import base64
import json
import logging
import os
import random
import re
from PIL import Image
import pytesseract
import io
import tempfile
import requests
import time
import openai
from dotenv import load_dotenv
from datetime import datetime, timezone, timedelta
from openai import OpenAI
from urllib.parse import quote
from bs4 import BeautifulSoup
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
import tweepy
import flickr_api
from foodie_config import (
    RECIPE_KEYWORDS, PROMO_KEYWORDS, HOME_KEYWORDS, PRODUCT_KEYWORDS, PERSONA_CONFIGS, 
    get_clean_source_name, AUTHORS, LIGHT_TASK_MODEL, SUMMARY_MODEL, X_API_CREDENTIALS,
    FLICKR_API_KEY, FLICKR_API_SECRET, PIXABAY_API_KEY
)

load_dotenv()
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

def load_json_file(file_path, expiration_hours):
    entries = []
    cutoff = datetime.now(timezone.utc) - timedelta(hours=expiration_hours)
    
    if not os.path.exists(file_path):
        logging.info(f"File {file_path} does not exist, returning empty list")
        return entries
    
    try:
        with open(file_path, 'r') as f:
            lines = f.readlines()
        
        for i, line in enumerate(lines, 1):
            try:
                entry = json.loads(line.strip())
                if not isinstance(entry, dict) or "title" not in entry or "timestamp" not in entry:
                    logging.warning(f"Skipping malformed entry in {file_path} at line {i}: {line.strip()}")
                    continue
                
                timestamp = datetime.fromisoformat(entry["timestamp"])
                if timestamp > cutoff:
                    entries.append(entry)
                else:
                    logging.debug(f"Entry expired in {file_path}: {entry['title']}")
            except json.JSONDecodeError as e:
                logging.warning(f"Skipping invalid JSON line in {file_path} at line {i}: {e}")
                continue
            except Exception as e:
                logging.warning(f"Skipping malformed entry in {file_path} at line {i}: {line.strip()}")
                continue
        
        logging.info(f"Loaded {len(entries)} entries from {file_path}, {len(entries)} valid after expiration check")
        return entries
    except Exception as e:
        logging.error(f"Failed to load {file_path}: {e}")
        return entries

def save_json_file(file_path, title, timestamp):
    try:
        entries = load_json_file(file_path, 24 if "posted_" in file_path else 7 * 24)  # 24 hours for titles, 7 days for images
        entry = {"title": title, "timestamp": timestamp}
        entries.append(entry)
        
        # Prune entries older than expiration period
        expiration_hours = 24 if "posted_" in file_path else 7 * 24
        cutoff = datetime.now(timezone.utc) - timedelta(hours=expiration_hours)
        pruned_entries = [e for e in entries if datetime.fromisoformat(e["timestamp"]) > cutoff]
        
        with open(file_path, 'w') as f:
            for entry in pruned_entries:
                f.write(json.dumps(entry) + '\n')
        
        logging.info(f"Saved '{title}' to {file_path}")
        logging.info(f"Pruned {file_path} to {len(pruned_entries)} entries (older than {expiration_hours//24} days removed)")
    except Exception as e:
        logging.error(f"Failed to save to {file_path}: {e}")

def load_post_counts():
    counts = []
    filename = '/home/shane/foodie_automator/x_post_counts.json'
    if os.path.exists(filename):
        try:
            with open(filename, 'r') as f:
                lines = f.readlines()
                for i, line in enumerate(lines, 1):
                    if line.strip():
                        try:
                            entry = json.loads(line.strip())
                            # Check for expected fields in x_post_counts.json
                            if not isinstance(entry, dict) or "username" not in entry or "month" not in entry or "monthly_count" not in entry or "day" not in entry or "daily_count" not in entry:
                                logging.warning(f"Skipping malformed entry in {filename} at line {i}: {entry}")
                                continue
                            counts.append(entry)
                        except json.JSONDecodeError as e:
                            logging.warning(f"Skipping invalid JSON line in {filename} at line {i}: {e}")
            logging.info(f"Loaded {len(counts)} entries from {filename}")
        except Exception as e:
            logging.error(f"Failed to load {filename}: {e}")
            counts = []  # Reset to empty on failure
    
    if not counts:
        counts = [{
            "username": author["username"],
            "month": datetime.now(timezone.utc).strftime("%Y-%m"),
            "monthly_count": 0,
            "day": datetime.now(timezone.utc).strftime("%Y-%m-%d"),
            "daily_count": 0
        } for author in AUTHORS]
    
    current_month = datetime.now(timezone.utc).strftime("%Y-%m")
    current_day = datetime.now(timezone.utc).strftime("%Y-%m-%d")
    for entry in counts:
        if entry["month"] != current_month:
            entry["month"] = current_month
            entry["monthly_count"] = 0
        if entry["day"] != current_day:
            entry["day"] = current_day
            entry["daily_count"] = 0
    return counts

def save_post_counts(counts):
    with open('/home/shane/foodie_automator/x_post_counts.json', 'w') as f:
        for item in counts:
            json.dump(item, f)
            f.write('\n')
    logging.info("Saved post counts to x_post_counts.json")

import re

def generate_article_tweet(author, post, persona):
    title = post["title"]
    url = post["url"]
    author_handle = f"@{author['username']}"
    
    prompt = (
        f"Craft a sharp tweet (under 230 characters) for {author_handle} with the voice of '{persona}'. "
        f"Distill the essence of the article '{title}' into a concise, engaging message. "
        f"Include the raw URL '{url}' at the end. "
        f"Do not wrap the tweet in quotation marks. "
        f"Make it bold, spark curiosity, and invite engagement with a human touch. "
        f"Swap 'elevate' for dynamic terms like 'ignite' or 'unleash'. "
        f"Absolutely do not include hashtags, emojis, or phrases like '[Read more]' or 'Read more'. "
        f"Skip any extra fluff or formatting around the URL—just append the raw URL after a space. "
        f"Example: 'Love food trends? Check this out! {url}'"
    )
    
    response = client.chat.completions.create(
        model=SUMMARY_MODEL,
        messages=[
            {"role": "system", "content": "You are a social media viral expert crafting engaging tweets."},
            {"role": "user", "content": prompt}
        ],
        max_tokens=80,
        temperature=0.7
    )
    
    tweet = response.choices[0].message.content.strip()
    
    # Post-generation check: Strip any emojis using regex
    tweet = re.sub(r'[\U0001F600-\U0001F64F\U0001F300-\U0001F5FF\U0001F680-\U0001F6FF\U0001F700-\U0001F77F\U0001F780-\U0001F7FF\U0001F800-\U0001F8FF\U0001F900-\U0001F9FF\U0001FA00-\U0001FA6F\U0001FA70-\U0001FAFF\U00002702-\U000027B0\U000024C2-\U0001F251]', '', tweet).strip()
    
    # Strip "[Read more]" or similar phrases as an additional failsafe
    tweet = re.sub(r'\[Read more\]\(.*?\)|\bRead more\b', '', tweet).strip()
    
    # Strip leading or trailing quotation marks
    tweet = tweet.strip('"\'')
    
    # Ensure tweet fits within 280 characters, accounting for URL (Twitter shortens to 23 chars)
    url_length = 23
    max_tweet_length = 280 - url_length - 1  # Subtract 1 for the space before URL
    if len(tweet) > max_tweet_length:
        tweet = tweet[:max_tweet_length-3] + "... " + url
    
    logging.info(f"Generated tweet: {tweet}")
    return tweet

def post_tweet(author, tweet):
    credentials = next((cred for cred in X_API_CREDENTIALS if cred["username"] == author["username"]), None)
    if not credentials:
        logging.error(f"No X credentials found for {author['username']}")
        return False
    
    post_counts = load_post_counts()
    author_count = next((entry for entry in post_counts if entry["username"] == author["username"]), None)
    if author_count["monthly_count"] >= 500:
        logging.warning(f"Monthly post limit (500) reached for {author['username']}")
        return False
    if author_count["daily_count"] >= 20:
        logging.warning(f"Daily post limit (20) reached for {author['username']}")
        return False
    
    try:
        client = tweepy.Client(
            consumer_key=credentials["api_key"],
            consumer_secret=credentials["api_secret"],
            access_token=credentials["access_token"],
            access_token_secret=credentials["access_token_secret"]
        )
        response = client.create_tweet(text=tweet)
        author_count["monthly_count"] += 1
        author_count["daily_count"] += 1
        save_post_counts(post_counts)
        logging.info(f"Posted tweet for {author['username']}: {tweet}")
        return True
    except Exception as e:
        logging.error(f"Failed to post tweet for {author['username']}: {e}")
        return False

def select_best_persona(interest_score, content=""):
    logging.info("Using select_best_persona with interest_score and content")
    personas = ["Visionary Editor", "Foodie Critic", "Trend Scout", "Culture Connoisseur"]
    content_lower = content.lower()
    
    if any(kw in content_lower for kw in ["tech", "ai", "innovation", "sustainability"]):
        return random.choice(["Trend Scout", "Visionary Editor"])
    elif any(kw in content_lower for kw in ["review", "critic", "taste", "flavor"]):
        return "Foodie Critic"
    elif any(kw in content_lower for kw in ["culture", "tradition", "history"]):
        return "Culture Connoisseur"
    
    if interest_score >= 8:
        return random.choice(personas[:2])
    elif interest_score >= 6:
        return random.choice(personas[2:])
    return random.choice(personas)

def get_image(search_query):
    headers = {'User-Agent': 'InsiderFoodieBot/1.0 (https://insiderfoodie.com; contact@insiderfoodie.com)'}
    
    # Try Pixabay with the original query
    try:
        pixabay_url = f"https://pixabay.com/api/?key={PIXABAY_API_KEY}&q={quote(search_query)}&image_type=photo&per_page=10"
        response = requests.get(pixabay_url, headers=headers, timeout=10)
        response.raise_for_status()
        data = response.json()
        
        for hit in data.get('hits', []):
            img_url = hit.get('webformatURL')
            if not img_url or img_url in used_images:
                continue
            uploader = hit.get('user', 'Unknown')
            page_url = hit.get('pageURL', img_url)
            
            used_images.add(img_url)
            save_used_images()
            
            logging.info(f"Selected Pixabay image: {img_url} by {uploader} for query '{search_query}'")
            return img_url, "Pixabay", uploader, page_url
        
        logging.info(f"No valid Pixabay image found for query '{search_query}'. Trying fallback query.")
    
    except Exception as e:
        logging.warning(f"Pixabay image fetch failed for query '{search_query}': {e}")
    
    # Fallback to a generic query
    fallback_query = "food dining"
    try:
        pixabay_url = f"https://pixabay.com/api/?key={PIXABAY_API_KEY}&q={quote(fallback_query)}&image_type=photo&per_page=10"
        response = requests.get(pixabay_url, headers=headers, timeout=10)
        response.raise_for_status()
        data = response.json()
        
        for hit in data.get('hits', []):
            img_url = hit.get('webformatURL')
            if not img_url or img_url in used_images:
                continue
            uploader = hit.get('user', 'Unknown')
            page_url = hit.get('pageURL', img_url)
            
            used_images.add(img_url)
            save_used_images()
            
            logging.info(f"Selected Pixabay fallback image: {img_url} by {uploader} for query '{fallback_query}'")
            return img_url, "Pixabay", uploader, page_url
        
        logging.warning(f"No valid Pixabay image found for fallback query '{fallback_query}'.")
    
    except Exception as e:
        logging.warning(f"Pixabay fallback image fetch failed for query '{fallback_query}': {e}")
    
    # Ultimate fallback: return None but log clearly
    logging.error(f"All image fetch attempts failed for query '{search_query}'. Returning None.")
    return None, None, None, None

def generate_image_query(title, summary):
    try:
        prompt = (
            "Given the following article title and summary, generate a concise image search query (max 5 words) to find a relevant image. "
            "Also provide a list of relevance keywords (max 5 words) that should be associated with the image. "
            "Return the result as a JSON object with 'search' and 'relevance' keys.\n\n"
            f"Title: {title}\n\n"
            f"Summary: {summary}\n\n"
            "Example output:\n"
            "```json\n"
            "{\"search\": \"Italian cuisine trends\", \"relevance\": \"pasta wine dining culture\"}\n"
            "```"
        )
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": prompt},
                {"role": "user", "content": "Generate an image search query and relevance keywords."}
            ],
            max_tokens=100,
            temperature=0.5
        )
        raw_response = response.choices[0].message.content
        json_match = re.search(r'```json\n([\s\S]*?)\n```', raw_response)
        if not json_match:
            logging.warning(f"Failed to parse image query JSON: {raw_response}")
            return title, [], True
        
        query_data = json.loads(json_match.group(1))
        search_query = query_data.get("search", title)
        relevance_keywords = query_data.get("relevance", "").split()
        
        # Log the JSON object in a single line
        log_json = json.dumps(query_data).replace('\n', ' ').replace('\r', ' ')
        logging.debug(f"Image query from content: {log_json}")
        
        return search_query, relevance_keywords, False
    except Exception as e:
        logging.warning(f"Image query generation failed: {e}. Using title as fallback.")
        return title, [], True

def smart_image_and_filter(title, summary):
    try:
        content = f"{title}\n\n{summary}"
        
        prompt = (
            "Analyze this article title and summary. Extract key entities (brands, locations, cuisines, or topics) "
            "for an image search about food industry trends or viral content. Prioritize specific terms if present, "
            "otherwise focus on the main theme. Also identify the main topic of the article (e.g., a specific food item or cuisine). "
            "Return 'SKIP' if the article is about home appliances, recipes, promotions, or contains 'homemade', else 'KEEP'. "
            "Return as JSON with double quotes for all property names and string values (e.g., "
            "{\"image_query\": \"specific term\", \"relevance\": [\"keyword1\", \"keyword2\"], \"main_topic\": \"main food item\", \"action\": \"KEEP\" or \"SKIP\"})."
        )
        
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": prompt},
                {"role": "user", "content": content}
            ],
            max_tokens=100
        )
        raw_result = response.choices[0].message.content.strip()
        logging.info(f"Raw GPT smart image/filter response: '{raw_result}'")
        
        cleaned_result = re.sub(r'```json\s*|\s*```', '', raw_result).strip()
        fixed_result = re.sub(r"(?<!\\)'(?=\s*[\w\s]*\])|(?<=\[|\{|\s)'|'(?=\s*[\]\},:])|(?<=\w)'(?=\s*:)", '"', cleaned_result)
        
        try:
            result = json.loads(fixed_result)
        except json.JSONDecodeError as e:
            logging.warning(f"JSON parsing failed: {e}, raw: '{fixed_result}'. Using fallback.")
            # Fallback: Extract main topic using simple keyword matching
            main_topic = extract_main_topic(title.lower() + " " + summary.lower())
            return main_topic, [main_topic, "food"], False
        
        if not isinstance(result, dict) or "image_query" not in result or "relevance" not in result or "action" not in result:
            logging.warning(f"Invalid GPT response format: {result}, using fallback")
            main_topic = extract_main_topic(title.lower() + " " + summary.lower())
            return main_topic, [main_topic, "food"], False
        
        image_query = result["image_query"]
        relevance_keywords = result["relevance"]
        main_topic = result.get("main_topic", extract_main_topic(title.lower() + " " + summary.lower()))
        skip_flag = result["action"] == "SKIP" or "homemade" in title.lower() or "homemade" in summary.lower()
        
        logging.info(f"Smart image query: {image_query}, Relevance: {relevance_keywords}, Main Topic: {main_topic}, Skip: {skip_flag}")
        
        specific_single_words = ["kimchi", "sushi", "pizza", "taco", "burger"]
        if not image_query:
            logging.warning(f"Image query is empty, using fallback")
            return main_topic, [main_topic, "food"], skip_flag
        if len(image_query.split()) < 2 and image_query.lower() not in specific_single_words:
            logging.warning(f"Image query '{image_query}' too vague, using fallback")
            return main_topic, [main_topic, "food"], skip_flag
        
        return image_query, relevance_keywords, main_topic, skip_flag
    
    except Exception as e:
        logging.error(f"Smart image/filter failed: {e}, using fallback")
        main_topic = extract_main_topic(title.lower() + " " + summary.lower())
        return main_topic, [main_topic, "food"], False

def extract_main_topic(text):
    # Common food-related keywords (expand as needed)
    food_keywords = ["kimchi", "sushi", "pizza", "taco", "burger", "ramen", "curry", "pasta", "salad", "soup"]
    for keyword in food_keywords:
        if keyword in text:
            return keyword
    # Fallback to a generic term if no specific food item is found
    return "food trends"

def upload_image_to_wp(image_url, post_title, wp_base_url, wp_username, wp_password, image_source="Pixabay", uploader=None, pixabay_url=None):
    try:
        safe_title = post_title.encode('ascii', 'ignore').decode('ascii').replace(' ', '_')[:50]
        headers = {
            "Authorization": f"Basic {base64.b64encode(f'{wp_username}:{wp_password}'.encode()).decode()}",
            "Content-Disposition": f"attachment; filename={safe_title}.jpg",
            "Content-Type": "image/jpeg"
        }
        image_headers = {
            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
        }
        logging.info(f"Fetching image from {image_url} for '{post_title}'")
        
        # Add rate limit handling for image download
        for attempt in range(3):
            try:
                image_response = requests.get(image_url, headers=image_headers, timeout=10)
                if image_response.status_code == 429:
                    wait_time = 10 * (2 ** attempt)  # 10s, 20s, 40s
                    logging.warning(f"Rate limit hit for {image_url}. Retrying after {wait_time}s (attempt {attempt+1}/3).")
                    time.sleep(wait_time)
                    continue
                image_response.raise_for_status()
                break
            except requests.exceptions.HTTPError as e:
                if e.response.status_code == 429:
                    wait_time = 10 * (2 ** attempt)
                    logging.warning(f"Rate limit hit for {image_url}. Retrying after {wait_time}s (attempt {attempt+1}/3).")
                    time.sleep(wait_time)
                    continue
                raise
        else:
            logging.warning(f"Rate limit hit for {image_url} after retries. Failing image upload.")
            return None

        response = requests.post(
            f"{wp_base_url}/media",
            headers=headers,
            data=image_response.content
        )
        response.raise_for_status()
        
        image_id = response.json()["id"]
        caption = f'<a href="{pixabay_url}">{image_source}</a> by {uploader}' if pixabay_url and uploader else image_source
        requests.post(
            f"{wp_base_url}/media/{image_id}",
            headers={"Authorization": headers["Authorization"], "Content-Type": "application/json"},
            json={"caption": caption}
        )
        
        logging.info(f"Uploaded image '{safe_title}.jpg' to WP (ID: {image_id}) with caption '{caption}'")
        return image_id
    except Exception as e:
        logging.error(f"Image upload to WP failed for '{post_title}': {e}")
        return None

def determine_paragraph_count(interest_score):
    if interest_score >= 9:
        return 5
    elif interest_score >= 7:
        return 4
    return 3

def is_interesting(summary):
    try:
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": (
                    "Rate this content from 0-10 based on its rarity, buzzworthiness, and engagement potential for food lovers, covering a wide range of food topics (skip recipes). "
                    "Score 8-10 for rare, highly shareable ideas that grab attention. "
                    "Score 5-7 for fresh, engaging updates with broad appeal. Score below 5 for common or unremarkable content. "
                    "Return only a number."
                )},
                {"role": "user", "content": f"Content: {summary}"}
            ],
            max_tokens=5
        )
        raw_score = response.choices[0].message.content.strip()
        score = int(raw_score) if raw_score.isdigit() else 0
        print(f"Interest Score for '{summary[:50]}...': {score} (raw: {raw_score})")
        logging.info(f"Interest Score: {score} (raw: {raw_score})")
        return score
    except Exception as e:
        logging.error(f"Interestingness scoring failed: {e}")
        print(f"Interest Error: {e}")
        return 0

def generate_title_from_summary(summary):
    banned_words = ["elevate", "elevating", "elevated"]
    for attempt in range(3):
        try:
            response = client.chat.completions.create(
                model=LIGHT_TASK_MODEL,
                messages=[
                    {"role": "system", "content": (
                        "Generate a concise, engaging title (under 100 characters) based on this summary, covering food topics. "
                        "Craft it with Upworthy/Buzzfeed flair—think ‘you won’t believe this’ or ‘this is nuts’—for food insiders. "
                        "Avoid quotes, emojis, special characters, or the words 'elevate', 'elevating', 'elevated'. "
                        "End with a question to spark shares."
                    )},
                    {"role": "user", "content": f"Summary: {summary}"}
                ],
                max_tokens=30
            )
            title = response.choices[0].message.content.strip().replace('"', '').replace("'", "")
            if ':' in title:
                title = title.split(':', 1)[1].strip()
            if len(title) > 100 or any(word in title.lower() for word in banned_words):
                reason = "length" if len(title) > 100 else "banned word"
                print(f"Rejected title (attempt {attempt + 1}/3): '{title}' due to {reason}")
                logging.info(f"Rejected title (attempt {attempt + 1}/3): '{title}' due to {reason}")
                continue
            logging.info(f"Generated title: {title}")
            return title
        except Exception as e:
            logging.error(f"Title generation failed (attempt {attempt + 1}/3): {e}")
            print(f"Title Error: {e}")
    print("Failed to generate valid title after 3 attempts")
    logging.info("Failed to generate valid title after 3 attempts")
    return None

def summarize_with_gpt4o(content, source_name, link, interest_score=0, extra_prompt=""):
    try:
        persona = select_best_persona(interest_score, content)
        persona_config = PERSONA_CONFIGS.get(persona, {
            "article_prompt": "Write a concise, engaging summary that captures the essence of the content for food lovers.",
            "description": "a generic food writer",
            "tone": "an engaging tone"
        })
        prompt = persona_config["article_prompt"].format(
            description=persona_config["description"],
            tone=persona_config["tone"],
            num_paragraphs=determine_paragraph_count(interest_score)
        )
        logging.info(f"Using {persona} with interest_score and content")
        
        full_prompt = (
            f"{prompt}\n\n"
            f"Do not include the article title in the summary.\n\n"
            f"{extra_prompt}\n\n"
            f"Avoid using the word 'elevate'—use more humanized language like 'level up' or 'bring to life'.\n"
            f"Content to summarize:\n{content}\n\n"
            f"Source: {source_name}\n"
            f"Link: {link}"
        )
        
        response = client.chat.completions.create(
            model=SUMMARY_MODEL,
            messages=[
                {"role": "system", "content": full_prompt},
                {"role": "user", "content": content}
            ],
            max_tokens=1000,
            temperature=0.7
        )
        
        summary = response.choices[0].message.content.strip()
        
        # Post-process to remove the original title if it still appears
        # Extract the title from the content (assuming it's the first line or part of the prompt)
        # For simplicity, we can pass the title as an additional parameter if needed
        # Here, we'll assume the title is passed via the calling function (e.g., from foodie_automator_rss.py)
        # For now, we'll use a placeholder for the title removal logic
        # In foodie_automator_rss.py, the title is available as entry.title
        # We'll handle the title removal in the calling script instead
        logging.info(f"Processed summary (Persona: {persona}): {summary}")
        return summary
    
    except Exception as e:
        logging.error(f"Summary generation failed with model {SUMMARY_MODEL}: {e}")
        return None

def insert_link_naturally(summary, source_name, source_url):
    try:
        logging.info(f"Input summary to insert_link_naturally: {summary!r}")

        prompt = (
            "Take this summary and insert a single HTML link into one paragraph (randomly chosen). "
            "Use the format '<a href=\"{source_url}\">{source_name}</a>' and weave it into the text naturally, "
            "e.g., 'According to {source_name}, ' or '{source_name} shares that '. "
            "Place the link at the end of a sentence (after a period). "
            "Preserve the original paragraph structure, maintaining all newlines exactly as they are (each paragraph separated by a single \\n). "
            "Return the modified summary with exactly one link.\n\n"
            "Summary:\n{summary}\n\n"
            "Source Name: {source_name}\nSource URL: {source_url}"
        ).format(summary=summary, source_name=source_name, source_url=source_url)
        
        # Add retry mechanism
        for attempt in range(3):
            try:
                response = client.chat.completions.create(
                    model=LIGHT_TASK_MODEL,
                    messages=[
                        {"role": "system", "content": prompt},
                        {"role": "user", "content": "Insert the link naturally into the summary."}
                    ],
                    max_tokens=1000,
                    temperature=0.7
                )
                new_summary = response.choices[0].message.content.strip()
                link_pattern = f'<a href="{source_url}">{source_name}</a>'
                if new_summary and new_summary.count(link_pattern) == 1:
                    paragraphs = new_summary.split('\n')
                    paragraphs = [p.strip() for p in paragraphs]
                    new_summary = '\n'.join(paragraphs)
                    logging.info(f"Summary with naturally embedded link (normalized): {new_summary!r}")
                    return new_summary
                else:
                    logging.warning(f"GPT attempt {attempt + 1}/3 failed to insert link correctly: {new_summary}")
            except Exception as e:
                logging.error(f"Link insertion attempt {attempt + 1}/3 failed: {e}")
        
        logging.warning(f"GPT failed to insert link after 3 attempts. Using fallback.")
    except Exception as e:
        logging.error(f"Link insertion failed: {e}")

    # Fallback path
    time_pattern = r'\b\d{1,2}\.\d{2}(?:am|pm)\b'
    protected_summary = re.sub(time_pattern, lambda m: m.group(0).replace('.', '@'), summary)
    paragraphs = protected_summary.split('\n')
    if not paragraphs or all(not p.strip() for p in paragraphs):
        logging.error("No valid paragraphs to insert link.")
        return summary
    
    target_para = random.choice([p for p in paragraphs if p.strip()])
    link_pattern = f'<a href="{source_url}">{source_name}</a>'
    phrases = [
        f"According to {link_pattern}",
        f"{link_pattern} notes this insight",
        f"Details shared by {link_pattern}",
        f"Source: {link_pattern}"
    ]
    insertion_phrase = random.choice(phrases)
    
    sentences = re.split(r'(?<=[.!?])\s+', target_para)
    insertion_point = -1
    for i, sent in enumerate(sentences):
        if sent.strip() and '@' not in sent:
            insertion_point = sum(len(s) + 1 for s in sentences[:i+1])
            break
    if insertion_point == -1:
        insertion_point = len(target_para)
    
    new_para = f"{target_para[:insertion_point]} {insertion_phrase}. {target_para[insertion_point:]}".strip()
    paragraphs[paragraphs.index(target_para)] = new_para
    new_summary = '\n'.join(paragraphs)
    
    new_summary = new_summary.replace('@', '.')
    logging.info(f"Fallback summary with link: {new_summary!r}")
    return new_summary

def generate_category_from_summary(summary):
    try:
        if not isinstance(summary, str) or not summary.strip():
            logging.warning(f"Invalid summary for category generation: {summary}. Defaulting to 'Trends'.")
            return "Trends"
        
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": (
                    "Based on this summary, select the most relevant category from: Food, Culture, Trends, Health, Lifestyle, Drink, Eats. "
                    "Return only the category name."
                )},
                {"role": "user", "content": summary}
            ],
            max_tokens=10
        )
        category = response.choices[0].message.content.strip()
        logging.info(f"Generated category: {category}")
        return category if category in ["Food", "Culture", "Trends", "Health", "Lifestyle", "Drink", "Eats"] else "Trends"
    except Exception as e:
        logging.error(f"Category generation failed: {e}")
        return "Trends"

def get_wp_category_id(category_name, wp_base_url, wp_username, wp_password):
    try:
        headers = {"Authorization": f"Basic {base64.b64encode(f'{wp_username}:{wp_password}'.encode()).decode()}"}
        response = requests.get(f"{wp_base_url}/categories", headers=headers, params={"search": category_name})
        response.raise_for_status()
        categories = response.json()
        for cat in categories:
            if cat["name"].lower() == category_name.lower():
                return cat["id"]
        return None
    except Exception as e:
        logging.error(f"Failed to get WP category ID for '{category_name}': {e}")
        return None

def create_wp_category(category_name, wp_base_url, wp_username, wp_password):
    try:
        headers = {
            "Authorization": f"Basic {base64.b64encode(f'{wp_username}:{wp_password}'.encode()).decode()}",
            "Content-Type": "application/json"
        }
        payload = {"name": category_name}
        response = requests.post(f"{wp_base_url}/categories", headers=headers, json=payload)
        response.raise_for_status()
        return response.json()["id"]
    except Exception as e:
        logging.error(f"Failed to create WP category '{category_name}': {e}")
        return None

def get_wp_tag_id(tag_name, wp_base_url, wp_username, wp_password):
    try:
        headers = {"Authorization": f"Basic {base64.b64encode(f'{wp_username}:{wp_password}'.encode()).decode()}"}
        response = requests.get(f"{wp_base_url}/tags", headers=headers, params={"search": tag_name})
        response.raise_for_status()
        tags = response.json()
        for tag in tags:
            if tag["name"].lower() == tag_name.lower():
                return tag["id"]
        return None
    except Exception as e:
        logging.error(f"Failed to get WP tag ID for '{tag_name}': {e}")
        return None

def post_to_wp(post_data, category, link, author, image_url, original_source, image_source="Pixabay", uploader=None, pixabay_url=None, interest_score=4, post_id=None, should_post_tweet=True):
    wp_base_url = "https://insiderfoodie.com/wp-json/wp/v2"
    logging.info(f"Starting post_to_wp for '{post_data['title']}', image_source: {image_source}")
    
    if not isinstance(author, dict) or "username" not in author or "password" not in author:
        raise ValueError(f"Invalid author data: {author}. Expected a dictionary with 'username' and 'password' keys.")
    
    wp_username = author["username"]
    wp_password = author["password"]
    
    if not isinstance(interest_score, int):
        logging.error(f"Invalid interest_score type: {type(interest_score)}, value: '{interest_score}'. Defaulting to 4.")
        interest_score = 4
    elif interest_score < 0 or interest_score > 10:
        logging.warning(f"interest_score out of valid range (0-10): {interest_score}. Clamping to 4.")
        interest_score = min(max(interest_score, 0), 10)
    
    try:
        headers = {
            "Authorization": f"Basic {base64.b64encode(f'{wp_username}:{wp_password}'.encode()).decode()}",
            "Content-Type": "application/json"
        }
        
        auth_test = requests.get(f"{wp_base_url}/users/me", headers=headers)
        auth_test.raise_for_status()
        logging.info(f"Auth test passed for {wp_username}: {auth_test.json()['id']}")
        
        category_id = get_wp_category_id(category, wp_base_url, wp_username, wp_password)
        if not category_id:
            category_id = create_wp_category(category, wp_base_url, wp_username, wp_password)
            logging.info(f"Created new category '{category}' with ID {category_id}")
        else:
            logging.info(f"Found existing category '{category}' with ID {category_id}")
        
        tags = [1]
        if interest_score >= 9:
            picks_tag_id = get_wp_tag_id("Picks", wp_base_url, wp_username, wp_password)
            if picks_tag_id and picks_tag_id not in tags:
                tags.append(picks_tag_id)
                logging.info(f"Added 'Picks' tag (ID: {picks_tag_id}) to post due to high interest score: {interest_score}")
        
        content = post_data["content"]
        if content is None:
            logging.error(f"Post content is None for title '{post_data['title']}' - using fallback")
            content = "Content unavailable. Check the original source for details."
        formatted_content = "\n".join(f"<p>{para}</p>" for para in content.split('\n') if para.strip())
        
        author_id_map = {
            "owenjohnson": 10,
            "javiermorales": 2,
            "aishapatel": 3,
            "trangnguyen": 12,
            "keishareid": 13,
            "lilamoreau": 7
        }
        author_id = author_id_map.get(author["username"], 5)
        
        # Handle image upload
        image_id = None
        if image_url:
            logging.info(f"Attempting image upload for '{post_data['title']}', URL: {image_url}, source: {image_source}")
            image_id = upload_image_to_wp(image_url, post_data["title"], wp_base_url, wp_username, wp_password, image_source, uploader, pixabay_url)
            if not image_id:
                logging.info(f"Flickr upload failed for '{post_data['title']}', falling back to Pixabay")
                pixabay_query = post_data["title"][:50]
                image_url, image_source, uploader, pixabay_url = get_image(pixabay_query)
                if image_url:
                    image_id = upload_image_to_wp(image_url, post_data["title"], wp_base_url, wp_username, wp_password, image_source, uploader, pixabay_url)
            if not image_id:
                logging.warning(f"All image uploads failed for '{post_data['title']}' - posting without image")
        
        payload = {
            "title": post_data["title"],
            "content": formatted_content,
            "status": "publish",
            "categories": [category_id],
            "tags": tags,
            "author": author_id,
            "meta": {
                "original_link": link,
                "original_source": original_source,
                "interest_score": interest_score
            }
        }
        
        if image_id:
            payload["featured_media"] = image_id
            logging.info(f"Set featured image for post '{post_data['title']}': Media ID={image_id}")
        
        endpoint = f"{wp_base_url}/posts/{post_id}" if post_id else f"{wp_base_url}/posts"
        method = requests.post
        
        logging.debug(f"Sending WP request to {endpoint} with payload: {json.dumps(payload, indent=2)}")
        
        response = method(endpoint, headers=headers, json=payload)
        response.raise_for_status()
        
        post_info = response.json()
        logging.debug(f"WP response: {json.dumps(post_info, indent=2)}")
        
        if not isinstance(post_info, dict) or "id" not in post_info:
            raise ValueError(f"Invalid WP response: {post_info}")
        
        post_id = post_info["id"]
        post_url = post_info["link"]

        # Save to recent_posts.json
        timestamp = datetime.now(timezone.utc).isoformat()
        save_post_to_recent(post_data["title"], post_url, author["username"], timestamp)

        # Post article tweet to X only if should_post_tweet is True
        if should_post_tweet:
            try:
                post = {"title": post_data["title"], "url": post_url}
                tweet = generate_article_tweet(author, post, author["persona"])
                if post_tweet(author, tweet):
                    logging.info(f"Successfully posted article tweet for {author['username']} on X")
                else:
                    logging.warning(f"Failed to post article tweet for {author['username']} on X")
            except Exception as e:
                logging.error(f"Error posting article tweet for {author['username']}: {e}")

        logging.info(f"Posted/Updated by {author['username']}: {post_data['title']} (ID: {post_id})")
        return post_id, post_url
    
    except requests.exceptions.RequestException as e:
        logging.error(f"WP API request failed: {e} - Response: {e.response.text if e.response else 'No response'}")
        print(f"WP Error: {e}")
        return None, None
    except KeyError as e:
        logging.error(f"WP payload error - Missing key: {e} - Author data: {author}")
        print(f"WP Error: {e}")
        return None, None
    except Exception as e:
        logging.error(f"WP posting failed: {e}")
        print(f"WP Error: {e}")
        return None, None

# Configure Flickr API with credentials
flickr_api.set_keys(api_key=FLICKR_API_KEY, api_secret=FLICKR_API_SECRET)
logging.info(f"Flickr API configured with key: {FLICKR_API_KEY[:4]}... and secret: {FLICKR_API_SECRET[:4]}...")

# Global variable to track the last Flickr request time
last_flickr_request_time = 0

# Flickr request counter
flickr_request_count = 0
flickr_request_start_time = time.time()

# Define exclude keywords for filtering unwanted image types
exclude_keywords = [
    "poster", "infographic", "chart", "graph", "data", "stats", "text", "typography",
    "design", "advertisement", "illustration", "diagram", "layout", "print"
]

# Initialize used_images as a set to track used image URLs
used_images_file = "/home/shane/foodie_automator/used_images.json"
used_images = set()

# Load used images from file if it exists
if os.path.exists(used_images_file):
    try:
        entries = load_json_file(used_images_file, IMAGE_EXPIRATION_DAYS * 24)  # Use load_json_file for consistency
        for entry in entries:
            if isinstance(entry, dict) and "title" in entry and entry["title"].startswith('https://'):
                used_images.add(entry["title"])
            else:
                logging.warning(f"Skipping invalid entry in {used_images_file}: {entry}")
        logging.info(f"Loaded {len(used_images)} used image URLs from {used_images_file}")
    except Exception as e:
        logging.warning(f"Failed to load used images from {used_images_file}: {e}. Resetting to empty set.")
        used_images = set()
        with open(used_images_file, 'w') as f:
            f.write("") 

# Function to save used_images to file
def save_used_images():
    try:
        # Load existing entries to preserve timestamps
        entries = load_json_file(used_images_file, IMAGE_EXPIRATION_DAYS * 24)
        existing_entries = {entry["title"]: entry for entry in entries if isinstance(entry, dict) and "title" in entry}
        
        # Create new entries for used_images
        timestamp = datetime.now(timezone.utc).isoformat()
        updated_entries = []
        for url in used_images:
            if url in existing_entries:
                updated_entries.append(existing_entries[url])
            else:
                updated_entries.append({"title": url, "timestamp": timestamp})
        
        with open(used_images_file, 'w') as f:
            for entry in updated_entries:
                f.write(json.dumps(entry) + '\n')
        logging.info(f"Saved {len(updated_entries)} used image URLs to {used_images_file}")
    except Exception as e:
        logging.warning(f"Failed to save used images to {used_images_file}: {e}")

def reset_flickr_request_count():
    global flickr_request_count, flickr_request_start_time
    if time.time() - flickr_request_start_time >= 3600:  # Reset every hour
        flickr_request_count = 0
        flickr_request_start_time = time.time()

def process_photo(photo, search_query):
    tags = [tag.text.lower() for tag in photo.getTags()]
    title = photo.title.lower() if photo.title else ""
    
    matched_keywords = [kw for kw in exclude_keywords if kw in tags or kw in title]
    if matched_keywords:
        logging.info(f"Skipping image with unwanted keywords: {photo.id} (tags: {tags}, title: {title}, matched: {matched_keywords})")
        return None
    
    # Try 'Large' size first, fall back to 'Medium' if unavailable
    img_url = None
    try:
        img_url = photo.getPhotoFile(size_label='Large')
    except flickr_api.flickrerrors.FlickrError as e:
        logging.info(f"Large size not available for photo {photo.id}: {e}, trying Medium")
        try:
            img_url = photo.getPhotoFile(size_label='Medium')
        except flickr_api.flickrerrors.FlickrError as e:
            logging.warning(f"Medium size not available for photo {photo.id}: {e}")
            return None
    
    if not img_url:
        logging.info(f"Image URL invalid for photo {photo.id}")
        return None
    
    # Check if the image is highly relevant to the query
    query_keywords = set(search_query.lower().split())
    photo_keywords = set(tags + title.split())
    is_relevant = bool(query_keywords & photo_keywords)  # Check if any query keyword is in tags or title
    
    # Allow reuse of highly relevant images
    if img_url in used_images and not is_relevant:
        logging.info(f"Image already used and not highly relevant for photo {photo.id}: {img_url}")
        return None
    
    uploader = photo.owner.username
    page_url = f"https://www.flickr.com/photos/{photo.owner.nsid}/{photo.id}"
    
    used_images.add(img_url)
    save_used_images()
    
    flickr_data = {
        "title": search_query,
        "image_url": img_url,
        "source": "Flickr",
        "uploader": uploader,
        "page_url": page_url,
        "timestamp": datetime.now(timezone.utc).isoformat()
    }
    flickr_file = "/home/shane/foodie_automator/flickr_images.json"
    with open(flickr_file, 'a') as f:
        json.dump(flickr_data, f)
        f.write('\n')
    logging.info(f"Saved Flickr image metadata to {flickr_file}: {img_url}")
    
    logging.info(f"Selected Flickr image: {img_url} by {uploader} for query '{search_query}' (tags: {tags})")
    return img_url, "Flickr", uploader, page_url

def search_flickr(query, per_page=5):
    try:
        photos = flickr_api.Photo.search(
            text=query,
            per_page=per_page,
            sort='relevance',
            safe_search=1,
            media='photos',
            license='4,5,9,10'
        )
        return photos
    except Exception as e:
        logging.warning(f"Flickr API error for query '{query}': {e}")
        return []

def fetch_photo_by_id(photo_id):
    try:
        photo = flickr_api.Photo(id=photo_id)
        return photo
    except Exception as e:
        logging.warning(f"Failed to fetch Flickr photo ID {photo_id}: {e}")
        return None

def search_ddg_for_flickr(query):
    ddg_query = f"{query} site:flickr.com"
    ddg_url = f"https://duckduckgo.com/?q={quote(ddg_query)}"
    try:
        response = requests.get(ddg_url, headers={'User-Agent': 'InsiderFoodieBot/1.0 (https://insiderfoodie.com; contact@insiderfoodie.com)'}, timeout=10)
        response.raise_for_status()
        soup = BeautifulSoup(response.text, 'html.parser')
        
        photo_ids = set()
        for link in soup.find_all('a', href=True):
            href = link['href']
            match = re.search(r'flickr\.com/photos/[^/]+/(\d+)', href)
            if match:
                photo_id = match.group(1)
                photo_ids.add(photo_id)
        
        photo_ids = list(photo_ids)[:2]  # Limit to 2 IDs
        logging.info(f"Found {len(photo_ids)} Flickr photo IDs via DDG: {photo_ids}")
        return photo_ids
    except Exception as e:
        logging.warning(f"DDG search failed for query '{ddg_query}': {e}")
        return set()

def classify_keywords(keywords):
    prompt = (
        "Given the following keywords from an image search query, classify each as 'specific' (e.g., brand names, unique entities like 'Taco Bell' or 'Paris') or 'generic' (e.g., common or abstract terms like 'dining' or 'trends'). "
        "Return a JSON object mapping each keyword to its classification.\n\n"
        "Keywords: " + ", ".join(keywords) + "\n\n"
        "Example output format (do not use these exact keywords in your response):\n"
        "```json\n"
        "{\n"
        "  \"keyword1\": \"specific\",\n"
        "  \"keyword2\": \"generic\"\n"
        "}\n```"
    )
    try:
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": "You are a helper that classifies keywords."},
                {"role": "user", "content": prompt}
            ],
            max_tokens=100,
            temperature=0.5
        )
        raw_response = response.choices[0].message.content
        json_match = re.search(r'```json\n([\s\S]*?)\n```', raw_response)
        if not json_match:
            logging.warning(f"Failed to parse keyword classification JSON: {raw_response}")
            return {kw: "specific" for kw in keywords}
        
        classifications = json.loads(json_match.group(1))
        return classifications
    except Exception as e:
        logging.warning(f"Keyword classification failed: {e}. Defaulting to all specific.")
        return {kw: "specific" for kw in keywords}

def get_flickr_image(search_query, relevance_keywords, main_topic):
    global last_flickr_request_time, flickr_request_count
    
    reset_flickr_request_count()
    flickr_request_count += 1
    logging.info(f"Flickr request count: {flickr_request_count}/3600")
    
    current_time = time.time()
    time_since_last_request = current_time - last_flickr_request_time
    if time_since_last_request < 10:
        time.sleep(10 - time_since_last_request)
    
    last_flickr_request_time = time.time()
    
    # Step 1: Search Flickr directly with the original query
    logging.info(f"Searching Flickr directly with query: '{search_query}'")
    photos = search_flickr(search_query)
    for photo in photos:
        result = process_photo(photo, search_query)
        if result:
            return result

    # Step 2: Search DDG to find Flickr photo IDs
    logging.info(f"Searching DDG with query: '{search_query} site:flickr.com'")
    photo_ids = search_ddg_for_flickr(search_query)
    if photo_ids:
        for photo_id in photo_ids:
            photo = fetch_photo_by_id(photo_id)
            if photo:
                result = process_photo(photo, search_query)
                if result:
                    return result

    # Step 3: Break down the query into keywords and classify them
    keywords = search_query.lower().split()
    if len(keywords) > 1:
        classifications = classify_keywords(keywords)
        logging.info(f"Keyword classifications: {classifications}")
        
        specific_keywords = [kw for kw, classification in classifications.items() if classification == "specific"]
        if specific_keywords:
            for keyword in specific_keywords:
                logging.info(f"Searching Flickr with specific keyword: '{keyword}'")
                photos = search_flickr(keyword)
                for photo in photos:
                    result = process_photo(photo, search_query)
                    if result:
                        return result

    # Step 4: Fallback using main topic
    logging.info(f"No results found. Falling back to main topic: '{main_topic}'")
    photos = search_flickr(main_topic)
    for photo in photos:
        result = process_photo(photo, main_topic)
        if result:
            return result

    # Step 5: Final fallback using relevance keywords
    fallback_query = " ".join(relevance_keywords) if isinstance(relevance_keywords, list) else relevance_keywords
    logging.info(f"No results with main topic. Falling back to relevance keywords: '{fallback_query}'")
    photos = search_flickr(fallback_query)
    for photo in photos:
        result = process_photo(photo, search_query)
        if result:
            return result

    logging.warning(f"No valid Flickr image found for query '{search_query}' after all attempts.")
    return None, None, None, None
    
def select_best_author(summary):
    try:
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": (
                    "Based on this restaurant/food industry trend summary, pick the most suitable author from: "
                    "owenjohnson, javiermorales, aishapatel, trangnguyen, keishareid, lilamoreau. "
                    "Consider their expertise: owenjohnson (global dining trends), javiermorales (food critique), "
                    "aishapatel (emerging food trends), trangnguyen (cultural dining), keishareid (soul food heritage), "
                    "lilamoreau (global street food). Return only the username."
                )},
                {"role": "user", "content": summary}
            ],
            max_tokens=20
        )
        author = response.choices[0].message.content.strip()
        valid_authors = ["owenjohnson", "javiermorales", "aishapatel", "trangnguyen", "keishareid", "lilamoreau"]
        logging.info(f"Selected author: {author}")
        return author if author in valid_authors else "owenjohnson"
    except Exception as e:
        logging.error(f"Author selection failed: {e}")
        return "owenjohnson"

def prepare_post_data(summary, title, main_topic=None):
    try:
        logging.info(f"Preparing post data for summary: {summary[:100]}...")
        
        prompt = (
            "Generate a concise, engaging title (5-15 words) for this food-related article summary. "
            "The title should be catchy, avoid emojis, and not reproduce the original title verbatim. "
            "Return the title as plain text."
        )
        
        response = client.chat.completions.create(
            model=LIGHT_TASK_MODEL,
            messages=[
                {"role": "system", "content": prompt},
                {"role": "user", "content": summary}
            ],
            max_tokens=50,
            temperature=0.7
        )
        new_title = response.choices[0].message.content.strip()
        logging.info(f"Generated new title: '{new_title}'")
        
        search_query, relevance_keywords, skip_flag = smart_image_and_filter(new_title, summary)
        if skip_flag:
            logging.info("Summary filtered out during post preparation")
            return None, None, None, None, None, None, None
        
        image_url, image_source, uploader, page_url = get_flickr_image(search_query, relevance_keywords, main_topic)
        if not image_url:
            image_url, image_source, uploader, page_url = get_image(search_query)
        
        if not image_url:
            logging.warning("No image found for post, skipping")
            return None, None, None, None, None, None, None
        
        pixabay_url = page_url if image_source == "Pixabay" else None
        
        authors = ["Aisha Patel", "Ravi Sharma", "Mei Lin", "Carlos Rivera"]
        author = random.choice(authors)
        
        categories = ["Food", "Trends", "Eats", "Culture"]
        category = random.choice(categories)
        
        post_data = {
            "title": new_title,
            "content": summary,
            "status": "publish",
            "author": author,
            "categories": [category]
        }
        
        logging.info(f"Post data prepared: Title: '{new_title}', Category: {category}, Author: {author}")
        return post_data, author, category, image_url, image_source, uploader, pixabay_url
    
    except Exception as e:
        logging.error(f"Failed to prepare post data: {e}")
        return None, None, None, None, None, None, None

def save_post_to_recent(post_title, post_url, author_username, timestamp):
    try:
        recent_posts = load_json_file('/home/shane/foodie_automator/recent_posts.json', 24)  # Added expiration_hours
        entry = {
            "title": post_title,
            "url": post_url,
            "author_username": author_username,
            "timestamp": timestamp
        }
        recent_posts.append(entry)
        with open('/home/shane/foodie_automator/recent_posts.json', 'w') as f:
            for item in recent_posts:
                json.dump(item, f)
                f.write('\n')
        logging.info(f"Saved post '{post_title}' to recent_posts.json")
    except Exception as e:
        logging.error(f"Failed to save post to recent_posts.json: {e}")

def prune_recent_posts():
    try:
        cutoff = (datetime.now(timezone.utc) - timedelta(hours=24)).isoformat()
        recent_posts = load_json_file('/home/shane/foodie_automator/recent_posts.json')
        recent_posts = [entry for entry in recent_posts if entry["timestamp"] > cutoff]
        with open('/home/shane/foodie_automator/recent_posts.json', 'w') as f:
            for item in recent_posts:
                json.dump(item, f)
                f.write('\n')
        logging.info(f"Pruned recent_posts.json to {len(recent_posts)} entries")
    except Exception as e:
        logging.error(f"Failed to prune recent_posts.json: {e}")