comate-cli 0.10.18__tar.gz → 0.10.18b2__tar.gz

{comate_cli-0.10.18 → comate_cli-0.10.18b2}/CHANGELOG.md

@@ -2,6 +2,19 @@
 
 ## Unreleased
 
+### Added
+
+- 输入框贴图：Ctrl+V（Windows Alt+V）读取系统剪贴板图片并挂为
+  `[Pasted image #k: label]` chip；粘贴或拖拽图片文件路径会自动升格为 chip。
+  手打 `@图片` 仍走 Read 工具；busy 时两种输入通道同样可用，提交后进入队列。
+- rewind 图片重挂：`/rewind`、Esc 打断召回、提交秒撤回时，文字与图片一起原样
+  回到输入框；`--resume` 后依然有效。
+- 队列撤回图片重挂：busy 排队的带图消息经 ↑/Esc 取回编辑时，文字与图片会一起
+  回到输入框；Ctrl+C 打断撤回仅在输入框为空时回填，已有草稿时会明确提示重新粘贴，
+  不再把图片压成 `[image]` 后静默丢失。
+- 非 vision 模型贴图时会前置拒绝并明示原因，不再静默丢图；slash command 或
+  feedback 草稿携带图片时会明确提示已丢弃。
+
 ### Performance
 
 - 启动重排：交互模式在参数解析后立即打印 logo，preflight、Agent 构造与 TUI 准备阶段
@@ -48,6 +61,9 @@
 
 ### Fixed
 
+- 修复：大段粘贴折叠不再遮蔽整段图片路径的 chip 升格（混合内容仍折叠，为已知限制）
+- 修复：图片 chip 计数器跨重启从输入历史播种，历史召回不再把新图错挂到旧提示；孤儿图片丢弃改为显式提示
+- 修复：BMP 剪贴板图片下沉到 SDK 统一预处理入口，炸弹图在安全边界内被拒且用户可见报错
 - 启动重排回归：`Preparing interface…` spinner（rich Live 会临时把 `sys.stdout`
   换成非 tty 的 FileProxy）窗口内构造了 prompt_toolkit Application，AppSession
   把降级 `PlainTextOutput` 永久缓存，导致 input 区域无法加载、底部带被反复追加进

{comate_cli-0.10.18 → comate_cli-0.10.18b2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: comate-cli
-Version: 0.10.18
+Version: 0.10.18b2
 Summary: Comate terminal CLI built on comate-agent-sdk
 Project-URL: Homepage, https://github.com/AndyLee1024/agent-sdk
 Project-URL: Repository, https://github.com/AndyLee1024/agent-sdk
@@ -15,7 +15,7 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Requires-Python: <3.14,>=3.11
 Requires-Dist: charset-normalizer==3.4.7
-Requires-Dist: comate-agent-sdk<0.12,>=0.11.14
+Requires-Dist: comate-agent-sdk==0.11.14b1
 Requires-Dist: concurrent-log-handler>=0.9.25
 Requires-Dist: curl-cffi==0.13.0
 Requires-Dist: httpx<0.29.0,>=0.28.1

comate_cli-0.10.18b2/comate_cli/terminal_agent/clipboard_image.py

@@ -0,0 +1,197 @@
+"""系统剪贴板图片读取 adapter。"""
+
+from __future__ import annotations
+
+import asyncio
+import logging
+import shutil
+import sys
+import tempfile
+from contextlib import suppress
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Awaitable, Callable, Sequence
+
+logger = logging.getLogger(__name__)
+
+_CLIPBOARD_TIMEOUT_S = 8.0
+_PROCESS_CLEANUP_TIMEOUT_S = 0.5
+
+CommandRunner = Callable[[Sequence[str]], Awaitable[tuple[int, bytes]]]
+
+
+@dataclass(frozen=True)
+class ClipboardImage:
+    data: bytes
+    suggested_label: str
+
+
+async def _terminate_process(proc: asyncio.subprocess.Process) -> None:
+    if proc.returncode is None:
+        with suppress(ProcessLookupError):
+            proc.kill()
+    try:
+        await asyncio.wait_for(
+            proc.communicate(),
+            timeout=_PROCESS_CLEANUP_TIMEOUT_S,
+        )
+    except ProcessLookupError:
+        return
+    except TimeoutError:
+        logger.warning("clipboard subprocess did not exit after kill")
+
+
+async def _run_command(argv: Sequence[str]) -> tuple[int, bytes]:
+    proc = await asyncio.create_subprocess_exec(
+        *argv,
+        stdin=asyncio.subprocess.DEVNULL,
+        stdout=asyncio.subprocess.PIPE,
+        stderr=asyncio.subprocess.DEVNULL,
+    )
+    try:
+        stdout, _ = await asyncio.wait_for(
+            proc.communicate(),
+            timeout=_CLIPBOARD_TIMEOUT_S,
+        )
+    except TimeoutError:
+        await _terminate_process(proc)
+        return 1, b""
+    except asyncio.CancelledError:
+        await _terminate_process(proc)
+        raise
+    return int(proc.returncode or 0), stdout or b""
+
+
+async def read_clipboard_image(
+    *,
+    runner: CommandRunner | None = None,
+    platform: str | None = None,
+) -> ClipboardImage | None:
+    run = runner or _run_command
+    plat = platform or sys.platform
+    if plat == "darwin":
+        return await _read_darwin(run)
+    if plat.startswith("linux"):
+        return await _read_linux(run)
+    if plat == "win32":
+        return await _read_win32(run)
+    return None
+
+
+async def _read_linux(run: CommandRunner) -> ClipboardImage | None:
+    for argv in (
+        ("xclip", "-selection", "clipboard", "-t", "image/png", "-o"),
+        ("wl-paste", "--type", "image/png"),
+        ("xclip", "-selection", "clipboard", "-t", "image/bmp", "-o"),
+        ("wl-paste", "--type", "image/bmp"),
+    ):
+        try:
+            code, data = await run(argv)
+        except OSError:
+            continue
+        if code == 0 and data:
+            return ClipboardImage(data=data, suggested_label="clipboard.png")
+    return None
+
+
+def _create_temp_target() -> tuple[Path, Path]:
+    root = Path(tempfile.mkdtemp(prefix="comate_cli_clipboard_"))
+    return root, root / "image.png"
+
+
+def _remove_temp_root(root: Path) -> None:
+    shutil.rmtree(root)
+
+
+async def _cleanup_temp_root(root: Path) -> None:
+    try:
+        await asyncio.to_thread(_remove_temp_root, root)  # to_thread: bounded-local, one finite private clipboard directory cleanup
+    except FileNotFoundError:
+        return
+    except OSError:
+        logger.warning("failed to clean clipboard temporary directory", exc_info=True)
+
+
+async def _read_saved_file(path: Path) -> bytes:
+    try:
+        return await asyncio.to_thread(path.read_bytes)  # to_thread: bounded-local, one finite local clipboard image read
+    except OSError:
+        return b""
+
+
+async def _read_darwin(run: CommandRunner) -> ClipboardImage | None:
+    try:
+        temp_root, target = await asyncio.to_thread(_create_temp_target)  # to_thread: bounded-local, one private clipboard directory creation
+    except OSError:
+        return None
+
+    argv: list[str] = ["osascript"]
+    for line in (
+        "on run argv",
+        "set target_path to item 1 of argv",
+        "set png_data to (the clipboard as «class PNGf»)",
+        "set fp to open for access POSIX file target_path with write permission",
+        "try",
+        "set eof fp to 0",
+        "write png_data to fp starting at eof",
+        "on error err_msg number err_num",
+        "try",
+        "close access fp",
+        "end try",
+        "error err_msg number err_num",
+        "end try",
+        "close access fp",
+        "end run",
+    ):
+        argv.extend(["-e", line])
+    argv.extend(["--", str(target)])
+    try:
+        try:
+            code, _ = await run(tuple(argv))
+        except OSError:
+            return None
+        if code != 0:
+            return None
+        data = await _read_saved_file(target)
+        if not data:
+            return None
+        return ClipboardImage(data=data, suggested_label="clipboard.png")
+    finally:
+        await _cleanup_temp_root(temp_root)
+
+
+async def _read_win32(run: CommandRunner) -> ClipboardImage | None:
+    try:
+        temp_root, target = await asyncio.to_thread(_create_temp_target)  # to_thread: bounded-local, one private clipboard directory creation
+    except OSError:
+        return None
+
+    script = (
+        "& { param([string]$Target) "
+        "$img = Get-Clipboard -Format Image; "
+        "if ($null -ne $img) { try { "
+        "$img.Save($Target, [System.Drawing.Imaging.ImageFormat]::Png) "
+        "} finally { $img.Dispose() } } }"
+    )
+    try:
+        try:
+            code, _ = await run(
+                (
+                    "powershell",
+                    "-NoProfile",
+                    "-NonInteractive",
+                    "-Command",
+                    script,
+                    str(target),
+                )
+            )
+        except OSError:
+            return None
+        if code != 0:
+            return None
+        data = await _read_saved_file(target)
+        if not data:
+            return None
+        return ClipboardImage(data=data, suggested_label="clipboard.png")
+    finally:
+        await _cleanup_temp_root(temp_root)

{comate_cli-0.10.18 → comate_cli-0.10.18b2}/comate_cli/terminal_agent/file_ref_hint.py

@@ -13,12 +13,14 @@ from comate_cli.terminal_agent.custom_slash_commands import FILE_REF_PATTERN
 
 # 超过该体积的文件不再逐行统计，避免大文件计数拖慢渲染。
 _FILE_REF_MAX_COUNT_BYTES = 10 * 1024 * 1024  # 10 MB
+_IMAGE_SUFFIXES = {".png", ".jpg", ".jpeg", ".gif", ".webp"}
 
 
 def build_file_ref_hints(text: str, project_root: Path | None) -> list[str]:
     """扫描 ``text`` 中的 ``@path`` 引用，返回展示用提示行列表。
 
-    每条形如 ``Read <path>  (<N> lines)`` 或 ``Listed directory <path>``；
+    每条形如 ``Read <path>  (<N> lines)``、``Read image <path>  (<N>KB)``
+    或 ``Listed directory <path>``；
     无效 / 不存在 / 不可访问的引用会被跳过。``project_root`` 为 ``None`` 时返回空列表。
     """
     if project_root is None:
@@ -32,6 +34,15 @@ def build_file_ref_hints(text: str, project_root: Path | None) -> list[str]:
                 hints.append(f"Listed directory {raw_path}")
                 continue
             if candidate.is_file():
+                if candidate.suffix.lower() in _IMAGE_SUFFIXES:
+                    hint = f"Read image {raw_path}"
+                    try:
+                        size_kb = candidate.stat().st_size / 1024
+                        hint += f"  ({size_kb:.1f}KB)"
+                    except OSError:
+                        pass
+                    hints.append(hint)
+                    continue
                 hint = f"Read {raw_path}"
                 if candidate.stat().st_size <= _FILE_REF_MAX_COUNT_BYTES:
                     try:

comate_cli-0.10.18b2/comate_cli/terminal_agent/image_chips.py

@@ -0,0 +1,74 @@
+"""输入框图片 chip 的纯函数工具。
+
+chip 字面量固定为 ``[Pasted image #k: label]``。本地编号只负责输入框
+字面量与图片 payload 对齐；SDK canonical ``[Image #N]`` 编号仍由 runtime
+会话级计数器负责。
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Sequence
+
+from comate_agent_sdk.llm.messages import ContentPartImageParam
+
+IMAGE_CHIP_RE = re.compile(r"\[Pasted image #([0-9]+): ([^\]\r\n]+)\]")
+
+
+def format_image_chip(number: int, label: str) -> str:
+    safe_label = " ".join(label.replace("]", "").splitlines()).strip() or "image"
+    return f"[Pasted image #{number}: {safe_label}]"
+
+
+def chip_numbers_in_text(text: str) -> list[int]:
+    """按出现顺序返回 chip 编号；重复编号只保留首个。"""
+    seen: set[int] = set()
+    ordered: list[int] = []
+    for match in IMAGE_CHIP_RE.finditer(text):
+        label = match.group(2)
+        if label != label.strip() or len(label.splitlines()) != 1:
+            continue
+        number = int(match.group(1))
+        if number in seen:
+            continue
+        seen.add(number)
+        ordered.append(number)
+    return ordered
+
+
+def collect_parts_for_submit(
+    text: str,
+    chips: dict[int, ContentPartImageParam],
+) -> tuple[ContentPartImageParam, ...]:
+    """按 chip 字面量顺序收集图片，过滤字面量已删除的孤儿。"""
+    return tuple(
+        chips[number] for number in chip_numbers_in_text(text) if number in chips
+    )
+
+
+def align_restored_chips(
+    text: str,
+    parts: Sequence[ContentPartImageParam],
+) -> tuple[str, dict[int, ContentPartImageParam], int]:
+    """按字面量顺序重挂图片，并把无字面量的多余图片追加为新 chip。"""
+    numbers = chip_numbers_in_text(text)
+    chips: dict[int, ContentPartImageParam] = {}
+    remaining = list(parts)
+    for number in numbers:
+        if not remaining:
+            break
+        chips[number] = remaining.pop(0)
+
+    max_used = max(numbers, default=0)
+    appended: list[str] = []
+    for part in remaining:
+        max_used += 1
+        label = part.image_url.label or "image"
+        chips[max_used] = part
+        appended.append(format_image_chip(max_used, label))
+
+    final_text = text
+    if appended:
+        separator = "" if (not final_text or final_text.endswith((" ", "\n"))) else " "
+        final_text = f"{final_text}{separator}{' '.join(appended)}"
+    return final_text, chips, max_used