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

list menu wip

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-07-09 19:11:05 +02:00
parent 03f833a2c3
commit 74fa87ea73
1 changed files with 20 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
lib/WPMenu.py
View File

@ -73,7 +73,7 @@ class WPMenu:
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, canalblog, err)) self._logger.error("{0} : Exception error for get url {1} : {2}".format(self._name, canalblog, err))
def _child(self, title, parent): def _getId(self, title, parent):
menu = {"id":"", "type":"", "title": "", "parent":"", "children": []} menu = {"id":"", "type":"", "title": "", "parent":"", "children": []}
exist = False exist = False
for i in ["categories", "tags"]: for i in ["categories", "tags"]:
@ -100,31 +100,31 @@ class WPMenu:
soup = BeautifulSoup(content, self._parser) soup = BeautifulSoup(content, self._parser)
ul = soup.find("ul", id="listsmooth") ul = soup.find("ul", id="listsmooth")
menus = list() menu = list()
menu = {"id":"", "type":"", "title": "", "parent":"", "children": []}
parents = list() parents = list()
children = list()
for anchor in ul.find_all("li"): for anchor in ul.find_all("li"):
parent = anchor.find("a").get_text()
parent = anchor.find("a").get_text().replace(" \xa0", "")
itemMenu = {"id":"", "type":"", "title": parent, "parent": parent, "children":[]}
itemChild = {"id":"", "type":"", "title": parent, "parent": parent}
href = anchor.find("a").get("href") href = anchor.find("a").get("href")
menu = self._child(parent, parent) #menu = self._child(parent, parent)
if href == "#": if href == "#":
li = anchor.find("ul").find_all("li") li = anchor.find("ul").find_all("li")
for child in li: for child in li:
a = child.find("a") a = child.find("a")
self._logger.info("{0} {1} : {2}".format(self._name, anchor.find("a").get_text(), a.get_text())) self._logger.info("{0} Parent {1} : Child {2}".format(self._name, parent, a.get_text()))
children.append(self._child(a.get_text(), parent)) itemChild["title"] = a.get_text()
self._logger.info(children) itemChild["parent"] = parent
menu["children"] = children itemMenu["children"].append(itemChild)
menus.append(menu) menu.append(itemMenu)
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]))
if menus[i]["title"] == parents[j]:
self._logger.info("{0} del : {1}".format(self._name, parents[j]))
del parents[j]
for i in parents: for i in range(0, len(menu)-1):
self._logger.info("{0} : Parent : {1}".format(self._name, i)) self._logger.info("{0} Menu : {1}".format(self._name, menu[i]))
#for j in menu[i]["children"]:
# if menu[i]["title"] == j["title"]:
# del menu[i]