#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Screenplay & Shot Validator for Tran_Empire Cinematic Epic Verifies: - 3-Act Structure (Hồi 1, Hồi 2, Hồi 3) - Scene counts and 10s atomic shot IDs (epXX_sceneYY_shotZZ) - 4-Stem Audio presence (Stem 1: BGM, Stem 2: Ambience, Stem 3: Foley, Stem 4: Vocal) - Duration estimation (~30 minutes / 1800 seconds) """ import os import re import sys import json import argparse from pathlib import Path def validate_screenplay(file_path, min_runtime_sec=0, check_flowkit=False): path = Path(file_path) if not path.exists(): return {"file": str(file_path), "valid": False, "error": "File does not exist"} with open(path, "r", encoding="utf-8") as f: content = f.read() lines = content.splitlines() total_lines = len(lines) # 1. Check Acts acts_found = [] for act_num in [1, 2, 3]: act_pattern = rf"(HỒI|Hồi|ACT|Act)\s*{act_num}[:\s]" if re.search(act_pattern, content): acts_found.append(act_num) # 2. Check Scenes scene_matches = re.findall(r"(CẢNH|Cảnh|SCENE|Scene)\s*(\d+)[:\s]", content) unique_scenes = sorted(list(set(int(m[1]) for m in scene_matches))) # 3. Check Shots and Dynamic Durations (2s - 10s Directorial Standard) # Pattern: ##### Shot 01: `ep01_scene01_shot01` (10s) or (3s - REACTION CUT) shot_matches = re.findall( r"#####\s+Shot\s+(\d+):\s*`([^`]+)`(?:\s*\(([^)]*)\))?", content, re.IGNORECASE ) durations = [] shot_ids = [] invalid_duration_shots = [] if shot_matches: for s_num, s_id, s_paren in shot_matches: shot_ids.append(s_id) sec = 10 if s_paren: dur_m = re.search(r"(\d+)\s*(?:s|giây)", s_paren, re.IGNORECASE) if dur_m: sec = int(dur_m.group(1)) durations.append(sec) if sec < 2 or sec > 10: invalid_duration_shots.append(f"{s_id} ({sec}s)") else: # Fallback regex found_ids = re.findall(r"ep\d+_scene\d+_shot\d+", content) if not found_ids: found_ids = re.findall(r"(?:Shot|shot)\s*(\d+)[:\s]", content) shot_ids = found_ids durations = [10] * len(shot_ids) total_shots = len(shot_ids) total_duration_sec = sum(durations) estimated_duration_min = round(total_duration_sec / 60, 2) avg_duration = round(total_duration_sec / total_shots, 1) if total_shots > 0 else 0 # 4. Check Audio Stems stem1_bgm = len(re.findall(r"(Stem 1|BGM|Nhạc nền)", content, re.IGNORECASE)) stem2_ambience = len(re.findall(r"(Stem 2|Ambience|Môi trường)", content, re.IGNORECASE)) stem3_foley = len(re.findall(r"(Stem 3|Foley|Tiếng động)", content, re.IGNORECASE)) stem4_vocal = len(re.findall(r"(Stem 4|Vocal|Thoại|Lời thoại)", content, re.IGNORECASE)) has_4stems = (stem1_bgm > 0 and stem2_ambience > 0 and stem3_foley > 0 and stem4_vocal > 0) # 5. Check FlowKit Criteria first_frames = len(re.findall(r"(First Frame|Start Frame|Frame 001|Đầu cú máy)", content, re.IGNORECASE)) last_frames = len(re.findall(r"(Last Frame|Tail Frame|Frame cuối|Kết cú máy)", content, re.IGNORECASE)) camera_motions = len(re.findall(r"(Chuyển động Camera|Camera Motion|Chuyển động máy)", content, re.IGNORECASE)) camera_angles = len(re.findall(r"(Góc máy|Camera Angle|Cỡ cảnh)", content, re.IGNORECASE)) characters = len(re.findall(r"(Nhân vật|Character)", content, re.IGNORECASE)) dialogues = len(re.findall(r"(Thoại|Lời thoại|Dialogue|Vocal|Closed Lips)", content, re.IGNORECASE)) flowkit_compliant = ( first_frames >= max(1, int(total_shots * 0.8)) and last_frames >= max(1, int(total_shots * 0.8)) and camera_motions >= max(1, int(total_shots * 0.8)) ) # Validation verdict is_valid = True issues = [] if len(acts_found) < 3: is_valid = False issues.append(f"Missing acts. Found: {acts_found}, expected [1, 2, 3]") if total_shots < 3: is_valid = False issues.append(f"Critically insufficient shots: {total_shots} (Target: ~180 shots for 30m episode)") elif total_shots < 30: issues.append(f"Pilot core draft: {total_shots} shots ({total_duration_sec}s, Target: ~180 shots for 30m episode)") if invalid_duration_shots: is_valid = False issues.append(f"Directorial duration violation (<2s or >10s): {', '.join(invalid_duration_shots[:5])}") if not has_4stems: issues.append(f"Missing complete 4 stems. Stem1: {stem1_bgm}, Stem2: {stem2_ambience}, Stem3: {stem3_foley}, Stem4: {stem4_vocal}") if min_runtime_sec > 0 and total_duration_sec < min_runtime_sec: is_valid = False issues.append(f"Runtime deficit: {total_duration_sec}s / required {min_runtime_sec}s (Deficit: {min_runtime_sec - total_duration_sec}s)") if check_flowkit and not flowkit_compliant: is_valid = False issues.append(f"FlowKit missing fields: FirstFrame={first_frames}/{total_shots}, LastFrame={last_frames}/{total_shots}, CameraMotion={camera_motions}/{total_shots}") report = { "file": str(path.name), "total_lines": total_lines, "acts_detected": acts_found, "scenes_count": len(unique_scenes), "total_shots": total_shots, "total_duration_sec": total_duration_sec, "estimated_duration_minutes": estimated_duration_min, "avg_duration_sec": avg_duration, "stems_presence": { "stem1_bgm": stem1_bgm, "stem2_ambience": stem2_ambience, "stem3_foley": stem3_foley, "stem4_vocal": stem4_vocal, "all_4_present": has_4stems }, "flowkit_stats": { "first_frames": first_frames, "last_frames": last_frames, "camera_motions": camera_motions, "camera_angles": camera_angles, "characters": characters, "dialogues": dialogues, "flowkit_compliant": flowkit_compliant }, "valid": is_valid, "warnings_or_errors": issues } return report def main(): parser = argparse.ArgumentParser(description="Validate Tran Empire Screenplay files") parser.add_argument("path", help="Path to screenplay file or directory containing screenplays") parser.add_argument("--min-runtime", type=int, default=0, help="Minimum required runtime in seconds (e.g. 1800 for 30 min)") parser.add_argument("--flowkit", action="store_true", help="Enforce strict FlowKit criteria (First Frame, Last Frame, Camera Motion)") args = parser.parse_args() target = Path(args.path) if target.is_file(): files = [target] elif target.is_dir(): files = list(target.rglob("*.md")) else: print(f"Error: {target} not found") sys.exit(1) print(f"=== TRAN EMPIRE SCREENPLAY VALIDATOR (FLOWKIT & RUNTIME EDITION) ===") if args.min_runtime > 0: print(f"Enforcing Minimum Runtime: {args.min_runtime}s (~{args.min_runtime / 60:.1f} min)") if args.flowkit: print(f"Enforcing Strict FlowKit Standards (First Frame, Last Frame, Camera Motion, 4 Stems)") print(f"Checking {len(files)} file(s)...\n") overall_valid = True for f in files: if "Bible" in str(f) or "Guides" in str(f) or "Prompts" in str(f): continue res = validate_screenplay(f, min_runtime_sec=args.min_runtime, check_flowkit=args.flowkit) status = "PASSED [OK]" if res["valid"] else "FAILED [X]" print(f"[{status}] {res['file']}") print(f" - Acts: {res['acts_detected']} | Scenes: {res['scenes_count']} | Shots: {res['total_shots']} (Avg: {res['avg_duration_sec']}s)") print(f" - Runtime: {res['total_duration_sec']}s (~{res['estimated_duration_minutes']} min) | Stems: {res['stems_presence']['all_4_present']}") fk = res['flowkit_stats'] print(f" - FlowKit Specs: FirstFrame={fk['first_frames']} | LastFrame={fk['last_frames']} | CamMotion={fk['camera_motions']} | Compliant: {fk['flowkit_compliant']}") if res["warnings_or_errors"]: for issue in res["warnings_or_errors"]: print(f" ! {issue}") print() if not res["valid"]: overall_valid = False if overall_valid: print("ALL CHECKED SCREENPLAYS MEET ARCHITECTURAL CRITERIA.") else: print("SOME SCREENPLAYS HAVE ISSUES TO RESOLVE.") sys.exit(1) if __name__ == "__main__": main()