galbook 0.0.0-stage → 1.2.4

package/galbook/importers/renpy.py

@@ -0,0 +1,333 @@
+"""Ren'Py（.rpy 源码）→ galbook/1。新游戏最常见的引擎之一；只吃 .rpy 源码（.rpyc 先用 unrpyc 反编译）。
+
+认的写法：
+  define e = Character("名字", …)          说话人
+  image eileen happy = "eileen_happy.png"  图片名 → 文件
+  default x = 0 / define persistent…       初始变量（persistent.* = 系统变量 → s.*）
+  label name:                              标签（全局，跨文件）
+  "旁白" / e "台词" / e happy "台词" / "名字" "台词"
+  menu: / menu name:  （可选一句提问）+ "选项" [if 条件]: 块
+  jump x / call x / return / pass
+  $ x = 1 / $ persistent.y = True
+  if 条件: / elif 条件: / else:
+  scene bg room / show eileen happy [at left] / hide eileen / play music "x"
+其它块（python:/init:/screen/transform/style/ATL）整块跳过。
+"""
+import os
+import re
+
+from .common import SceneB, Assets, read_text, is_h_title
+
+STR = r'"(?:[^"\\]|\\.)*"|\'(?:[^\'\\]|\\.)*\''
+SAY = re.compile(r"^(?:(%s)|([A-Za-z_]\w*)(?:\s+[A-Za-z_][\w\s]*?)?)\s+(%s)(?:\s+with\s+\w+)?\s*$" % (STR, STR))
+NARR = re.compile(r"^(%s)(?:\s+with\s+\w+)?\s*$" % STR)
+SKIP_BLOCK = re.compile(r"^(init\b.*|python\b.*|screen\b.*|transform\b.*|style\b.*|translate\b.*|image\s+[^=]+:\s*$|"
+                        r"layeredimage\b.*|testcase\b.*)$")
+
+
+def _unq(s):
+    s = s.strip()
+    if len(s) >= 2 and s[0] in "\"'" and s[-1] == s[0]:
+        s = s[1:-1]
+    return s.replace('\\"', '"').replace("\\'", "'").replace("\\n", "\n")
+
+
+def conv_expr(e):
+    e = e.strip()
+    parts = re.split(r"(%s)" % STR, e)
+    out = []
+    for i, p in enumerate(parts):
+        if i % 2:
+            out.append(p)
+            continue
+        p = re.sub(r"\bpersistent\.(\w+)", r"s.\1", p)
+        p = re.sub(r"\bTrue\b", "true", p)
+        p = re.sub(r"\bFalse\b", "false", p)
+        p = re.sub(r"\bNone\b", "0", p)
+
+        def rep(m):
+            w = m.group(0)
+            if w in ("and", "or", "not", "true", "false", "s", "f"):
+                return w
+            return "f." + w
+        p = re.sub(r"(?<![\w.])(?!s\.)([A-Za-z_]\w*)(?![\w(])", rep, p)
+        p = p.replace("f.s.", "s.")
+        out.append(p)
+    return "".join(out)
+
+
+class _Line(object):
+    __slots__ = ("ind", "text", "no")
+
+    def __init__(self, ind, text, no):
+        self.ind, self.text, self.no = ind, text, no
+
+
+def _lines(src):
+    out = []
+    for no, raw in enumerate(src.splitlines(), 1):
+        if not raw.strip():
+            continue
+        # 去掉行尾注释（不在字符串里的 #）
+        s, q, buf = raw.rstrip(), None, []
+        for ch in s:
+            if q:
+                buf.append(ch)
+                if ch == q:
+                    q = None
+            elif ch in "\"'":
+                q = ch
+                buf.append(ch)
+            elif ch == "#":
+                break
+            else:
+                buf.append(ch)
+        t = "".join(buf).rstrip()
+        if not t.strip():
+            continue
+        ind = len(t) - len(t.lstrip(" \t").replace("\t", "    ")) if "\t" in t else len(t) - len(t.lstrip())
+        out.append(_Line(ind, t.strip(), no))
+    return out
+
+
+class Importer(object):
+    def __init__(self, src, assets=None, entry="start"):
+        self.src, self.assets, self.entry = src, assets or Assets(), entry
+        self.chars, self.images, self.defaults = {}, {}, []
+        self.label_scene = {}
+        self.scenes = {}
+        self.warnings = []
+        self.unknown = {}
+        self.stats = {"files": 0, "says": 0, "choices": 0, "options": 0, "jumps": 0, "calls": 0, "sets": 0, "ifs": 0}
+        self.system_vars = set()
+        self.n = 0
+
+    def build(self):
+        files = []
+        for dp, _, fns in os.walk(self.src):
+            for fn in sorted(fns):
+                if fn.endswith(".rpy"):
+                    files.append(os.path.join(dp, fn))
+        if not files:
+            raise ValueError("没找到 .rpy")
+        parsed = [(os.path.splitext(os.path.relpath(p, self.src))[0].replace(os.sep, "/"), _lines(read_text(p))) for p in files]
+        for sid, ls in parsed:          # 第一遍：角色、图片、默认值、标签归属
+            for ln in ls:
+                if ln.ind == 0:
+                    self._decl(ln.text, sid)
+        for sid, ls in parsed:
+            self.stats["files"] += 1
+            b = SceneB(sid, sid, tags=["h"] if is_h_title(sid) else [])
+            self._block(b, ls, 0, len(ls), top=True)
+            b.add("end", kind="title", name="%s 走到文件末尾（Ren'Py 回到主菜单）" % sid)
+            self.scenes[sid] = b.data()
+        if self.entry not in self.label_scene:
+            raise ValueError("没有入口标签 label %s" % self.entry)
+        st = SceneB("__start__", "开始", tags=["system"])
+        st.add("label", name="top")
+        for name, val in self.defaults:
+            st.add("set", expr="%s = %s" % (name, val))
+        st.add("jump", scene=self.label_scene[self.entry], label=self.entry)
+        self.scenes["__start__"] = st.data()
+        meta = {"format": "galbook/1", "title": "", "engine": "renpy", "importer": "galbook-renpy-1",
+                "entry": {"scene": "__start__", "label": "top"},
+                "scenes": {k: {"title": v["title"], "tags": v["tags"]} for k, v in self.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, self.scenes
+
+    def _decl(self, t, sid):
+        m = re.match(r"^define\s+([A-Za-z_]\w*)\s*=\s*(?:Character|DynamicCharacter)\s*\(\s*(?:_\()?\s*(%s|None)" % STR, t)
+        if m:
+            self.chars[m.group(1)] = "" if m.group(2) == "None" else _unq(m.group(2))
+            return
+        m = re.match(r"^image\s+([^=]+?)\s*=\s*(%s)" % STR, t)
+        if m:
+            self.images[m.group(1).strip()] = _unq(m.group(2))
+            return
+        m = re.match(r"^(?:default|define)\s+(persistent\.)?([A-Za-z_]\w*)\s*=\s*(.+)$", t)
+        if m and "Character" not in t:
+            scope = "s." if m.group(1) else "f."
+            try:
+                val = conv_expr(m.group(3))
+                if scope == "s.":
+                    self.system_vars.add(m.group(2))
+                    val = "%s%s == 0 && %s || %s%s" % (scope, m.group(2), val, scope, m.group(2)) if False else val
+                if scope == "f.":
+                    self.defaults.append((scope + m.group(2), val))
+            except Exception:  # noqa
+                pass
+            return
+        m = re.match(r"^label\s+([A-Za-z_][\w.]*)\s*(\(.*\))?\s*:", t)
+        if m:
+            self.label_scene.setdefault(m.group(1), sid)
+
+    def _img(self, name):
+        name = name.strip()
+        f = self.images.get(name)
+        if f:
+            return self.assets.find(f) or f
+        return self.assets.find(name.replace(" ", "_")) or self.assets.find(name) or ("image:" + name)
+
+    def _end_of(self, ls, i, j):
+        ind = ls[i].ind
+        k = i + 1
+        while k < j and ls[k].ind > ind:
+            k += 1
+        return k
+
+    def _lab(self, kind):
+        self.n += 1
+        return "__%s%d" % (kind, self.n)
+
+    def _block(self, b, ls, i, j, top=False):
+        while i < j:
+            ln = ls[i]
+            t = ln.text
+            end = self._end_of(ls, i, j)
+            if top and ln.ind == 0 and not t.startswith("label "):
+                i = end
+                continue
+            m = re.match(r"^label\s+([A-Za-z_][\w.]*)", t)
+            if m:
+                b.add("label", name=m.group(1))
+                self._block(b, ls, i + 1, end)
+                i = end
+                continue
+            if SKIP_BLOCK.match(t) or t.startswith("python:") or t.startswith("$ python"):
+                i = end
+                continue
+            if re.match(r"^menu(\s+[A-Za-z_]\w*)?\s*:", t):
+                self._menu(b, ls, i, end)
+                i = end
+                continue
+            m = re.match(r"^(if|elif|else)\b\s*(.*?)\s*:$", t)
+            if m and m.group(1) == "if":
+                k = i
+                done = self._lab("ifend")
+                while k < j:
+                    mm = re.match(r"^(if|elif|else)\b\s*(.*?)\s*:$", ls[k].text)
+                    if not mm or ls[k].ind != ln.ind or (k != i and mm.group(1) == "if"):
+                        break
+                    e2 = self._end_of(ls, k, j)
+                    nxt = self._lab("ifnext")
+                    if mm.group(1) != "else":
+                        b.add("jump", label=nxt, **{"if": "!(%s)" % conv_expr(mm.group(2))})
+                        self.stats["ifs"] += 1
+                    self._block(b, ls, k + 1, e2)
+                    b.add("jump", label=done)
+                    b.add("label", name=nxt)
+                    k = e2
+                b.add("label", name=done)
+                i = k
+                continue
+            self._stmt(b, t)
+            i = end if end > i + 1 and t.endswith(":") else i + 1
+
+    def _menu(self, b, ls, i, end):
+        prompt = ""
+        opts = []
+        done = self._lab("menuend")
+        k = i + 1
+        bodies = []
+        while k < end:
+            t = ls[k].text
+            e2 = self._end_of(ls, k, end)
+            m = re.match(r"^(%s)\s*(?:if\s+(.+?))?\s*:$" % STR, t)
+            if m:
+                lb = self._lab("opt")
+                o = {"text": _unq(m.group(1)), "goto": {"label": lb}}
+                if m.group(2):
+                    o["if"] = conv_expr(m.group(2))
+                opts.append(o)
+                bodies.append((lb, k + 1, e2))
+            elif SAY.match(t) or NARR.match(t):
+                self._stmt(b, t)          # 菜单的提问句照常显示
+                prompt = _unq((SAY.match(t) or NARR.match(t)).group(3 if SAY.match(t) else 1))
+            elif t.startswith("set "):
+                pass
+            else:
+                self.unknown["menu:" + t.split()[0]] = self.unknown.get("menu:" + t.split()[0], 0) + 1
+            k = e2
+        b.add("choice", id=self._lab("menu"), prompt=prompt, options=opts)
+        self.stats["choices"] += 1
+        self.stats["options"] += len(opts)
+        for lb, s, e in bodies:
+            b.add("label", name=lb)
+            self._block(b, ls, s, e)
+            b.add("jump", label=done)
+        b.add("label", name=done)
+
+    def _stmt(self, b, t):
+        m = NARR.match(t)
+        if m:
+            b.add("say", who="", text=_unq(m.group(1)))
+            self.stats["says"] += 1
+            return
+        m = SAY.match(t)
+        if m and (m.group(1) or m.group(2) in self.chars or m.group(2) not in
+                  ("scene", "show", "hide", "play", "stop", "queue", "voice", "jump", "call", "return", "with", "window", "pause", "nvl")):
+            who = _unq(m.group(1)) if m.group(1) else self.chars.get(m.group(2), m.group(2))
+            b.add("say", who=who, text=_unq(m.group(3)))
+            self.stats["says"] += 1
+            return
+        w = t.split()
+        k = w[0] if w else ""
+        if k == "jump":
+            if w[1] == "expression":
+                self.warnings.append("jump expression 动态跳转没法静态解析：%s" % t)
+                return
+            sc = self.label_scene.get(w[1])
+            if sc is None:
+                self.warnings.append("跳到不存在的标签 %s" % w[1])
+                return
+            b.add("jump", scene=None if sc == b.id else sc, label=w[1])
+            self.stats["jumps"] += 1
+        elif k == "call":
+            if len(w) > 1 and w[1] != "screen" and w[1] in self.label_scene:
+                sc = self.label_scene[w[1]]
+                b.add("call", scene=None if sc == b.id else sc, label=w[1])
+                self.stats["calls"] += 1
+        elif k == "return":
+            b.add("ret")
+            b.add("end", kind="title", how="return（回到主菜单）")
+        elif k == "$":
+            e = t[1:].strip()
+            m2 = re.match(r"^(persistent\.)?([A-Za-z_]\w*)\s*(\+=|-=|=)\s*(.+)$", e)
+            if m2:
+                scope = "s." if m2.group(1) else "f."
+                if scope == "s.":
+                    self.system_vars.add(m2.group(2))
+                b.add("set", expr="%s%s %s %s" % (scope, m2.group(2), m2.group(3), conv_expr(m2.group(4))))
+                self.stats["sets"] += 1
+            else:
+                self.unknown["$"] = self.unknown.get("$", 0) + 1
+        elif k == "scene":
+            nm = re.sub(r"\s+(with|at|onlayer|behind|as)\b.*$", "", t[5:].strip())
+            b.add("sprites_clear")
+            b.add("bg", img=self._img(nm) if nm else None)
+        elif k == "show":
+            nm = re.sub(r"\s+(with|at|onlayer|behind|zorder|as)\b.*$", "", t[4:].strip())
+            pos = re.search(r"\bat\s+(\w+)", t)
+            if nm.startswith("bg "):
+                b.add("bg", img=self._img(nm))
+            elif nm and not nm.startswith("screen"):
+                b.add("sprite", name=nm.split()[0], img=self._img(nm), pos=pos.group(1) if pos else "center")
+        elif k == "hide":
+            if len(w) > 1:
+                b.add("sprite", name=w[1], img=None)
+        elif k == "play" and len(w) > 2 and w[1] == "music":
+            b.add("bgm", name=_unq(" ".join(w[2:]).split(" fade")[0]))
+        elif k == "stop" and len(w) > 1 and w[1] == "music":
+            b.add("bgm", name=None)
+        elif k in ("pass", "with", "pause", "window", "voice", "play", "stop", "queue", "nvl", "centered", "extend",
+                   "define", "default", "image", "init", "python", "camera", "renpy", "show_screen"):
+            pass
+        else:
+            self.unknown[k] = self.unknown.get(k, 0) + 1
+
+
+def build(src, assets=None, entry="start", **_):
+    return Importer(src, assets, entry or "start").build()

package/galbook/importers/textdump.py

@@ -0,0 +1,128 @@
+"""文本导出（线性）→ galbook/1：给「暂时没有专用导入器」的引擎兜底。
+
+很多引擎已经有成熟的文本抽取工具，抽出来的东西能直接喂进来，先「能读」：
+  * VNTextPatch（arcusmaximus，MIT）`extractlocal <游戏脚本目录> <输出目录>` 产出的 .json
+    格式：[{"name": "灯", "message": "「…」"}, {"names": ["A","B"], "message": "…"}, {"message": "旁白"}]
+    它支持 AdvHD / Artemis / BGI(Ethornell) / CatSystem2 / Majiro / Musica / RealLive / Softpal /
+    SystemNNN / YU-RIS / Silky's 等 20 多种引擎。
+  * 任意 .csv / .tsv：表头里有 name/speaker/character/who/名前/名字 之一 + text/message/line/original/原文/台词 之一
+  * .txt：一行一句，「名字：台词」「名字「台词」」或旁白（要 --engine textdump 显式指定，避免和 NScripter 0.txt 混）
+
+限制（如实说）：这些工具只抽文字，**不抽选项的跳转关系**（选项文字会混在台词里），所以导出来是一条直线，mode=linear。
+要真分支，需要该引擎的专用导入器（见 docs/IMPORTERS.md，欢迎贡献）。
+文件顺序：目录里有 order.txt 就按它（一行一个文件名），否则按文件名自然排序。"""
+import csv
+import io
+import json
+import os
+import re
+
+from .common import SceneB, Assets, read_text
+
+NAME = "textdump"
+NAME_KEYS = ("name", "speaker", "character", "chara", "who", "名前", "名字", "角色", "说话人")
+TEXT_KEYS = ("message", "text", "line", "original", "source", "原文", "台词", "本文", "セリフ")
+SAY = re.compile(r"^([^\s：:「『。，、！？!?,…—]{1,12})(?:：|:\s)\s*(.+)$")
+SAY_Q = re.compile(r"^([^\s：:「『（(。，、！？!?,…—]{1,12})([「『].*)$")
+
+
+def _natkey(s):
+    return [int(x) if x.isdigit() else x.lower() for x in re.split(r"(\d+)", s)]
+
+
+def _is_vntp_json(path):
+    try:
+        with open(path, encoding="utf-8-sig") as f:
+            d = json.load(f)
+        return isinstance(d, list) and (not d or (isinstance(d[0], dict) and "message" in d[0]))
+    except Exception:  # noqa
+        return False
+
+
+def detect(src):
+    js = []
+    for dp, _, fn in os.walk(src):
+        js += [os.path.join(dp, f) for f in fn if f.lower().endswith(".json")]
+        for f in fn:
+            if f.lower().endswith((".csv", ".tsv")):
+                head = read_text(os.path.join(dp, f)).splitlines()[:1]
+                if head and any(k in head[0].lower() for k in TEXT_KEYS):
+                    return True, "有带表头的台词表（.csv/.tsv）：按线性文本导入"
+    if js and any(_is_vntp_json(p) for p in js[:5]):
+        return True, "有 VNTextPatch 格式的 .json 文本导出：按线性文本导入（没有分支）"
+    return False, ""
+
+
+def _rows(path):
+    low = path.lower()
+    if low.endswith(".json"):
+        with open(path, encoding="utf-8-sig") as f:
+            d = json.load(f)
+        if not isinstance(d, list):
+            raise ValueError("%s 不是 JSON 数组" % path)
+        for e in d:
+            if isinstance(e, dict):
+                who = e.get("name") or "・".join(e.get("names") or []) or ""
+                yield who, e.get("message") or ""
+            elif isinstance(e, str):
+                yield "", e
+    elif low.endswith((".csv", ".tsv")):
+        txt = read_text(path)
+        rd = csv.reader(io.StringIO(txt), delimiter="\t" if low.endswith(".tsv") else ",")
+        head = [h.strip().lower() for h in next(rd, [])]
+        ni = next((head.index(k) for k in NAME_KEYS if k in head), None)
+        ti = next((head.index(k) for k in TEXT_KEYS if k in head), None)
+        if ti is None:
+            raise ValueError("%s 表头里找不到台词列（认：%s）" % (path, "/".join(TEXT_KEYS)))
+        for row in rd:
+            if len(row) > ti:
+                yield (row[ni].strip() if ni is not None and len(row) > ni else ""), row[ti]
+    else:
+        for line in read_text(path).splitlines():
+            s = line.strip()
+            if not s:
+                continue
+            m = SAY.match(s) or SAY_Q.match(s)
+            yield (m.group(1), m.group(2)) if m else ("", s)
+
+
+def build(src, assets=None, **_):
+    assets = assets or Assets()
+    exts = (".json", ".csv", ".tsv", ".txt")
+    files = []
+    for dp, _, fn in os.walk(src):
+        files += [os.path.relpath(os.path.join(dp, f), src) for f in fn if f.lower().endswith(exts) and f != "order.txt"]
+    op = os.path.join(src, "order.txt")
+    if os.path.isfile(op):
+        want = [l.strip() for l in read_text(op).splitlines() if l.strip()]
+        missing = [w for w in want if w not in files]
+        if missing:
+            raise ValueError("order.txt 里列的文件不存在：%s" % "、".join(missing[:5]))
+        files = want
+    else:
+        files.sort(key=_natkey)
+    if not files:
+        raise ValueError("%s 里没有 .json/.csv/.tsv/.txt 文本" % src)
+    scenes, order, stats = {}, [], {"files": len(files), "says": 0, "choices": 0, "options": 0, "jumps": 0}
+    for rel in files:
+        sid = re.sub(r"[\\/]", "__", os.path.splitext(rel)[0])
+        b = SceneB(sid, os.path.splitext(os.path.basename(rel))[0])
+        b.add("label", name="top")
+        for who, text in _rows(os.path.join(src, rel)):
+            text = (text or "").replace("\r\n", "\n").strip()
+            if text:
+                b.add("say", who=who.strip(), text=text)
+                stats["says"] += 1
+        scenes[sid] = b
+        order.append(sid)
+    for a, c in zip(order, order[1:]):
+        scenes[a].add("jump", scene=c, label="top")
+        stats["jumps"] += 1
+    scenes[order[-1]].add("end", kind="ending", name="全文完")
+    data = {k: v.data() for k, v in scenes.items()}
+    meta = {"format": "galbook/1", "title": "", "engine": "textdump", "importer": "galbook-textdump-1", "mode": "linear",
+            "entry": {"scene": order[0], "label": "top"},
+            "scenes": {k: {"title": v["title"], "tags": v["tags"]} for k, v in data.items()},
+            "routes": [], "system_vars": [], "stats": dict(stats, unknown=[], assets=assets.report()),
+            "warnings": ["线性文本：原作的选项文字（如果有）混在台词里，没有分支"]}
+    return meta, data