optimized sitemap search

src/staticwordpress/core/sitemaps.py

@@ -41,7 +41,7 @@ from bs4 import BeautifulSoup
 # +++++++++++++++++++++++++++++++++++++++++++++++++++++
 
 from ..core.constants import CONFIGS
-from ..core.utils import get_clean_url, get_remote_content
+from ..core.utils import get_clean_url, get_remote_content, is_url_valid
 
 # +++++++++++++++++++++++++++++++++++++++++++++++++++++
 # IMPLEMENATIONS
@@ -57,27 +57,29 @@ def find_sitemap_location(home_url_: str) -> str:
     Returns:
         str: Location of Sitemap
     """
-    for sitemap_path in CONFIGS["SITEMAP"]["SEARCH_PATHS"]:
-        sitemap_url = get_clean_url(home_url_, sitemap_path)
-        response = get_remote_content(sitemap_url)
-        if response.status_code < 400:
-            return parse.urlparse(response.url).path
+    if is_url_valid(home_url_):
+        for sitemap_path in CONFIGS["SITEMAP"]["SEARCH_PATHS"]:
+            sitemap_url = get_clean_url(home_url_, sitemap_path)
+            response = get_remote_content(sitemap_url)
+            if response.status_code < 400:
+                return parse.urlparse(response.url).path
 
-    # robots.txt
-    robots_txt = get_clean_url(home_url_, "robots.txt")
-    response = get_remote_content(robots_txt)
-    if response:
-        for item in response.text.split("\n"):
-            if item.startswith("Sitemap:"):
-                return item.split("Sitemap:")[-1].strip()
+        # robots.txt
+        robots_txt = get_clean_url(home_url_, "robots.txt")
+        response = get_remote_content(robots_txt)
+        if response:
+            for item in response.text.split("\n"):
+                if item.startswith("Sitemap:"):
+                    return item.split("Sitemap:")[-1].strip()
 
-    # check home page for link rel=sitemap
-    response = get_remote_content(home_url_)
-    if response:
-        soup = BeautifulSoup(response.text, features="xml")
-        for link in soup.find_all("link"):
-            if link.has_attr("sitemap"):
-                return link["href"]
+        # check home page for link rel=sitemap
+        response = get_remote_content(home_url_)
+        if response:
+            soup = BeautifulSoup(response.text, features="xml")
+            for link in soup.find_all("link"):
+                if link.has_attr("sitemap"):
+                    return link["href"]
+        return ""
     return ""
 
 

src/staticwordpress/core/utils.py

@@ -228,6 +228,12 @@ def is_url_valid(url_: str) -> bool:
     url_parsed_ = parse.urlparse(url_)
 
     if all([url_parsed_.scheme, url_parsed_.netloc]):
-        return get_remote_content(url_parsed_, max_retires=1).status_code < 399
+        from urllib.request import urlopen
 
+        # print(url_parsed_)
+        # # return get_remote_content(url_parsed_, max_retires=1).status_code < 399
+        try:
+            return urlopen(url_).getcode() < 399
+        except:
+            return False
     return False

src/staticwordpress/gui/mainwindow.py

@@ -378,7 +378,7 @@ class StaticWordPressGUI(QMainWindow):
                 self._project.open(project_path)
                 if self._project.is_open():
                     project_dialog = ProjectDialog(
-                        self, self._project, title_="Project Properties"
+                        parent=self, project_=self._project, title_="Project Properties"
                     )
 
                     if project_dialog.exec_():
@@ -756,7 +756,8 @@ class StaticWordPressGUI(QMainWindow):
     def update_widgets(self) -> None:
         self.findChild(QMenu, "menu_github").setEnabled(self._project.has_github())
         self.findChild(QMenu, "menu_wordpress").setEnabled(
-            self._project.has_wordpress() or self._project.can_crawl()
+            self._project.is_open()
+            and (self._project.has_wordpress() or self._project.can_crawl())
         )
         self.findChild(QToolBar, "toolbar_github").setEnabled(
             self._project.has_github()

src/staticwordpress/gui/project.py

@@ -323,7 +323,7 @@ class ProjectDialog(QDialog):
         self.pushbutton_verify.setIcon(
             QIcon(f"{SHARE_FOLDER_PATH}/icons/check_project.svg")
         )
-        self.pushbutton_verify.clicked.connect(self.check_project)
+        self.pushbutton_verify.clicked.connect(self.verify_project_settings)
 
         self.pushbutton_save = QPushButton("&Save")
         self.pushbutton_save.setIcon(QIcon(f"{SHARE_FOLDER_PATH}/icons/ok.svg"))
@@ -398,10 +398,11 @@ class ProjectDialog(QDialog):
         self._bg_worker.emit_sitemap_location.connect(self.update_sitemap_location)
         self._bg_thread.start()
 
-    def update_sitemap_location(self, sitemap_location):
-        self.lineedit_sitemap.setText(sitemap_location)
+    def update_sitemap_location(self, sitemap_location_):
+        if sitemap_location_:
+            self.lineedit_sitemap.setText(sitemap_location_)
 
-    def check_project(self):
+    def verify_project_settings(self):
         """"""
         # TODO: Add checks for WP_API and Gh_API and if not present then disable them.
         # TODO: Move these checks to background thread e.g. for WP_API or  SRC_URL or SRC or DST Path
@@ -455,7 +456,7 @@ class ProjectDialog(QDialog):
             [
                 self.lineedit_project_name.text(),
                 self.lineedit_output.text(),
-                is_url_valid(self.lineedit_src_url.text()),
+                # is_url_valid(self.lineedit_src_url.text()),
                 Path(self.lineedit_output.text()).is_dir(),
             ]
         ):