File size: 20,800 Bytes
15f6bf4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
"""
Unified SRT processing module combining resegmentation and translation functionality.
"""

import os
import re
import concurrent.futures
from typing import List, Tuple, Optional
from dotenv import load_dotenv
from openai import OpenAI

# Load environment variables from .env if present
load_dotenv(override=True)


# ============================================================================
# Core SRT Utilities
# ============================================================================


def read_srt(file_path: str) -> str:
    """Read SRT file content."""
    with open(file_path, "r", encoding="utf-8") as f:
        return f.read()


def write_srt(file_path: str, content: str) -> None:
    """Write content to SRT file."""
    with open(file_path, "w", encoding="utf-8") as f:
        f.write(content)


def parse_srt_blocks(srt_content: str) -> List[Tuple[str, str, List[str]]]:
    """
    Parse SRT content into blocks.
    Returns list of (index, time, text_lines).
    """
    blocks = re.split(r"\n\s*\n", srt_content.strip(), flags=re.MULTILINE)
    parsed: List[Tuple[str, str, List[str]]] = []
    for block in blocks:
        lines = block.strip().splitlines()
        if len(lines) < 3:
            continue
        index = lines[0].strip()
        time_line = lines[1].strip()
        text_lines = [line.rstrip() for line in lines[2:]]
        parsed.append((index, time_line, text_lines))
    return parsed


def parse_srt_block(block: str) -> Optional[Tuple[str, str, List[str]]]:
    """Parse a single SRT block."""
    lines = block.strip().splitlines()
    if len(lines) < 3:
        return None
    index = lines[0]
    time = lines[1]
    text_lines = lines[2:]
    return index, time, text_lines


def build_srt_block(index: int, start_time: str, end_time: str, text: str) -> str:
    """Build SRT block with index, time range, and text."""
    return f"{index}\n{start_time} --> {end_time}\n{text}"


def build_srt_block_from_lines(index: str, time: str, text_lines: List[str]) -> str:
    """Build SRT block from parsed components."""
    return f"{index}\n{time}\n" + "\n".join(text_lines)


# ============================================================================
# Time Utilities
# ============================================================================


def extract_times(time_line: str) -> Tuple[str, str]:
    """Extract start and end times from time line."""
    # Expected format: HH:MM:SS,mmm --> HH:MM:SS,mmm
    parts = [p.strip() for p in time_line.split("-->")]
    if len(parts) != 2:
        raise ValueError(f"Invalid time line: {time_line}")
    return parts[0], parts[1]


def time_str_to_ms(t: str) -> int:
    """Convert time string to milliseconds."""
    # HH:MM:SS,mmm
    hms, ms = t.split(",")
    hours, minutes, seconds = hms.split(":")
    total_ms = (
        int(hours) * 3600 * 1000
        + int(minutes) * 60 * 1000
        + int(seconds) * 1000
        + int(ms)
    )
    return total_ms


def ms_to_time_str(ms: int) -> str:
    """Convert milliseconds to time string."""
    if ms < 0:
        ms = 0
    hours = ms // (3600 * 1000)
    ms %= 3600 * 1000
    minutes = ms // (60 * 1000)
    ms %= 60 * 1000
    seconds = ms // 1000
    millis = ms % 1000
    return f"{hours:02d}:{minutes:02d}:{seconds:02d},{millis:03d}"


# ============================================================================
# Text Processing Utilities
# ============================================================================


def ends_with_preferred_punctuation(text: str) -> bool:
    """Check if text ends with preferred punctuation."""
    stripped = text.rstrip()
    return stripped.endswith(".") or stripped.endswith(",")


def normalize_whitespace(text: str) -> str:
    """Normalize whitespace in text."""
    return re.sub(r"\s+", " ", text).strip()


def count_chars(text: str) -> int:
    """Count characters including spaces after normalization."""
    return len(text)


def split_text_into_chunks_by_chars_with_punctuation(
    text: str, max_chars: int
) -> List[str]:
    """Split text into chunks respecting punctuation boundaries."""
    text = normalize_whitespace(text)
    chunks: List[str] = []
    i = 0
    n = len(text)
    while i < n:
        remaining = text[i:]
        if len(remaining) <= max_chars:
            chunks.append(remaining.strip())
            break
        window = remaining[:max_chars]
        # Prefer last '.' or ',' within the window
        last_dot = window.rfind(".")
        last_comma = window.rfind(",")
        cut_at = max(last_dot, last_comma)
        if cut_at != -1:
            end = cut_at + 1
        else:
            # If no punctuation found, look for the last space to avoid cutting words
            last_space = window.rfind(" ")
            if last_space != -1:
                end = last_space
            else:
                # If no space found, we have to cut at max_chars (single long word)
                end = max_chars
        chunk = remaining[:end].strip()
        if chunk:
            chunks.append(chunk)
        i += end
        # Skip any following spaces before next chunk
        while i < n and text[i] == " ":
            i += 1
    return [c for c in chunks if c]


# ============================================================================
# Translation Functionality
# ============================================================================


def translate_text(
    text: str, target_lang: str, model: str, router: str = "dashscope"
) -> str:
    """Translate text using specified provider."""
    if router == "dashscope":
        client = OpenAI(
            api_key=os.getenv("DASHSCOPE_API_KEY"),
            base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
        )
        prompt = (
            f"Translate the following subtitle text to {target_lang}. "
            "Do not translate timestamps or numbers. Only translate the spoken text. "
            "Return only the translated text, no explanations or formatting.\n\n"
            f"{text}"
        )
        response = client.chat.completions.create(
            model=model,
            messages=[
                {
                    "role": "system",
                    "content": "You are a helpful assistant that translates subtitles.",
                },
                {"role": "user", "content": prompt},
            ],
            temperature=0.3,
            max_tokens=1024,
        )
        return response.choices[0].message.content.strip()

    elif router == "openrouter":
        client = OpenAI(
            api_key=os.getenv("OPENROUTER_API_KEY"),
            base_url="https://openrouter.ai/api/v1",
        )
        prompt = (
            f"Translate the following subtitle text to {target_lang}. "
            "Do not translate timestamps or numbers. Only translate the spoken text. "
            "Return only the translated text, no explanations or formatting.\n\n"
            f"{text}"
        )
        # Optional attribution headers
        extra_headers = {}
        referer = os.getenv("OPENROUTER_SITE_URL")
        app_title = os.getenv("OPENROUTER_APP_TITLE")
        if referer:
            extra_headers["HTTP-Referer"] = referer
        if app_title:
            extra_headers["X-Title"] = app_title
        response = client.chat.completions.create(
            model=model,
            messages=[
                {
                    "role": "system",
                    "content": "You are a helpful assistant that translates subtitles.",
                },
                {"role": "user", "content": prompt},
            ],
            temperature=0.3,
            max_tokens=1024,
            extra_headers=extra_headers,
        )
        return response.choices[0].message.content.strip()

    elif router == "openai":
        client = OpenAI()
        prompt = (
            f"Translate the following subtitle text to {target_lang}. "
            "Do not translate timestamps or numbers. Only translate the spoken text. "
            "Return only the translated text, no explanations or formatting.\n\n"
            f"{text}"
        )
        try:
            # Use Responses API for newer models (e.g., gpt-4.1, gpt-4o)
            if model and (model.startswith("gpt-4.1") or model.startswith("gpt-4o")):
                response = client.responses.create(
                    model=model,
                    input=prompt,
                    instructions="You are a helpful assistant that translates subtitles.",
                    temperature=0.3,
                    max_output_tokens=1024,
                )
                # Prefer helper if available
                try:
                    return response.output_text.strip()
                except Exception:
                    # Fallback parsing if helper is unavailable
                    try:
                        segments = []
                        if hasattr(response, "output") and response.output:
                            for content_item in response.output[0].content:
                                text_val = getattr(content_item, "text", None)
                                if text_val:
                                    segments.append(text_val)
                        if segments:
                            return "\n".join(segments).strip()
                    except Exception:
                        pass
                    return str(response).strip()
            else:
                # Backward compatibility: use Chat Completions for older models
                response = client.chat.completions.create(
                    model=model,
                    messages=[
                        {
                            "role": "system",
                            "content": "You are a helpful assistant that translates subtitles.",
                        },
                        {"role": "user", "content": prompt},
                    ],
                    temperature=0.3,
                    max_tokens=1024,
                )
                return response.choices[0].message.content.strip()
        except Exception as e:
            # Last-resort fallback to ensure we return something
            return str(e)
    else:
        return f"Unsupported provider: {router}"


def translate_block(args: Tuple[str, str, str, str]) -> str:
    """Translate a single SRT block."""
    block, target_lang, model, router = args
    parsed = parse_srt_block(block)
    if not parsed:
        return block
    index, time, text_lines = parsed
    text = "\n".join(text_lines)
    if text.strip():
        translated_text = translate_text(text, target_lang, model=model, router=router)
        translated_text_lines = translated_text.splitlines() or [translated_text]
    else:
        translated_text_lines = text_lines
    translated_block = build_srt_block_from_lines(index, time, translated_text_lines)
    return translated_block


def translate_srt(
    input_path: str,
    output_path: str,
    target_lang: str,
    model: Optional[str] = None,
    workers: int = 15,
    router: str = "dashscope",
    max_chars: int = 125,
) -> str:
    """Translate SRT file using specified provider with resegmentation."""
    # Check API keys based on router
    if router == "openai":
        api_key = os.getenv("OPENAI_API_KEY")
        if not api_key:
            raise RuntimeError(
                "Error: OPENAI_API_KEY not found in environment variables."
            )
        if not model:
            model = os.getenv("MODEL") or "gpt-4.1"
    elif router == "openrouter":
        openrouter_key = os.getenv("OPENROUTER_API_KEY")
        if not openrouter_key:
            raise RuntimeError(
                "Error: OPENROUTER_API_KEY not found in environment variables."
            )
        if not model:
            model = os.getenv("MODEL") or "openai/gpt-4o"
    elif router == "dashscope":
        dashscope_key = os.getenv("DASHSCOPE_API_KEY")
        if not dashscope_key:
            raise RuntimeError(
                "Error: DASHSCOPE_API_KEY not found in environment variables."
            )
        if not model:
            model = os.getenv("MODEL") or "qwen-max"
    else:
        raise RuntimeError(
            f"Error: Unknown provider '{router}'. Expected one of: openai, openrouter, dashscope."
        )

    # First resegment the SRT to get optimal chunks for translation
    srt_content = read_srt(input_path)
    parsed_blocks = parse_srt_blocks(srt_content)
    resegmented_blocks = resegment_blocks(parsed_blocks, max_chars)

    # Now translate the resegmented blocks
    block_args = [(block, target_lang, model, router) for block in resegmented_blocks]
    translated_blocks = []
    with concurrent.futures.ThreadPoolExecutor(max_workers=workers) as executor:
        for translated_block in executor.map(translate_block, block_args):
            translated_blocks.append(translated_block)
    translated_content = "\n\n".join(translated_blocks)
    write_srt(output_path, translated_content)
    return output_path


# ============================================================================
# Resegmentation Functionality
# ============================================================================


def resegment_blocks(
    parsed_blocks: List[Tuple[str, str, List[str]]], max_chars: int
) -> List[str]:
    """Resegment SRT blocks based on character limit."""
    output_blocks: List[str] = []

    current_index = 1
    group_start_time: str = ""
    group_end_time: str = ""
    group_text_parts: List[str] = []
    group_char_count = 0

    def flush_group():
        nonlocal current_index, group_start_time, group_end_time, group_text_parts, group_char_count
        if group_char_count > 0 and group_text_parts:
            block_text = normalize_whitespace(" ".join(group_text_parts))
            output_blocks.append(
                build_srt_block(
                    current_index, group_start_time, group_end_time, block_text
                )
            )
            current_index += 1
            group_start_time = ""
            group_end_time = ""
            group_text_parts = []
            group_char_count = 0

    for _, time_line, text_lines in parsed_blocks:
        start_time_str, end_time_str = extract_times(time_line)
        start_ms = time_str_to_ms(start_time_str)
        end_ms = time_str_to_ms(end_time_str)
        duration_ms = max(0, end_ms - start_ms)

        text = normalize_whitespace(" ".join(text_lines))
        if not text:
            continue

        this_count = count_chars(text)

        # If adding this block would exceed the limit, flush the current group first
        if group_char_count > 0 and (group_char_count + this_count) > max_chars:
            flush_group()

        # If the single block itself exceeds max_chars, split it internally
        if this_count > max_chars:
            # Ensure any pending group is flushed before inserting split pieces
            flush_group()
            sub_texts = split_text_into_chunks_by_chars_with_punctuation(
                text, max_chars
            )
            # Distribute timings proportionally by character count
            total_chars = sum(count_chars(st) for st in sub_texts) or 1
            accumulated_ms = 0
            for idx, st in enumerate(sub_texts):
                chars_in_chunk = count_chars(st) or 1
                # compute chunk duration (last chunk takes remaining to avoid rounding drift)
                if idx < len(sub_texts) - 1:
                    chunk_ms = int(duration_ms * (chars_in_chunk / total_chars))
                else:
                    chunk_ms = max(0, duration_ms - accumulated_ms)
                chunk_start_ms = start_ms + accumulated_ms
                chunk_end_ms = chunk_start_ms + chunk_ms
                accumulated_ms += chunk_ms

                output_blocks.append(
                    build_srt_block(
                        current_index,
                        ms_to_time_str(chunk_start_ms),
                        ms_to_time_str(chunk_end_ms),
                        st,
                    )
                )
                current_index += 1
            # Done with this overlong block
            continue

        # Otherwise, safe to merge this whole block into the group
        if group_char_count == 0:
            group_start_time = start_time_str
        group_text_parts.append(text)
        group_end_time = end_time_str
        group_char_count += this_count

        # Prefer flushing on punctuation at the end of this block
        if ends_with_preferred_punctuation(text):
            flush_group()
        elif group_char_count >= max_chars:
            flush_group()

    # Flush any remaining group
    if group_char_count > 0:
        flush_group()

    return output_blocks


def resegment_srt(input_path: str, output_path: str, max_chars: int = 125) -> str:
    """Resegment SRT file based on character limit."""
    srt_content = read_srt(input_path)
    parsed = parse_srt_blocks(srt_content)
    merged_blocks = resegment_blocks(parsed, max_chars=max_chars)
    output_content = "\n\n".join(merged_blocks) + "\n"
    write_srt(output_path, output_content)
    return output_path


# ============================================================================
# Combined Processing Functions
# ============================================================================


def process_srt_file(
    input_path: str,
    output_path: str,
    operation: str = "resegment",
    max_chars: int = 125,
    target_lang: Optional[str] = None,
    model: Optional[str] = None,
    workers: int = 15,
    router: str = "dashscope",
) -> str:
    """
    Process SRT file with specified operation.

    Args:
        input_path: Path to input SRT file
        output_path: Path to output SRT file
        operation: "resegment" or "translate"
        max_chars: Maximum characters per segment (for resegmentation)
        target_lang: Target language code (for translation)
        model: Model to use for translation
        workers: Number of concurrent workers for translation
        router: Translation provider ("dashscope", "openai", "openrouter")

    Returns:
        Path to output file
    """
    if operation == "resegment":
        return resegment_srt(input_path, output_path, max_chars)
    elif operation == "translate":
        if not target_lang:
            raise ValueError("target_lang is required for translation")
        return translate_srt(
            input_path, output_path, target_lang, model, workers, router, max_chars
        )
    else:
        raise ValueError(
            f"Unknown operation: {operation}. Must be 'resegment' or 'translate'"
        )


# ============================================================================
# CLI Interface (for backward compatibility)
# ============================================================================

if __name__ == "__main__":
    import argparse

    parser = argparse.ArgumentParser(
        description="Unified SRT processing tool for resegmentation and translation. Translation automatically includes resegmentation for optimal chunk sizes."
    )
    parser.add_argument("input", help="Input SRT file path")
    parser.add_argument("output", help="Output SRT file path")
    parser.add_argument(
        "--operation",
        choices=["resegment", "translate"],
        default="resegment",
        help="Operation to perform (default: resegment)",
    )
    parser.add_argument(
        "--max-chars",
        dest="max_chars",
        type=int,
        default=125,
        help="Maximum characters per segment (default: 125)",
    )
    parser.add_argument(
        "--target-lang", help="Target language code (e.g., fr, es, de, zh)"
    )
    parser.add_argument(
        "--model", help="Model to use for translation (default: value of MODEL in .env)"
    )
    parser.add_argument(
        "--workers",
        type=int,
        default=25,
        help="Number of concurrent workers for translation (default: 25)",
    )
    parser.add_argument(
        "--provider",
        choices=["openai", "dashscope", "openrouter"],
        default="dashscope",
        help="Translation provider (default: dashscope)",
    )

    args = parser.parse_args()

    try:
        result = process_srt_file(
            args.input,
            args.output,
            operation=args.operation,
            max_chars=args.max_chars,
            target_lang=args.target_lang,
            model=args.model,
            workers=args.workers,
            router=args.provider,
        )
        print(f"Processing complete. Output written to {result}")
    except Exception as e:
        print(f"Error: {e}")
        exit(1)