None, ".pdf")[0] == "note"  # was #3
+    assert C.detect_kind("Physics final.pdf", "", "", None, None, ".pdf")[0] == "exam"
+    assert C.detect_kind("Phys mid.pdf", "", "", None, None, ".pdf")[0] == "exam"
     assert C.detect_kind("scan.pdf", "", "Time allowed: 2 hours. Answer all questions", None, None, ".pdf")[0] == "exam"
     assert C.detect_kind("whatever.pdf", "", "", None, None, ".pdf")[0] == "note"