mirror of
https://github.com/MemeMeow-Studio/MemeMeow.git
synced 2026-10-18 13:42:36 +08:00
infstellar
64 lines
2.4 KiB
Python
64 lines
2.4 KiB
Python
# 图片文本向量的共享输入规则，供上传、名称更新、搜索和修复共同使用。
|
|
|
|
from __future__ import annotations
|
|
|
|
import hashlib
|
|
import re
|
|
import unicodedata
|
|
from dataclasses import dataclass
|
|
|
|
from backend.image_naming import SUPPORTED_IMAGE_EXTENSIONS
|
|
from backend.metadata import MAX_SEMANTIC_DOCUMENT_LENGTH, MemeContext, semantic_document
|
|
|
|
|
|
TEXT_INPUT_VERSION = "name-context-v1"
|
|
_HASH_NAME = re.compile(r"(?:[0-9a-f]{32}|[0-9a-f]{40}|[0-9a-f]{64})", re.IGNORECASE)
|
|
_SEPARATORS = re.compile(r"[\s_.-]*")
|
|
|
|
|
|
def filename_text(display_name: str) -> str:
|
|
"""规范化展示名称并过滤完整哈希或纯分隔符，返回可贡献的名称文本。"""
|
|
name = " ".join(unicodedata.normalize("NFKC", display_name).split())
|
|
while True:
|
|
extension = next((item for item in SUPPORTED_IMAGE_EXTENSIONS if name.lower().endswith(item)), None)
|
|
if extension is None:
|
|
break
|
|
name = name[:-len(extension)].rstrip()
|
|
if _HASH_NAME.fullmatch(name) or _SEPARATORS.fullmatch(name):
|
|
return ""
|
|
return name
|
|
|
|
|
|
@dataclass(frozen=True)
|
|
class TextInput:
|
|
"""当前名称和有效语境产生的文本、版本指纹及等待原因。"""
|
|
|
|
text: str
|
|
fingerprint: str | None
|
|
status: str
|
|
|
|
|
|
def build_text_input(display_name: str, context: object, context_status: str) -> TextInput:
|
|
"""根据当前名称及语境资格构造输入；图片身份验证由调用入口负责。"""
|
|
if context_status == "repair_required":
|
|
return TextInput("", None, "repair_required")
|
|
chunks = []
|
|
name = filename_text(display_name)
|
|
if name:
|
|
chunks.append(f"名称：{name}")
|
|
if context_status in {"partial", "ready"}:
|
|
try:
|
|
parsed = context if isinstance(context, MemeContext) else MemeContext.model_validate(context)
|
|
except (TypeError, ValueError):
|
|
# 无效语境不贡献文本；经过身份校验的名称仍有独立索引资格。
|
|
parsed = None
|
|
if parsed is not None:
|
|
document = semantic_document(parsed)
|
|
if document:
|
|
chunks.append(document)
|
|
document = "\n".join(chunks)[:MAX_SEMANTIC_DOCUMENT_LENGTH]
|
|
if not document:
|
|
return TextInput("", None, "waiting_for_text")
|
|
fingerprint = hashlib.sha256(f"{TEXT_INPUT_VERSION}\n{document}".encode("utf-8")).hexdigest()
|
|
return TextInput(document, fingerprint, "eligible")
|