algorithms/get_results_fl_medical.py

"""
Get Reddit Results | Cannlytics
Copyright (c) 2023-2024 Cannlytics

Authors: Keegan Skeate <https://github.com/keeganskeate>
Created: 12/1/2023
Updated: 5/21/2024
License: MIT License <https://github.com/cannlytics/cannlytics/blob/main/LICENSE>

Description:

    This tool collects cannabis product reviews and associated product
    images to perform research.
    
    Product data from the product label images, as well as natural
    language data from the review, such as sentiment rating, can be used
    to analyze how product data may affect how the product is reviewed.

"""
# Standard imports:
from datetime import datetime
import json
import os
import shutil
from time import sleep

# External imports:
from bs4 import BeautifulSoup
from cannlytics.data.cache import Bogart
from cannlytics.data.coas import CoADoc
from cannlytics.data.web import initialize_selenium
from cannlytics.utils.constants import DEFAULT_HEADERS
from cannlytics.utils.utils import (
    download_file_with_selenium,
    remove_duplicate_files,
)
from dotenv import dotenv_values
import logging
import pandas as pd
import praw
import requests
import tempfile


#-----------------------------------------------------------------------
# Setup.
#-----------------------------------------------------------------------

# Create a directory to store the downloaded images.
images_directory = 'D://data/reddit/FLMedicalTrees/images'
os.makedirs(images_directory, exist_ok=True)

# Set up logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')


#-----------------------------------------------------------------------
# Get Reddit posts with Selenium.
#-----------------------------------------------------------------------

# Get the Subreddit page.
# Note: This required being logged-in in the browser.
# Note: This step is currently done manually with repeated reading
# of the `page_source` to append posts to the `data`.
driver = initialize_selenium(headless=False)
query = 'COA'
queries = [
    'COA',
    'COA attached',
    'COA in', # e.g. in the comments, included, etc.
    'Certificate',
    'Certificate of Analysis',
    'lab results',
    'test results',
    'results',
    'effect',
    'aroma',
    'taste',
    'smell',
    'flavor',
]
sort_by = 'new'
subreddit = 'FLMedicalTrees'
driver.get(f"https://www.reddit.com/r/{subreddit}/search/?q={query}&sort={sort_by}")
sleep(5)

# # Scroll to load posts (manually or automatically).
# for _ in range(10):
#     driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")
#     sleep(2)

# Collect post details.
data = []
recorded_posts = []

# Manual iteration of queries here.
page_source = driver.page_source
soup = BeautifulSoup(page_source, 'html.parser')
posts = soup.find_all('faceplate-tracker', {'data-testid': 'search-post'})
for post in posts:
    context = post.get('data-faceplate-tracking-context')
    context = json.loads(context)
    post_context = context['post']
    post_id = post_context['id']
    if post_id in recorded_posts:
        continue
    recorded_posts.append(post_id)
    data.append({
        'title': post_context['title'],
        'url': 'https://www.reddit.com' + post_context['url'],
        'created_timestamp': post_context['created_timestamp'],
        'author_id': post_context['author_id'],
        'post_id': post_id,
        'number_comments': post_context['number_comments'],
        'subreddit_id': post_context['subreddit_id'],
        'subreddit_name': post_context['subreddit_name'],
    })
print(f'Number of posts: {len(data)}')

# Close the driver.
driver.close()
driver.quit()

# Save the post data.
data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"
timestamp = datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
datafile = os.path.join(data_dir, f'fl-medical-trees-posts-{timestamp}.xlsx')
df = pd.DataFrame(data)
df.to_excel(datafile, index=False)
print('Saved post data:', datafile)


#-----------------------------------------------------------------------
# Get Reddit post data with the Reddit API.
#-----------------------------------------------------------------------

# DEV:
data = pd.read_excel(r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data\fl-medical-trees-posts-2024-06-25-18-59-50.xlsx")
data = data.to_dict(orient='records')
recorded_posts = [x['post_id'] for x in data]


def initialize_reddit(config):
    reddit = praw.Reddit(
        client_id=config['REDDIT_CLIENT_ID'],
        client_secret=config['REDDIT_SECRET'],
        password=config['REDDIT_PASSWORD'],
        user_agent=config['REDDIT_USER_AGENT'],
        username=config['REDDIT_USERNAME'],
    )
    return reddit


# # Read already collected posts.
collected_posts = []
# data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"
# post_datafiles = [os.path.join(data_dir, x) for x in os.listdir(data_dir) if 'posts' in x and 'results' not in x]
# posts = pd.concat([pd.read_excel(x) for x in post_datafiles])
# posts.drop_duplicates(subset=['post_id', 'coa_url', 'redirect_url'], inplace=True)
# collected_posts = list(set(posts['post_id'].values) - set(recorded_posts))
# print('Total number of already collected posts:', len(collected_posts))
# print('Number of posts to collect:', len(data) - len(collected_posts))

# Initialize Reddit.
config = dotenv_values('.env')
reddit = initialize_reddit(config)

# Get each post page and data for each post.
all_posts = []
for n, post_data in enumerate(data[len(all_posts):]):

    # Retrieve the post content.
    post_id = post_data['post_id'].split('_')[-1]
    if post_id in collected_posts:
        print('Post already collected:', post_id)
        continue
    print('Getting data for post:', post_id)
    try:
        submission = reddit.submission(id=post_id)
    except:
        try:
            print('Failed to retrieve post:', post_id)
            print('Waiting 60 seconds to retry...')
            sleep(61)
            reddit = initialize_reddit(config)
            submission = reddit.submission(id=post_id)
        except:
            print('Failed to retrieve post:', post_id)
            print('Waiting 60 seconds to retry...')
            sleep(61)
            reddit = initialize_reddit(config)
            submission = reddit.submission(id=post_id)
    post_content = submission.selftext

    # Retrieve images.
    images = []
    if 'imgur.com' in submission.url or submission.url.endswith(('.jpg', '.jpeg', '.png', '.gif')):
        images.append(submission.url)

    try:
        if submission.is_gallery:
            image_dict = submission.media_metadata
            for image_item in image_dict.values():
                try:
                    largest_image = image_item['s']
                    image_url = largest_image['u']
                    images.append(image_url)
                except KeyError:
                    pass
    except AttributeError:
        pass

    # Download images.
    for i, image_url in enumerate(images, start=1):
        file_extension = os.path.splitext(image_url)[-1].split('?')[0]
        filename = f"{post_id}_image_{i}{file_extension}"
        if file_extension not in ['.jpg', '.jpeg', '.png', '.gif']:
            filename = f"{post_id}_image_{i}.jpg"
        outfile = os.path.join(images_directory, filename)
        if os.path.exists(outfile):
            continue
        try:
            response = requests.get(image_url, headers={'User-agent': 'CannBot'})
        except:
            try:
                print('Failed to download image:', image_url)
                print('Waiting 60 seconds to retry...')
                sleep(60)
                response = requests.get(image_url, headers={'User-agent': 'CannBot'})
            except:
                print('Failed to download image:', image_url)
                print('Waiting 60 seconds to retry...')
                sleep(60)
                response = requests.get(image_url, headers={'User-agent': 'CannBot'})
        sleep(3.33)
        if response.status_code != 200:
            print('Unsuccessful request for image:', image_url)
            continue
        with open(outfile, 'wb') as file:
            file.write(response.content)
        print(f"Downloaded image: {outfile}")

    # Retrieve comments.
    comments = []
    submission.comments.replace_more(limit=None)
    for comment in submission.comments.list():
        comments.append({
            'comment_id': comment.id,
            'comment_author': comment.author.name if comment.author else None,
            'comment_body': comment.body,
            'comment_created_utc': datetime.utcfromtimestamp(comment.created_utc).strftime('%Y-%m-%d %H:%M:%S')
        })

    # Update post_data with the retrieved information.
    post_data['post_content'] = post_content
    post_data['upvotes'] = submission.ups
    post_data['downvotes'] = submission.downs
    post_data['images'] = images
    post_data['comments'] = comments
    print('Post data retrieved:', submission.title)
    all_posts.append(post_data)
    sleep(3.33)

# Optional: Try downloading all of the images after the post data is retrieved?

# Save the post data.
try:
    df = pd.DataFrame(all_posts)
    data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"
    timestamp = datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
    datafile = os.path.join(data_dir, f'fl-medical-trees-coa-posts-{timestamp}.xlsx')
    df.to_excel(datafile, index=False)
    print('Saved post data:', datafile)
except:
    print('No posts to curate.')


#-----------------------------------------------------------------------
# Parse COA URLs from images.
#-----------------------------------------------------------------------

# DEV: Extract all COA URLs for posts from logs.
text_file = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\158-reported-effects\data\scanned-images.txt"
with open(text_file, 'r') as file:
    lines = file.readlines()
    coa_urls = {}
    for line in lines:
        if 'COA URL found for post' in line:
            post_id, coa_url = line.split(':', maxsplit=1)
            post_id = 't3_' + post_id.split(' ')[-1].strip()
            coa_url = coa_url.strip()
            post_urls = coa_urls.get(post_id, [])
            post_urls.append(coa_url)
            coa_urls[post_id] = list(set(post_urls))
print('Number of COA URLs:', len(coa_urls))

# Define the image directory.
images_directory = 'D://data/reddit/FLMedicalTrees/images'

# Initialize CoADoc.
parser = CoADoc()
temp_path = tempfile.mkdtemp()

# Scan all images for COA URLs.
# coa_urls = {}
image_files = os.listdir(images_directory)
image_files = [os.path.join(images_directory, x) for x in image_files]
print('Number of images:', len(image_files))
for image_file in image_files:
    post_id = os.path.basename(image_file).split('_')[0]
    if post_id in coa_urls:
        continue
    print('Scanning:', image_file)
    post_urls = coa_urls.get(post_id, [])
    try:
        coa_url = parser.scan(
            image_file,
            temp_path=temp_path,
        )
    except:
        print('Failed to scan:', image_file)
        continue
    if coa_url:
        print(f"COA URL found for post {post_id}: {coa_url}")
        post_urls.append(coa_url)
        coa_urls[post_id] = list(set(post_urls))

# Clean up the temporary directory.
try:
    shutil.rmtree(temp_path)
except:
    pass


#-----------------------------------------------------------------------
# Download COA PDFs using the COA URLs.
#-----------------------------------------------------------------------

# Define the minimum file size for a PDF.
MIN_FILE_SIZE = 21 * 1024

# Download all PDFs.
pdf_dir = 'D://data/reddit/FLMedicalTrees/pdfs'
redirect_urls = {}
for post_id, urls in coa_urls.items():
    print(f"Downloading COA for post {post_id}: {urls}")
    for i, url in enumerate(urls, start=1):
        filename = f"{post_id}-coa-{i}.pdf"
        outfile = os.path.join(pdf_dir, filename)
        if os.path.exists(outfile):
            print('Cached:', outfile)
            continue

        # Download Kaycha Labs COAs.
        if 'yourcoa.com' in url:
            base = 'https://yourcoa.com'
            sample_id = url.split('/')[-1].split('?')[0].split('&')[0]
            if sample_id == 'coa-download':
                sample_id = url.split('sample=')[-1]
            try:
                coa_url = f'{base}/coa/download?sample={sample_id}'
                response = requests.get(coa_url, headers=DEFAULT_HEADERS)
                if response.status_code == 200:
                    if len(response.content) < MIN_FILE_SIZE:
                        print('File size is small, retrying with Selenium:', url)
                        # coa_url = f'{base}/coa/coa-view?sample={sample_id}'
                        response = requests.get(url, allow_redirects=True)
                        if response.status_code == 200:
                            redirected_url = response.url
                            download_file_with_selenium(
                                redirected_url,
                                download_dir=pdf_dir,
                            )
                            print('Downloaded with Selenium:', redirected_url)
                    else:
                        with open(outfile, 'wb') as pdf:
                            pdf.write(response.content)
                        print('Downloaded:', outfile)
                else:
                    print('Failed to download, retrying with Selenium:', url)
                    response = requests.get(url, allow_redirects=True)
                    if response.status_code == 200:
                        redirected_url = response.url
                        download_file_with_selenium(
                            redirected_url,
                            download_dir=pdf_dir,
                        )
                        print('Downloaded with Selenium:', redirected_url)
            except:
                coa_url = f'{base}/coa/coa-view?sample={sample_id}'
                response = requests.get(coa_url, allow_redirects=True)
                if response.status_code == 200:
                    redirected_url = response.url
                    download_file_with_selenium(
                        redirected_url,
                        download_dir=pdf_dir,
                    )
                    print('Downloaded with Selenium:', redirected_url)

        # Download Method Testing Labs COAs.
        elif 'mete.labdrive.net' in url:
            download_file_with_selenium(
                url,
                download_dir=pdf_dir,
                method='a',
                tag_name='a',
                filename=f"{post_id}-coa-{i}.pdf",
            )
            print('Downloaded with Selenium:', url)

        # Download regular PDFs.
        # Note: Ensure ModernCanna, ACS, etc. COAs are being downloaded.
        elif url.startswith('http'):
            response = requests.get(url, allow_redirects=True)
            if response.status_code == 200:
                filename = f"{post_id}-coa-{i}.pdf"
                outfile = os.path.join(pdf_dir, filename)
                with open(outfile, 'wb') as file:
                    file.write(response.content)
                print(f"Downloaded COA: {outfile}")
            sleep(1)
        
        # Skip invalid URLs.
        else:
            print('Invalid URL:', url)
            continue

# Remove duplicate PDFs.
remove_duplicate_files(pdf_dir, verbose=True)

try:

    # Tie the COA URLs to the posts.
    df = pd.DataFrame(all_posts)
    df['coa_url'] = df['post_id'].map(coa_urls)
    df['redirect_url'] = df['post_id'].map(redirect_urls)

    # Save the post data.
    data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"
    timestamp = datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
    datafile = os.path.join(data_dir, f'fl-medical-trees-coa-posts-{timestamp}.xlsx')
    df.to_excel(datafile, index=False)
    print('Saved post data:', datafile)

except:
    print('No posts to curate.')


#-----------------------------------------------------------------------
# Parse COA data from the PDFs.
#-----------------------------------------------------------------------

# Parse COA data from the PDFs.
parser = CoADoc()
pdf_dir = r'D:\data\reddit\FLMedicalTrees\pdfs'
pdf_files = os.listdir(pdf_dir)
pdf_files = [os.path.join(pdf_dir, x) for x in pdf_files]
logging.info('Parsing %i COA PDFs...' % len(pdf_files))
all_coa_data = {}
failed = []
temp_path = tempfile.mkdtemp()
for pdf_file in pdf_files:
    try:
        coa_data = parser.parse_pdf(
            pdf_file,
            temp_path=temp_path,
            verbose=True,
            use_qr_code=False,
        )
    except:
        logging.info('Failed to parse: %s' % pdf_file)
        failed.append(pdf_file)
        continue
    if coa_data:
        logging.info('Parsed: %s' % pdf_file)
        coa_id = os.path.basename(pdf_file).split(' ')[0]
        if isinstance(coa_data, list):
            all_coa_data[coa_id] = coa_data[0]
        elif isinstance(coa_data, dict):
            all_coa_data[coa_id] = coa_data
    else:
        logging.info('Found no data: %s' % pdf_file)
try:
    shutil.rmtree(temp_path)
except:
    pass

# Compile the COA data and remove duplicates.
coa_df = pd.DataFrame(list(all_coa_data.values()))
coa_df['coa_id'] = [x.replace('.pdf', '') for x in list(all_coa_data.keys())]
coa_df.drop_duplicates(subset=['coa_id', 'sample_id', 'results_hash'], inplace=True)

# Save the COA data.
data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"
timestamp = datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
datafile = os.path.join(data_dir, f'fl-medical-trees-coa-data-{timestamp}.xlsx')
coa_df.to_excel(datafile, index=False)
logging.info('Saved %i COA data: %s' % (len(coa_df), datafile))

# Save all of the COAs that failed to be parsed.
with open(os.path.join(data_dir, f'unidentified-coas-{timestamp}.json'), 'w') as f:
    json.dump(failed, f, indent=4)
logging.info('Saved list of failed PDFs.')


#-----------------------------------------------------------------------
# Create a sample of posts that have COA URls.
#-----------------------------------------------------------------------

# Define where the data lives.
data_dir = r"C:\Users\keega\Documents\cannlytics\cannabis-data-science\season-4\155-seed-to-smoke\data"

# Read all of the posts.
post_datafiles = [os.path.join(data_dir, x) for x in os.listdir(data_dir) if 'posts' in x and 'results' not in x]
posts = pd.concat([pd.read_excel(x) for x in post_datafiles])
posts.drop_duplicates(subset=['post_id', 'coa_url', 'redirect_url'], inplace=True)
print('Number of posts:', len(posts))

# Read in all of the COA datafiles.
coa_datafiles = [os.path.join(data_dir, x) for x in os.listdir(data_dir) if 'coa-data' in x]
results = pd.concat([pd.read_excel(x) for x in coa_datafiles])
results.drop_duplicates(subset=['sample_id', 'results_hash'], inplace=True)
print('Number of COA results:', len(results))

# Find the post ID for the lab results.
post_coa_data = {}
for index, row in results.iterrows():
    coa_id = row['coa_id'].replace('t3_', '').split('-coa')[0]
    matches = posts.loc[posts['post_id'].str.contains(coa_id)]
    try:
        if matches.empty:
            matches = posts.loc[posts['coa_url'].str.contains(coa_id)]
        if matches.empty:
            matches = posts.loc[posts['redirect_url'].str.contains(coa_id)]
    except:
        pass
    if not matches.empty:
        post_id = matches['post_id'].values[0]
        post_coa_data[post_id] = row.to_dict()

# Merge the lab results with the post data.
coa_data_df = pd.DataFrame.from_dict(post_coa_data, orient='index')
coa_data_df.reset_index(inplace=True)
coa_data_df.rename(columns={'index': 'post_id'}, inplace=True)
merged_df = posts.merge(coa_data_df, on='post_id', how='left')
merged_df = merged_df.loc[~merged_df['coa_id'].isna()]
print('Number of posts with COA data:', len(merged_df))

# Save the updated post data.
timestamp = datetime.now().strftime('%Y-%m-%d-%H-%M-%S')
datafile = os.path.join(data_dir, f'fl-medical-trees-posts-with-results-{timestamp}.xlsx')
merged_df.to_excel(datafile, index=False)
print('Saved %i posts with COA data:' % len(merged_df), datafile)