Upload folder using huggingface_hub

tools/clean.py

@@ -0,0 +1,62 @@
+import json
+import re
+
+def Clean_output_file(input_file_path, output_file_path):
+    with open(input_file_path, 'r') as input_file, open(output_file_path, 'w') as output_file:
+        for line in input_file:
+            # Load the line as a JSON object
+            item = json.loads(line)
+
+            # Check if 'text' field exists
+            if 'text' in item:
+                patterns_to_remove = [
+                    r'Author:.*?\n',  # Remove author part
+                    r'\+ More articles by.*?\n',  # Remove "+ More articles" lines
+                    r'- Hide list.*?\n',  # Remove "- Hide list" lines
+                    r'Rating:.*?\n',  # Remove lines starting with "Rating:"
+                    r'\n\uf129\n.*?\n',  # Remove lines with special characters and any text until the next newline                    
+                ]
+
+                # Apply each pattern removal
+                for pattern in patterns_to_remove:
+                    item['text'] = re.sub(pattern, '', item['text'], flags=re.DOTALL)
+                # Replace specific unicode characters with a space or equivalent character
+                item['text'] = item['text'].replace('\u00a0', ' ')
+                item['text'] = item['text'].replace('\u2008', ' ')
+                item['text'] = item['text'].replace('\u2019', '\'')
+
+                # Remove extra newline characters
+                item['text'] = re.sub(r'\n\s*\n', '\n\n', item['text'])
+
+                # Check for unwanted phrases and skip the line if any are found
+                if ("This page doesn" in item['text'] or
+                    "+ Show component code" in item['text'] or
+                    "position: relative" in item['text'] or 
+                    "Don't forget to fill out this" in item['text']or 
+                    "Table of Contents" in item['text'] or
+                    "Tales by SCP Series" in item['text']) :
+                    continue  # Skip this line
+
+                # Remove "rating: +" and any text until the next newline
+                item['text'] = re.sub(r'rating: \+.*?\n', '', item['text'], flags=re.DOTALL)
+
+            # Write the updated item back to the file if it doesn't contain unwanted phrases
+            json.dump(item, output_file)
+            output_file.write('\n')
+
+
+# stories 
+for num in range(1,9):
+     input_file_path = f'scp_stories{num}.jsonl'  # Replace with your input file path
+     output_file_path = f'stories{num}.jsonl'  # Replace with your desired output file path
+     Clean_output_file(input_file_path, output_file_path)
+
+# tales
+input_file_path = f'scp_tales.jsonl'  # Replace with your input file path
+output_file_path = f'scp_tales_cleaned.jsonl'  # Replace with your desired output file path            
+Clean_output_file(input_file_path, output_file_path)
+
+# jokes                
+input_file_path = f'scp_jokes.jsonl'  # Replace with your input file path
+output_file_path = f'jokes-cleaned.jsonl'  # Replace with your desired output file path
+Clean_output_file(input_file_path, output_file_path)                

tools/crawl jokes.py

@@ -0,0 +1,67 @@
+import requests
+from bs4 import BeautifulSoup
+import json
+def crawl_scp_series(url,num):
+    response = requests.get(url)
+    soup = BeautifulSoup(response.content, 'html.parser')
+
+    # Find the div with id 'page-content'
+    content_div = soup.find('div', id='page-content')
+
+    # Extract all the links within this div
+    links = [a['href'] for a in content_div.find_all('a', href=True) if a.text]
+
+    # Initialize a list to store the extracted texts
+    stories = []
+
+    for link in links:
+        # Ensure the link is absolute
+        if not link.startswith('http'):
+            link = f"https://scp-wiki.wikidot.com{link}"
+
+        # Fetch each link
+        try:
+            story_response = requests.get(link)
+            story_soup = BeautifulSoup(story_response.content, 'html.parser')
+            if story_soup:
+                # Extract the text from the div with id 'page-content'
+                page_content = story_soup.find('div', id='page-content')
+                if page_content:
+                    first_div = page_content.find('div', style="text-align: right;")
+                    if first_div:
+                        first_div.decompose()
+
+                    # Remove the div with class 'licensebox' if it exists
+                    licensebox_div = page_content.find('div', class_='licensebox')
+                    if licensebox_div:
+                        licensebox_div.decompose()
+                    print("Found page-content div")
+                else:
+                    print(f"Could not find page-content div for {link}")
+
+            if page_content:
+                story_text = page_content.get_text().strip()
+                stories.append(story_text)
+
+                # Check if 10 stories have been collected
+                if len(stories) == 10:
+                    # Write these stories to a JSONL file
+                    with open(f"scp_jokes.jsonl", 'a') as file:
+                        for story in stories:
+                            json_record = json.dumps({'text': story})
+                            file.write(json_record + '\n')
+
+                    # Reset the stories list for the next batch
+                    stories = []
+        except requests.exceptions.RequestException as e:
+            print(f"Error fetching {link}: {e}")
+        
+
+# URL of the SCP series page
+urls = ['https://scp-wiki.wikidot.com/joke-scps']
+num=1
+# Start crawling
+for url in urls:
+    crawl_scp_series(url,num)
+    print(url)
+    num+=1

tools/crawl-stories.py

@@ -0,0 +1,67 @@
+import requests
+from bs4 import BeautifulSoup
+import json
+def crawl_scp_series(url,num):
+    response = requests.get(url)
+    soup = BeautifulSoup(response.content, 'html.parser')
+
+    # Find the div with id 'page-content'
+    content_div = soup.find('div', id='page-content')
+
+    # Extract all the links within this div
+    links = [a['href'] for a in content_div.find_all('a', href=True) if a.text]
+
+    # Initialize a list to store the extracted texts
+    stories = []
+
+    for link in links:
+        # Ensure the link is absolute
+        if not link.startswith('http'):
+            link = f"https://scp-wiki.wikidot.com{link}"
+
+        # Fetch each link
+        try:
+            story_response = requests.get(link)
+            story_soup = BeautifulSoup(story_response.content, 'html.parser')
+            if story_soup:
+                # Extract the text from the div with id 'page-content'
+                page_content = story_soup.find('div', id='page-content')
+                if page_content:
+                    first_div = page_content.find('div', style="text-align: right;")
+                    if first_div:
+                        first_div.decompose()
+
+                    # Remove the div with class 'licensebox' if it exists
+                    licensebox_div = page_content.find('div', class_='licensebox')
+                    if licensebox_div:
+                        licensebox_div.decompose()
+                    print("Found page-content div")
+                else:
+                    print(f"Could not find page-content div for {link}")
+
+            if page_content:
+                story_text = page_content.get_text().strip()
+                stories.append(story_text)
+
+                # Check if 10 stories have been collected
+                if len(stories) == 10:
+                    # Write these stories to a JSONL file
+                    with open(f"scp_stories{num}.jsonl", 'a') as file:
+                        for story in stories:
+                            json_record = json.dumps({'text': story})
+                            file.write(json_record + '\n')
+
+                    # Reset the stories list for the next batch
+                    stories = []
+        except requests.exceptions.RequestException as e:
+            print(f"Error fetching {link}: {e}")
+        
+
+# URL of the SCP series page
+urls = ['https://scp-wiki.wikidot.com/scp-series-1','https://scp-wiki.wikidot.com/scp-series-2','https://scp-wiki.wikidot.com/scp-series-3','https://scp-wiki.wikidot.com/scp-series-4','https://scp-wiki.wikidot.com/scp-series-5','https://scp-wiki.wikidot.com/scp-series-6','https://scp-wiki.wikidot.com/scp-series-7','https://scp-wiki.wikidot.com/scp-series-8']
+num=1
+# Start crawling
+for url in urls:
+    crawl_scp_series(url,num)
+    print(url)
+    num+=1

tools/crawl-tales.py

@@ -0,0 +1,55 @@
+import requests
+from bs4 import BeautifulSoup
+import json
+
+def scrape_tales_by_year(start_year, end_year):
+    base_url = "https://scp-wiki.wikidot.com/tales-by-date-"
+    all_tales = []
+
+    for year in range(start_year, end_year + 1):
+        url = f"{base_url}{year}"
+        response = requests.get(url)
+        soup = BeautifulSoup(response.content, 'html.parser')
+
+        # Find all tables, then find all table rows within each table
+        for table in soup.find_all('table', class_='wiki-content-table'):
+            for row in table.find_all('tr'):
+                cells = row.find_all('td')
+                if cells:
+                    a_tag = cells[0].find('a', href=True)
+                    if a_tag:
+                        tale_url = f"https://scp-wiki.wikidot.com{a_tag['href']}"
+                        tale_response = requests.get(tale_url)
+                        tale_soup = BeautifulSoup(tale_response.content, 'html.parser')
+
+                        # Extract the text from the div with id 'page-content'
+                        page_content = tale_soup.find('div', id='page-content')
+                        if page_content:
+                            # Remove the first div with specific style if it exists
+                            first_div = page_content.find('div', style="text-align: right;")
+                            if first_div:
+                                first_div.decompose()
+
+                            # Remove the div with class 'licensebox' if it exists
+                            licensebox_div = page_content.find('div', class_='licensebox')
+                            if licensebox_div:
+                                licensebox_div.decompose()
+
+                            # Now get the text from the rest of the page-content
+                            tale_text = page_content.get_text().strip()
+
+                            all_tales.append({'text': tale_text})
+
+                            # Find the first non-empty line in the tale's text
+                            first_line = next((line for line in tale_text.splitlines() if line.strip()), "")
+                            print(first_line)  # Print the first non-empty line of the tale
+                            # Optionally, write tale to a file or process it further here
+                        else:
+                            print(f"Could not find page-content div for {tale_url}")
+    # Write the tales to a JSONL file
+    with open('scp_tales.jsonl', 'w') as file:
+        for tale in all_tales:
+            json_record = json.dumps(tale)  # tale is already a dict with the key "text"
+            file.write(json_record + '\n')
+# Call the function with the range of years you want to scrape
+scrape_tales_by_year(2008, 2022)