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

add href

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-07-13 23:53:54 +02:00
parent 0445054dc8
commit 5fe4b1f786
1 changed files with 25 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
lib/WPMenu.py
View File

@ -74,6 +74,24 @@ 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 _getIdFromPost(self, href):
idMenu = 0
return idMenu
def _getIdFromReverse(self, title, href):
self._logger.info("{0} : get title {1} from href {2}".format(self._name, title, href))
idMenu = 0
if href != "#":
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)
return idMenu
def _getId(self, title): def _getId(self, title):
idMenu = 0 idMenu = 0
exist = False exist = False
@ -138,7 +156,7 @@ class WPMenu:
for j in range(0, len(menu)): for j in range(0, len(menu)):
self._logger.info("{0} : Children for : {1}".format(self._name, menu[j]["title"])) self._logger.info("{0} : Children for : {1}".format(self._name, menu[j]["title"]))
if menu[j]["title"] == children[i]["parent"]: if menu[j]["title"] == children[i]["parent"]:
menu[j]["children"].append({"id":"", "title":children[i]["title"], "parent": children[i]["parent"]}) menu[j]["children"].append({"id":"", "title":children[i]["title"], "parent": children[i]["parent"], "href":children[i]["href"]})
for i in range(0, len(menu)): for i in range(0, len(menu)):
self._logger.info("{0} : Menu {1} {2}".format(self._name, menu[i]["title"], len(menu[i]["children"]))) self._logger.info("{0} : Menu {1} {2}".format(self._name, menu[i]["title"], len(menu[i]["children"])))
@ -146,17 +164,14 @@ class WPMenu:
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: if idMenu == 0:
title = href[::-1] self._logger.debug("{0} : content children {1}".format(self._name, menu[i]["children"][j]))
link = title.split("/")[0] idMenu = self._getIdFromReverse(menu[i]["children"][j]["title"], menu[i]["children"][j]["href"])
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"])
self._logger.debug("{0} : content parent {1}".format(self._name, menu[i]))
if idMenu == 0:
idMenu = self._getIdFromReverse(menu[i]["title"], menu[i]["href"])
if idMenu != 0: if idMenu != 0:
menu[i] = {"id":idMenu, "title":menu[i]["title"], "children": menu[i]["children"]} menu[i] = {"id":idMenu, "title":menu[i]["title"], "children": menu[i]["children"]}