model/Deep/model.py

import pandas as pd
import re

# =============================================================================
# CONFIGURATION - Update these paths as needed
# =============================================================================
ASR_RESULTS_PATH = "asr_results (2).csv"
SUBMISSION_PATH = "75.csv"

# =============================================================================
# LOAD DATA
# =============================================================================
print("Loading data files...")
df_asr = pd.read_csv(ASR_RESULTS_PATH)
df_submission = pd.read_csv(SUBMISSION_PATH)


# =============================================================================
# FUNCTION DEFINITIONS
# =============================================================================
def has_request_time(text):
    """
    Check if text contains time request pattern like "ขอเวลา ... นาที" or "ขอเวลา ... ชั่วโมง"

    Args:
        text: Input text to search

    Returns:
        bool: True if pattern is found, False otherwise
    """
    return bool(re.search(r"ขอเวลา\s*\d+\s*(นาที|ชั่วโมง)", str(text)))


def name_match(row, transcription):
    """
    Check if both first and last name appear in transcription

    Args:
        row: DataFrame row containing first_name and last_name
        transcription: Text to search in

    Returns:
        bool: True if both names are found, False otherwise
    """
    first = str(row["first_name"])
    last = str(row["last_name"])
    text = str(transcription)
    return (first in text) and (last in text)


# =============================================================================
# TASK 1: Find rows containing "สวัสดี" (greeting)
# =============================================================================
print("\n=== TASK 1: Finding greeting patterns ===")
rows_with_sawasdee = df_asr[
    df_asr.apply(lambda row: row.astype(str).str.contains("สวัสดี").any(), axis=1)
]
print(f"Found {len(rows_with_sawasdee)} rows with greeting patterns")
print(rows_with_sawasdee)

# Update submission file for greeting column
matching_ids_greeting = set(rows_with_sawasdee["id"])
mask_greeting = df_submission["id"].isin(matching_ids_greeting) & (
    df_submission["กล่าวสวัสดี"] == False
)
num_changed_greeting = mask_greeting.sum()

# Apply changes to submission file
df_submission.loc[mask_greeting, "กล่าวสวัสดี"] = True
print(f"Number of rows updated for greeting: {num_changed_greeting}")

# =============================================================================
# TASK 2: Find rows with time request patterns
# =============================================================================
print("\n=== TASK 2: Finding time request patterns ===")
rows_with_request_time = df_asr[df_asr["transcription"].apply(has_request_time)]
print(f"Found {len(rows_with_request_time)} rows with time request patterns")
print(rows_with_request_time[["id", "transcription"]])

# Update submission file for time request column
request_time_ids = set(rows_with_request_time["id"])
mask_time = df_submission["id"].isin(request_time_ids) & (
    df_submission["บอกระยะเวลาที่ใช้ในการเข้าพบ"] == False
)
num_changed_time = mask_time.sum()

# Apply changes to submission file
df_submission.loc[mask_time, "บอกระยะเวลาที่ใช้ในการเข้าพบ"] = True
print(f"Number of rows updated for time request: {num_changed_time}")

# =============================================================================
# TASK 3: Name matching analysis
# =============================================================================
print("\n=== TASK 3: Name matching analysis ===")

# Find name matches using the updated submission file
matched_rows = []
for _, sub_row in df_submission.iterrows():
    # Find all rows in df_asr where both first and last name appear in transcription
    matches = df_asr[df_asr["transcription"].apply(lambda t: name_match(sub_row, t))]
    for _, asr_row in matches.iterrows():
        matched_rows.append(
            {
                "id": asr_row["id"],
                "first_name": sub_row["first_name"],
                "last_name": sub_row["last_name"],
                "transcription": asr_row["transcription"],
            }
        )

# Convert to DataFrame and display results
df_matched = pd.DataFrame(matched_rows)
print(f"Found {len(df_matched)} name matches")
print(df_matched)

# =============================================================================
# SAVE UPDATED SUBMISSION FILE
# =============================================================================
print("\n=== SAVING UPDATED SUBMISSION FILE ===")
df_submission.to_csv(SUBMISSION_PATH, index=False)
print(f"Updated submission file saved as: {SUBMISSION_PATH}")

# =============================================================================
# SUMMARY
# =============================================================================
print("\n=== PROCESSING SUMMARY ===")
print(f"✓ Greeting patterns found: {len(rows_with_sawasdee)}")
print(f"✓ Greeting updates applied: {num_changed_greeting}")
print(f"✓ Time request patterns found: {len(rows_with_request_time)}")
print(f"✓ Time request updates applied: {num_changed_time}")
print(f"✓ Name matches found: {len(df_matched)}")
print(f"✓ All changes applied to: {SUBMISSION_PATH}")
print(f"✓ Source data from: {ASR_RESULTS_PATH}")