File size: 11,282 Bytes
f98dc3e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
from __future__ import annotations

from typing import Any, Dict, List, Literal, Optional, Union

from pydantic import BaseModel, Field, ConfigDict, field_validator, model_validator


class StrictSchemaModel(BaseModel):
    model_config = ConfigDict(extra="forbid")


class RetrieveToolArgs(StrictSchemaModel):
    question: str = Field(..., min_length=1, description="Canonical retrieval question for memory search.")

    @field_validator("question")
    @classmethod
    def _normalize_question(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("question must not be empty")
        return text


class RetrieveToolCall(StrictSchemaModel):
    type: Literal["tool"]
    name: Literal["retrieve"]
    args: RetrieveToolArgs


class SummarizeToolArgs(StrictSchemaModel):
    min_time: float = Field(..., ge=0.0)
    max_time: Optional[float] = Field(None, ge=0.0)
    time_mode: Literal["relative", "absolute", "auto"] = "auto"
    granularity_seconds: float = Field(..., gt=0.0, le=86400.0)
    summary_structure: Optional[str] = Field(None, min_length=1, max_length=64)
    prompt: str = Field(..., min_length=1)

    @field_validator("summary_structure")
    @classmethod
    def _normalize_summary_structure(cls, value: Optional[str]) -> Optional[str]:
        if value is None:
            return None
        text = value.strip().lower()
        if not text:
            raise ValueError("summary_structure must not be empty")
        return text

    @field_validator("prompt")
    @classmethod
    def _normalize_prompt(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("prompt must not be empty")
        return text

    @model_validator(mode="after")
    def _validate_bounds(self) -> "SummarizeToolArgs":
        if self.max_time is not None and self.max_time < self.min_time:
            raise ValueError("max_time must be >= min_time")
        return self


class SummarizeToolCall(StrictSchemaModel):
    type: Literal["tool"]
    name: Literal["summarize"]
    args: SummarizeToolArgs


class FinalAnswerCall(StrictSchemaModel):
    type: Literal["final"]
    text: str = Field(..., min_length=1)

    @field_validator("text")
    @classmethod
    def _normalize_text(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("text must not be empty")
        return text


WsAgentDecision = Union[RetrieveToolCall, SummarizeToolCall, FinalAnswerCall]


class PlannerBestRef(StrictSchemaModel):
    turn_idx: int = Field(..., ge=1)
    result_idx: int = Field(..., ge=0)


class PlannerTimeRange(StrictSchemaModel):
    min: Optional[float] = None
    max: Optional[float] = None
    mode: Literal["relative", "absolute", "auto"] = "auto"
    anchor: Optional["PlannerTimeAnchor"] = None
    start_anchor: Optional["PlannerTimeAnchor"] = None
    end_anchor: Optional["PlannerTimeAnchor"] = None

    @model_validator(mode="after")
    def _validate_bounds(self) -> "PlannerTimeRange":
        if self.min is not None and self.max is not None and self.max < self.min:
            raise ValueError("time_range.max must be >= time_range.min")
        has_single_anchor = self.anchor is not None
        has_range_anchor = self.start_anchor is not None or self.end_anchor is not None
        if has_single_anchor and has_range_anchor:
            raise ValueError("time_range cannot mix anchor with start_anchor/end_anchor")
        if (self.start_anchor is None) != (self.end_anchor is None):
            raise ValueError("time_range requires both start_anchor and end_anchor when using anchored ranges")
        return self


class PlannerVisualRef(StrictSchemaModel):
    turn_idx: int = Field(..., ge=1)
    result_idx: int = Field(..., ge=0)


class PlannerTimeAnchor(StrictSchemaModel):
    query: Optional[str] = Field(None, min_length=1)
    query_variants: Optional[List[str]] = None
    ref: Optional[PlannerBestRef] = None
    before_seconds: float = Field(0.0, ge=0.0, le=604800.0)
    after_seconds: float = Field(0.0, ge=0.0, le=604800.0)
    sources: Optional[List[Literal["frame", "event", "summary"]]] = None
    candidate_source_groups: Optional[List[List[Literal["frame", "event", "summary"]]]] = None
    top_k: int = Field(5, ge=1, le=50)
    inspect_k: int = Field(3, ge=0, le=20)
    threshold: float = Field(0.0, ge=0.0)
    occurrence_index: Optional[int] = Field(None, ge=1, le=100)
    verification_prompt: Optional[str] = Field(None, min_length=1, max_length=400)

    @field_validator("query")
    @classmethod
    def _normalize_query(cls, value: Optional[str]) -> Optional[str]:
        if value is None:
            return None
        text = value.strip()
        if not text:
            raise ValueError("anchor query must not be empty")
        return text

    @field_validator("query_variants")
    @classmethod
    def _normalize_query_variants(cls, value: Optional[List[str]]) -> Optional[List[str]]:
        if value is None:
            return None
        normalized: List[str] = []
        seen: set[str] = set()
        for item in value:
            text = str(item or "").strip()
            if not text:
                continue
            key = text.casefold()
            if key in seen:
                continue
            normalized.append(text)
            seen.add(key)
        if not normalized:
            raise ValueError("query_variants must contain at least one non-empty string")
        return normalized

    @field_validator("candidate_source_groups")
    @classmethod
    def _normalize_candidate_source_groups(
        cls,
        value: Optional[List[List[Literal["frame", "event", "summary"]]]],
    ) -> Optional[List[List[Literal["frame", "event", "summary"]]]]:
        if value is None:
            return None
        normalized: List[List[Literal["frame", "event", "summary"]]] = []
        seen: set[tuple[str, ...]] = set()
        for group in value:
            cleaned = [str(item).strip().lower() for item in (group or []) if str(item).strip()]
            if not cleaned:
                continue
            deduped = list(dict.fromkeys(cleaned))
            key = tuple(deduped)
            if key in seen:
                continue
            normalized.append(deduped)  # type: ignore[list-item]
            seen.add(key)
        if not normalized:
            raise ValueError("candidate_source_groups must contain at least one non-empty group")
        return normalized

    @field_validator("verification_prompt")
    @classmethod
    def _normalize_verification_prompt(cls, value: Optional[str]) -> Optional[str]:
        if value is None:
            return None
        text = value.strip()
        if not text:
            raise ValueError("verification_prompt must not be empty")
        return text

    @model_validator(mode="after")
    def _validate_anchor(self) -> "PlannerTimeAnchor":
        if self.inspect_k > self.top_k:
            self.inspect_k = self.top_k
        has_query = self.query is not None or bool(self.query_variants)
        if not has_query and self.ref is None:
            raise ValueError("time_range.anchor requires query, query_variants, or ref")
        return self


class PlannerQuerySpec(StrictSchemaModel):
    q: str = Field(..., min_length=1)
    top_k: int = Field(5, ge=1, le=200)
    inspect_k: int = Field(5, ge=0, le=50)
    threshold: float = Field(0.65, ge=0.0)

    @field_validator("q")
    @classmethod
    def _normalize_q(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("query text must not be empty")
        return text

    @model_validator(mode="after")
    def _validate_inspect_k(self) -> "PlannerQuerySpec":
        if self.inspect_k > self.top_k:
            self.inspect_k = self.top_k
        return self


class PlannerSummaryFilter(StrictSchemaModel):
    summary_structure: Optional[str] = Field(None, min_length=1, max_length=64)
    granularity_seconds: Optional[float] = Field(None, gt=0.0, le=86400.0)

    @field_validator("summary_structure")
    @classmethod
    def _normalize_summary_structure(cls, value: Optional[str]) -> Optional[str]:
        if value is None:
            return None
        text = value.strip().lower()
        if not text:
            raise ValueError("summary_structure must not be empty")
        return text


class PlannerAnswerAction(StrictSchemaModel):
    action: Literal["answer"]
    response: str = Field(..., min_length=1)
    best_ref: Optional[PlannerBestRef] = None
    thought: str = ""

    @field_validator("response")
    @classmethod
    def _normalize_response(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("response must not be empty")
        return text


class PlannerSearchAction(StrictSchemaModel):
    action: Literal["search"]
    queries: List[PlannerQuerySpec] = Field(..., min_length=1)
    time_range: Optional[PlannerTimeRange] = None
    sources: Optional[List[Literal["frame", "event", "summary"]]] = None
    summary_filter: Optional[PlannerSummaryFilter] = None
    visual_ref: Optional[PlannerVisualRef] = None
    joint_inspection: bool = False
    inspection_prompt: str = ""
    thought: str = ""


class PlannerInspectAction(StrictSchemaModel):
    action: Literal["inspect"]
    prompt: str = Field(..., min_length=1)
    time_range: Optional[PlannerTimeRange] = None
    ref: Optional[PlannerBestRef] = None
    max_frames: int = Field(6, ge=1, le=16)
    thought: str = ""

    @field_validator("prompt")
    @classmethod
    def _normalize_prompt(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("prompt must not be empty")
        return text

    @model_validator(mode="after")
    def _validate_target(self) -> "PlannerInspectAction":
        if self.time_range is None and self.ref is None:
            raise ValueError("inspect requires time_range or ref")
        return self


class PlannerSummarizeAction(StrictSchemaModel):
    action: Literal["summarize"]
    time_range: PlannerTimeRange
    granularity_seconds: float = Field(..., gt=0.0, le=86400.0)
    summary_structure: Optional[str] = Field(None, min_length=1, max_length=64)
    prompt: str = Field(..., min_length=1)
    thought: str = ""

    @field_validator("summary_structure")
    @classmethod
    def _normalize_summary_structure(cls, value: Optional[str]) -> Optional[str]:
        if value is None:
            return None
        text = value.strip().lower()
        if not text:
            raise ValueError("summary_structure must not be empty")
        return text

    @field_validator("prompt")
    @classmethod
    def _normalize_prompt(cls, value: str) -> str:
        text = value.strip()
        if not text:
            raise ValueError("prompt must not be empty")
        return text


PlannerAction = Union[PlannerAnswerAction, PlannerSearchAction, PlannerInspectAction, PlannerSummarizeAction]


class PlannerActionAdapter(StrictSchemaModel):
    payload: PlannerAction

    @classmethod
    def parse_payload(cls, raw: Dict[str, Any]) -> PlannerAction:
        return cls.model_validate({"payload": raw}).payload