#!/usr/bin/python3

from bs4 import BeautifulSoup
from urllib.parse import urlparse
import requests, os, logging, re, json
from requests.adapters import HTTPAdapter
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"):
        self._name = name
        self._basic = basic
        self._wordpress = wordpress
        self._logger = logger
        self._parser = parser
        self._headers_json = {'Content-Type': 'application/json', 'Accept':'application/json'}

        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))

    # Public method

    def setUrl(self, wordpress):
        self._wordpress = wordpress

    def fromUrl(self, webpage):
        for i in range(0, len(webpage)):
            try:
                r = self._request.get(webpage[i])
            except Exception as err:
                self._logger.error("{0} : Connection error for get url {1} : {2}".format(self._name, webpage[i], err))
                exit(1)
            if r.status_code == 200:
                self._logger.info("{0} : ({1}/{2} : Page is importing : {3}".format(self._name, i+1, len(webpage), webpage[i]))
                soup = BeautifulSoup(r.content, self._parser)
                articlebody = soup.find_all("div", class_="articlebody")
                if len(articlebody) > 0:
                    self._addOrUpdatePost(soup)
                else:
                    self._addOrUpdateFeaturedMedia(soup)
            else:
                self._logger.error("{0} : Connection error for get url {1} with status code : {2}".format(self._name, webpage[i], r.status_code))
                self._logger.debug("{0} : {1}".format(self._name, r.content))


    def fromDirectory(self, directory="", number_thread=1, max_thread=1):
        directory = "{0}/archives".format(directory)
        directories = self._getDirectories([], "{0}".format(directory))
        if len(directories) > 0:
            files = self._getFiles(directories)
            self.fromFile(files, number_thread, max_thread)
        else:
            self._logger.error("{0} : No files for {1}".format(self._name, directory))
        

    def fromFile(self, files=[], number_thread=1, max_thread=1):
        divFiles = int(len(files) / max_thread)
        currentRangeFiles = int(divFiles * (number_thread+1))
        firstRange = int(currentRangeFiles - divFiles)
        self._logger.debug("{0} : index : {1}".format(self._name,number_thread))

        self._logger.debug("{0} : first range : {1}".format(self._name,firstRange))
        self._logger.debug("{0} : last range : {1}".format(self._name,currentRangeFiles))

        for i in range(firstRange, currentRangeFiles):
            if os.path.exists(files[i]):
                self._logger.info("{0} : File is being processed : {1}".format(self._name, files[i]))
                with open(files[i], 'r') as f:
                    content = f.read()
                    soup = BeautifulSoup(content, self._parser)
                    articlebody = soup.find_all("div", class_="articlebody")
                    self._logger.debug("{0} : Number of article : {1}".format(self._name, len(articlebody)))
                    if len(articlebody) > 0:
                        self._addOrUpdatePost(soup)
                    else:
                        self._addOrUpdateFeaturedMedia(soup)
    
    # Private method

    ## Get all files

    def _getFiles(self, item):
        files = []
        for i in item:
            for j in os.listdir(i):
                if os.path.isfile("{0}/{1}".format(i, j)):
                    files.append("{0}/{1}".format(i, j))
        return files

    ## Get directories

    def _getDirectories(self, subdirectory, item):
        sub = subdirectory
        for i in os.listdir(item):
            if os.path.isdir("{0}/{1}".format(item, i)):
                sub.append("{0}/{1}".format(item, i))
                subdirectory = self._getDirectories(sub, "{0}/{1}".format(item, i))
        return subdirectory

    ## Add or update featured media

    def _addOrUpdateFeaturedMedia(self, soup):
        item_div = soup.find_all("div", {"data-edittype": "post"})
        for i in item_div:
            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)
            except Exception as err:
                self._logger.error("{0} : Connection error : {1}".format(self._name, err))
                exit(1)
            if page.status_code == 200:
                result = page.json()
                if len(result) > 0:
                    if h2 == result[0]["title"]:
                        img = i.find_all("img")
                        if len(img) > 0:
                            img_src = img[0].get("src")
                            try:
                                page = self._request.get(img_src)
                            except Exception as err:
                                self._logger.error("{0} : Connection error for get featured media : {1}".format(self._name, err))
                                exit(1)
                            if page.status_code == 200:
                                name_img = img_src.replace("_q", "")
                                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)
                                except Exception as err:
                                    self._logger.error("{0} : Connection error search featured media : {1}".format(self._name, err))
                                    exit(1)
                                if page.status_code == 200:
                                    res = page.json()
                                    if len(res) > 0:
                                        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))
                                        except Exception as err:
                                            self._logger.error("{0} : Connection error for post media featured : {1}".format(self._name, err))
                                            exit(1)
                                        if r.status_code == 200:
                                            self._logger.info("{0} : Add media featured : {1}".format(self._name, r.json()["title"]["raw"]))
                                        else:
                                            self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, r.status_code))
                                            self._logger.debug("{0} : {1}".format(self._name, r.content))

                                    else:
                                        self._logger.info("{0} : No media found for {1}".format(self._name, h2))
                                else:
                                    self._logger.error("{0} : Connection error with status code for search featured media: {1}".format(self._name, page.status_code))
                                    self._logger.debug("{0} : {1}".format(self._name, page.content))

                            else:
                                self._logger.error("{0} : Connection error for get featured media with status code : {1}".format(self._name, page.status_code))
                                self._logger.debug("{0} : {1}".format(self._name, page.content))

            else:
                self._logger.error("{0} : Connection error with status code for featured media : {1}".format(self._name, page.status_code))
                self._logger.debug("{0} : {1}".format(self._name, page.content))


    ## Association image to post

    def _linkImgPost(self, title, list_img, post_id):
        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)
            except Exception as err:
                self._logger.error("{0} : Connection error for link image to post : {1}".format(self._name, err))
                exit(1)
            if r.status_code == 200:
                self._logger.info("{0} : Link image to post {1}".format(self._name, title))
            else:
                self._logger.error("{0} Connection error with status code for link image to post : {1}".format(self._name, r.status_code))
                self._logger.debug("{0} : {1}".format(self._name, r.content))

    
    ## Add or update img 

    def _addOrUpdateMedia(self, href_img, page):
        media = {"id":"", "rendered":""}
        split_fileimg = href_img.split("/")
        img_name = split_fileimg[len(split_fileimg)-1]
        params = { "search": img_name}
        try:
            r = self._request.get("http://{0}/wp-json/wp/v2/media".format(self._wordpress), 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)
        if r.status_code == 200:
            res = r.json()
            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)
                except Exception as err:
                    self._logger.error("{0} Connection error for delete image : {1}".format(self._name, err))
                    exit(1)
                if r.status_code == 200:
                    self._logger.info("{0} : Image removed {1}".format(self._name, img_name))
                else:
                    self._logger.error("{0} : Image not removed due status code : {1}".format(self._name, r.status_code))
                    self._logger.debug("{0} : {1}".format(self._name, r.content))

            data = page.content
            img_type = "image/png"
            if img_name.split(".")[1] == "jpg" or img_name.split(".")[1] == "jpeg":
                img_type = "image/jpg"
            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)
            except Exception as err:
                self._logger.error("{0} : Connection error for add image : {1}".format(self._name, err))
                exit(1)
            if r.status_code == 201:
                self._logger.info("{0} : Image added {1}".format(self._name, img_name))
                res = r.json()
                media["id"] = res["id"]
                media["rendered"] = res["guid"]["rendered"]
            else:
                self._logger.error("{0} : Image not added due status code : {1}".format(self._name, r.status_code))
                self._logger.debug(r.content)

        else:
            self._logger.error("{0} : Connection error for search image with status code : {1}".format(self._name, r.status_code))
            self._logger.debug("{0} : {1}".format(self._name, r.content))

        return media

    ## Add or update comment

    def _addOrUpdateComment(self, post, comment, title):
            for i in comment:

                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)
                except Exception as err:
                    self._logger.error("{0} : Connection error for search comment : {1}".format(self._name, err))
                    exit(1)
                if page.status_code == 200:
                    result = page.json()
                    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)
                        except Exception as err:
                            self._logger.error("{0} : Connection error for delete comment : {1}".format(self._name, err))
                            exit(1)
                        if page.status_code == 200:
                            self._logger.info("{0} : Comment deleted for {1}".format(self._name, title))
                            self._logger.debug("{0} : Comment deleted : {1}".format(self._name, j))
                        else:
                            self._logger.error("{0} : Comment not deleted for {1} due status code : {2}".format(self._name, title, page.status_code))
                            self._logger.debug("{0} : {1}".format(self._name, page.content))

                else:
                    self._logger.error("{0} : Comment not listed for {1} due status code : {2}".format(self._name, title, page.status_code))
                    self._logger.debug("{0} : {1}".format(self._name, page.content))
   
            for i in comment:           
                data = {"post": post, "content": i["content"], "date": i["date"], "author_name": i["author"], "status": "approved"}
    
                if i["parent_id"] != -1:
                    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)
                    except Exception as err:
                        self._logger.error("{0} : Connection error for parent comment : {1}".format(self._name, err))
                        exit(1)
                    if page.status_code == 200:
                        result = page.json()
                        if len(result) > 0:
                            data["parent"]=result[0]["id"]
                    else:
                        self._logger.error("{0} : Connection error for parent comment with status code : {1}".format(self._name, page.status_code))
                        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)
                except Exception as err:
                    self._logger.error("{0} : Connection error for add comment : {1}".format(self._name, err))
                    exit(1)
                if page.status_code == 201:
                    self._logger.info("{0} : Comment added for {1}".format(self._name, title))
                    self._logger.debug("{0} : Data : {1}".format(self._name, data))
                else:
                    self._logger.error("{0} : Comment not added for {1} due status code : {2}".format(self._name, title, page.status_code))
                    self._logger.debug("{0} : {1}".format(self._name, page.content))
    
    ## Check class name 

    def _hasClassName(self, tag, className):
        for i in tag["class"]:
            if i == className:
                return True
        return False
    
    ## Get class name 

    def _getClassName(self, tag, className):
        for i in tag["class"]:
            if re.match(className, i):
                return i
        return ""
    
    ## Get all comments

    def _getComment(self, comment):
        comment_post = []
        for i in range(0, len(comment)):
            comment_div = comment[i].find("div", class_="comment_item")
            comment_item = comment_div.text.split("\n")
            footer = comment_div.find_all("div", class_="itemfooter")
            comment_author = footer[0].text.split(",")[0].replace("Posté par ", "")
            comment_date = footer[0].find_all("abbr")[0].get("title")
            comment_content = "<p>"
            for j in range(0, len(comment_item)-2):
                if len(comment_item[j]) > 0:
                    comment_content = comment_content + comment_item[j] + "<br />"
            comment_content = comment_content + "</p>"
            parent = -1
            if self._hasClassName(comment[i], "level-1") is False:
                block = False
                className = self._getClassName(comment[i], "level-").split("-")
                level = 1
                if len(className) > 0:
                    level = int(className[1])
                for j in range(i-1, 0, -1):
                    if block is False:
                        levelName = "level-{0}".format(level - 1)
                        if self._hasClassName(comment[j], levelName) is True:
                            parent = j
                            block = True

            comment_post.append({"author": comment_author, "date": comment_date, "content": comment_content, "parent_id":parent})
        return comment_post
    ## Add or Update post 

    def _addOrUpdatePost(self, soup):
        tags = []
        month = {"janvier":"01", "février": "02", "mars": "03", "avril":"04", "mai": "05", "juin": "06", "juillet": "07", "août": "08", "septembre": "09", "octobre": "10", "novembre": "11", "décembre": "12"}
        liste = ["categories", "tags"]
        elements = {}
        element = {}
        listelement = {}

        for i in liste:
            element[i] = []
            listelement[i] = []

        articletitle = soup.find_all("h2", class_="articletitle")
        articlebody = soup.find_all("div", class_="articlebody")
        articledate = soup.find_all("span", class_="articledate")
        articleacreator = soup.find_all("span", class_="articlecreator")
        dateheader = soup.find_all("div", class_="dateheader")
        itemfooter = soup.find_all("div", class_="itemfooter")
        comment = soup.find_all("li", class_="comment")
        img_a = articlebody[0].find_all("a", {"target": "_blank"})
        list_img = []
        for i in img_a:
            new_img = {}
            img = i.find_all("img")
            if len(img) > 0:
                href_a = i.get("href")
                href_img = img[0].get("src")
                new_img["old_src"]=href_img
                new_img["old_href"]=href_a
                try:
                    page_img = self._request.get(href_img)
                except Exception as err:
                    self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
                    exit(1)
                if page_img.status_code == 404:
                    href_img = href_a
                    try:
                        page_img = self._request.get(href_a)
                    except Exception as err:
                        self._logger.error("{0} : Connection error for get image : {1}".format(self._name, err))
                        exit(1)
                
                if page_img.status_code == 200:
                    media=self._addOrUpdateMedia(href_img, page_img)
                    new_img["id"]=media["id"]
                    new_img["new_src"]=media["rendered"]
                    list_img.append(new_img)
                    if href_img != href_a:
                        media=self._addOrUpdateMedia(href_a, page_img)
                        new_img["id"]=media["id"]
                        new_img["new_src"]=media["rendered"]
                        list_img.append(new_img)
                if page_img.status_code not in [200, 404]:
                    self._logger.error("{0} : Connection error with status code for get image : {1}".format(self._name, page_img.status_code))
                    self._logger.debug("{0} : {1}".format(self._name, page_img.content))

                
        comment_post = self._getComment(comment)

        a = itemfooter[0].find_all("a", {"rel": True})
        for i in a:
            rel = i.get("rel")
            if rel[0] == 'tag':
                href = i.get("href")
                if re.search(r'/tag/', href):
                    element["tags"].append(i.text)
                if re.search(r'/archives/', href):
                    element["categories"].append(i.text)
        for i in liste:
            for j in element[i]:
                element_exist = False
                try:
                    params = {"params":j}
                    page = self._request.get("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, params=params)
                except Exception as err:
                    self._logger.error("{0} : Connection error for {1} : {2}".format(self._name, i, err))
                    exit(1)
                if page.status_code == 200:
                    element_exist = True
                    result = page.json()
                    listelement[i].append(result[0]["id"])

                else:
                    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))


                if element_exist is False:
                    data = {"name": j}
                    self._logger.debug("{0} : URL : {1} ".format("http://{1}/wp-json/wp/v2/{2}".format(self._name, self._wordpress, i)))
                    self._logger.debug("{0} : data : {1}".format(self._name, data))
                    self._logger.debug("{0} : headers : {1}".format(self._name, self._headers_form))
                    try:
                        page = self._request.post("http://{0}/wp-json/wp/v2/{1}".format(self._wordpress, i), auth=self._basic, headers=self._headers_json, data=data)
                    except Exception as err:
                        self._logger.error("{0} : Connection error for post {1} : {2}".format(self._name, i, err))
                        exit(1)
                    if page.status_code == 201:
                        result = page.json()
                        listelement[i].append(result["id"])
                    else:
                        self._logger.error("{0} : {1} not added due status code : {2}".format(self._name, i, page.status_code))
                        self._logger.debug("{0} : {1}".format(self._name, page.content))

        title = articletitle[0].text
        author = articleacreator[0].text.lower()        
        body = articlebody[0].find_all("p")
        bodyhtml = "<p>"
        for i in body:
            if len(i.text) == 1:
                bodyhtml = bodyhtml + "<br />"
            else:
                bodyhtml = bodyhtml + str(i).replace("<p>", "").replace("</p>", "").replace("<br>", "<br />") + "<br />"
        bodyhtml = bodyhtml + "</p>"
        for i in list_img:
            o = urlparse(i["new_src"])
            bodyhtml = bodyhtml.replace(i["old_href"], o.path)
            bodyhtml = bodyhtml.replace(i["old_src"], o.path)
        hour = articledate[0].text
        time = dateheader[0].text.split(" ")
        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"]}
        params = {"search":author}
        try:
            page = self._request.get("http://{0}/wp-json/wp/v2/users".format(self._wordpress), auth=self._basic, params=params)
        except Exception as err:
            self._logger.error("{0} : Connection error for get author : {1}".format(self._name, err))
            exit(1)
        if page.status_code == 200:
            result = page.json()
            data["author"] = result[0]["id"]
        else:
            self._logger.error("{0} : Connection error with status code for get author : {1}".format(self._name, page.status_code))
            self._logger.debug("{0} : {1}".format(page.content))


        params = {"search":title}
        try:
            page = self._request.get("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, params=params)
        except Exception as err:
            self._logger.error("{0} : Connection error for search post : {1}".format(self._name, err))
            exit(1)
        page_exist = True
        headers = {'Content-Type': 'application/json', 'Accept':'application/json'}
        if page.status_code == 200:
            result = page.json()
            if len(result) == 0:
                page_exist = False
            else:
                self._logger.info("{0} : Page {1} already exist and going to update".format(self._name, title))
                post_id = result[0]["id"]
                try:
                    page = self._request.post("http://{0}/wp-json/wp/v2/posts/{1}".format(self._wordpress, post_id), auth=self._basic, headers=headers, data=json.dumps(data))
                except Exception as err:
                    self._logger.error("{0} : Connection error for update post : {1}".format(self._name, err))
                    exit(1)
                if page.status_code == 200:
                    result = page.json()
                    self._logger.info("{0} : Post updated : {1}".format(self._name, result["title"]["raw"]))
                    self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
                    self._linkImgPost(result["title"]["raw"], list_img, result["id"])
                else:
                    self._logger.error("{0} : Post not updated due status code : {1}".format(self._name, page.status_code))
                    self._logger.debug("{0} : {1}".format(self._name, page.content))

        else:
            self._logger.error("{0} : Connection for update post error with status code : {1}".format(self._name, page.status_code))
            self._logger.debug("{0} : {1}".format(self._name, page.content))

                    
        if page_exist == False:
            try:
                page = self._request.post("http://{0}/wp-json/wp/v2/posts".format(self._wordpress), auth=self._basic, headers=headers, data=json.dumps(data))
            except Exception as err:
                self._logger.error("{0} : Connection error for create post : {1}".format(self._name, err))
                exit(1)
            if page.status_code == 201:
                result = page.json()
                self._logger.info("{0} : Post added : {1}".format(self._name, result["title"]["raw"]))
                self._addOrUpdateComment(result["id"], comment_post, result["title"]["raw"])
                self._linkImgPost(result["title"]["raw"], list_img, result["id"])
            else:
                self._logger.error("{0} : Post not added due status code : {1}".format(self._name, r.status_code))
                self._logger.debug("{0} : {1}".format(self._name, r.content))