style: format summarizer.py

This commit is contained in:
2026-07-07 05:55:40 -04:00
parent ba7fdaddf6
commit ca3ea1cdf2
+6 -18
View File
@@ -93,9 +93,7 @@ def _expand_anchor_ids(group_text: str) -> list[str]:
return ids return ids
def check_anchors( def check_anchors(summary: str, anchor_map: dict[str, str]) -> tuple[str, dict[str, int]]:
summary: str, anchor_map: dict[str, str]
) -> tuple[str, dict[str, int]]:
"""校验行号引注:非法行号删锚不删断言。 """校验行号引注:非法行号删锚不删断言。
参数: 参数:
@@ -235,9 +233,7 @@ async def _call_llm(
{"role": "system", "content": system_prompt}, {"role": "system", "content": system_prompt},
{"role": "user", "content": user_text}, {"role": "user", "content": user_text},
] ]
response = await llm.chat( response = await llm.chat(messages, session_id=session_id, parent_call_id=parent_call_id)
messages, session_id=session_id, parent_call_id=parent_call_id
)
return response.content return response.content
@@ -315,9 +311,7 @@ async def summarize_node(
verify_result = "跳过(调用失败)" verify_result = "跳过(调用失败)"
if anchor_map is not None: if anchor_map is not None:
raw_summary, asm_stats = assemble_anchored_output( raw_summary, asm_stats = assemble_anchored_output(raw_summary, anchor_map, assemble_mode)
raw_summary, anchor_map, assemble_mode
)
anchor_stats.update(asm_stats) anchor_stats.update(asm_stats)
result = f"[内容摘要] {raw_summary}\n[核实] {verify_result}" result = f"[内容摘要] {raw_summary}\n[核实] {verify_result}"
@@ -358,9 +352,7 @@ async def summarize_children(
lines = [] lines = []
for child in children_info: for child in children_info:
t_start, t_end = child["time_range"] t_start, t_end = child["time_range"]
lines.append( lines.append(f"- {child['id']} ({t_start:.0f}-{t_end:.0f}s): {child['summary']}")
f"- {child['id']} ({t_start:.0f}-{t_end:.0f}s): {child['summary']}"
)
children_text = "\n".join(lines) children_text = "\n".join(lines)
extract_input = f"问题: {question}\n\n{children_text}" extract_input = f"问题: {question}\n\n{children_text}"
@@ -417,9 +409,7 @@ async def _summarize_search_result(
返回: 返回:
"[内容摘要] {提取结果}\\n[核实] {验证结果}" 或错误信息。 "[内容摘要] {提取结果}\\n[核实] {验证结果}" 或错误信息。
""" """
extract_input = ( extract_input = f"问题: {question}\n\n以下是语义搜索命中的视频节点描述和字幕:\n{raw_text}"
f"问题: {question}\n\n以下是语义搜索命中的视频节点描述和字幕:\n{raw_text}"
)
try: try:
raw_summary = await _call_llm( raw_summary = await _call_llm(
llm, llm,
@@ -487,9 +477,7 @@ async def summarize_nodes_batch(
) )
return idx, node_id, summary return idx, node_id, summary
tasks = [ tasks = [_worker(i, nid, text) for i, (nid, text, _) in enumerate(items)]
_worker(i, nid, text) for i, (nid, text, _) in enumerate(items)
]
results_raw = await asyncio.gather(*tasks) results_raw = await asyncio.gather(*tasks)
results: dict[int, tuple[str, str]] = {} results: dict[int, tuple[str, str]] = {}