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

fix get id menu

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-07-13 22:35:38 +02:00
parent be9ad9b934
commit 0445054dc8
1 changed files with 16 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
lib/WPMenu.py
View File

@ -82,13 +82,15 @@ class WPMenu:
if exist is False: if exist is False:
try: try:
params = {"search":title, "per_page":"100", "page":index} params = {"search":title, "per_page":"100", "page":index}
self._logger.info("{0} Get menu {1} {2} {3}".format(self._name, "{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), index, title))
page = self._request_wordpress.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol_wordpress), auth=self._basic, params=params) 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: if page.status_code == 200:
result = page.json() result = page.json()
if len(result) > 0: if len(result) > 0:
for j in result: for j in result:
self._logger.info("{0} info : {1} {2}".format(self._name, j["name"], title)) self._logger.info("{0} info : {1} {2} {3}".format(self._name, j["name"], j["slug"], title))
if j["name"] == title: if j["name"] == title or j["slug"] == title:
self._logger.info("{0} : comparaison ok : {1} {2}".format(self._name, j["id"], i))
idMenu = j["id"] idMenu = j["id"]
exist = True exist = True
@ -114,13 +116,14 @@ class WPMenu:
href = anchor.find("a").get("href") href = anchor.find("a").get("href")
if href == "{0}://{1}/".format(self._protocol_canalblog, self._canalblog): if href == "{0}://{1}/".format(self._protocol_canalblog, self._canalblog):
parent = "home" parent = "home"
itemMenu = {"id":"", "type":"", "title": parent, "children":list()} itemMenu = {"id":"", "type":"", "title": parent, "href":href, "children":list()}
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")
href = a.get("href")
self._logger.info("{0} Parent {1} : Child {2}".format(self._name, parent, a.get_text())) self._logger.info("{0} Parent {1} : Child {2}".format(self._name, parent, a.get_text()))
children.append({"title": a.get_text(), "parent": parent}) children.append({"title": a.get_text(), "parent": parent, "href":href})
menu.append(itemMenu) menu.append(itemMenu)
@ -142,6 +145,15 @@ class WPMenu:
if menu[i]["title"] != "home": if menu[i]["title"] != "home":
for j in range(0, len(menu[i]["children"])): for j in range(0, len(menu[i]["children"])):
idMenu = self._getId(menu[i]["children"][j]["title"]) idMenu = self._getId(menu[i]["children"][j]["title"])
if idMenu == 0:
title = href[::-1]
link = title.split("/")[0]
link = link[::-1]
title = title.split("/")[1]
title = title[::-1]
self._logger.info("{0} link {1} title {2}".format(self._name, link, title))
if link == "index.html":
idMenu = self._getId(title)
if idMenu != 0: if idMenu != 0:
menu[i]["children"][j] = {"id":idMenu, "title": menu[i]["children"][j]["title"], "parent": menu[i]["children"][j]["parent"]} menu[i]["children"][j] = {"id":idMenu, "title": menu[i]["children"][j]["title"], "parent": menu[i]["children"][j]["parent"]}
idMenu = self._getId(menu[i]["title"]) idMenu = self._getId(menu[i]["title"])