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

fix bug import

Browse Source
This commit is contained in:
Valentin CZERYBA 2023-09-11 22:40:41 +02:00
parent eaec1ba9d4
commit 41e192f903
1 changed files with 8 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
lib/WPImport.py
View File

@ -302,7 +302,7 @@ class WPimport:
exit(1) exit(1)
except Exception as err: except Exception as err:
self._logger.error("{0} : Exception error for update page : {1}".format(self._name, err)) self._logger.error("{0} : Exception error for update page : {1}".format(self._name, err))
if page.status_code == 400: elif page.status_code == 400:
self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code)) self._logger.error("{0} : Connection for update post unauthorized : {1}".format(self._name, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content)) self._logger.debug("{0} : {1}".format(self._name, page.content))
break break
@ -817,7 +817,7 @@ class WPimport:
listelement[i].append(k["id"]) listelement[i].append(k["id"])
else: else:
break break
if page.status_code == 400: elif page.status_code == 400:
self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code)) self._logger.error("{0} : {1} not found due status code : {2}".format(self._name, i, page.status_code))
self._logger.debug("{0} : {1}".format(self._name, page.content)) self._logger.debug("{0} : {1}".format(self._name, page.content))
break break
@ -867,12 +867,15 @@ class WPimport:
bodyhtml = bodyhtml.replace(i["old_href"], o.path) bodyhtml = bodyhtml.replace(i["old_href"], o.path)
bodyhtml = bodyhtml.replace(i["old_src"], o.path) bodyhtml = bodyhtml.replace(i["old_src"], o.path)
hour = articledate[0].text hour = articledate[0].text
time = dateheader[0].text.split(" ") time = ""
if len(dateheader) > 0:
time = dateheader[0].text.split(" ")
self._logger.debug("{0} : Title post : |{1}|".format(self._name, title)) self._logger.debug("{0} : Title post : |{1}|".format(self._name, title))
title = self._removeSpace(title) title = self._removeSpace(title)
self._logger.debug("{0} : Rendered Title post : |{1}|".format(self._name, title)) self._logger.debug("{0} : Rendered Title post : |{1}|".format(self._name, title))
data = {"title":title, "content":bodyhtml, "status":"publish", "tags": listelement["tags"], "categories": listelement["categories"]}
data = {"title":title, "content":bodyhtml, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]} if len(dateheader) > 0:
data = {"title":title, "content":bodyhtml, "status":"publish", "date": "{0}-{1}-{2}T{3}:00".format(time[2],month[time[1]],time[0], hour), "tags": listelement["tags"], "categories": listelement["categories"]}
self._logger.debug("{0} : Data for post : |{1}| : {2}" .format(self._name, title, data)) self._logger.debug("{0} : Data for post : |{1}| : {2}" .format(self._name, title, data))
data["author"] = self._getAuthor(author) data["author"] = self._getAuthor(author)