v4l3n71n/web_scrap
1
0
Fork 0
Code Issues Pull Requests Projects Releases 10 Wiki Activity

Menu and sub-menu

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-07-09 15:27:20 +02:00
parent 9acb620f93
commit 03f833a2c3
1 changed files with 37 additions and 30 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

67
lib/WPMenu.py
View File

@ -73,49 +73,56 @@ class WPMenu:
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, canalblog, err))
def _child(self, title, parent):
menu = {"id":"", "type":"", "title": "", "parent":"", "children": []}
exist = False
for i in ["categories", "tags"]:
if exist is False:
try:
params = {"search":title, "per_page":"100"}
page = self._request_wordpress.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
if len(result) > 0:
menu = {"id":result[0]["id"], "type":i, "title": title, "parent":parent, "children": []}
exist = True
else:
self._logger.error("{0} : {2} didn't get due status code : {1}".format(self._name, page.status_code, i))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
return menu
def _menu(self, content):
soup = BeautifulSoup(content, self._parser)
ul = soup.find_all("ul", id="listsmooth")
menu = list()
ul = soup.find("ul", id="listsmooth")
menus = list()
menu = {"id":"", "type":"", "title": "", "parent":"", "children": []}
parents = list()
for anchor in ul[0].find_all("li"):
children = list()
for anchor in ul.find_all("li"):
parent = anchor.find("a").get_text()
href = anchor.find("a").get("href")
menu = self._child(parent, parent)
if href == "#":
li = anchor.find_all("li")
li = anchor.find("ul").find_all("li")
for child in li:
a = child.find("a")
self._logger.info("{0} {1} : {2}".format(self._name, anchor.find("a").get_text(), a.get_text()))
for i in ["categories", "tags"]:
exist == False
if exist is False:
try:
params = {"search":a.get_text(), "per_page":"100"}
page = self._request_wordpress.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), auth=self._basic, params=params)
if page.status_code == 200:
result = page.json()
if len(result) > 0:
menu.append({"id":result[0]["id"], "type":i, "title": a.get_text(), "parent":parent})
exist = True
else:
self._logger.error("{0} : {2} didn't get due status code : {1}".format(self._name, page.status_code, i))
self._logger.debug("{0} : {1}".format(self._name, page.content))
except ConnectionError as err:
self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
exit(1)
except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), err))
else:
self._logger.info("{0} : Parent title : {1}".format(self._name, parent))
parents.append(parent)
for i in range(0, len(menu)-1):
self._logger.info("{0} : Menu : {1} {2}".format(self._name, menu[i], len(parents)))
children.append(self._child(a.get_text(), parent))
self._logger.info(children)
menu["children"] = children
menus.append(menu)
for i in range(0, len(menus)-1):
self._logger.info("{0} : Menu : {1} {2}".format(self._name, menus[i], len(parents)))
for j in range(0, len(parents)-1):
self._logger.info("{0} : Comparaison : {1} {2}".format(self._name, menu[i]["title"], parents[j]))
#self._logger.info("{0} : Comparaison : {1} {2}".format(self._name, menu[i]["title"], parents[j]))
if menu[i]["title"] == parents[j]:
if menus[i]["title"] == parents[j]:
self._logger.info("{0} del : {1}".format(self._name, parents[j]))
del parents[j]