File size: 7,333 Bytes
7245599
 
 
 
c8b5814
 
6d0bcdd
7245599
 
 
 
 
c7b2544
 
7245599
 
 
 
ec10f07
 
7245599
 
 
 
6d0bcdd
 
c8b5814
 
 
7245599
 
 
 
 
 
 
 
6d0bcdd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c8b5814
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6d0bcdd
 
 
 
 
 
 
 
 
c8b5814
6d0bcdd
 
 
 
 
 
 
 
 
c8b5814
 
 
 
6d0bcdd
 
7245599
ec10f07
 
7245599
 
 
 
 
 
 
 
 
c8b5814
 
 
 
 
 
 
 
 
 
 
c7b2544
7245599
 
 
 
c7b2544
7245599
 
 
 
 
 
 
 
 
 
 
 
 
c7b2544
 
 
 
7245599
 
 
 
c7b2544
7245599
c7b2544
 
 
 
 
7245599
c7b2544
 
 
 
 
 
 
 
 
 
 
 
7245599
6d0bcdd
 
 
 
 
 
c7b2544
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7245599
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
from __future__ import annotations

import uuid
from dataclasses import dataclass
from typing import Any
from typing import AsyncIterator
from typing import Iterable

from .bootstrap import bootstrap_environment

bootstrap_environment()

from google.adk.agents.run_config import RunConfig
from google.adk.agents.run_config import StreamingMode
from google.adk.runners import Runner
from google.adk.sessions import InMemorySessionService
from google.genai import types

from .agent import FACT_RETRIEVER
from .agent import PERSONA_RETRIEVER
from .agent import root_agent


APP_NAME = "megumin_rag_app"
MAX_TURNS_IN_CONTEXT = 6
SUMMARY_MAX_CHARS = 800
SUMMARY_USER_LIMIT = 3
SUMMARY_ASSISTANT_LIMIT = 2
SUMMARY_ITEM_CHARS = 42


@dataclass
class ChatServices:
    runner: Runner
    session_service: InMemorySessionService


def _event_texts(events: Iterable) -> list[str]:
    lines: list[str] = []
    for event in events:
        if not getattr(event, "content", None) or not event.content.parts:
            continue
        text_parts = [
            getattr(part, "text", None)
            for part in event.content.parts
            if getattr(part, "text", None)
        ]
        if not text_parts:
            continue
        author = "user" if event.author == "user" else "assistant"
        lines.append(f"{author}: {' '.join(text_parts).strip()}")
    return lines


def _compact_summary_item(text: str, limit: int = SUMMARY_ITEM_CHARS) -> str:
    compact = " ".join(str(text or "").split()).strip()
    if len(compact) <= limit:
        return compact
    return compact[: limit - 3].rstrip() + "..."


def _parse_summary_map(value: Any) -> dict[str, list[str]]:
    if not isinstance(value, dict):
        return {
            "user_topics": [],
            "assistant_points": [],
        }
    return {
        "user_topics": [
            str(item) for item in value.get("user_topics", []) if str(item).strip()
        ],
        "assistant_points": [
            str(item)
            for item in value.get("assistant_points", [])
            if str(item).strip()
        ],
    }


def _merge_unique_tail(previous: list[str], additions: list[str], limit: int) -> list[str]:
    merged: list[str] = []
    for item in [*previous, *additions]:
        if not item or item in merged:
            continue
        merged.append(item)
    return merged[-limit:]


def _compress_summary(
    previous_summary_map: Any,
    new_lines: list[str],
) -> dict[str, list[str]]:
    summary_map = _parse_summary_map(previous_summary_map)
    user_lines = [
        _compact_summary_item(line.removeprefix("user:").strip())
        for line in new_lines
        if line.startswith("user:")
    ]
    assistant_lines = [
        _compact_summary_item(line.removeprefix("assistant:").strip())
        for line in new_lines
        if line.startswith("assistant:")
    ]
    summary_map["user_topics"] = _merge_unique_tail(
        summary_map["user_topics"],
        user_lines,
        SUMMARY_USER_LIMIT,
    )
    summary_map["assistant_points"] = _merge_unique_tail(
        summary_map["assistant_points"],
        assistant_lines,
        SUMMARY_ASSISTANT_LIMIT,
    )
    return summary_map


def _render_summary(summary_map: dict[str, list[str]]) -> str:
    chunks: list[str] = []
    if summary_map.get("user_topics"):
        chunks.append("user_topics: " + " ; ".join(summary_map["user_topics"]))
    if summary_map.get("assistant_points"):
        chunks.append("assistant_points: " + " ; ".join(summary_map["assistant_points"]))
    rendered = "\n".join(chunks).strip()
    if len(rendered) <= SUMMARY_MAX_CHARS:
        return rendered
    return rendered[: SUMMARY_MAX_CHARS - 3].rstrip() + "..."


def _trim_session_history(
    services: ChatServices,
    *,
    user_id: str,
    session_id: str,
) -> None:
    session_store = services.session_service.sessions
    storage_session = session_store.get(APP_NAME, {}).get(user_id, {}).get(session_id)
    if storage_session is None:
        return

    max_events = MAX_TURNS_IN_CONTEXT * 2
    if len(storage_session.events) <= max_events:
        return

    overflow = storage_session.events[:-max_events]
    storage_session.events = storage_session.events[-max_events:]
    previous_summary_map = storage_session.state.get("conversation_summary_map", {})
    summary_map = _compress_summary(previous_summary_map, _event_texts(overflow))
    storage_session.state["conversation_summary_map"] = summary_map
    storage_session.state["conversation_summary"] = _render_summary(summary_map)


def create_chat_services() -> ChatServices:
    PERSONA_RETRIEVER.warmup()
    FACT_RETRIEVER.warmup()
    session_service = InMemorySessionService()
    runner = Runner(
        agent=root_agent,
        app_name=APP_NAME,
        session_service=session_service,
    )
    return ChatServices(runner=runner, session_service=session_service)


def _extract_text(event: Any) -> str:
    if not getattr(event, "content", None) or not getattr(event.content, "parts", None):
        return ""
    texts = [
        getattr(part, "text", "")
        for part in event.content.parts
        if getattr(part, "text", "")
    ]
    return "".join(texts).strip()


async def stream_chat(
    user_message: str,
    services: ChatServices,
    session_id: str | None = None,
    user_id: str = "local-user",
) -> AsyncIterator[tuple[str, str]]:
    active_session_id = session_id or str(uuid.uuid4())
    existing_session = await services.session_service.get_session(
        app_name=APP_NAME,
        user_id=user_id,
        session_id=active_session_id,
    )
    if existing_session is None:
        await services.session_service.create_session(
            app_name=APP_NAME,
            user_id=user_id,
            session_id=active_session_id,
        )

    streamed_text = ""
    final_text = ""
    run_config = RunConfig(streaming_mode=StreamingMode.SSE)

    async for event in services.runner.run_async(
        user_id=user_id,
        session_id=active_session_id,
        new_message=types.UserContent(parts=[types.Part(text=user_message)]),
        run_config=run_config,
    ):
        if getattr(event, "author", None) == "user":
            continue

        text = _extract_text(event)
        if not text:
            continue

        if getattr(event, "partial", None) is True:
            streamed_text += text
            yield streamed_text, active_session_id
            continue

        if getattr(event, "is_final_response", None) and event.is_final_response():
            final_text = text

    if final_text and final_text != streamed_text:
        streamed_text = final_text
        yield streamed_text, active_session_id

    _trim_session_history(
        services,
        user_id=user_id,
        session_id=active_session_id,
    )


async def chat_once(
    user_message: str,
    services: ChatServices,
    session_id: str | None = None,
    user_id: str = "local-user",
) -> tuple[str, str]:
    last_text = ""
    active_session_id = session_id or str(uuid.uuid4())
    async for chunk_text, active_session_id in stream_chat(
        user_message=user_message,
        services=services,
        session_id=active_session_id,
        user_id=user_id,
    ):
        last_text = chunk_text
    return last_text, active_session_id