File size: 25,483 Bytes
190894a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
# File: prompts.py

DOCUMENT_OUTLINE_PROMPT_SYSTEM = """You are a document generator. Provide the outline of the document requested in <prompt></prompt> in JSON format.
Include sections and subsections if required. Use the "Content" field to provide a specific prompt or instruction for generating content for that particular section or subsection.
make sure the Sections follow a logical flow and each prompt's content does not overlap with other sections.
OUTPUT IN FOLLOWING JSON FORMAT enclosed in <output> tags
<output>
{
    "Document": {
      "Title": "Document Title",
      "Author": "Author Name",
      "Date": "YYYY-MM-DD",
      "Version": "1.0",

      "Sections": [
        {
          "SectionNumber": "1",
          "Title": "Section Title",
          "Content": "Specific prompt or instruction for generating content for this section",
          "Subsections": [
            {
              "SectionNumber": "1.1",
              "Title": "Subsection Title",
              "Content": "Specific prompt or instruction for generating content for this subsection"
            }
          ]
        }
      ]
    }
  }
</output>"""

DOCUMENT_OUTLINE_PROMPT_USER = """<prompt>{query}</prompt>"""

DOCUMENT_SECTION_PROMPT_SYSTEM = """You are a document generator, You need to output only the content requested in the section in the prompt.
FORMAT YOUR OUTPUT AS MARKDOWN ENCLOSED IN <response></response> tags
<overall_objective>{overall_objective}</overall_objective>
<document_layout>{document_layout}</document_layout>"""

DOCUMENT_SECTION_PROMPT_USER = """<prompt>Output the content for the section "{section_or_subsection_title}" formatted as markdown. Follow this instruction: {content_instruction}</prompt>"""

##########################################

DOCUMENT_TEMPLATE_OUTLINE_PROMPT_SYSTEM = """You are a document template generator. Provide the outline of the document requested in <prompt></prompt> in JSON format.
Include sections and subsections if required. Use the "Content" field to provide a specific prompt or instruction for generating template with placeholder text /example content for that particular section or subsection. Specify in each prompt to output as a template and use placeholder text/ tables as necessory.
make sure the Sections follow a logical flow and each prompt's content does not overlap with other sections.
OUTPUT IN FOLLOWING JSON FORMAT enclosed in <output> tags
<output>
{
    "Document": {
      "Title": "Document Title",
      "Author": "Author Name",
      "Date": "YYYY-MM-DD",
      "Version": "1.0",

      "Sections": [
        {
          "SectionNumber": "1",
          "Title": "Section Title",
          "Content": "Specific prompt or instruction for generating template for this section",
          "Subsections": [
            {
              "SectionNumber": "1.1",
              "Title": "Subsection Title",
              "Content": "Specific prompt or instruction for generating template for this subsection"
            }
          ]
        }
      ]
    }
  }
</output>"""

DOCUMENT_TEMPLATE_PROMPT_USER = """<prompt>{query}</prompt>"""

DOCUMENT_TEMPLATE_SECTION_PROMPT_SYSTEM = """You are a document template generator,You need to output only the content requested in the section in the prompt, Use placeholder text/examples/tables wherever required.
FORMAT YOUR OUTPUT AS A TEMPLATE ENCLOSED IN <response></response> tags
<overall_objective>{overall_objective}</overall_objective>
<document_layout>{document_layout}</document_layout>"""

DOCUMENT_TEMPLATE_SECTION_PROMPT_USER = """<prompt>Output the content for the section "{section_or_subsection_title}" formatted as markdown. Follow this instruction: {content_instruction}</prompt>"""


# File: llm_observability.py

import sqlite3
import json
from datetime import datetime
from typing import Dict, Any, List, Optional

class LLMObservabilityManager:
    def __init__(self, db_path: str = "llm_observability.db"):
        self.db_path = db_path
        self.create_table()

    def create_table(self):
        with sqlite3.connect(self.db_path) as conn:
            cursor = conn.cursor()
            cursor.execute('''
                CREATE TABLE IF NOT EXISTS llm_observations (
                    id TEXT PRIMARY KEY,
                    conversation_id TEXT,
                    created_at DATETIME,
                    status TEXT,
                    request TEXT,
                    response TEXT,
                    model TEXT,
                    total_tokens INTEGER,
                    prompt_tokens INTEGER,
                    completion_tokens INTEGER,
                    latency FLOAT,
                    user TEXT
                )
            ''')

    def insert_observation(self, response: Dict[str, Any], conversation_id: str, status: str, request: str, latency: float, user: str):
        created_at = datetime.fromtimestamp(response['created'])
        
        with sqlite3.connect(self.db_path) as conn:
            cursor = conn.cursor()
            cursor.execute('''
                INSERT INTO llm_observations 
                (id, conversation_id, created_at, status, request, response, model, total_tokens, prompt_tokens, completion_tokens, latency, user)
                VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
            ''', (
                response['id'],
                conversation_id,
                created_at,
                status,
                request,
                json.dumps(response['choices'][0]['message']),
                response['model'],
                response['usage']['total_tokens'],
                response['usage']['prompt_tokens'],
                response['usage']['completion_tokens'],
                latency,
                user
            ))

    def get_observations(self, conversation_id: Optional[str] = None) -> List[Dict[str, Any]]:
        with sqlite3.connect(self.db_path) as conn:
            cursor = conn.cursor()
            if conversation_id:
                cursor.execute('SELECT * FROM llm_observations WHERE conversation_id = ? ORDER BY created_at', (conversation_id,))
            else:
                cursor.execute('SELECT * FROM llm_observations ORDER BY created_at')
            rows = cursor.fetchall()

            column_names = [description[0] for description in cursor.description]
            return [dict(zip(column_names, row)) for row in rows]

    def get_all_observations(self) -> List[Dict[str, Any]]:
        return self.get_observations()


# File: app.py
import os
import json
import re
import asyncio
import time
from typing import List, Dict, Optional, Any, Callable, Union
from openai import OpenAI
import logging
import functools
from fastapi import APIRouter, HTTPException, Request, UploadFile, File, Depends
from fastapi.responses import StreamingResponse
from pydantic import BaseModel
from fastapi_cache.decorator import cache
import psycopg2
from datetime import datetime
import base64
from fastapi import Form 

logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
logger = logging.getLogger(__name__)

def log_execution(func: Callable) -> Callable:
    @functools.wraps(func)
    def wrapper(*args: Any, **kwargs: Any) -> Any:
        logger.info(f"Executing {func.__name__}")
        try:
            result = func(*args, **kwargs)
            logger.info(f"{func.__name__} completed successfully")
            return result
        except Exception as e:
            logger.error(f"Error in {func.__name__}: {e}")
            raise
    return wrapper

# aiclient.py

class AIClient:
    def __init__(self):
        self.client = OpenAI(
            base_url="https://openrouter.ai/api/v1",
            api_key="sk-or-v1-" + os.environ['OPENROUTER_API_KEY']
        )
        self.observability_manager = LLMObservabilityManager()

    @log_execution
    def generate_response(
        self,
        messages: List[Dict[str, str]],
        model: str = "openai/gpt-4o-mini",
        max_tokens: int = 32000,
        conversation_id: str = None,
        user: str = "anonymous"
    ) -> Optional[str]:
        if not messages:
            return None
        
        start_time = time.time()
        response = self.client.chat.completions.create(
            model=model,
            messages=messages,
            max_tokens=max_tokens,
            stream=False
        )
        end_time = time.time()
        latency = end_time - start_time

        # Log the observation
        self.observability_manager.insert_observation(
            response=response.dict(),
            conversation_id=conversation_id or "default",
            status="success",
            request=json.dumps(messages),
            latency=latency,
            user=user
        )

        return response.choices[0].message.content

    @log_execution
    def generate_vision_response(
        self,
        messages: List[Dict[str, Union[str, List[Dict[str, Union[str, Dict[str, str]]]]]]],
        model: str = "google/gemini-flash-1.5-8b",
        max_tokens: int = 32000,
        conversation_id: str = None,
        user: str = "anonymous"
    ) -> Optional[str]:
        if not messages:
            return None
        
        start_time = time.time()
        response = self.client.chat.completions.create(
            model=model,
            messages=messages,
            max_tokens=max_tokens,
            stream=False
        )
        end_time = time.time()
        latency = end_time - start_time

        # Log the observation
        self.observability_manager.insert_observation(
            response=response.dict(),
            conversation_id=conversation_id or "default",
            status="success",
            request=json.dumps(messages),
            latency=latency,
            user=user
        )

        return response.choices[0].message.content


class VisionTools:
    def __init__(self, ai_client):
        self.ai_client = ai_client

    async def extract_images_info(self, images: List[UploadFile]) -> str:
        try:
            image_contents = []
            for image in images:
                image_content = await image.read()
                base64_image = base64.b64encode(image_content).decode('utf-8')
                image_contents.append({
                    "type": "image_url",
                    "image_url": {
                        "url": f"data:image/jpeg;base64,{base64_image}"
                    }
                })
            
            messages = [
                {
                    "role": "user",
                    "content": [
                        {
                            "type": "text",
                            "text": "Extract the contents of these images in detail in a structured format, focusing on any text, tables, diagrams, or visual elements that might be relevant for document generation."
                        },
                        *image_contents
                    ]
                }
            ]
            
            image_context = self.ai_client.generate_vision_response(messages)
            return image_context
        except Exception as e:
            print(f"Error processing images: {str(e)}")
            return ""


class DatabaseManager:
    """Manages database operations."""

    def __init__(self):
        self.db_params = {
            "dbname": "postgres",
            "user": os.environ['SUPABASE_USER'],
            "password": os.environ['SUPABASE_PASSWORD'],
            "host": "aws-0-us-west-1.pooler.supabase.com",
            "port": "5432"
        }

    @log_execution
    def update_database(self, user_id: str, user_query: str, response: str) -> None:
        with psycopg2.connect(**self.db_params) as conn:
            with conn.cursor() as cur:
                insert_query = """
                INSERT INTO ai_document_generator (user_id, user_query, response)
                VALUES (%s, %s, %s);
                """
                cur.execute(insert_query, (user_id, user_query, response))
        
class DocumentGenerator:
    def __init__(self, ai_client: AIClient):
        self.ai_client = ai_client
        self.document_outline = None
        self.content_messages = []

    @staticmethod
    def extract_between_tags(text: str, tag: str) -> str:
        pattern = f"<{tag}>(.*?)</{tag}>"
        match = re.search(pattern, text, re.DOTALL)
        return match.group(1).strip() if match else ""

    @staticmethod
    def remove_duplicate_title(content: str, title: str, section_number: str) -> str:
        patterns = [
            rf"^#+\s*{re.escape(section_number)}(?:\s+|\s*:\s*|\.\s*){re.escape(title)}",
            rf"^#+\s*{re.escape(title)}",
            rf"^{re.escape(section_number)}(?:\s+|\s*:\s*|\.\s*){re.escape(title)}",
            rf"^{re.escape(title)}",
        ]
        
        for pattern in patterns:
            content = re.sub(pattern, "", content, flags=re.MULTILINE | re.IGNORECASE)
        
        return content.lstrip()

    @log_execution
    def generate_document_outline(self, query: str, template: bool = False, image_context: str = "", max_retries: int = 3) -> Optional[Dict]:
        messages = [
            {"role": "system", "content": DOCUMENT_OUTLINE_PROMPT_SYSTEM if not template else DOCUMENT_TEMPLATE_OUTLINE_PROMPT_SYSTEM},
            {"role": "user", "content": DOCUMENT_OUTLINE_PROMPT_USER.format(query=query) if not template else DOCUMENT_TEMPLATE_PROMPT_USER.format(query=query, image_context=image_context)}
        ]
        # Update user content to include image context if provided
        if image_context:
            messages[1]["content"] += f"<attached_images>\n\n{image_context}\n\n</attached_images>"
        
        for attempt in range(max_retries):
            outline_response = self.ai_client.generate_response(messages, model="openai/gpt-4o")
            outline_json_text = self.extract_between_tags(outline_response, "output")
            
            try:
                self.document_outline = json.loads(outline_json_text)
                return self.document_outline
            except json.JSONDecodeError as e:
                if attempt < max_retries - 1:
                    logger.warning(f"Failed to parse JSON (attempt {attempt + 1}): {e}")
                    logger.info("Retrying...")
                else:
                    logger.error(f"Failed to parse JSON after {max_retries} attempts: {e}")
                    return None

    @log_execution
    def generate_content(self, title: str, content_instruction: str, section_number: str, template: bool = False) -> str:
        SECTION_PROMPT_USER = DOCUMENT_SECTION_PROMPT_USER if not template else DOCUMENT_TEMPLATE_SECTION_PROMPT_USER
        self.content_messages.append({
            "role": "user",
            "content": SECTION_PROMPT_USER.format(
                section_or_subsection_title=title,
                content_instruction=content_instruction
            )
        })
        section_response = self.ai_client.generate_response(self.content_messages)
        content = self.extract_between_tags(section_response, "response")
        content = self.remove_duplicate_title(content, title, section_number)
        self.content_messages.append({
            "role": "assistant",
            "content": section_response
        })
        return content

class MarkdownConverter:
    @staticmethod
    def slugify(text: str) -> str:
        return re.sub(r'\W+', '-', text.lower())

    @classmethod
    def generate_toc(cls, sections: List[Dict]) -> str:
        toc = "<div style='page-break-before: always;'></div>\n\n"
        toc += "<h2 style='color: #2c3e50; text-align: center;'>Table of Contents</h2>\n\n"
        toc += "<nav style='background-color: #f8f9fa; padding: 20px; border-radius: 5px; line-height: 1.6;'>\n\n"
        for section in sections:
            section_number = section['SectionNumber']
            section_title = section['Title']
            toc += f"<p><a href='#{cls.slugify(section_title)}' style='color: #3498db; text-decoration: none;'>{section_number}. {section_title}</a></p>\n\n"
            
            for subsection in section.get('Subsections', []):
                subsection_number = subsection['SectionNumber']
                subsection_title = subsection['Title']
                toc += f"<p style='margin-left: 20px;'><a href='#{cls.slugify(subsection_title)}' style='color: #2980b9; text-decoration: none;'>{subsection_number} {subsection_title}</a></p>\n\n"
        
        toc += "</nav>\n\n"
        return toc

    @classmethod
    def convert_to_markdown(cls, document: Dict) -> str:
        markdown = "<div style='text-align: center; padding-top: 33vh;'>\n\n"
        markdown += f"<h1 style='color: #2c3e50; border-bottom: 2px solid #3498db; padding-bottom: 10px; display: inline-block;'>{document['Title']}</h1>\n\n"
        markdown += f"<p style='color: #7f8c8d;'><em>By {document['Author']}</em></p>\n\n"
        markdown += f"<p style='color: #95a5a6;'>Version {document['Version']} | {document['Date']}</p>\n\n"
        markdown += "</div>\n\n"
        
        markdown += cls.generate_toc(document['Sections'])
        
        markdown += "<div style='max-width: 800px; margin: 0 auto; font-family: \"Segoe UI\", Arial, sans-serif; line-height: 1.6;'>\n\n"
        
        for section in document['Sections']:
            markdown += "<div style='page-break-before: always;'></div>\n\n"
            section_number = section['SectionNumber']
            section_title = section['Title']
            markdown += f"<h2 id='{cls.slugify(section_title)}' style='color: #2c3e50; border-bottom: 1px solid #bdc3c7; padding-bottom: 5px;'>{section_number}. {section_title}</h2>\n\n"
            markdown += f"<div style='color: #34495e; margin-bottom: 20px;'>\n\n{section['Content']}\n\n</div>\n\n"
            
            for subsection in section.get('Subsections', []):
                subsection_number = subsection['SectionNumber']
                subsection_title = subsection['Title']
                markdown += f"<h3 id='{cls.slugify(subsection_title)}' style='color: #34495e;'>{subsection_number} {subsection_title}</h3>\n\n"
                markdown += f"<div style='color: #34495e; margin-bottom: 20px;'>\n\n{subsection['Content']}\n\n</div>\n\n"
        
        markdown += "</div>"
        return markdown

router = APIRouter()

class JsonDocumentResponse(BaseModel):
    json_document: Dict

# class JsonDocumentRequest(BaseModel):
#     query: str
#     template: bool = False
#     images: Optional[List[UploadFile]] = File(None)
#     documents: Optional[List[UploadFile]] = File(None)
#     conversation_id: str = ""

class MarkdownDocumentRequest(BaseModel):
    json_document: Dict
    query: str

MESSAGE_DELIMITER = b"\n---DELIMITER---\n"

def yield_message(message):
    message_json = json.dumps(message, ensure_ascii=False).encode('utf-8')
    return message_json + MESSAGE_DELIMITER

async def generate_document_stream(document_generator: DocumentGenerator, document_outline: Dict, query: str, template: bool = False):
    document_generator.document_outline = document_outline
    db_manager = DatabaseManager()
    overall_objective = query
    document_layout = json.dumps(document_generator.document_outline, indent=2)

    SECTION_PROMPT_SYSTEM = DOCUMENT_SECTION_PROMPT_SYSTEM if not template else DOCUMENT_TEMPLATE_SECTION_PROMPT_SYSTEM
    document_generator.content_messages = [
        {
            "role": "system",
            "content": SECTION_PROMPT_SYSTEM.format(
                overall_objective=overall_objective,
                document_layout=document_layout
            )
        }
    ]

    for section in document_generator.document_outline["Document"].get("Sections", []):
        section_title = section.get("Title", "")
        section_number = section.get("SectionNumber", "")
        content_instruction = section.get("Content", "")
        logging.info(f"Generating content for section: {section_title}")
        content = document_generator.generate_content(section_title, content_instruction, section_number, template)
        section["Content"] = content
        yield yield_message({
            "type": "document_section",
            "content": {
                "section_number": section_number,
                "section_title": section_title,
                "content": content
            }
        })

        for subsection in section.get("Subsections", []):
            subsection_title = subsection.get("Title", "")
            subsection_number = subsection.get("SectionNumber", "")
            subsection_content_instruction = subsection.get("Content", "")
            logging.info(f"Generating content for subsection: {subsection_title}")
            content = document_generator.generate_content(subsection_title, subsection_content_instruction, subsection_number, template)
            subsection["Content"] = content
            yield yield_message({
                "type": "document_section",
                "content": {
                    "section_number": subsection_number,
                    "section_title": subsection_title,
                    "content": content
                }
            })

    markdown_document = MarkdownConverter.convert_to_markdown(document_generator.document_outline["Document"])
    
    yield yield_message({
        "type": "complete_document",
        "content": {
            "markdown": markdown_document,
            "json": document_generator.document_outline
                    },
            });

    db_manager.update_database("elevatics", query, markdown_document)

@router.post("/generate-document/markdown-stream")
async def generate_markdown_document_stream_endpoint(request: MarkdownDocumentRequest):
    ai_client = AIClient()
    document_generator = DocumentGenerator(ai_client)
    
    async def stream_generator():
        try:
            async for chunk in generate_document_stream(document_generator, request.json_document, request.query, request.template):
                yield chunk
        except Exception as e:
            yield yield_message({
                "type": "error",
                "content": str(e)
            })

    return StreamingResponse(stream_generator(), media_type="application/octet-stream")


@cache(expire=600*24*7)
@router.post("/generate-document/json", response_model=JsonDocumentResponse)
async def generate_document_outline_endpoint(
    query: str = Form(...),
    template: bool = Form(False),
    conversation_id: str = Form(...),
    images: Optional[List[UploadFile]] = File(None),
    documents: Optional[List[UploadFile]] = File(None)
):
    ai_client = AIClient()
    document_generator = DocumentGenerator(ai_client)
    vision_tools = VisionTools(ai_client)
    
    try:
        image_context = ""
        if images:
            image_context = await vision_tools.extract_images_info(images)
        
        json_document = document_generator.generate_document_outline(
            query,
            template,
            image_context=image_context
        )
        
        if json_document is None:
            raise HTTPException(status_code=500, detail="Failed to generate a valid document outline")
        
        return JsonDocumentResponse(json_document=json_document)
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))



## OBSERVABILITY
from uuid import uuid4
import csv
from io import StringIO

class ObservationResponse(BaseModel):
    observations: List[Dict]
    
def create_csv_response(observations: List[Dict]) -> StreamingResponse:
    def iter_csv(data):
        output = StringIO()
        writer = csv.DictWriter(output, fieldnames=data[0].keys() if data else [])
        writer.writeheader()
        for row in data:
            writer.writerow(row)
        output.seek(0)
        yield output.read()

    headers = {
        'Content-Disposition': 'attachment; filename="observations.csv"'
    }
    return StreamingResponse(iter_csv(observations), media_type="text/csv", headers=headers)
    

@router.get("/last-observations/{limit}")
async def get_last_observations(limit: int = 10, format: str = "json"):
    observability_manager = LLMObservabilityManager()
    
    try:
        # Get all observations, sorted by created_at in descending order
        all_observations = observability_manager.get_observations()
        all_observations.sort(key=lambda x: x['created_at'], reverse=True)
        
        # Get the last conversation_id
        if all_observations:
            last_conversation_id = all_observations[0]['conversation_id']
            
            # Filter observations for the last conversation
            last_conversation_observations = [
                obs for obs in all_observations
                if obs['conversation_id'] == last_conversation_id
            ][:limit]
            
            if format.lower() == "csv":
                return create_csv_response(last_conversation_observations)
            else:
                return ObservationResponse(observations=last_conversation_observations)
        else:
            if format.lower() == "csv":
                return create_csv_response([])
            else:
                return ObservationResponse(observations=[])
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Failed to retrieve observations: {str(e)}")