Added deHTML

.gitignore

@@ -128,4 +128,5 @@ dmypy.json
 # PyCharm
 .idea/
 
-upload.sh
+upload.sh
+test.py

tryhackme/path.py

@@ -1,5 +1,5 @@
 from .task import PathTask
-
+from . import utils
 
 class Path:
     def __init__(self, state, data):
@@ -14,9 +14,9 @@ def __init__(self, state, data):
 
     def _from_data(self, data):
         self.code = data.get("code")
-        self.description = data.get("description")
+        self.raw_description = data.get("description")
         self.color = data.get("color")
-        self.intro = data.get("intro")
+        self.raw_intro = data.get("intro")
         self.type = data.get("contentType")
         self.public = data.get("public", False)
         self.room_count = data.get("roomNo")
@@ -34,6 +34,12 @@ def _sync(self, data):
         self.user = self._state.store_user(data.get('username'))
 
     @property
+    def description(self):
+        return utils.HTML_parse(self.raw_description)
+    @property
+    def intro(self):
+        return utils.HTML_parse(self.raw_intro)
+    @property
     def tasks(self):
         return [PathTask(state=self._state, data=task) for task in self._tasks]
     @property

tryhackme/question.py

@@ -1,3 +1,4 @@
+from . import utils
 
 # TODO: de HTML all the things
 class Question:
@@ -6,12 +7,22 @@ def __init__(self, state, data):
         self._from_data(data)
 
     def _from_data(self, data):
-        self.question = data.get("question")
+        self.raw_question = data.get("question")
         self.number = data.get("questionNo")
-        self.hint = data.get("hint")
-        self.description = data.get("answerDesc", None)
+        self.raw_hint = data.get("hint")
+        self.raw_description = data.get("answerDesc", "")
         self.extra_points = data.get("extraPoints", None)
         self.correct = data.get("correct", False)
-        self.attempts = data.get("attempts", None)
-        self.submission = data.get("submission", None)
-        self.has_answer = data.get("noAnswer", False)
+        self.attempts = data.get("attempts", 0)
+        self.submission = data.get("submission", "")
+        self.has_answer = data.get("noAnswer", False)
+
+    @property
+    def question(self):
+        return utils.HTML_parse(self.raw_question)
+    @property
+    def description(self):
+        return utils.HTML_parse(self.raw_description)
+    @property
+    def hint(self):
+        return utils.HTML_parse(self.raw_hint)

tryhackme/room.py

@@ -59,7 +59,7 @@ def scoreboard(self):
     def tasks(self):
         # TODO: add sessionless http client for no session task gathering
         if self.freeToUse or self._state.subscribed:
-            return [RoomTask(state=self._state, data=task) for task in self._state.http.get_room_tasks(room_code=self.name)]
+            return [RoomTask(state=self._state, data=task) for task in self._state.http.get_room_tasks(room_code=self.name).get('data')]
         else:
             return [RoomTask(state=self._state, data=task) for task in self._state.http.get_room_tasks(room_code=self.name, settings={"static": True})]
     @property

tryhackme/task.py

@@ -1,6 +1,7 @@
 from .question import Question
+from . import utils
+
 
-# TODO: html elements removing
 class RoomTask:
     def __init__(self, state, data):
         self._state = state
@@ -9,15 +10,19 @@ def __init__(self, state, data):
         self._from_data(data)
 
     def _from_data(self, data):
-        self.title = data.get('taskTitle')
-        self.description = data.get('taskDesc')
+        self.raw_title = data.get('taskTitle')
+        self.raw_description = data.get('taskDesc')
         self.type = data.get('taskType')
         self.number = data.get('taskNo')
         self.created = data.get('taskCreated')
         self.deadline = data.get('taskDeadline')
         self.uploadId = data.get('uploadId')
         self._questions = data.get('tasksInfo', []) if self._state.authenticated else data.get('questions', [])
 
+    @property
+    def title(self):
+        return utils.HTML_parse(self.raw_title, "*")
+
     @property
     def question_count(self):
         return self._questions.__len__()

tryhackme/utils.py

@@ -1,6 +1,6 @@
 import json
-
-
+import re
+import html
 
 def to_json(obj):
     return json.dumps(obj, separators=(',', ':'), ensure_ascii=True)
@@ -12,7 +12,14 @@ def response_to_json_or_text(response):
         if response.headers['content-type'].startswith('application/json'):
             return json.loads(text)
     except KeyError:
-        # Thanks Cloudflare
+        # $ Thanks Cloudflare
         pass
-
     return text
+
+
+_HTML_TAGS_ = re.compile("<[^<]*>")
+def HTML_parse(text, replace=""):
+    text = text.replace("\n", "")
+    text = html.unescape(text)
+    text = re.sub(_HTML_TAGS_, replace, text)
+    return text