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

add protocol https/http

Browse Source
This commit is contained in:
Valentin CZERYBA
2023-05-25 00:31:34 +02:00
parent f3b04f9459
commit cd2fbd5372
4 changed files with 76 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
lib/WPExport.py
View File

@@ -6,20 +6,22 @@ from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
class WPExport:
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup"):
def __init__(self, name = "Thread-0", url = "", logger = None, parser = "html.parser", directory = "backup", ssl_canalblog=True):
self._url = url
self._logger = logger
self._parser = parser
self._dir = directory
self._name = name
self._protocol = "https"
if ssl_canalblog is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(total=10,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -89,7 +91,7 @@ class WPExport:
for i in page_url:
section = "publications"
o = urlparse(i)
o = o._replace(scheme="https")
o = o._replace(scheme=self._protocol)
i = o.geturl().replace(":///", "://")
if i == "{0}/".format(self._url):
section = "principal"

46
lib/WPImport.py
View File

@@ -8,20 +8,22 @@ from requests.packages.urllib3.util.retry import Retry
class WPimport:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser"):
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, parser="html.parser", ssl_wordpress=True):
self._name = name
self._basic = basic
self._wordpress = wordpress
self._logger = logger
self._parser = parser
self._headers_json = {'Content-Type': 'application/json; charset=utf-8', 'Accept':'application/json'}
self._protocol = "https"
if ssl_wordpress is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(connect=10, read=10, redirect=5,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -134,7 +136,7 @@ class WPimport:
h2 = i.find_all("h2")[0].text
params = {"search":h2, "type":"post"}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/search".format(self._wordpress), auth=self._basic, params=params)
page = self._request.get("{1}://{0}/wp-json/wp/v2/search".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error : {1}".format(self._name, err))
exit(1)
@@ -155,7 +157,7 @@ class WPimport:
name_img = name_img.split("/")[len(name_img.split("/"))-1]
params = {"search": name_img}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, params=params)
page = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
exit(1)
@@ -165,7 +167,7 @@ class WPimport:
id_media = res[0]["id"]
data = {"featured_media": id_media}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"]), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
r = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, result[0]["id"], self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except Exception as err:
self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
exit(1)
@@ -196,7 +198,7 @@ class WPimport:
for i in list_img:
data = {"post": post_id}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"]), auth=self._basic, data=data)
r = self._request.post("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, i["id"], self._protocol), auth=self._basic, data=data)
except Exception as err:
self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err))
exit(1)
@@ -220,10 +222,10 @@ class WPimport:
self._logger.error("{0} : Element {1} is not image".format(self._name,img_name))
is_img = False
if is_img is True:
self._logger.debug("{0} : Search for image {1} with URL {2}".format(self._name, img_name, "http://{0}/wp-json/wp/v2/media".format(self._wordpress)))
self._logger.debug("{0} : Search for image {1} with URL {2}".format(self._name, img_name, "{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol)))
params = { "search": img_name}
try:
r = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, params=params)
r = self._request.get("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error for search media : {1}".format(self._name, err))
exit(1)
@@ -234,7 +236,7 @@ class WPimport:
if len(res) > 0:
params = {"force":1}
try:
r = self._request.delete("http://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"]), auth=self._basic, params=params)
r = self._request.delete("{2}://{0}/wp-json/wp/v2/media/{1}".format(self._wordpress, res[0]["id"], self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
exit(1)
@@ -250,7 +252,7 @@ class WPimport:
img_type = "image/jpeg"
headers={ 'Content-Type': img_type,'Content-Disposition' : 'attachment; filename={0}'.format(img_name)}
try:
r = self._request.post("http://{0}/wp-json/wp/v2/media".format(self._wordpress), auth=self._basic, headers=headers, data=data)
r = self._request.post("{1}://{0}/wp-json/wp/v2/media".format(self._wordpress, self._protocol), auth=self._basic, headers=headers, data=data)
except Exception as err:
self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
exit(1)
@@ -276,7 +278,7 @@ class WPimport:
try:
params = {"post": post, "author_name":i["author"], "date":i["date"]}
page = self._request.get("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, params=params)
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
exit(1)
@@ -285,7 +287,7 @@ class WPimport:
for j in result:
try:
params = {"force":1}
page = self._request.delete("http://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"]), params=params, auth=self._basic)
page = self._request.delete("{2}://{0}/wp-json/wp/v2/comments/{1}".format(self._wordpress, j["id"], self._protocol), params=params, auth=self._basic)
except Exception as err:
self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
exit(1)
@@ -307,7 +309,7 @@ class WPimport:
parent_id = int(i["parent_id"])
params = {"post": post, "author_name":comment[parent_id]["author"], "date":comment[parent_id]["date"]}
try:
page = self._request.get("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, params=params)
page = self._request.get("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, params=params)
except Exception as err:
self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
exit(1)
@@ -320,7 +322,7 @@ class WPimport:
self._logger.debug("{0} : {1}".format(self._name, page.content))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/comments".format(self._wordpress), auth=self._basic, data=data)
page = self._request.post("{1}://{0}/wp-json/wp/v2/comments".format(self._wordpress, self._protocol), auth=self._basic, data=data)
except Exception as err:
self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
exit(1)
@@ -459,7 +461,7 @@ class WPimport:
self._logger.info("{0} : search {1} with index {2} : {3}".format(self._name, i, index, title_element))
try:
params = {"search":title_element, "per_page":"100", "page":index}
page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, params=params)
page = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, params=params)
except ConnectionError as err:
self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
exit(1)
@@ -496,7 +498,7 @@ class WPimport:
self._logger.info("{0} : Create {1} : {2}".format(self._name, i, title_element))
self._logger.debug("{0} : Data : {1}".format(self._name, data))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
page = self._request.post("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except ConnectionError as err:
self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err))
exit(1)
@@ -536,7 +538,7 @@ class WPimport:
params = {"search":author, "per_page":100}
try:
self._logger.info("{0} : Search author : {1}".format(self._name, author))
page = self._request.get("http://{0}/wp-json/wp/v2/users".format(self._wordpress), auth=self._basic, headers=self._headers_json, params=params)
page = self._request.get("{1}://{0}/wp-json/wp/v2/users".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
self._logger.debug("{0} : End Search author : {1}".format(self._name, author))
self._logger.debug("{0} : Debug requests : {1}".format(self._name, page.content))
@@ -560,7 +562,7 @@ class WPimport:
params = {"search": title, "per_page":100, "page": index}
try:
self._logger.info("{0} : Search post with index {2} : {1}".format(self._name, title, index))
page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params, headers=self._headers_json)
page = self._request.get("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
except ConnectionError as err:
self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
exit(1)
@@ -589,7 +591,7 @@ class WPimport:
self._logger.info("{0} : Page {1} is double and going to delete".format(self._name, title))
try:
params = {"force":1}
page = self._request.delete("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=self._headers_json, params=params)
page = self._request.delete("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, params=params)
except ConnectionError as err:
self._logger.error("{0} : Connection error for deleted post : {1}".format(self._name, err))
exit(1)
@@ -606,7 +608,7 @@ class WPimport:
self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
try:
page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
page = self._request.post("{2}://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except ConnectionError as err:
self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
exit(1)
@@ -631,7 +633,7 @@ class WPimport:
if page_is_exist is False:
try:
self._logger.info("{0} : Creating posts : {1}".format(self._name, data["title"]))
page = self._request.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
page = self._request.post("{1}://{0}/wp-json/wp/v2/posts".format(self._wordpress, self._protocol), auth=self._basic, headers=self._headers_json, data=json.dumps(data))
except ConnectionError as err:
self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
exit(1)

13
lib/WPRemove.py
View File

@@ -8,19 +8,22 @@ from requests.packages.urllib3.util.retry import Retry
class WPRemove:
# Constructor
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None):
def __init__(self, name="Thread-0", basic=None, wordpress="", logger=None, ssl_wordpress=True):
self._name = name
self._basic = basic
self._wordpress = wordpress
self._logger = logger
self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}
self._protocol = "https"
if ssl_wordpress is False:
self._protocol = "http"
self._request = requests.Session()
retries = Retry(connect=10, read=10, redirect=5,
status_forcelist=[429, 500, 502, 503, 504], backoff_factor=2)
self._request.mount('http://', HTTPAdapter(max_retries=retries))
self._request.mount('{0}://'.format(self._protocol), HTTPAdapter(max_retries=retries))
# Destructor
@@ -51,7 +54,7 @@ class WPRemove:
try:
self._logger.info("{0} : List {2} to remove for url : {1}".format(self._name, self._wordpress, composant))
r = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant), auth=self._basic, params=params, headers=self._headers_json)
r = self._request.get("{2}://{0}/wp-json/wp/v2/{1}".format(self._wordpress, composant, self._protocol), auth=self._basic, params=params, headers=self._headers_json)
except Exception as err:
self._logger.error("{0} : Connection error for list {1} to remove : {2}".format(self._name, composant, err))
if r.status_code == 200:
@@ -61,7 +64,7 @@ class WPRemove:
self._logger.info("{0} : Remove {2} for url {1} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"]))
params = {"force":1}
try:
r = self._request.delete("http://{0}/wp-json/wp/v2/{1}/{2}".format(self._wordpress, composant, i["id"]), auth=self._basic, headers=self._headers_json , params=params)
r = self._request.delete("{3}://{0}/wp-json/wp/v2/{1}/{2}".format(self._wordpress, composant, i["id"], self._protocol), auth=self._basic, headers=self._headers_json , params=params)
if r.status_code == 200:
self._logger.info("{0} : Post removed for URL {1} {2} : {3}".format(self._name, self._wordpress, composant, i["title"]["rendered"]))
else: