Migrated from GitHub

data/P0_get_categories.py

@@ -0,0 +1,29 @@
+import requests
+import pathlib
+import bs4
+
+dest = pathlib.Path("data")
+dest.mkdir(exist_ok=True)
+
+f_save = dest / "categories.txt"
+
+url = "https://www.literotica.com/stories/"
+r = requests.get(url)
+
+if not r.ok:
+    msg = f"Failed to download {url}"
+    raise ValueError(msg)
+
+html = r.content
+soup = bs4.BeautifulSoup(html, "lxml")
+
+with open(f_save, "w") as FOUT:
+    for a in soup.find_all("a", href=True):
+        link = a["href"]
+
+        if "www.literotica.com/c/" not in link:
+            continue
+
+        print(link)
+
+        FOUT.write(link + "\n")

data/P1_get_page_counts.py

@@ -0,0 +1,47 @@
+import requests
+import bs4
+import pandas as pd
+from dspipe import Pipe
+import time
+
+
+categories = open("data/categories.txt").read().strip().split("\n")
+sess = requests.session()
+
+data = []
+
+
+def compute(url):
+    print(url)
+
+    # Append this so we start the first page
+    first_page_url = url + "/1-page"
+
+    r = sess.get(first_page_url)
+
+    if not r.ok:
+        msg = f"Failed to download {url}"
+        raise ValueError(msg)
+
+    soup = bs4.BeautifulSoup(r.content, "lxml")
+    button = soup.find("select")
+
+    # If this fails assume we only have 3 pages
+    try:
+        options = [int(x["value"]) for x in button.find_all("option", value=True)]
+    except AttributeError:
+        options = [1, 2, 3]
+
+    data.append(
+        {"url": url, "last_page": max(options),}
+    )
+
+    print(data[-1])
+
+    time.sleep(1)
+
+
+Pipe(categories, limit=None, shuffle=True)(compute, 1)
+
+df = pd.DataFrame(data).set_index("url")
+df.to_csv("data/categories_with_pages.csv")

data/P2_get_story_links.py

@@ -0,0 +1,41 @@
+import requests
+import bs4
+import pandas as pd
+from dspipe import Pipe
+import time
+
+sess = requests.session()
+
+
+def page_iterator():
+    df = pd.read_csv("data/categories_with_pages.csv")
+
+    for url, n_pages in zip(df.url, df.last_page):
+        for n in range(n_pages):
+            yield (url, n)
+
+
+def compute(item, f1):
+    base_url, n = item
+    category = base_url.split("/")[-1]
+    url = f"{base_url}/{n}-page"
+
+    f1 = f1.parent / f"{category}_{n:04d}.html"
+    if f1.exists():
+        return False
+
+    r = sess.get(url)
+
+    if not r.ok:
+        msg = f"Failed to download {url}"
+        raise ValueError(msg)
+
+    with open(f1, "wb") as FOUT:
+        FOUT.write(r.content)
+
+    time.sleep(2)
+
+
+Pipe(page_iterator(), "data/story_listings", limit=None, shuffle=True, autoname=False)(
+    compute, 1
+)

data/P3_download_stories.py

@@ -0,0 +1,93 @@
+import requests
+import bs4
+import pandas as pd
+from dspipe import Pipe
+import time
+from pathlib import Path
+from itertools import chain
+
+
+"""
+Downloads only the first page of each story, later stages of the pipeline
+check and download additionally pages if needed.
+"""
+
+
+n_downloads = 2
+fail_time = 120
+
+sess = requests.session()
+
+
+def proxy(url):
+    # Comment the line below to use the proxy
+    return sess.get(url)
+    
+    port = 8000
+    host = "" # Fill this in with a proper host
+    proxy_url = f"{host}:{port}"
+    return requests.get(proxy_url, params={"url": url})
+
+
+def collect(f_html):
+
+    with open(f_html) as FIN:
+        soup = bs4.BeautifulSoup(FIN.read(), "lxml")
+
+    links = []
+    for a in soup.find_all("a", href=True):
+        link = a["href"]
+        # 'www.literotica.com/i/' are image stories
+
+        if "www.literotica.com/s/" not in link:
+            continue
+
+        link = link.split("/s/")[-1]
+
+        links.append(link)
+
+    # if not len(links):
+    #    print(f"Note: {f_html} has no text links")
+
+    return links
+
+
+def download(f0, f1):
+    print(f0, f1)
+
+    url = "https://www.literotica.com/s/" + f0
+
+    try:
+        # r = sess.get(url)
+        r = proxy(url)
+
+    except:
+        print(f"Failed to download {url}")
+        time.sleep(fail_time)
+        return False
+
+    if not r.ok:
+        print(f"Failed to download {url}")
+        time.sleep(fail_time)
+        return False
+
+    html = r.content
+    soup = bs4.BeautifulSoup(html, "lxml")
+
+    # Remove the scripting and styling
+    for name in ["script", "style"]:
+        for block in soup.find_all(name):
+            block.decompose()
+
+    with open(f1, "w") as FOUT:
+        FOUT.write(str(soup))
+
+    time.sleep(1)
+
+
+P = Pipe("data/story_listings/", input_suffix=".html", limit=None)
+links = P(collect, -1)
+links = list(chain(*links))
+
+
+Pipe(links, "data/stories", output_suffix=".html", shuffle=True)(download, n_downloads)

data/P4_download_extra_pages.py

@@ -0,0 +1,108 @@
+import requests
+import bs4
+import json
+import pandas as pd
+from dspipe import Pipe
+import time
+from pathlib import Path
+from itertools import chain
+
+"""
+Downloads additional pages if needed for each story. Collects the text
+for each story into a single file along with the meta information from
+the first page.
+"""
+
+n_downloads = 4
+fail_time = 120
+sess = requests.session()
+
+
+def proxy(url):
+    # Comment the line below to use the proxy
+    return sess.get(url)
+    
+    port = 8000
+    host = "" # Fill this in with a proper host
+    proxy_url = f"{host}:{port}"
+    return requests.get(proxy_url, params={"url": url})
+
+
+def download(f0, f1):
+
+    with open(f0) as FIN:
+        raw = FIN.read()
+
+    soup = bs4.BeautifulSoup(raw, "lxml")
+
+    page_text = soup.find("div", class_="b-pager-pages").span.text
+
+    title = soup.find("div", class_="b-story-header").h1.get_text()
+    author = soup.find("div", class_="b-story-header").a.get_text()
+
+    stats = soup.find("span", class_="b-story-stats").text.split()
+    cat = soup.find("div", class_="b-breadcrumbs").children
+
+    meta = {
+        "title": title,
+        "author": author,
+        "category": list(cat)[1].text,
+        "n_comments": int(stats[0]),
+        "n_views": int(stats[2]),
+        "n_favorites": int(stats[4]),
+        "n_pages": int(page_text.split()[0]),
+    }
+
+    next_btn = soup.find("a", text="Next")
+    story = soup.find("div", class_="b-story-body-x").get_text()
+
+    print(meta)
+
+    while next_btn is not None:
+        link = next_btn["href"]
+
+        r = proxy(link)
+
+        if not r.ok:
+            print(f"Failed to download {url}")
+            time.sleep(fail_time)
+            return False
+
+        soup = bs4.BeautifulSoup(r.content, "lxml")
+
+        page_text = soup.find("div", class_="b-story-body-x")
+        next_btn = soup.find("a", text="Next")
+
+        # print(page_text)
+
+        if page_text:
+            story += page_text.get_text()
+
+    item = {
+        "text": story,
+        "meta": meta,
+    }
+
+    js = json.dumps(item, indent=2)
+
+    with open(f1, "w") as FOUT:
+        FOUT.write(js)
+
+    time.sleep(1)
+
+
+def safe(f0, f1):
+    try:
+        download(f0, f1)
+    except Exception as EX:
+        print(f"FAILED {f0}, {EX}")
+
+
+Pipe(
+    "data/stories/",
+    "data/full_text",
+    input_suffix=".html",
+    output_suffix=".json",
+    shuffle=True,
+    limit=None,
+)(safe, n_downloads)

data/P5_collate_stories.py

@@ -0,0 +1,16 @@
+import jsonlines
+import json
+from dspipe import Pipe
+
+f_save = "data/Literotica.jsonl"
+
+
+def compute(f0):
+    with open(f0) as FIN:
+        js = json.load(FIN)
+
+    FOUT.write(js)
+
+
+with jsonlines.open(f_save, "w") as FOUT:
+    Pipe("data/full_text/")(compute, 1)