File size: 9,550 Bytes
e2b8b61
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import argparse
import json
import re
from pathlib import Path
from typing import Any, Dict, List

from src.config import DECODED_DIR, PROMPTS_DIR
from src.generator import Generator
from src.region_registry import get_region_description


def parse_args() -> argparse.Namespace:
    parser = argparse.ArgumentParser(
        description="Decode localized DAS into natural dialogue."
    )
    parser.add_argument(
        "--input_path",
        type=str,
        required=True,
        help="Path to localized JSON file.",
    )
    parser.add_argument(
        "--output_path",
        type=str,
        default=None,
        help="Path to save decoded output JSON.",
    )
    parser.add_argument(
        "--language",
        type=str,
        required=True,
        help="Target language label, e.g. 'Swahili'.",
    )
    parser.add_argument(
        "--region",
        type=str,
        default="",
        help="Optional target region/community, e.g. 'Kenya - Nairobi'.",
    )
    parser.add_argument(
        "--decode_prompt",
        type=str,
        default=str(PROMPTS_DIR / "das_decode.md"),
        help="Path to DAS decode prompt.",
    )
    parser.add_argument(
        "--model",
        type=str,
        default=None,
        help="Model alias from model_registry.py",
    )
    parser.add_argument(
        "--max_instances",
        type=int,
        default=None,
        help="Optional cap on number of dialogues to process.",
    )
    parser.add_argument(
        "--start_idx",
        type=int,
        default=0,
        help="Optional start index for slicing input data.",
    )
    parser.add_argument(
        "--end_idx",
        type=int,
        default=None,
        help="Optional end index for slicing input data.",
    )
    parser.add_argument(
        "--dont_use_cached",
        action="store_true",
        help="Disable cached prompt responses.",
    )
    return parser.parse_args()


def load_json(path: str) -> Any:
    return json.loads(Path(path).read_text(encoding="utf-8"))


def save_json(path: str, data: Any) -> None:
    output_path = Path(path)
    output_path.parent.mkdir(parents=True, exist_ok=True)
    output_path.write_text(
        json.dumps(data, indent=2, ensure_ascii=False),
        encoding="utf-8",
    )


def normalize_speaker_id(speaker_id: Any) -> str:
    speaker = str(speaker_id)
    if speaker == "1":
        return "A"
    if speaker == "2":
        return "B"
    return speaker


def preprocess_localized_conversation(localized_das: List[Dict[str, Any]]) -> str:
    formatted_turns: List[str] = []

    for idx, turn in enumerate(localized_das, start=1):
        speaker = normalize_speaker_id(turn.get("speaker_id", "A"))
        functions = turn.get("functions", "")

        if isinstance(functions, list):
            functions_str = "; ".join(str(f) for f in functions)
        else:
            functions_str = str(functions)

        formatted_turns.append(f"{idx}: {speaker}.{functions_str}")

    return "\n".join(formatted_turns)


def get_original_dialogue(item: Dict[str, Any]) -> List[str]:
    if "original" in item and isinstance(item["original"], list):
        return item["original"]

    if "conversation" in item and isinstance(item["conversation"], list):
        return item["conversation"]

    if "dialogue" in item and isinstance(item["dialogue"], list):
        return item["dialogue"]

    if "utterances" in item and isinstance(item["utterances"], list):
        return item["utterances"]

    return []


def preprocess_decode_input(
    data: List[Dict[str, Any]],
    language: str,
    region: str,
) -> List[Dict[str, Any]]:
    decode_desc = get_region_description(region, "decode", language) or ""
    processed: List[Dict[str, Any]] = []

    for item in data:
        if "localized_das" not in item:
            raise ValueError("Each input item must contain 'localized_das'.")
        if "localized_context" not in item:
            raise ValueError("Each input item must contain 'localized_context'.")

        if not isinstance(item["localized_das"], list):
            raise ValueError(
                f"localized_das must be a list, got {type(item['localized_das']).__name__}"
            )

        new_item = dict(item)
        new_item["language"] = language
        new_item["region"] = region
        new_item["region_description"] = decode_desc
        new_item["turns"] = preprocess_localized_conversation(item["localized_das"])
        new_item["localized_context"] = item["localized_context"]
        new_item["context"] = item["localized_context"]
        processed.append(new_item)

    return processed


def strip_code_fences(text: str) -> str:
    text = text.strip()
    if text.startswith("```"):
        text = re.sub(r"^```[a-zA-Z0-9_+-]*\n?", "", text)
        text = re.sub(r"\n?```$", "", text)
    return text.strip()


def parse_numbered_dialogue_string(raw: str) -> List[str]:
    raw = strip_code_fences(raw)

    # Split on numbered turns like "1: ...", "2. ..."
    matches = list(re.finditer(r"(?m)^\s*(\d+)[\:\.]\s*", raw))
    if not matches:
        lines = [line.strip() for line in raw.splitlines() if line.strip()]
        return lines

    turns: List[str] = []
    for idx, match in enumerate(matches):
        start = match.end()
        end = matches[idx + 1].start() if idx + 1 < len(matches) else len(raw)
        turn_text = raw[start:end].strip()
        if turn_text:
            turns.append(turn_text)

    return turns


def normalize_generated_conversation(generated_conversation: Any) -> List[str]:
    # Case 1: expected list of objects with text
    if isinstance(generated_conversation, list):
        text_only_dialogue: List[str] = []

        for turn in generated_conversation:
            if isinstance(turn, dict) and "text" in turn:
                text_only_dialogue.append(str(turn["text"]).strip())
            elif isinstance(turn, str):
                text_only_dialogue.append(turn.strip())
            else:
                raise ValueError(
                    f"Unsupported generated_conversation list item: {turn}"
                )

        return text_only_dialogue

    # Case 2: model returned one big numbered string
    if isinstance(generated_conversation, str):
        return parse_numbered_dialogue_string(generated_conversation)

    raise ValueError(
        f"Unsupported generated_conversation type: {type(generated_conversation).__name__}"
    )


def merge_decoded_responses(
    base_data: List[Dict[str, Any]],
    responses: List[str],
    language: str,
) -> List[Dict[str, Any]]:
    merged: List[Dict[str, Any]] = []
    decoded_key = f"decoded_{language.strip().lower().replace(' ', '_')}"

    for item, response_text in zip(base_data, responses):
        if response_text is None:
            print(f"[Decode] Skipping item with failed generation")
            continue

        response_json = Generator.parse_json_response(response_text)

        if "generated_conversation" not in response_json:
            raise ValueError(
                f"Missing 'generated_conversation' in model response:\n{response_text}"
            )

        generated_conversation = response_json["generated_conversation"]
        text_only_dialogue = normalize_generated_conversation(generated_conversation)

        output_item: Dict[str, Any] = {}

        if "dialogue_id" in item:
            output_item["dialogue_id"] = item["dialogue_id"]
        elif "id" in item:
            output_item["dialogue_id"] = item["id"]

        output_item["original"] = get_original_dialogue(item)
        output_item[decoded_key] = text_only_dialogue

        merged.append(output_item)

    return merged


def default_output_path(input_path: str, language: str, region: str) -> str:
    stem = Path(input_path).stem
    suffix_parts = [language.strip().lower().replace(" ", "_")]

    if region.strip():
        suffix_parts.append(region.strip().lower().replace(" ", "_").replace("/", "_"))

    suffix = "_".join(suffix_parts)
    return str(DECODED_DIR / f"{stem}_{suffix}_decoded.json")


def main() -> None:
    args = parse_args()

    raw_data = load_json(args.input_path)
    if not isinstance(raw_data, list):
        raise ValueError("Input JSON must be a list of dialogue objects.")

    sliced_data = raw_data[args.start_idx:args.end_idx]
    if args.max_instances is not None:
        sliced_data = sliced_data[: args.max_instances]

    output_path = args.output_path or default_output_path(
        args.input_path,
        args.language,
        args.region,
    )

    generator = Generator(
        model_alias=args.model,
        use_cache=not args.dont_use_cached,
    )

    processed_data = preprocess_decode_input(
        data=sliced_data,
        language=args.language,
        region=args.region,
    )

    print(f"[Decode] Building decoded dialogue for {len(processed_data)} items...")
    decode_prompts, decode_response_format = generator.build_prompts(
        args.decode_prompt,
        processed_data,
    )

    decode_responses = generator.prompt(
        prompts=decode_prompts,
        response_format=decode_response_format,
        dont_use_cached=args.dont_use_cached,
        skip_failures=True,
    )

    final_data = merge_decoded_responses(
        processed_data,
        decode_responses,
        args.language,
    )

    save_json(output_path, final_data)
    print(f"Saved decoded data to: {output_path}")
    generator.print_usage_summary(stage="Decode")


if __name__ == "__main__":
    main()