File size: 21,074 Bytes
3f5c41b
 
 
876a77e
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
eaaf050
876a77e
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
 
876a77e
 
eaaf050
876a77e
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
 
876a77e
 
eaaf050
876a77e
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
eaaf050
 
 
876a77e
eaaf050
 
 
 
876a77e
 
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
 
 
 
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
eaaf050
 
876a77e
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
eaaf050
876a77e
 
 
 
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
 
eaaf050
876a77e
 
 
eaaf050
876a77e
 
 
 
 
eaaf050
876a77e
eaaf050
876a77e
 
 
3f5c41b
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
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
#!/usr/bin/env python
# coding=utf-8
# Copyright 2024 The Footscray Coding Collective. All rights reserved.
"""
Zhou Protocol FLUX-LoRA Integration Tool

This module provides a Smolagents Tool implementation for interacting with FLUX-LoRA-DLC API.
It enables agents to generate high-quality images with customizable LoRA models.

Usage:
    flux_tool = FluxLoRATool()
    agent = CodeAgent(tools=[flux_tool], ...)
"""

import logging
import os
import tempfile
import uuid
from dataclasses import dataclass
from typing import Any, Dict, Optional

# Third-party
import requests
from gradio_client import Client
from PIL import Image
from smolagents import Tool

# -----------------------------------------------------------------------------
# CONSTANTS AND TYPE DEFINITIONS
# -----------------------------------------------------------------------------


@dataclass
class LoRAModelInfo:
    """Value object representing LoRA model information."""

    name: str
    description: Optional[str] = None
    example_image_url: Optional[str] = None


@dataclass
class ImageGenerationResult:
    """Value object representing a generated image result."""

    image_path: str
    seed: int
    metadata: Optional[Dict[str, Any]] = None


# -----------------------------------------------------------------------------
# CORE TOOL IMPLEMENTATION
# -----------------------------------------------------------------------------


class FluxLoRATool(Tool):
    """
    Tool for generating images using FLUX-LoRA-DLC API.

    This tool implements the Zhou Protocol integration patterns to provide
    a clean, efficient interface for image generation using LoRA models.
    """

    name = "flux_lora_generator"
    description = """
    Generates high-quality images using FLUX-LoRA models.
    Can use custom LoRA models, adjust image parameters, and handle image inputs.
    """
    inputs = {
        "prompt": {
            "type": "string",
            "description": "Detailed description of the desired image.",
        },
        "image_input": {
            "type": "string",
            "description": "Optional URL or file path to input image for img2img generation.",
            "optional": True,
        },
        "image_strength": {
            "type": "float",
            "description": "Strength of input image influence (0.0-1.0), where 1.0 maintains more of original image.",
            "optional": True,
            "default": 0.75,
        },
        "cfg_scale": {
            "type": "float",
            "description": "Guidance scale for prompt adherence (1.0-30.0).",
            "optional": True,
            "default": 3.5,
        },
        "steps": {
            "type": "integer",
            "description": "Number of sampling steps (10-100).",
            "optional": True,
            "default": 28,
        },
        "seed": {
            "type": "integer",
            "description": "Random seed for reproducibility. Use -1 for random seed.",
            "optional": True,
            "default": -1,
        },
        "width": {
            "type": "integer",
            "description": "Image width in pixels.",
            "optional": True,
            "default": 1024,
        },
        "height": {
            "type": "integer",
            "description": "Image height in pixels.",
            "optional": True,
            "default": 1024,
        },
        "lora_scale": {
            "type": "float",
            "description": "LoRA influence scale (0.0-1.0).",
            "optional": True,
            "default": 0.95,
        },
        "custom_lora": {
            "type": "string",
            "description": "Custom LoRA model to use. Leave empty for default.",
            "optional": True,
        },
    }
    output_type = "string"

    def __init__(
        self,
        api_url: str = "xkerser/FLUX-LoRA-DLC",
        image_save_dir: Optional[str] = None,
        connection_timeout: int = 60,
        verbose: bool = False,
    ):
        """
        Initialize the FLUX-LoRA Tool with Zhou Protocol connection patterns.

        Args:
            api_url: URL or endpoint ID for the FLUX-LoRA-DLC API
            image_save_dir: Directory to save generated images (created if doesn't exist)
            connection_timeout: API connection timeout in seconds
            verbose: Enable detailed logging
        """
        super().__init__()

        # Initialize logging
        self.logger = logging.getLogger("flux_lora_tool")
        self.logger.setLevel(logging.DEBUG if verbose else logging.INFO)

        # Set up client and storage directories
        self.api_url = api_url
        self.connection_timeout = connection_timeout
        self._client = None  # Lazy initialization

        # Set up image storage directory
        self.image_save_dir = image_save_dir or os.path.join(
            tempfile.gettempdir(), "flux_lora_images"
        )
        os.makedirs(self.image_save_dir, exist_ok=True)
        self.logger.info(
            f"FluxLoRATool initialized. Images will be saved to: {self.image_save_dir}"
        )

    @property
    def client(self) -> Client:
        """
        Get or initialize the Gradio client with proper connection handling.

        Returns:
            Initialized Gradio client

        Raises:
            ConnectionError: If client initialization fails
        """
        if self._client is None:
            try:
                self._client = Client(self.api_url, timeout=self.connection_timeout)
                self.logger.debug(f"Gradio client initialized for: {self.api_url}")
            except Exception as e:
                error_msg = f"Failed to initialize FLUX-LoRA client: {str(e)}"
                self.logger.error(error_msg)
                raise ConnectionError(error_msg) from e

        return self._client

    def _validate_inputs(self, **kwargs) -> Dict[str, Any]:
        """
        Validate and normalize input parameters with Zhou Protocol validation patterns.

        Args:
            **kwargs: Input parameters

        Returns:
            Validated and normalized parameters

        Raises:
            ValueError: If input validation fails
        """
        validated = {}

        # Required parameter: prompt
        if not kwargs.get("prompt"):
            raise ValueError("Prompt is required for image generation")
        validated["prompt"] = kwargs["prompt"]

        # Image input handling
        if "image_input" in kwargs and kwargs["image_input"]:
            input_image = kwargs["image_input"]
            # Handle URL vs. local file
            if input_image.startswith(("http://", "https://")):
                # We'll need to download and process this
                validated["image_input"] = self._download_image(input_image)
            else:
                # Check if file exists
                if not os.path.exists(input_image):
                    raise ValueError(f"Image file not found: {input_image}")
                validated["image_input"] = input_image

        # Numeric parameter validation with constraints
        numeric_params = {
            "image_strength": {"min": 0.0, "max": 1.0, "default": 0.75},
            "cfg_scale": {"min": 1.0, "max": 30.0, "default": 3.5},
            "steps": {"min": 10, "max": 100, "default": 28},
            "width": {"min": 128, "max": 2048, "default": 1024},
            "height": {"min": 128, "max": 2048, "default": 1024},
            "lora_scale": {"min": 0.0, "max": 1.0, "default": 0.95},
        }

        for param, constraints in numeric_params.items():
            if param in kwargs and kwargs[param] is not None:
                value = kwargs[param]

                # Type conversion if needed
                if param in ["steps", "width", "height"]:
                    try:
                        value = int(value)
                    except (ValueError, TypeError):
                        raise ValueError(f"Parameter '{param}' must be an integer")
                else:
                    try:
                        value = float(value)
                    except (ValueError, TypeError):
                        raise ValueError(f"Parameter '{param}' must be a number")

                # Range validation
                if value < constraints["min"] or value > constraints["max"]:
                    raise ValueError(
                        f"Parameter '{param}' must be between {constraints['min']} and {constraints['max']}"
                    )

                validated[param] = value
            else:
                validated[param] = constraints["default"]

        # Special handling for seed
        if "seed" in kwargs and kwargs["seed"] is not None:
            try:
                seed = int(kwargs["seed"])
                # -1 indicates random seed
                if seed == -1:
                    try:
                        seed = self._get_random_seed()
                    except Exception as e:
                        self.logger.warning(f"Failed to get random seed from API: {e}")
                        # Fallback to Python's random
                        import random

                        seed = random.randint(0, 2**32 - 1)
                validated["seed"] = seed
            except (ValueError, TypeError):
                raise ValueError("Seed must be an integer")
        else:
            # Default to random seed
            validated["seed"] = self._get_random_seed()

        # Custom LoRA handling
        if "custom_lora" in kwargs and kwargs["custom_lora"]:
            validated["custom_lora"] = kwargs["custom_lora"]

        return validated

    def _download_image(self, url: str) -> str:
        """
        Download image from URL and save to local file.

        Args:
            url: Image URL

        Returns:
            Local file path

        Raises:
            ConnectionError: If download fails
        """
        try:
            response = requests.get(url, stream=True, timeout=30)
            response.raise_for_status()

            # Generate temporary file path
            file_ext = self._guess_extension(response.headers.get("Content-Type", ""))
            temp_path = os.path.join(
                self.image_save_dir, f"input_{uuid.uuid4().hex}{file_ext}"
            )

            # Save image
            with open(temp_path, "wb") as f:
                for chunk in response.iter_content(chunk_size=8192):
                    f.write(chunk)

            self.logger.debug(f"Downloaded image from {url} to {temp_path}")
            return temp_path

        except Exception as e:
            error_msg = f"Failed to download image from {url}: {str(e)}"
            self.logger.error(error_msg)
            raise ConnectionError(error_msg) from e

    def _guess_extension(self, content_type: str) -> str:
        """
        Guess file extension from content type.

        Args:
            content_type: HTTP Content-Type header

        Returns:
            File extension (with dot)
        """
        content_type = content_type.lower()
        if "jpeg" in content_type or "jpg" in content_type:
            return ".jpg"
        elif "png" in content_type:
            return ".png"
        elif "webp" in content_type:
            return ".webp"
        elif "gif" in content_type:
            return ".gif"
        else:
            return ".png"  # Default to PNG

    def _get_random_seed(self) -> int:
        """
        Get a random seed from the API.

        Returns:
            Random seed value

        Raises:
            RuntimeError: If random seed retrieval fails
        """
        try:
            result = self.client.predict(api_name="/get_random_value")
            if isinstance(result, (int, float)):
                return int(result)
            else:
                raise ValueError(f"Unexpected result type: {type(result)}")
        except Exception as e:
            # Just log and re-raise as we have fallback in the validation method
            self.logger.warning(f"Failed to get random seed: {e}")
            raise

    def _handle_custom_lora(self, custom_lora: Optional[str]) -> None:
        """
        Add or remove custom LoRA model.

        Args:
            custom_lora: Custom LoRA model string

        Raises:
            RuntimeError: If LoRA handling fails
        """
        if not custom_lora:
            # Remove any existing custom LoRA
            try:
                self.client.predict(api_name="/remove_custom_lora")
                self.logger.debug("Removed custom LoRA")
            except Exception as e:
                error_msg = f"Failed to remove custom LoRA: {str(e)}"
                self.logger.error(error_msg)
                raise RuntimeError(error_msg) from e
        else:
            # Add custom LoRA
            try:
                self.client.predict(
                    custom_lora=custom_lora, api_name="/add_custom_lora"
                )
                self.logger.debug(f"Added custom LoRA: {custom_lora}")
            except Exception as e:
                error_msg = f"Failed to add custom LoRA '{custom_lora}': {str(e)}"
                self.logger.error(error_msg)
                raise RuntimeError(error_msg) from e

    def forward(
        self,
        prompt: str,
        image_input: Optional[str] = None,
        image_strength: Optional[float] = None,
        cfg_scale: Optional[float] = None,
        steps: Optional[int] = None,
        seed: Optional[int] = None,
        width: Optional[int] = None,
        height: Optional[int] = None,
        lora_scale: Optional[float] = None,
        custom_lora: Optional[str] = None,
    ) -> str:
        """
        Generate an image with FLUX-LoRA.

        Args:
            prompt: Text description of the desired image
            image_input: Optional path or URL to input image for img2img
            image_strength: Strength of input image influence (0.0-1.0)
            cfg_scale: Guidance scale (1.0-30.0)
            steps: Number of sampling steps (10-100)
            seed: Random seed (-1 for random)
            width: Image width in pixels (128-2048)
            height: Image height in pixels (128-2048)
            lora_scale: LoRA influence scale (0.0-1.0)
            custom_lora: Custom LoRA model to use

        Returns:
            Formatted string with image generation results

        Raises:
            ValueError: If input validation fails
            ConnectionError: If API communication fails
            RuntimeError: If image generation fails
        """
        # Step 1: Validate and normalize inputs
        try:
            params = self._validate_inputs(
                prompt=prompt,
                image_input=image_input,
                image_strength=image_strength,
                cfg_scale=cfg_scale,
                steps=steps,
                seed=seed,
                width=width,
                height=height,
                lora_scale=lora_scale,
                custom_lora=custom_lora,
            )
            self.logger.debug(f"Validated parameters: {params}")
        except ValueError as e:
            return f"Parameter validation failed: {str(e)}"

        # Step 2: Handle custom LoRA if specified
        if "custom_lora" in params:
            try:
                custom_lora_value = params.pop("custom_lora")
                self._handle_custom_lora(custom_lora_value)
            except RuntimeError as e:
                return f"Custom LoRA setup failed: {str(e)}"

        # Step 3: Generate image
        try:
            # Prepare image input if provided
            img_param = None
            if "image_input" in params and params["image_input"]:
                from gradio_client import handle_file

                img_param = handle_file(params.pop("image_input"))

            # Call the API
            generation_args = {
                "prompt": params["prompt"],
                "image_strength": params["image_strength"],
                "cfg_scale": params["cfg_scale"],
                "steps": params["steps"],
                "randomize_seed": False,  # We handle seed explicitly
                "seed": params["seed"],
                "width": params["width"],
                "height": params["height"],
                "lora_scale": params["lora_scale"],
            }

            # Add image input if available
            if img_param:
                generation_args["image_input"] = img_param

            self.logger.info(f"Generating image with params: {generation_args}")
            result = self.client.predict(api_name="/run_lora", **generation_args)

            # Process result
            if isinstance(result, tuple) and len(result) >= 2:
                image_path, actual_seed = result[0], result[1]

                # Save image to our directory
                try:
                    output_path = self._save_image(image_path)
                    image_result = ImageGenerationResult(
                        image_path=output_path, seed=int(actual_seed)
                    )
                    return self._format_result(image_result, params["prompt"])
                except Exception as e:
                    self.logger.error(f"Failed to save generated image: {e}")
                    return f"Image generated but failed to save: {str(e)}"
            else:
                raise ValueError(f"Unexpected API response format: {result}")

        except Exception as e:
            error_msg = f"Image generation failed: {str(e)}"
            self.logger.error(error_msg)
            return error_msg

    def _save_image(self, image_path: str) -> str:
        """
        Save generated image to specified directory.

        Args:
            image_path: Path to generated image from API

        Returns:
            Path to saved image

        Raises:
            IOError: If image saving fails
        """
        try:
            # Load the image
            img = Image.open(image_path)

            # Generate timestamp-based filename
            timestamp = uuid.uuid4().hex[:8]
            output_filename = f"flux_lora_{timestamp}.png"
            output_path = os.path.join(self.image_save_dir, output_filename)

            # Save to our directory
            img.save(output_path)
            self.logger.debug(f"Saved image to {output_path}")

            return output_path

        except Exception as e:
            error_msg = f"Failed to save image: {str(e)}"
            self.logger.error(error_msg)
            raise IOError(error_msg) from e

    def _format_result(self, result: ImageGenerationResult, prompt: str) -> str:
        """
        Format the image generation result as a string.

        Args:
            result: Image generation result
            prompt: Original prompt

        Returns:
            Formatted string with generation details
        """
        lines = [
            "๐Ÿ“ท Image generated successfully!",
            f"๐Ÿ–ผ๏ธ Image saved to: {result.image_path}",
            f"๐ŸŒฑ Seed used: {result.seed}",
            f"๐Ÿ“ Original prompt: {prompt}",
        ]

        # Add metadata if available
        if result.metadata:
            lines.append("๐Ÿ“Š Additional metadata:")
            for key, value in result.metadata.items():
                lines.append(f"   - {key}: {value}")

        return "\n".join(lines)


# -----------------------------------------------------------------------------
# UTILITY FUNCTIONS
# -----------------------------------------------------------------------------


def download_image(url: str, output_dir: Optional[str] = None) -> str:
    """
    Standalone utility to download an image from a URL.

    Args:
        url: Image URL
        output_dir: Directory to save image (created if doesn't exist)

    Returns:
        Path to downloaded image

    Raises:
        ValueError: If URL is invalid
        ConnectionError: If download fails
        IOError: If saving fails
    """
    if not url.startswith(("http://", "https://")):
        raise ValueError(f"Invalid URL: {url}")

    # Setup output directory
    if output_dir is None:
        output_dir = os.path.join(tempfile.gettempdir(), "flux_lora_images")
    os.makedirs(output_dir, exist_ok=True)

    try:
        # Download image
        response = requests.get(url, stream=True, timeout=30)
        response.raise_for_status()

        # Determine file extension
        content_type = response.headers.get("Content-Type", "")
        ext = ".jpg" if "jpeg" in content_type.lower() else ".png"

        # Save image
        output_path = os.path.join(output_dir, f"download_{uuid.uuid4().hex}{ext}")
        with open(output_path, "wb") as f:
            for chunk in response.iter_content(chunk_size=8192):
                f.write(chunk)

        return output_path

    except requests.RequestException as e:
        raise ConnectionError(f"Failed to download image: {str(e)}")
    except IOError as e:
        raise IOError(f"Failed to save image: {str(e)}")