improve #4, fix #5

- now all files are imported & functions are called automatically, so no need to import your module! Just drop the file in the folder and watch the magic happen
- add factor of importance: after importing the files, the script executes the functions with the greatest importance first

Author: corentinbettiol

django-check-seo/checks/__init__.py

@@ -0,0 +1,18 @@
+# Standard Library
+import glob
+from os.path import basename, dirname, isfile, join
+
+
+# list files
+modules = glob.glob(join(dirname(__file__), "*.py"))
+
+__all__ = []
+
+# add them to __all__ so they can be imported
+for module in modules:
+    if (
+        isfile(module)
+        and not module.endswith("__init__.py")
+        and not module.endswith("launch_checks.py")
+    ):
+        __all__.append(basename(module)[:-3])

django-check-seo/checks/check_description.py

@@ -5,7 +5,16 @@
 from django.utils.translation import gettext as _
 
 
-def check_description(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     meta = site.soup.find_all("meta")
     for tag in meta:
         if (

django-check-seo/checks/check_h1.py

@@ -5,7 +5,16 @@
 from django.utils.translation import gettext as _
 
 
-def check_h1(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """Check all h1-related conditions
     """
 

django-check-seo/checks/check_h2.py

@@ -5,10 +5,19 @@
 from django.utils.translation import gettext as _
 
 
-def check_h2(self):
-    h2 = self.soup.find_all("h2")
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
+    h2 = site.soup.find_all("h2")
     if not h2:
-        self.warnings.append(
+        site.warnings.append(
             {
                 "name": _("No h2 tag"),
                 "settings": _("at least 1"),
@@ -20,7 +29,7 @@ def check_h2(self):
     else:
         occurence = []
         # check if each keyword
-        for keyword in self.keywords:
+        for keyword in site.keywords:
             # is present at least
             for single_h2 in h2:
                 occurence.append(
@@ -34,7 +43,7 @@ def check_h2(self):
                 )
         # if no keyword is found in h2
         if not any(i > 0 for i in occurence):
-            self.warnings.append(
+            site.warnings.append(
                 {
                     "name": _("No keyword in h2"),
                     "settings": _("at least 1"),

django-check-seo/checks/check_images.py

@@ -1,13 +1,26 @@
 # Third party
+import bs4
 from django.utils.translation import gettext as _
 
 
-def check_images(self):
-    images = self.content.find_all("img")
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
+    images = bs4.element.ResultSet(None)
+
+    for c in site.content:
+        images += c.find_all("img")
 
     for image in images:
         if "alt" not in image.attrs or image.attrs["alt"] == "None":
-            self.problems.append(
+            site.problems.append(
                 {
                     "name": _("Img lack alt tag"),
                     "settings": _("all images"),

django-check-seo/checks/check_keyword_url.py

@@ -2,7 +2,16 @@
 from django.utils.translation import gettext as _
 
 
-def check_keyword_url(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """Check presence of keywords in url
     """
     for keyword in site.keywords:

django-check-seo/checks/check_keywords.py

@@ -2,8 +2,18 @@
 from django.utils.translation import gettext as _
 
 
-def check_keywords(site):
-    """Ensure that all keywords are present.
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 5
+
+
+def run(site):
+    """Ensure that meta tag exists and contain at least one keyword.
+    Populate site.keywords list with keywords found.
     """
     meta = site.soup.find_all("meta")
     for tag in meta:

django-check-seo/checks/check_links.py

@@ -2,13 +2,28 @@
 import os
 
 # Third party
+import bs4
 from django.utils.translation import gettext as _
 
 
-def check_links(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """Check all link-related conditions
     """
-    links = site.content.find_all("a")
+
+    links = bs4.element.ResultSet(None)
+
+    for c in site.content:
+        links = c.find_all("a")
+
     internal_links = 0
     external_links = 0
 

django-check-seo/checks/check_title.py

@@ -2,7 +2,16 @@
 from django.utils.translation import gettext as _
 
 
-def check_title(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """Check all title-related conditions.
     """
     # title presence

django-check-seo/checks/check_url.py

@@ -2,7 +2,16 @@
 from django.utils.translation import gettext as _
 
 
-def check_url(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """All the url-related checks.
     """
 

django-check-seo/checks/content_words_number.py

@@ -1,17 +1,21 @@
-# Standard Library
-import re
-
 # Third party
 from django.utils.translation import gettext as _
 
 
-def content_words_number(site):
-    """Count number of words in content.
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
     """
+    return 1
 
-    content = re.findall(r"\w+", site.content.text.lower())
 
-    nb_words = len(content)
+def run(site):
+    """Count number of words in content.
+    """
+
+    nb_words = len(site.content_text)
 
     # too few words
     if nb_words < site.settings.SEO_SETTINGS["content_words_number"][0]:

django-check-seo/checks/keyword_present_first_paragraph.py

@@ -2,15 +2,24 @@
 from django.utils.translation import gettext as _
 
 
-def keyword_present_first_paragraph(site):
+def importance():
+    """Scripts with higher importance will be executed in first.
+
+    Returns:
+        int -- Importance of the script.
+    """
+    return 1
+
+
+def run(site):
     """Get [keywords_in_first_words] first words of the content, and ensure that there is a keyword among them.
     """
-    content = site.content.text.lower().split()[
+    first_N_words = site.content_text.split()[
         : site.settings.SEO_SETTINGS["keywords_in_first_words"]
     ]
 
     for keyword in site.keywords:
-        if keyword in content:
+        if keyword in first_N_words:
             return
 
     site.problems.append(

django-check-seo/checks/launch_checks.py

@@ -1,33 +1,33 @@
+# Standard Library
+import importlib
+import sys
+
+# Local application / specific library imports
+from . import *  # noqa: F403,F401
+
+
 def launch_checks(site):
     """All the checks are performed here. Called in get_context_data().
-    All function should add a dict in site.problems or site.warnings
+    All functions should do its test(s), then add a dict in site.problems or site.warnings.
 
     Arguments:
-        site {Site} -- A set of useful vars (including problems & warnings, two lists of dict).
+        site {Site} -- A set of useful vars that can be used by the functions (including problems & warnings, two lists of dict).
     """
-    from .check_description import check_description
-    from .check_h1 import check_h1
-    from .check_h2 import check_h2
-    from .check_images import check_images
-    from .check_keywords import check_keywords
-    from .check_keyword_url import check_keyword_url
-    from .check_links import check_links
-    from .check_title import check_title
-    from .check_url import check_url
-    from .content_words_number import content_words_number
-    from .keyword_present_first_paragraph import keyword_present_first_paragraph
-
-    # add your file here pls
-
-    check_keywords(site)
-    check_description(site)
-    check_h1(site)
-    check_h2(site)
-    check_images(site)
-    check_keyword_url(site)
-    check_links(site)
-    check_title(site)
-    check_url(site)
-    content_words_number(site)
-    keyword_present_first_paragraph(site)
-    # add your function here pls
+
+    modules_order = []
+
+    # only get modules in ...checks.*
+    for module_name in sys.modules:
+        if (
+            "django-check-seo.checks." in module_name
+            and module_name != "django-check-seo.checks.launch_checks"
+        ):
+            module = importlib.import_module(module_name)
+            get_module_order = getattr(module, "importance")
+
+            # get the importance
+            modules_order.append([module, get_module_order()])
+
+    # execute modules with higher importance first from sorted list
+    for module in sorted(modules_order, key=lambda x: x[1], reverse=True):
+        getattr(module[0], "run")(site)

django-check-seo/views.py

@@ -43,10 +43,9 @@ def get_context_data(self, *args, **kwargs):
 
         (context["problems"], context["warnings"]) = (site.problems, site.warnings)
 
-        context["parsehtml"] = r.text
         context["settings"] = json.dumps(settings.SEO_SETTINGS, indent=4)
-        context["contenthtml"] = site.content
-        context["content"] = site.content_text
+        context["html"] = site.content
+        context["text"] = site.content_text
 
         return context
 
@@ -73,20 +72,25 @@ def __init__(self, soup, full_url):
         self.soup = soup
 
         # Get content of the page (exclude header/footer)
-        self.content = self.soup.find("div", {"class": "container"})
+        self.content = self.soup.select(".container")
+
         if self.content is None:
             self.content = ""
 
-        # remove ul with nav class from content (<ul class="nav">, <ul class="navbar">, or <nav>)
-        if self.content.find("ul", {"class": "nav"}):
-            self.content.find("ul", {"class": "nav"}).extract()
-        elif self.content.find("ul", {"class": "navbar"}):
-            self.content.find("ul", {"class": "navbar"}).extract()
-        elif self.content.find("nav"):
-            self.content.find("nav").extract()
+        for c in self.content:
+            # remove ul with nav class from content (<ul class="nav">, <ul class="navbar">, or <nav>)
+            if c.find("ul", {"class": "nav"}):
+                c.find("ul", {"class": "nav"}).extract()
+            elif c.find("ul", {"class": "navbar"}):
+                c.find("ul", {"class": "navbar"}).extract()
+            elif c.find("nav"):
+                c.find("nav").extract()
+
+        # get content without doublewords thx to custom separator ("<h1>Title</h1><br /><p>Content</p>" -> TitleContent)
+        self.content_text = ""
+        for c in self.content:
+            self.content_text += c.get_text(separator=" ")
 
-        # get content without doublewords thx to separator ("<h1>Title</h1><br /><p>Content</p>" -> TitleContent)
-        self.content_text = self.content.get_text(separator=" ")
         # strip multiple carriage return (with optional space) to only one
         self.content_text = re.sub(r"(\n( ?))+", "\n", self.content_text)
         # strip multiples spaces (>3) to only 2 (for title readability)