Thread reading grouped annotations

reading_grouped_annotations.py

@@ -174,6 +174,8 @@ def apply_actions_and_regenerate_grouped(root_dir, data, student_id, actions, la
 from utils import read_all_labels
 import argparse
 
+
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Read grouped annotations and compile PDFs")
     parser.add_argument("input_path", help="Directory path")
@@ -195,9 +197,67 @@ if __name__ == "__main__":
     # Load original data
     original_data = annotating.make_dictionary(root_dir)
 
+    lock = threading.Lock()
     actions_by_student = collections.defaultdict(list)
     notes_by_student = collections.defaultdict(dict)
 
+
+    def process_bgnot_entry(entry):
+        gdir = os.path.join(bgnot_dir, entry)
+        if not os.path.isdir(gdir) or entry.startswith("Copie"):
+            return
+
+        actions, notes_img = detect_checks_and_notes(gdir)
+        bnote_path = os.path.join(gdir, "bnote.json")
+        if not os.path.exists(bnote_path) or notes_img is None:
+            return
+
+        with open(bnote_path, "r") as f:
+            bnote_data = json.load(f)
+
+        with lock:
+            for act in actions:
+                sid = str(act.get("student_id"))
+                if sid: actions_by_student[sid].append(act)
+
+            for img_info in bnote_data.get("images", []):
+                sid, lbl = str(img_info.get("id")), img_info.get("label")
+                hmin, hmax = img_info.get("hmin", 0), img_info.get("hmax", 0)
+                if hmax > hmin:
+                    crop = notes_img.crop((0, hmin, notes_img.width, hmax))
+                    if has_significant_notes(crop):
+                        notes_by_student[sid][lbl] = {'img': crop, 'old_header_h': img_info.get("header_height", 0)}
+
+
+    def process_refaire_entry(sid, r_labels):
+        s_bnot_dir = os.path.join(root_dir, "Bnot", f"Copie{sid}")
+        if not os.path.exists(s_bnot_dir): return
+        if not r_labels: r_labels = list(original_data.get(sid, {}).keys())
+
+        with lock:
+            actions_by_student[sid] = [a for a in actions_by_student[sid] if a.get('label') not in r_labels]
+            for lbl in r_labels: notes_by_student[sid].pop(lbl, None)
+
+        b_actions, b_notes_img = detect_checks_and_notes(s_bnot_dir)
+        b_bnote_path = os.path.join(s_bnot_dir, "bnote.json")
+        if os.path.exists(b_bnote_path):
+            with open(b_bnote_path, "r") as f:
+                b_bnote_data = json.load(f)
+            with lock:
+                for act in b_actions:
+                    act["student_id"] = sid
+                    actions_by_student[sid].append(act)
+                if b_notes_img:
+                    for img_info in b_bnote_data.get("images", []):
+                        lbl = img_info.get("label")
+                        hmin, hmax = img_info.get("hmin", 0), img_info.get("hmax", 0)
+                        if hmax > hmin:
+                            crop = b_notes_img.crop((0, hmin, b_notes_img.width, hmax))
+                            if has_significant_notes(crop):
+                                notes_by_student[sid][lbl] = {'img': crop, 'old_header_h': img_info.get("header_height", 0)}
+
+
+
     # --- 0. Read refaire.json if requested ---
     refaire_dict = {}
     if args.with_refaire:
@@ -210,92 +270,15 @@ if __name__ == "__main__":
                 refaire_dict[sid] = labels
         else:
             print(f"Warning: --with-refaire flag used, but {refaire_path} not found.")
-    # --- 1. Scan BGnot grouped directories and extract all checks & notes ---
-    for entry in os.listdir(bgnot_dir):
-        gdir = os.path.join(bgnot_dir, entry)
 
-        if not os.path.isdir(gdir) or entry.startswith("Copie"):
+    # Part 1 : lecture des bgnot
-            continue  # Ignore files and already compiled student folders
+    with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:
+        executor.map(process_bgnot_entry, os.listdir(bgnot_dir))
 
-        print(f"\nScanning grouped annotations in {entry}")
+    # Part 1.5: Refaire
-        actions, notes_img = detect_checks_and_notes(gdir)
-
-        bnote_path = os.path.join(gdir, "bnote.json")
-        if not os.path.exists(bnote_path) or notes_img is None:
-            continue
-
-        with open(bnote_path, "r") as f:
-            bnote_data = json.load(f)
-
-        # Route actions to specific students
-        for act in actions:
-            sid = str(act.get("student_id"))
-            if sid:
-                actions_by_student[sid].append(act)
-
-        # Route manual note crops to specific students and labels
-        for img_info in bnote_data.get("images", []):
-            sid = str(img_info.get("id"))
-            lbl = img_info.get("label")
-            hmin = img_info.get("hmin", 0)
-            hmax = img_info.get("hmax", 0)
-
-            if hmax > hmin:
-                crop = notes_img.crop((0, hmin, notes_img.width, hmax))
-                # Store it if there are pen marks on it
-                if has_significant_notes(crop):
-                    notes_by_student[sid][lbl] = {
-                        'img': crop,
-                        'old_header_h': img_info.get("header_height", 0)
-                    }
-
-    # --- 1.5. Override BGnot actions with Bnot actions for Refaire targets ---
     if args.with_refaire and refaire_dict:
-        bnot_dir = os.path.join(root_dir, "Bnot")
+        for sid, labels in refaire_dict.items():
-        for sid, r_labels in refaire_dict.items():
+            process_refaire_entry(sid, labels)
-            s_bnot_dir = os.path.join(bnot_dir, f"Copie{sid}")
-            if not os.path.exists(s_bnot_dir):
-                continue
-
-            # If empty list, it targets all labels for this Copie
-            if not r_labels:
-                r_labels = list(original_data.get(sid, {}).keys())
-
-            # Clear out existing BGnot actions & notes for the refaire labels
-            actions_by_student[sid] = [
-                a for a in actions_by_student[sid] if a.get('label') not in r_labels
-            ]
-            for lbl in r_labels:
-                notes_by_student[sid].pop(lbl, None)
-
-            print(f"\nScanning refaire annotations in Bnot for Copie{sid}")
-            b_actions, b_notes_img = detect_checks_and_notes(s_bnot_dir)
-
-            b_bnote_path = os.path.join(s_bnot_dir, "bnote.json")
-            if os.path.exists(b_bnote_path):
-                with open(b_bnote_path, "r") as f:
-                    b_bnote_data = json.load(f)
-
-                # Inject actions
-                for act in b_actions:
-                    # Bnot checkboxes don't natively have student_id, so we inject it
-                    act["student_id"] = sid
-                    actions_by_student[sid].append(act)
-
-                # Inject notes
-                if b_notes_img is not None:
-                    for img_info in b_bnote_data.get("images", []):
-                        lbl = img_info.get("label")
-                        hmin = img_info.get("hmin", 0)
-                        hmax = img_info.get("hmax", 0)
-
-                        if hmax > hmin:
-                            crop = b_notes_img.crop((0, hmin, b_notes_img.width, hmax))
-                            if has_significant_notes(crop):
-                                notes_by_student[sid][lbl] = {
-                                    'img': crop,
-                                    'old_header_h': img_info.get("header_height", 0)
-                                }
 
 
     def process_student(sid):