Improved URL handling

py34/list.py

@@ -1,5 +1,6 @@
 from .post import Post
-from .scraper import scraper
+from .scraper import scraper, ScraperException
+from .url import parse_thumbnail_url
 from .dockid import is_nochick, is_toodeep
 import urllib.parse
 from threading import Thread
@@ -34,9 +35,9 @@ class List:
                 # Extract image
                 img = entry.find_all("img")[0]
                 if "src" in img.attrs:
-                    img_src = img["src"].split("?")[0].split("/")[-2:]
+                    img_src = parse_thumbnail_url(img["src"])
                 else:
-                    img_src = img["data-cfsrc"].split("?")[0].split("/")[-2:]
+                    img_src = parse_thumbnail_url(img["data-cfsrc"])
 
                 # Append post
                 def _thread_proc(*argv, **kwargs):
@@ -45,8 +46,8 @@ class List:
                     target=_thread_proc,
                     args=(
                         int(entry["id"][1:]),
-                        int(img_src[0]),
+                        img_src.dir,
-                        img_src[1].split("_")[1].split(".")[0], # Thumbnail_[id].jpg
+                        img_src.id,
                         img["alt"].split(" "),
                     )
                 ))

py34/url.py

@@ -0,0 +1,48 @@
+from urllib.parse import urlparse
+from os.path import splitext
+
+
+class ImageURL:
+    def __init__(self, image_dir: int, image_id: str, image_format: str):
+        self.dir:    int = image_dir
+        self.id:     str = image_id
+        self.format: str = image_format
+
+
+class ThumbnailURL:
+    def __init__(self, image_dir: int, image_id: str):
+        self.dir: int = image_dir
+        self.id:  str = image_id
+
+
+def parse_image_url(url: str) -> ImageURL:
+    url = urlparse(url)
+    if url.hostname != "wimg.rule34.xxx":
+        raise Exception("Invalid URL hostname")
+
+    path = list(filter(bool, url.path.split("/")))
+    if len(path) != 3 or path[0] != "images":
+        raise Exception("Invalid URL path")
+
+    file = splitext(path[2])
+    return ImageURL(
+        int(path[1]),
+        file[0],
+        file[1],
+    )
+
+
+def parse_thumbnail_url(url: str) -> ThumbnailURL:
+    url = urlparse(url)
+    if url.hostname != "wimg.rule34.xxx":
+        raise Exception("Invalid URL hostname")
+
+    path = list(filter(bool, url.path.split("/")))
+    if len(path) != 3 or path[0] != "thumbnails":
+        raise Exception("Invalid URL path")
+
+    file = splitext(path[2].split("_")[1])
+    return ThumbnailURL(
+        int(path[1]),
+        file[0],
+    )