1hew
161b87e870
- feat(conversion): 新增 Any to Any 通配透传节点 - feat(int): 新增 Int Mask/Audio/Text Count，并统一计数共享层 - feat(logic): Multi Switch Select 支持 a/all 全选 - feat(io): Load PS 预览默认开启并抑制 psd_tools INFO 噪音；Save Audio 默认 mp3，播放器兼容 Nodes 2.0 - fix(multi): Multi String Join 的 string_1 改为纯插槽，修复动态端口错位 - refactor(shared): 复用 parse_pad_color / pad_to_rgb - docs: 同步中英文节点文档、README 列表与 v4.1.0 日志 - test: 新增/更新相关回归测试
93 lines
2.7 KiB
Python
93 lines
2.7 KiB
Python
"""Shared counting helpers for the Int * Count nodes.
|
|
|
|
Presence is exactly the ``is_empty`` rule used by ``Any Empty Bool``:
|
|
``None``, empty or all-zero tensors, blank strings, ``0`` / ``False`` scalars, and
|
|
empty containers are absent. Counting is per item: lists/tuples are flattened
|
|
first (an OUTPUT_IS_LIST upstream contributes one item per element), and an
|
|
IMAGE/MASK batch contributes one count per frame.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
from typing import Any, Iterator
|
|
|
|
import torch
|
|
|
|
from ._logic_ops import is_empty
|
|
|
|
_IMAGE_BATCH_NDIM = 4
|
|
_MASK_BATCH_NDIM = 3
|
|
|
|
|
|
def iter_count_items(value: Any) -> Iterator[Any]:
|
|
"""Yield leaf values, flattening nested lists/tuples from list outputs."""
|
|
if isinstance(value, (list, tuple)):
|
|
for item in value:
|
|
yield from iter_count_items(item)
|
|
return
|
|
yield value
|
|
|
|
|
|
def _count_tensor_items(value: Any, batch_ndim: int) -> int:
|
|
"""Count non-empty entries of an IMAGE/MASK tensor, frame by frame."""
|
|
if not isinstance(value, torch.Tensor):
|
|
return 0
|
|
if value.ndim < batch_ndim:
|
|
return 0 if is_empty(value) else 1
|
|
frames = int(value.shape[0])
|
|
if frames <= 0:
|
|
return 0
|
|
return sum(1 for index in range(frames) if not is_empty(value[index : index + 1]))
|
|
|
|
|
|
def count_image_items(value: Any) -> int:
|
|
return sum(
|
|
_count_tensor_items(item, _IMAGE_BATCH_NDIM)
|
|
for item in iter_count_items(value)
|
|
)
|
|
|
|
|
|
def count_mask_items(value: Any) -> int:
|
|
return sum(
|
|
_count_tensor_items(item, _MASK_BATCH_NDIM)
|
|
for item in iter_count_items(value)
|
|
)
|
|
|
|
|
|
def count_video_items(value: Any) -> int:
|
|
return sum(1 for item in iter_count_items(value) if not is_empty(item))
|
|
|
|
|
|
def count_audio_items(value: Any) -> int:
|
|
return sum(1 for item in iter_count_items(value) if not is_empty(item))
|
|
|
|
|
|
def count_text_items(value: Any) -> int:
|
|
"""Count non-empty STRING / INT / FLOAT leaves; other types are ignored."""
|
|
return sum(
|
|
1
|
|
for item in iter_count_items(value)
|
|
if isinstance(item, (str, int, float)) and not is_empty(item)
|
|
)
|
|
|
|
|
|
def collect_prefixed_values(
|
|
prefix: str,
|
|
first_value: Any,
|
|
kwargs: dict[str, Any],
|
|
) -> list[Any]:
|
|
"""Return the ``<prefix>1..N`` values ordered by numeric suffix."""
|
|
indexed: list[tuple[int, Any]] = [(1, first_value)]
|
|
for key, value in kwargs.items():
|
|
if not isinstance(key, str) or not key.startswith(prefix):
|
|
continue
|
|
suffix = key[len(prefix) :]
|
|
if not suffix.isdigit():
|
|
continue
|
|
index = int(suffix)
|
|
if index == 1:
|
|
continue
|
|
indexed.append((index, value))
|
|
indexed.sort(key=lambda item: item[0])
|
|
return [value for _, value in indexed]
|