File size: 12,612 Bytes
5536b2b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
from click import prompt
from langchain_core.runnables import RunnableParallel
from langchain_core.runnables import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser
from streamlit import text_input
from langchain_community.vectorstores import FAISS
from langchain_ollama import OllamaEmbeddings
from langchain_ollama import ChatOllama
from langchain_core.prompts import PromptTemplate
from ollama import chat
from langchain_core.runnables import RunnableParallel 

#### ************************************************************************************SQL***********************************************************************************

def sql_text(text_input):
    embedding = OllamaEmbeddings(model="nomic-embed-text")

    vector_store = FAISS.load_local(
        "faiss_index",
        embedding,
        allow_dangerous_deserialization=True
    )

    retriever = vector_store.as_retriever(search_type="similarity", search_kwargs={"k": 1})


    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )

    prompt = PromptTemplate(
        template="""

    You are an SQL assistant.



    Use ONLY the provided context.



    Rules:

    1. If the SQL query exists in the context, return ONLY that SQL query.

    2. Do not explain.

    3. Do not rewrite the question.

    4. Do not generate a new query.

    5. If the SQL query is missing, reply exactly:

    I don't know.



    Context:

    {context}



    Question:

    {question}



    SQL:

    """,
        input_variables=["context", "question"],
    )


    parallel_chain = RunnableParallel({
        'context': retriever ,
        'question': RunnablePassthrough()
    })

    parser = StrOutputParser()

    main_chain = parallel_chain | prompt | llm | parser


    response = main_chain.invoke(text_input)
    return response






# def image_input( image_path):

#     system_prompt = """
#     Act as an expert OCR and data extraction assistant.

#     Please format your response clearly using the following headings:

#     ### User Statement
#     Extract the introductory text written by the user at the very top of the image.

#     ### Question Description
#     Extract the problem statement from the left-hand panel under the Description tab.

#     ### SQL Query
#     Extract the exact SQL code written inside the dark-themed code editor on the right.

#     ### Error Message
#     Extract the red error message displayed at the bottom right.
#     """

#     response = chat(
#         model="qwen2.5vl:7b",
#         messages=[
#             {
#                 "role": "system",
#                 "content": system_prompt
#             },
#             {
#                 "role": "user",
#                 "content": "Extract all requested information from this image.",
#                 "images": ["sql_eorror.png"]
#             }
#         ]
#     )

#     response_image_test =response.message.content


#     llm = ChatOllama(
#         model="llama3.2",
#         temperature=0
#     )


#     embedding = OllamaEmbeddings(model="nomic-embed-text")

#     vector_store = FAISS.load_local(
#         "faiss_index",
#         embedding,
#         allow_dangerous_deserialization=True
#     )

#     retriever = vector_store.as_retriever(
#         search_type="similarity",
#         search_kwargs={"k": 1}
#     )




#     prompt = PromptTemplate(
#         template="""
#     You are an SQL assistant.

#     Use ONLY the provided context.

#     Rules:
#     1. If the SQL query exists in the context, return ONLY that SQL query.
#     2. Do not explain.
#     3. Do not rewrite the question.
#     4. Do not generate a new query.
#     5. If the SQL query is missing, reply exactly:
#     I don't know.

#     Context:
#     {context}

#     Question:
#     {question}

#     SQL:
#     """,
#         input_variables=["context", "question"],
#     )


#     parallel_chain = RunnableParallel({
#         'context': retriever ,
#         'question': RunnablePassthrough()
#     })

#     parser = StrOutputParser()

#     main_chain = parallel_chain | prompt | llm | parser


#     response = main_chain.invoke(response_image_test)
#     return response




def sql_image(image_path):

    # --------------------------------------------------
    # STEP 1 : Extract ONLY the SQL question
    # --------------------------------------------------
    system_prompt = """

You are an OCR assistant.



Your job is to read the screenshot and extract ONLY the SQL problem statement.



Rules:

- Ignore the SQL editor.

- Ignore the error message.

- Ignore buttons.

- Ignore the database output.

- Ignore everything except the Question Description.



Return ONLY the question text.

Do not add headings.

Do not explain anything.

"""

    response = chat(
        model="qwen2.5vl:7b",
        messages=[
            {
                "role": "system",
                "content": system_prompt
            },
            {
                "role": "user",
                "content": "Extract only the SQL question.",
                "images": [image_path]
            }
        ]
    )

    extracted_question = response.message.content.strip()

    print("\nExtracted Question:\n")
    print(extracted_question)

    # --------------------------------------------------
    # STEP 2 : Load LLM
    # --------------------------------------------------

    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )

    # --------------------------------------------------
    # STEP 3 : Load Embeddings
    # --------------------------------------------------

    embedding = OllamaEmbeddings(
        model="nomic-embed-text"
    )

    vector_store = FAISS.load_local(
        "faiss_index",
        embedding,
        allow_dangerous_deserialization=True
    )

    retriever = vector_store.as_retriever(
        search_type="similarity",
        search_kwargs={"k":3}
    )

    # --------------------------------------------------
    # STEP 4 : Prompt
    # --------------------------------------------------

    prompt = PromptTemplate(
        template="""

You are an SQL assistant.



Use ONLY the retrieved context.



Rules:



1. Return ONLY the SQL query.

2. No explanation.

3. No markdown.

4. Do not rewrite the question.

5. If the answer is not present in the context, reply exactly:



I don't know.



Context:

{context}



Question:

{question}



SQL:

""",
        input_variables=["context", "question"],
    )

    # --------------------------------------------------
    # STEP 5 : Chain
    # --------------------------------------------------

    parallel_chain = RunnableParallel(
        {
            "context": retriever,
            "question": RunnablePassthrough()
        }
    )

    parser = StrOutputParser()

    main_chain = (
        parallel_chain
        | prompt
        | llm
        | parser
    )

    # --------------------------------------------------
    # STEP 6 : Retrieve using ONLY the extracted question
    # --------------------------------------------------

    response = main_chain.invoke(extracted_question)

    return response





def sql_text_image(question, image_path):
    # --------------------------------------------------
    # STEP 1 : Extract ONLY the SQL question from the image
    # --------------------------------------------------
    sql_answer1 = sql_image(image_path)
    sql_answer2 = sql_text(question)

    # --------------------------------------------------
    # STEP 2 : Combine the extracted question and the text input
    # --------------------------------------------------
    # Build the Parallel Chain
    prompt = f"""

You are an SQL expert.



Image SQL:

{sql_answer1}



User Question:

{sql_answer2}



Use BOTH pieces of information to produce the final SQL answer.



Output exactly in this format:



Kindly go through the provided information.

<final_answer> -> By Prince Kushwaha



Do not include any explanation or markdown.

Replace <final_answer> with the actual SQL answer.

"""
    

    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )
    response = llm.invoke(prompt)

    return response.content





### *************************************************************************Excel**************************************************************************



def excel_text(question):
        # Load LLM
    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )

    # Prompt
    prompt = PromptTemplate(
        template="""

    You are an Excel expert.



    Answer the following Excel question.



    Rules:

    1. Return only the final answer.

    2. If the answer is an Excel formula, return only the formula.

    3. Do not use markdown.

    4. Keep the answer concise.



    Question:

    {question}



    Answer:

    """,
        input_variables=["question"],
    )

    # Output Parser
    parser = StrOutputParser()

    # Chain
    main_chain = prompt | llm | parser

    # Generate Response
    response = main_chain.invoke({
        "question": text_input
    })

    return response.content.strip()









def excel_image(image_path):
    # --------------------------------------------------
    # STEP 1 : Extract ONLY the Excel question
    # --------------------------------------------------

    system_prompt = """

You are an OCR assistant.



Your task is to read the screenshot and extract ONLY the Excel question.



Rules:

- Ignore the Excel ribbon.

- Ignore row and column headers.

- Ignore formulas already written.

- Ignore buttons and menus.

- Ignore answer choices if not part of the question.

- Return ONLY the question text.



Do not explain anything.

Do not add headings.

"""

    response = chat(
        model="qwen2.5vl:7b",
        messages=[
            {
                "role": "system",
                "content": system_prompt
            },
            {
                "role": "user",
                "content": "Extract only the Excel question.",
                "images": [image_path]
            }
        ]
    )

    extracted_question = response.message.content.strip()

    print("\nExtracted Question:\n")
    print(extracted_question)

    # --------------------------------------------------
    # STEP 2 : Load LLM
    # --------------------------------------------------

    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )

    # --------------------------------------------------
    # STEP 3 : Prompt
    # --------------------------------------------------

    prompt = PromptTemplate(
        template="""

You are an Excel expert.



Answer the following Excel question.



Rules:

1. Provide the correct answer only.

2. If a formula is required, return only the formula.

3. If an explanation is required, keep it concise.

4. Do not use markdown.



Question:

{question}



Answer:

""",
        input_variables=["question"],
    )

    parser = StrOutputParser()

    chain = prompt | llm | parser

    # --------------------------------------------------
    # STEP 4 : Generate Answer
    # --------------------------------------------------

    answer = chain.invoke({
        "question": extracted_question
    })

    return answer 

def excel_text_image(question, image_path):
    """

    question   : Extracted text from OCR

    image_path : Path to the Excel screenshot (optional, not used here)

    """

    # Load LLM
    llm = ChatOllama(
        model="llama3.2",
        temperature=0
    )

    # Prompt
    prompt = PromptTemplate(
        template="""

You are an Excel expert.



Answer the following Excel question.



Rules:

1. Return only the final answer.

2. If the answer is an Excel formula, return only the formula.

3. If the question asks for a value, return only the value.

4. If an explanation is needed, keep it brief.

5. Do not use markdown.



Question:

{question}



Answer:

""",
        input_variables=["question"],
    )

    parser = StrOutputParser()

    # Chain
    chain = prompt | llm | parser

    # Generate answer
    response = chain.invoke({
        "question": question
    })

    return response