galbook 0.0.0-stage → 1.2.4

package/galbook/importers/legacy.py

@@ -0,0 +1,122 @@
+"""老式书稿（book.json + chNNN.json：按卷分章、一页一句、没有原始选项数据）→ 可玩剧本。
+
+没有脚本里的真实分支，所以：有「共通」卷就先读共通，读完让 agent 选走哪条线（按卷）；
+每条线读完算一个结局。mode = "linear"，并在 warnings 里写清楚这是「按卷生成的选项」。"""
+import json
+import os
+import threading
+
+from ..ir import story_from_dict
+from .common import is_h_title
+
+_cache = {}
+_lock = threading.Lock()
+
+
+def _load(p):
+    with open(p, encoding="utf-8") as f:
+        return json.load(f)
+
+
+def _sig(book_dir):
+    bj = os.path.join(book_dir, "book.json")
+    st = os.stat(bj)
+    return (st.st_mtime_ns, st.st_size)
+
+
+def load(book_dir):
+    sig = _sig(book_dir)
+    with _lock:
+        hit = _cache.get(book_dir)
+        if hit and hit[0] == sig:
+            return hit[1]
+    meta, scenes = build(book_dir)
+    story = story_from_dict(meta, scenes)
+    story.mode = "linear"
+    with _lock:
+        _cache[book_dir] = (sig, story)
+    return story
+
+
+def _sprites(step):
+    out = {}
+    for i, s in enumerate(step.get("sprites") or []):
+        if isinstance(s, str):
+            out["#%d" % i] = {"img": s, "pos": "center"}
+        elif isinstance(s, dict) and s.get("img"):
+            out[s.get("name") or "#%d" % i] = {"img": s["img"], "pos": s.get("pos") or "center"}
+    return out
+
+
+def build(book_dir):
+    b = _load(os.path.join(book_dir, "book.json"))
+    vols = b.get("volumes")
+    if not vols:
+        vols = [{"title": "", "chapters": b.get("chapters", [])}]
+    scenes, smeta = {}, {}
+    vol_entries = []
+    for vi, v in enumerate(vols):
+        chs = v.get("chapters") or []
+        first = None
+        prev = None
+        for ci, c in enumerate(chs):
+            sid = "v%d_%s" % (vi, c.get("id") or ("c%d" % ci))
+            path = os.path.join(book_dir, c["file"])
+            try:
+                d = _load(path)
+            except (OSError, ValueError):
+                continue
+            title = d.get("title") or c.get("title") or sid
+            ops = [{"op": "label", "name": "top"}]
+            bg = cg = None
+            sp = {}
+            for st in d.get("steps") or []:
+                if st.get("bg") != bg:
+                    bg = st.get("bg")
+                    ops.append({"op": "bg", "img": bg})
+                    cg = None
+                if st.get("cg") != cg:
+                    cg = st.get("cg")
+                    ops.append({"op": "cg", "img": cg})
+                ns = _sprites(st)
+                if ns != sp:
+                    ops.append({"op": "sprites_clear"})
+                    for name, x in ns.items():
+                        ops.append({"op": "sprite", "name": name, "img": x["img"], "pos": x["pos"]})
+                    sp = ns
+                who = st.get("speaker") or ""
+                if who == "旁白":
+                    who = ""
+                ops.append({"op": "say", "who": who.split("＠")[0], "text": st.get("text", "")})
+            scenes[sid] = {"id": sid, "title": title, "ops": ops}
+            smeta[sid] = {"title": title, "route": v.get("title", ""), "tags": ["h"] if is_h_title(title) else []}
+            if prev:
+                scenes[prev]["ops"].append({"op": "jump", "scene": sid})
+            first = first or sid
+            prev = sid
+        if first:
+            vol_entries.append({"title": v.get("title") or b.get("title", ""), "first": first, "last": prev})
+    common = [e for e in vol_entries if "共通" in e["title"] or "common" in e["title"].lower()]
+    others = [e for e in vol_entries if e not in common]
+    start_ops = [{"op": "label", "name": "top"}]
+    if len(vol_entries) == 1:
+        start_ops.append({"op": "jump", "scene": vol_entries[0]["first"]})
+        scenes[vol_entries[0]["last"]]["ops"].append({"op": "end", "kind": "ending", "name": vol_entries[0]["title"]})
+    else:
+        chooser = {"op": "choice", "id": "route", "prompt": "选一条线读下去（老书稿没有原始选项，按卷生成）",
+                   "options": [{"text": e["title"], "goto": {"scene": e["first"]}} for e in others]}
+        if common:
+            c = common[0]
+            start_ops.append({"op": "jump", "scene": c["first"]})
+            scenes[c["last"]]["ops"].append(chooser)
+        else:
+            start_ops.append(chooser)
+        for e in others:
+            scenes[e["last"]]["ops"].append({"op": "end", "kind": "ending", "name": e["title"]})
+    scenes["__start__"] = {"id": "__start__", "title": b.get("title", ""), "ops": start_ops}
+    smeta["__start__"] = {"title": b.get("title", "")}
+    meta = {"format": "galbook/1", "title": b.get("title", ""), "engine": "book-legacy",
+            "importer": "galbook-legacy-1", "mode": "linear", "entry": {"scene": "__start__", "label": "top"},
+            "scenes": smeta, "routes": [{"id": e["first"], "title": e["title"]} for e in vol_entries],
+            "warnings": ["老书稿没有原始脚本里的选项与跳转；这里的选项是按卷（路线）生成的，不代表游戏原本的分支。"]}
+    return meta, scenes

package/galbook/importers/nscripter.py

@@ -0,0 +1,314 @@
+"""NScripter / ONScripter（0.txt…9.txt 或 nscript.dat）→ galbook/1。老游戏、同人作品很多是这个。
+
+nscript.dat 每字节 XOR 0x84 就是文本。认的写法：
+  *label   ;注释   一行多条用 : 分隔
+  文字行（不以英文指令开头）；行尾 \\ 换页、@ 等点击；【名字】「…」 或 名字「…」 认说话人
+  select "文字",*l1,"文字",*l2 …（可跨行）  selgosub 同上但是调用  selnum %v,"a","b"（把序号写进变量）
+  goto *l  gosub *l  return  jumpf/jumpb（~ 标记）不支持会报
+  mov %0,1  add/sub/mul/div %0,1  inc/dec %0   if 条件 指令 / notif 条件 指令
+  bg "x.bmp",1 / bg black,1   ld l|c|r,":a;x.bmp",1   cl l|c|r|a,1   bgm "x" / stop
+  end → 结局；reset / definereset → 回标题
+%0-%199 是本局变量 f.v*；%200 起（globalon 的全局区）当系统变量 s.v*；$n 字符串 → f.str*
+"""
+import os
+import re
+
+from .common import SceneB, Assets, is_h_title
+
+
+def _readb(p):
+    with open(p, "rb") as f:
+        return f.read()
+
+CMD = re.compile(r"^([A-Za-z_][A-Za-z0-9_]*)\b\s*(.*)$")
+GLOBAL_BORDER = 200
+
+
+def load_script(src):
+    """返回脚本全文（str）和来源说明。"""
+    p = None
+    for name in ("nscript.dat", "nscript.___"):
+        for dp, _, fns in os.walk(src):
+            for fn in fns:
+                if fn.lower() == name:
+                    p = os.path.join(dp, fn)
+                    break
+    if p:
+        with open(p, "rb") as fh:
+            raw = bytes(b ^ 0x84 for b in fh.read())
+        return _dec(raw), os.path.basename(p)
+    parts = []
+    for dp, _, fns in os.walk(src):
+        for fn in sorted(fns):
+            if re.match(r"^\d{1,2}\.txt$", fn.lower()):
+                parts.append((int(fn.split(".")[0]), os.path.join(dp, fn)))
+    if not parts:
+        raise ValueError("没找到 nscript.dat 或 0.txt")
+    return "\n".join(_dec(_readb(p)) for _, p in sorted(parts)), "%d 个 N.txt" % len(parts)
+
+
+def _dec(raw):
+    for e in ("utf-8-sig", "cp932", "gbk"):
+        try:
+            return raw.decode(e)
+        except UnicodeDecodeError:
+            pass
+    return raw.decode("cp932", "replace")
+
+
+def var(tok):
+    tok = tok.strip()
+    m = re.match(r"^%(\d+)$", tok)
+    if m:
+        n = int(m.group(1))
+        return ("s.v%d" if n >= GLOBAL_BORDER else "f.v%d") % n
+    m = re.match(r"^\$(\d+)$", tok)
+    if m:
+        return "f.str%s" % m.group(1)
+    return None
+
+
+def conv_expr(e):
+    e = re.sub(r"%(\d+)", lambda m: var("%" + m.group(1)), e)
+    e = re.sub(r"\$(\d+)", lambda m: var("$" + m.group(1)), e)
+    e = re.sub(r"(?<![=!<>])=(?!=)", "==", e)
+    e = e.replace("<>", "!=")
+    e = re.sub(r"(?<!&)&(?!&)", "&&", e)
+    return e
+
+
+def _split_colon(line):
+    out, cur, q = [], [], False
+    for ch in line:
+        if ch == '"':
+            q = not q
+        if ch == ":" and not q:
+            out.append("".join(cur))
+            cur = []
+        elif ch == ";" and not q:
+            break
+        else:
+            cur.append(ch)
+    out.append("".join(cur))
+    return [x.strip() for x in out if x.strip()]
+
+
+def _args(s):
+    out, cur, q = [], [], False
+    for ch in s:
+        if ch == '"':
+            q = not q
+            cur.append(ch)
+        elif ch == "," and not q:
+            out.append("".join(cur).strip())
+            cur = []
+        else:
+            cur.append(ch)
+    if cur or out:
+        out.append("".join(cur).strip())
+    return out
+
+
+def _unq(s):
+    s = s.strip()
+    return s[1:-1] if len(s) >= 2 and s[0] == '"' and s[-1] == '"' else s
+
+
+class Importer(object):
+    def __init__(self, src, assets=None, entry="start"):
+        self.src, self.assets, self.entry = src, assets or Assets(), entry
+        self.warnings, self.unknown = [], {}
+        self.stats = {"files": 1, "says": 0, "choices": 0, "options": 0, "jumps": 0, "calls": 0, "sets": 0, "ifs": 0}
+        self.system_vars = set()
+        self.buf = []
+
+    def build(self):
+        text, origin = load_script(self.src)
+        self.stats["origin"] = origin
+        b = SceneB("script", "script")
+        lines = text.splitlines()
+        i = 0
+        in_define = False
+        while i < len(lines):
+            s = lines[i].strip()
+            i += 1
+            if not s or s.startswith(";"):
+                continue
+            if s.startswith("*"):
+                self._flush(b)
+                name = s[1:].split()[0].split(":")[0]
+                in_define = name.lower() == "define"
+                b.add("label", name=name.lower())
+                continue
+            if in_define:
+                continue
+            # select 跨行：行尾逗号就接下一行
+            while re.match(r"^(select|selgosub|selnum|csel)\b", s, re.I) and s.rstrip().endswith(",") and i < len(lines):
+                s = s.rstrip() + lines[i].strip()
+                i += 1
+            if CMD.match(s) and not re.match(r"^[A-Za-z]+[「『]", s):
+                for part in _split_colon(s):
+                    self._cmd(b, part)
+            else:
+                self._text(b, s)
+        self._flush(b)
+        start = SceneB("__start__", "开始", tags=["system"])
+        start.add("label", name="top")
+        labels = {op["name"] for op in b.ops if op["op"] == "label"}
+        ent = self.entry if self.entry in labels else ("start" if "start" in labels else None)
+        if not ent:
+            raise ValueError("脚本里没有 *start")
+        start.add("jump", scene="script", label=ent)
+        scenes = {"__start__": start.data(), "script": b.data()}
+        meta = {"format": "galbook/1", "title": "", "engine": "nscripter", "importer": "galbook-nscripter-1",
+                "entry": {"scene": "__start__", "label": "top"},
+                "scenes": {k: {"title": v["title"], "tags": v["tags"]} for k, v in scenes.items()},
+                "routes": [], "system_vars": sorted(self.system_vars),
+                "stats": dict(self.stats, unknown=sorted(self.unknown.items(), key=lambda x: -x[1])[:20],
+                              assets=self.assets.report()),
+                "warnings": self.warnings[:200]}
+        return meta, scenes
+
+    def _text(self, b, s):
+        page = s.endswith("\\") or s.endswith("@")
+        t = s.rstrip("\\@").replace("@", "")
+        t = re.sub(r"#[0-9A-Fa-f]{6}", "", t)
+        self.buf.append(t)
+        if page or True:
+            self._flush(b)
+
+    def _flush(self, b):
+        t = "".join(self.buf).strip()
+        self.buf = []
+        if not t:
+            return
+        who = ""
+        m = re.match(r"^【([^】]{1,16})】\s*(.*)$", t) or re.match(r"^([^「『\s]{1,8})([「『].*)$", t)
+        if m:
+            who, t = m.group(1), m.group(2)
+        b.add("say", who=who, text=t)
+        self.stats["says"] += 1
+
+    def _jump_label(self, tok):
+        tok = tok.strip()
+        return tok[1:].lower() if tok.startswith("*") else None
+
+    def _cmd(self, b, s):
+        m = CMD.match(s)
+        if not m:
+            return self._text(b, s)
+        c, rest = m.group(1).lower(), m.group(2).strip()
+        a = _args(rest) if rest else []
+        if c in ("goto", "gosub"):
+            lb = self._jump_label(rest)
+            if lb:
+                self._flush(b)
+                b.add("jump" if c == "goto" else "call", label=lb)
+                self.stats["jumps" if c == "goto" else "calls"] += 1
+        elif c == "return":
+            self._flush(b)
+            b.add("ret")
+        elif c in ("select", "selgosub", "csel"):
+            self._flush(b)
+            opts = []
+            for k in range(0, len(a) - 1, 2):
+                lb = self._jump_label(a[k + 1])
+                if lb:
+                    opts.append({"text": _unq(a[k]), "goto": {"label": lb}})
+            if c == "selgosub":
+                # 调用式：选完 gosub，回来接着往下
+                ret = "__selret%d" % len(b.ops)
+                for o in opts:
+                    o["goto"] = {"label": "__selcall_%s_%d" % (o["goto"]["label"], len(b.ops))}
+                b.add("choice", id="c%d" % len(b.ops), options=opts)
+                b.add("jump", label=ret)
+                for o, k in zip(opts, range(0, len(a) - 1, 2)):
+                    b.add("label", name=o["goto"]["label"])
+                    b.add("call", label=self._jump_label(a[k + 1]))
+                    b.add("jump", label=ret)
+                b.add("label", name=ret)
+            else:
+                b.add("choice", id="c%d" % len(b.ops), options=opts)
+            self.stats["choices"] += 1
+            self.stats["options"] += len(opts)
+        elif c == "selnum":
+            self._flush(b)
+            v = var(a[0]) if a else None
+            opts = [{"text": _unq(x), "set": "%s = %d" % (v, k)} for k, x in enumerate(a[1:])] if v else []
+            if opts:
+                b.add("choice", id="c%d" % len(b.ops), options=opts)
+                self.stats["choices"] += 1
+                self.stats["options"] += len(opts)
+        elif c in ("mov", "add", "sub", "mul", "div", "mod"):
+            v = var(a[0]) if a else None
+            if v and len(a) > 1:
+                rhs = conv_expr(a[1])
+                e = {"mov": "%s = %s", "add": "%s += %s", "sub": "%s -= %s", "mul": "%s *= %s"}.get(c)
+                e = (e % (v, rhs)) if e else ("%s = %s %s %s" % (v, v, "/" if c == "div" else "%", rhs))
+                b.add("set", expr=e)
+                self.stats["sets"] += 1
+                if v.startswith("s."):
+                    self.system_vars.add(v[2:])
+        elif c in ("inc", "dec"):
+            v = var(rest)
+            if v:
+                b.add("set", expr="%s %s= 1" % (v, "+" if c == "inc" else "-"))
+        elif c in ("if", "notif"):
+            mm = re.match(r"^(.*?)\s+((?:goto|gosub|mov|add|sub|inc|dec|end|return|reset|select)\b.*)$", rest, re.I)
+            if not mm:
+                self.unknown["if"] = self.unknown.get("if", 0) + 1
+                return
+            cond = conv_expr(mm.group(1))
+            if c == "notif":
+                cond = "!(%s)" % cond
+            n0 = len(b.ops)
+            self._cmd(b, mm.group(2))
+            for op in b.ops[n0:]:
+                op["if"] = cond if "if" not in op else "(%s) && (%s)" % (cond, op["if"])
+            self.stats["ifs"] += 1
+        elif c == "bg":
+            self._flush(b)
+            nm = _unq(a[0]) if a else ""
+            b.add("sprites_clear")
+            b.add("bg", img=None if nm.lower() in ("black", "white", "") or nm.startswith("#") else (self.assets.find(nm) or nm))
+        elif c == "ld":
+            self._flush(b)
+            if len(a) >= 2:
+                f = _unq(a[1]).split(";")[-1]
+                b.add("sprite", name=a[0], img=self.assets.find(f) or f, pos={"l": "left", "c": "center", "r": "right"}.get(a[0], a[0]))
+        elif c == "cl":
+            if a and a[0] == "a":
+                b.add("sprites_clear")
+            elif a:
+                b.add("sprite", name=a[0], img=None)
+        elif c in ("bgm", "mp3loop", "mp3", "play"):
+            b.add("bgm", name=_unq(a[0]) if a else None)
+        elif c in ("stop", "mp3stop", "playstop"):
+            b.add("bgm", name=None)
+        elif c == "end":
+            self._flush(b)
+            b.add("end", kind="ending", how="end")
+        elif c in ("reset", "definereset"):
+            self._flush(b)
+            b.add("end", kind="title", how="reset")
+        elif c in ("jumpf", "jumpb", "tablegoto"):
+            self.warnings.append("%s 不支持（只能人工改成 goto）: %s" % (c, s))
+        elif c in ("wait", "delay", "click", "lr_trap", "trap", "print", "effect", "dwave", "dwaveloop", "wave",
+                   "waveloop", "wavestop", "btndef", "btn", "btnwait", "btnwait2", "spbtn", "lsp", "csp", "msp",
+                   "amsp", "vsp", "texton", "textoff", "textclear", "erasetextwindow", "setwindow", "setwindow2",
+                   "monocro", "nega", "quake", "quakex", "quakey", "locate", "br", "saveon", "saveoff", "rmode",
+                   "skipoff", "clickvoice", "voicevol", "bgmvol", "sevol", "humanz", "windoweffect", "puttext",
+                   "game", "caption", "versionstr", "globalon", "nsa", "ns2", "ns3", "numalias", "stralias",
+                   "defaultspeed", "menuselectcolor", "menusetwindow", "savenumber", "rmenu", "kidokuskip",
+                   "mode_saya", "effectblank", "maxkaisoupage", "underline", "transmode", "textspeed",
+                   "setcursor", "abssetcursor", "linepage", "linepage2", "systemcall", "savegame", "loadgame",
+                   "chvol", "selectcolor", "selectvoice", "automode", "filelog", "labellog", "kidokumode",
+                   "shadedistance", "usewheel", "useescspc", "getcursorpos", "time", "date", "getparam", "itoa",
+                   "atoi", "len", "mid", "cmp", "add_", "for", "next", "break", "dim", "movl", "rnd", "rnd2"):
+            pass
+        else:
+            self.unknown[c] = self.unknown.get(c, 0) + 1
+
+
+def build(src, assets=None, entry="start", **_):
+    return Importer(src, assets, entry or "start").build()