services/whisper/utils/variations.py

"""
variations.py
---------------------------------------
Hook & Script Variation Engine (V8)

Purpose:
- Generate multiple viral hooks
- Create alternative script directions
- Support A/B testing of edits
- Enable Multi-Version Render pipeline

Works fully CPU-only.
No external API required.
"""

import random
import hashlib


# =====================================================
# HOOK TEMPLATES (VIRAL PATTERNS)
# =====================================================

HOOK_PATTERNS = [
    "You are not going to believe this...",
    "This is what nobody tells you about {}",
    "Stop scrolling if you want to understand {}",
    "The truth about {} will shock you",
    "Most people get {} wrong",
    "If you understand this, your {} changes forever",
    "I wish I knew this before about {}",
    "This is how you actually win at {}",
    "Everyone is lying about {}",
    "Watch this before it's too late..."
]


# =====================================================
# TEXT CLEANER
# =====================================================

def extract_keywords(words):
    """
    Extract simple keyword candidates from transcript
    """

    freq = {}

    for w in words:
        word = w["word"].lower().strip()
        if len(word) < 3:
            continue
        freq[word] = freq.get(word, 0) + 1

    sorted_words = sorted(freq.items(), key=lambda x: x[1], reverse=True)

    return [w[0] for w in sorted_words[:5]]


# =====================================================
# HOOK GENERATION
# =====================================================

def generate_hooks(words, count=5):
    """
    Generate multiple viral hooks from transcript
    """

    keywords = extract_keywords(words)

    hooks = []

    for i in range(count):

        template = random.choice(HOOK_PATTERNS)

        keyword = random.choice(keywords) if keywords else "this"

        try:
            hook = template.format(keyword)
        except:
            hook = template

        hooks.append(hook)

    return hooks


# =====================================================
# SCRIPT VARIATION ENGINE
# =====================================================

def generate_script_variations(words):
    """
    Creates alternative narrative directions
    """

    base_text = " ".join([w["word"] for w in words])

    variations = []

    variations.append({
        "style": "direct",
        "script": base_text
    })

    variations.append({
        "style": "emotional",
        "script": "Imagine this... " + base_text
    })

    variations.append({
        "style": "urgent",
        "script": "You need to hear this: " + base_text
    })

    variations.append({
        "style": "story",
        "script": "Let me tell you something important. " + base_text
    })

    return variations


# =====================================================
# CAPTION VARIATION ENGINE
# =====================================================

def generate_caption_variations(captions):
    """
    Creates multiple caption styles for rendering
    """

    styles = []

    for c in captions:

        styles.append({
            "style": "bold_center",
            "text": c["text"].upper()
        })

        styles.append({
            "style": "minimal",
            "text": c["text"]
        })

        styles.append({
            "style": "emphasis_words",
            "text": highlight_keywords(c["text"])
        })

    return styles


# =====================================================
# KEYWORD HIGHLIGHTER
# =====================================================

def highlight_keywords(text):
    """
    Emphasizes strong words in captions
    """

    keywords = ["you", "this", "stop", "now", "secret", "important"]

    words = text.split()

    output = []

    for w in words:
        if w.lower() in keywords:
            output.append(w.upper())
        else:
            output.append(w)

    return " ".join(output)


# =====================================================
# MULTI VERSION RENDER ENGINE
# =====================================================

def generate_render_variations(video_path, hooks=None):
    """
    Creates multiple render variants metadata
    (actual rendering happens in render.py)
    """

    if not hooks:
        hooks = ["Hook 1", "Hook 2", "Hook 3"]

    outputs = []

    for i, hook in enumerate(hooks):

        outputs.append({
            "version": i + 1,
            "hook": hook,
            "output_file": f"render_variant_{i+1}.mp4"
        })

    return outputs


# =====================================================
# DETERMINISTIC VIRAL HASH
# =====================================================

def viral_signature(text):
    """
    Creates deterministic ID for A/B testing consistency
    """

    return hashlib.md5(text.encode()).hexdigest()[:10]


# =====================================================
# PUBLIC API
# =====================================================

def generate_hooks_only(words):
    return generate_hooks(words)


def generate_full_variations(words):
    """
    Full pipeline for V8 Multi-Version system
    """

    hooks = generate_hooks(words)

    scripts = generate_script_variations(words)

    return {
        "hooks": hooks,
        "scripts": scripts
    }