script_parser.py

# -*- coding: utf-8 -*-
"""
结构化台本解析器 v1.3
台词（可混用）：
  【角色名】台词   /   角色名：台词
指令行：
  #BGM=音乐名                    设置背景音乐（后续段落生效）
  #BGM=音乐名,前奏秒,尾奏秒      同时指定淡入/淡出
  #BGM=无                        清除背景音乐
  #SFX=音效名                    穿插音效段
  #SFX=音效名,音量               音量 0~1.5，默认 0.9
  #MUSIC=名称,秒数               纯音乐留白段
  #END=名称,秒数                 片尾曲段（默认20秒）
注释：// 或 # 开头（指令除外）
段落类型：voice / sfx / music / end
"""
import re

BGM_DIRECTIVE_RE = re.compile(r'^#\s*BGM\s*[=:]\s*(.*?)\s*$', re.IGNORECASE)
SFX_DIRECTIVE_RE = re.compile(r'^#\s*SFX\s*[=:]\s*(.*?)\s*$', re.IGNORECASE)
MUSIC_DIRECTIVE_RE = re.compile(r'^#\s*(MUSIC|END)\s*[=:]\s*(.*?)\s*$', re.IGNORECASE)
BRACKET_RE = re.compile(r'^【\s*([^】]{1,32}?)\s*】\s*(.+)$', re.DOTALL)
COLON_RE = re.compile(r'^([^：:【】（）()《》"\'\s][^：:【】（）()]{0,30})\s*[：:]\s*(.+)$', re.DOTALL)
DECO_RE = re.compile(r'^[\s═━─﹘—=_\-~*·•▪#│┃║▏▎▕▍+.\u3000]{3,}$')
NONE_WORDS = {"", "无", "none", "null", "关闭", "off", "false"}

DEFAULT_CHARACTER = "旁白"
DEFAULT_MUSIC_SECONDS = 10.0
DEFAULT_END_SECONDS = 20.0
DEFAULT_SFX_VOLUME = 0.9


def _num(s, default=None):
    if s is None:
        return default
    s = str(s).strip()
    if not s:
        return default
    m = re.fullmatch(r'([0-9]+(?:\.[0-9]+)?)(s|秒|S)?', s)
    if not m:
        return default
    try:
        v = float(m.group(1))
        return v if v > 0 else default
    except ValueError:
        return default


def _parse_music_directive(kind, arg, current_bgm):
    arg = (arg or "").strip().strip('"\'')
    bgm = current_bgm
    dur = DEFAULT_END_SECONDS if kind.lower() == "end" else DEFAULT_MUSIC_SECONDS
    if arg:
        parts = [p for p in re.split(r'[,，\s]+', arg) if p]
        if parts:
            if _num(parts[0]):
                dur = _num(parts[0]) or dur
            else:
                bgm = parts[0]
                if len(parts) >= 2 and _num(parts[1]):
                    dur = _num(parts[1]) or dur
    dur = max(1.0, min(dur, 300.0))
    seg_type = "end" if kind.lower() == "end" else "music"
    return {"type": seg_type,
            "character": "片尾曲" if seg_type == "end" else "纯音乐",
            "text": "", "bgm": bgm, "duration": dur}


def _parse_sfx_directive(arg):
    arg = (arg or "").strip().strip('"\'')
    if not arg:
        return None
    parts = [p for p in re.split(r'[,，\s]+', arg) if p]
    name = parts[0]
    vol = DEFAULT_SFX_VOLUME
    if len(parts) >= 2 and _num(parts[1]) is not None:
        vol = max(0.05, min(_num(parts[1]), 1.5))
    return {"type": "sfx", "character": "音效", "text": "",
            "sfx": name, "volume": round(vol, 2)}


def parse_script(text):
    segments = []
    warnings = []
    current_bgm = None
    current_pre = None
    current_post = None

    if text is None:
        text = ""
    lines = text.replace("\r\n", "\n").replace("\r", "\n").split("\n")

    for lineno, raw in enumerate(lines, 1):
        line = raw.strip()
        if not line:
            continue

        m = SFX_DIRECTIVE_RE.match(line)
        if m:
            seg = _parse_sfx_directive(m.group(1))
            if seg:
                segments.append(seg)
            else:
                warnings.append(f"第 {lineno} 行 #SFX 指令缺少音效名，已忽略。")
            continue

        m = MUSIC_DIRECTIVE_RE.match(line)
        if m and m.group(1).upper() in ("MUSIC", "END"):
            seg = _parse_music_directive(m.group(1), m.group(2), current_bgm)
            if seg:
                segments.append(seg)
            continue

        m = BGM_DIRECTIVE_RE.match(line)
        if m:
            val = m.group(1).strip().strip('"\'')
            if val.lower() in {w.lower() for w in NONE_WORDS}:
                current_bgm = None
                current_pre = None
                current_post = None
            else:
                parts = [p.strip() for p in re.split(r'[,，]', val) if p.strip()]
                current_bgm = parts[0]
                current_pre = _num(parts[1]) if len(parts) >= 2 else None
                current_post = _num(parts[2]) if len(parts) >= 3 else None
            continue

        if line.startswith("//") or line.startswith("#"):
            continue
        if DECO_RE.match(line):
            continue

        m = BRACKET_RE.match(line)
        if m:
            segments.append({"type": "voice", "character": m.group(1).strip(),
                             "text": m.group(2).strip(), "bgm": current_bgm,
                             "pre_seconds": current_pre, "post_seconds": current_post})
            continue

        m = COLON_RE.match(line)
        if m and m.group(1).strip() and m.group(2).strip() and not re.search(r'\s', m.group(1)):
            segments.append({"type": "voice", "character": m.group(1).strip(),
                             "text": m.group(2).strip(), "bgm": current_bgm,
                             "pre_seconds": current_pre, "post_seconds": current_post})
            continue

        if segments and segments[-1].get("type", "voice") == "voice":
            segments[-1]["text"] += "\n" + line
            warnings.append(f"第 {lineno} 行未识别到角色前缀，已并入上一段台词。")
        elif segments:
            warnings.append(f"第 {lineno} 行未识别到角色前缀，且上一段是非台词段，该行已忽略。")
        else:
            segments.append({"type": "voice", "character": DEFAULT_CHARACTER,
                             "text": line, "bgm": current_bgm,
                             "pre_seconds": current_pre, "post_seconds": current_post})
            warnings.append(f"第 {lineno} 行未识别到角色前缀，已默认归入『旁白』。")

    for seg in segments:
        if seg.get("type") != "voice":
            seg["text"] = ""
            continue
        seg.setdefault("type", "voice")
        seg["text"] = seg["text"].strip()
        seg["character"] = seg["character"].strip() or DEFAULT_CHARACTER

    segments = [s for s in segments if s.get("type") != "voice" or s["text"]]
    return segments, warnings


def extract_characters(segments):
    seen = []
    for s in segments:
        if s.get("type") != "voice":
            continue
        if s["character"] not in seen:
            seen.append(s["character"])
    return seen


if __name__ == "__main__":
    demo = """【旁白】深夜，旧书店的铃铛忽然响了。
#SFX=门铃
【店主】这么晚，还来买书？
#SFX=脚步声,1.2
【女孩】[轻声] 我来找一本…不会老去的书。
#BGM=夜色钢琴,3,8
【旁白】灯光昏黄。
#MUSIC=12
【店主】[叹息] 这本书，等了你四十年。
#END=月光摇篮曲,15
"""
    segs, warns = parse_script(demo)
    types = [s["type"] for s in segs]
    assert types.count("sfx") == 2 and types.count("music") == 1 and types.count("end") == 1
    fade = [s for s in segs if s.get("pre_seconds")]
    assert fade and fade[0]["pre_seconds"] == 3.0
    print("PARSER v1.3 SELF-TEST PASSED:", types)