Merge pull request #46 from ivansaul/refactor

feat: refine selectors to handle dynamic HTML structures

src/facilito/async_api.py

@@ -100,9 +100,9 @@ async def download(self, url: str, **kwargs):
 
         from .downloaders import download_course, download_unit
         from .models import TypeUnit
-        from .utils import is_course, is_lecture, is_video
+        from .utils import is_course, is_lecture, is_quiz, is_video
 
-        if is_video(url) or is_lecture(url):
+        if is_video(url) or is_lecture(url) or is_quiz(url):
             unit = await self.fetch_unit(url)
             extension = ".mp4" if unit.type == TypeUnit.VIDEO else ".mhtml"
             await download_unit(

src/facilito/collectors/course.py

@@ -11,7 +11,7 @@
 
 async def _fetch_course_chapters(page: Page) -> list[Chapter]:
     CHAPTERS_SELECTOR = (
-        ".collapsible.no-box-shadow.no-border.f-topics.no-time div.f-top-16"
+        ".collapsible.no-box-shadow.no-border.f-topics.no-time > .f-top-16"
     )
 
     try:
@@ -32,7 +32,7 @@ async def _fetch_course_chapters(page: Page) -> list[Chapter]:
         for i in range(chapters_count):
             CHAPTER_NAME_SELECTOR = "header h4"
             UNITS_SELECTOR = ".collapsible-body ul a"
-            UNIT_NAME_SELECTOR = ".box p.ibm"
+            UNIT_NAME_SELECTOR = "p.ibm"
 
             chapter_name = (
                 await chapters_selectors.nth(i)
@@ -86,7 +86,7 @@ async def _fetch_course_chapters(page: Page) -> list[Chapter]:
 
 
 async def fetch_course(context: BrowserContext, url: str) -> Course:
-    NAME_SELECTOR = ".cover-with-image h1"
+    NAME_SELECTOR = ".f-course-presentation h1, .cover-with-image h1"
 
     try:
         page = await context.new_page()

src/facilito/collectors/unit.py

@@ -2,13 +2,27 @@
 
 from ..errors import UnitError
 from ..helpers import slugify
-from ..models import Unit
+from ..models import TypeUnit, Unit
 from ..utils import get_unit_type
 
 
 async def fetch_unit(context: BrowserContext, url: str):
     NAME_SELECTOR = ".title-section header h1"
 
+    try:
+        type = get_unit_type(url)
+
+        if type == TypeUnit.QUIZ:
+            # TODO: implement quiz fetching
+            return Unit(
+                type=type,
+                url=url,
+                name="quiz",
+                slug="quiz",
+            )
+    except Exception:
+        raise UnitError()
+
     try:
         page = await context.new_page()
         await page.goto(url)

src/facilito/downloaders/course.py

@@ -42,7 +42,7 @@ async def download_course(context: BrowserContext, course: Course, **kwargs):
                     **kwargs,
                 )
 
-            if unit.type == TypeUnit.LECTURE:
+            else:
                 await download_unit(
                     context,
                     unit,

src/facilito/downloaders/unit.py

@@ -30,5 +30,5 @@ async def download_unit(context: BrowserContext, unit: Unit, path: Path, **kwarg
             **kwargs,
         )  # type: ignore
 
-    if unit.type == TypeUnit.LECTURE:
+    else:
         await save_page(context, unit.url, path)

src/facilito/models.py

@@ -15,6 +15,7 @@ class Quality(str, Enum):
 class TypeUnit(str, Enum):
     LECTURE = "lecture"
     VIDEO = "video"
+    QUIZ = "quiz"
 
 
 class Resource(BaseModel):

src/facilito/utils.py

@@ -143,6 +143,21 @@ def is_course(url: str) -> bool:
     return "/cursos/" in url
 
 
+def is_quiz(url: str) -> bool:
+    """
+    Check if a URL is a quiz.
+
+    :param str url: URL to check.
+    :return bool: True if the URL is a quiz, False otherwise.
+
+    Example
+    -------
+    >>> is_quiz("https: ..../quizzes/...")
+    True
+    """
+    return "/quizzes/" in url
+
+
 def get_unit_type(url: str) -> TypeUnit:
     """
     Get the type of a unit from its URL.
@@ -163,4 +178,7 @@ def get_unit_type(url: str) -> TypeUnit:
     if is_lecture(url):
         return TypeUnit.LECTURE
 
+    if is_quiz(url):
+        return TypeUnit.QUIZ
+
     raise UnitError()