Losita/mai-bot
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat：升级jargon，提取更快速，更精准

Browse Source
This commit is contained in:
SengokuCola
2025-11-25 01:58:31 +08:00
parent a1dd26d578
commit b684a95fbc
5 changed files with 297 additions and 52 deletions
Download Patch File Download Diff File Expand all files Collapse all files

61
src/chat/utils/chat_message_builder.py
View File

@@ -352,6 +352,7 @@ def _build_readable_messages_internal(
pic_counter: int = 1,
show_pic: bool = True,
message_id_list: Optional[List[Tuple[str, DatabaseMessages]]] = None,
pic_single: bool = False,
) -> Tuple[str, List[Tuple[float, str, str]], Dict[str, str], int]:
# sourcery skip: use-getitem-for-re-match-groups
"""
@@ -378,6 +379,7 @@ def _build_readable_messages_internal(
if pic_id_mapping is None:
pic_id_mapping = {}
current_pic_counter = pic_counter
pic_description_cache: Dict[str, str] = {}
# 创建时间戳到消息ID的映射，用于在消息前添加[id]标识符
timestamp_to_id_mapping: Dict[float, str] = {}
@@ -400,6 +402,17 @@ def _build_readable_messages_internal(
nonlocal current_pic_counter
nonlocal pic_counter
pic_id = match.group(1)
if pic_single:
if pic_id not in pic_description_cache:
description = "内容正在阅读，请稍等"
try:
image = Images.get_or_none(Images.image_id == pic_id)
if image and image.description:
description = image.description
except Exception:
pass
pic_description_cache[pic_id] = description
return f"[图片：{pic_description_cache[pic_id]}]"
if pic_id not in pic_id_mapping:
pic_id_mapping[pic_id] = f"图片{current_pic_counter}"
current_pic_counter += 1
@@ -603,6 +616,7 @@ async def build_readable_messages_with_list(
replace_bot_name: bool = True,
timestamp_mode: str = "relative",
truncate: bool = False,
pic_single: bool = False,
) -> Tuple[str, List[Tuple[float, str, str]]]:
"""
将消息列表转换为可读的文本格式，并返回原始(时间戳, 昵称, 内容)列表。
@@ -613,10 +627,16 @@ async def build_readable_messages_with_list(
replace_bot_name,
timestamp_mode,
truncate,
pic_id_mapping=None,
pic_counter=1,
show_pic=True,
message_id_list=None,
pic_single=pic_single,
)
if pic_mapping_info := build_pic_mapping_info(pic_id_mapping):
formatted_string = f"{pic_mapping_info}\n\n{formatted_string}"
if not pic_single:
if pic_mapping_info := build_pic_mapping_info(pic_id_mapping):
formatted_string = f"{pic_mapping_info}\n\n{formatted_string}"
return formatted_string, details_list
@@ -630,6 +650,7 @@ def build_readable_messages_with_id(
show_actions: bool = False,
show_pic: bool = True,
remove_emoji_stickers: bool = False,
pic_single: bool = False,
) -> Tuple[str, List[Tuple[str, DatabaseMessages]]]:
"""
将消息列表转换为可读的文本格式，并返回原始(时间戳, 昵称, 内容)列表。
@@ -647,6 +668,7 @@ def build_readable_messages_with_id(
read_mark=read_mark,
message_id_list=message_id_list,
remove_emoji_stickers=remove_emoji_stickers,
pic_single=pic_single,
)
return formatted_string, message_id_list
@@ -662,6 +684,7 @@ def build_readable_messages(
show_pic: bool = True,
message_id_list: Optional[List[Tuple[str, DatabaseMessages]]] = None,
remove_emoji_stickers: bool = False,
pic_single: bool = False,
) -> str: # sourcery skip: extract-method
"""
将消息列表转换为可读的文本格式。
@@ -769,14 +792,14 @@ def build_readable_messages(
truncate,
show_pic=show_pic,
message_id_list=message_id_list,
pic_single=pic_single,
)
# 生成图片映射信息并添加到最前面
pic_mapping_info = build_pic_mapping_info(pic_id_mapping)
if pic_mapping_info:
return f"{pic_mapping_info}\n\n{formatted_string}"
else:
return formatted_string
if not pic_single:
pic_mapping_info = build_pic_mapping_info(pic_id_mapping)
if pic_mapping_info:
return f"{pic_mapping_info}\n\n{formatted_string}"
return formatted_string
else:
# 按 read_mark 分割消息
messages_before_mark = [msg for msg in copy_messages if (msg.time or 0) <= read_mark]
@@ -796,6 +819,7 @@ def build_readable_messages(
pic_counter,
show_pic=show_pic,
message_id_list=message_id_list,
pic_single=pic_single,
)
formatted_after, _, pic_id_mapping, _ = _build_readable_messages_internal(
messages_after_mark,
@@ -806,15 +830,19 @@ def build_readable_messages(
pic_counter,
show_pic=show_pic,
message_id_list=message_id_list,
pic_single=pic_single,
)
read_mark_line = "\n--- 以上消息是你已经看过，请关注以下未读的新消息---\n"
# 生成图片映射信息
if pic_id_mapping:
pic_mapping_info = f"图片信息：\n{build_pic_mapping_info(pic_id_mapping)}\n聊天记录信息：\n"
if not pic_single:
if pic_id_mapping:
pic_mapping_info = f"图片信息：\n{build_pic_mapping_info(pic_id_mapping)}\n聊天记录信息：\n"
else:
pic_mapping_info = "聊天记录信息：\n"
else:
pic_mapping_info = "聊天记录信息：\n"
pic_mapping_info = ""
# 组合结果
result_parts = []
@@ -832,7 +860,7 @@ def build_readable_messages(
return "".join(result_parts)
async def build_anonymous_messages(messages: List[DatabaseMessages]) -> str:
async def build_anonymous_messages(messages: List[DatabaseMessages], show_ids: bool = False) -> str:
"""
构建匿名可读消息，将不同人的名称转为唯一占位符（A、B、C...），bot自己用SELF。
处理 回复<aaa:bbb> 和 @<aaa:bbb> 字段，将bbb映射为匿名占位符。
@@ -889,7 +917,7 @@ async def build_anonymous_messages(messages: List[DatabaseMessages]) -> str:
current_char += 1
return person_map[person_id]
for msg in messages:
for i, msg in enumerate(messages):
try:
platform = msg.chat_info.platform
user_id = msg.user_info.user_id
@@ -910,7 +938,12 @@ async def build_anonymous_messages(messages: List[DatabaseMessages]) -> str:
content = replace_user_references(content, platform, anon_name_resolver, replace_bot_name=False)
header = f"{anon_name}说 "
# 构建消息头，如果启用show_ids则添加序号
if show_ids:
header = f"[{i + 1}] {anon_name}说 "
else:
header = f"{anon_name}说 "
output_lines.append(header)
stripped_line = content.strip()
if stripped_line: