#!/usr/bin/env  python
# License: GPLv3 Copyright: 2008, Kovid Goyal <kovid at kovidgoyal.net>

try:
    from http.cookiejar import Cookie
except ImportError:
    from cookielib import Cookie

import json
from html5_parser import parse
from lxml import etree
from collections import defaultdict

from calibre import replace_entities
from calibre.ebooks.BeautifulSoup import NavigableString, Tag
from calibre.utils.date import parse_only_date
from calibre.web.feeds.news import BasicNewsRecipe


def E(parent, name, text='', **attrs):
    ans = parent.makeelement(name, **attrs)
    ans.text = text
    parent.append(ans)
    return ans


def process_node(node, html_parent):
    ntype = node.get('type')
    if ntype == 'tag':
        c = html_parent.makeelement(node['name'])
        c.attrib.update({k: v or '' for k, v in node.get('attribs', {}).items()})
        html_parent.append(c)
        for nc in node.get('children', ()):
            process_node(nc, c)
    elif ntype == 'text':
        text = node.get('data')
        if text:
            text = replace_entities(text)
            if len(html_parent):
                t = html_parent[-1]
                t.tail = (t.tail or '') + text
            else:
                html_parent.text = (html_parent.text or '') + text


def safe_dict(data, *names):
    ans = data
    for x in names:
        ans = ans.get(x) or {}
    return ans


class JSONHasNoContent(ValueError):
    pass


def load_article_from_json(raw, root):
    # open('/t/raw.json', 'w').write(raw)
    try:
        data = json.loads(raw)['props']['pageProps']['content']
    except KeyError as e:
        raise JSONHasNoContent(e)
    if isinstance(data, list):
        data = data[0]
    body = root.xpath('//body')[0]
    for child in tuple(body):
        body.remove(child)
    article = E(body, 'article')
    E(article, 'h4', data['subheadline'], style='color: red; margin: 0')
    E(article, 'h1', data['headline'], style='font-size: x-large')
    E(article, 'div', data['description'], style='font-style: italic')
    E(article, 'div', (data['datePublishedString'] or '') + ' | ' + (data['dateline'] or ''), style='color: gray; margin: 1em')
    main_image_url = safe_dict(data, 'image', 'main', 'url').get('canonical')
    if main_image_url:
        div = E(article, 'div')
        try:
            E(div, 'img', src=main_image_url)
        except Exception:
            pass
    for node in data.get('text') or ():
        process_node(node, article)


def cleanup_html_article(root):
    main = root.xpath('//main')[0]
    body = root.xpath('//body')[0]
    for child in tuple(body):
        body.remove(child)
    body.append(main)
    main.set('id', '')
    main.tag = 'article'
    for x in root.xpath('//*[@style]'):
        x.set('style', '')
    for x in root.xpath('//button'):
        x.getparent().remove(x)


def classes(classes):
    q = frozenset(classes.split(' '))
    return dict(attrs={
        'class': lambda x: x and frozenset(x.split()).intersection(q)})


def new_tag(soup, name, attrs=()):
    impl = getattr(soup, 'new_tag', None)
    if impl is not None:
        return impl(name, attrs=dict(attrs))
    return Tag(soup, name, attrs=attrs or None)


class NoArticles(Exception):
    pass


def process_url(url):
    if url.startswith('/'):
        url = 'https://www.economist.com' + url
    return url


class Economist(BasicNewsRecipe):

    title = 'The Economist - Search'
    language = 'en'
    encoding = 'utf-8'
    __author__ = "Kovid Goyal"
    description = (
        'In the Account section below, type your search term in the Username field.'
    )
    extra_css = '''
        .calibre_navbar {opacity: 0 !important;}
        .headline {font-size: x-large;}
        h2 { font-size: small;  }
        h1 { font-size: medium;  }
        em.Bold {font-weight:bold;font-style:normal;}
        em.Italic {font-style:italic;}
        p.xhead {font-weight:bold;}
        .pullquote {
            float: right;
            font-size: larger;
            font-weight: bold;
            font-style: italic;
            page-break-inside:avoid;
            border-bottom: 3px solid black;
            border-top: 3px solid black;
            width: 228px;
            margin: 0px 0px 10px 15px;
            padding: 7px 0px 9px;
        }
        .flytitle-and-title__flytitle {
            display: block;
            font-size: smaller;
            color: red;
        }
        '''
    
    resolve_internal_links = True
    remove_tags = [
        dict(name=['script', 'noscript', 'title', 'iframe', 'cf_floatingcontent', 'aside', 'footer']),
        dict(attrs={'aria-label': "Article Teaser"}),
        dict(attrs={
                'class': [
                    'dblClkTrk', 'ec-article-info', 'share_inline_header',
                    'related-items', 'main-content-container', 'ec-topic-widget',
                    'teaser', 'blog-post__bottom-panel-bottom', 'blog-post__comments-label',
                    'blog-post__foot-note', 'blog-post__sharebar', 'blog-post__bottom-panel',
                    'newsletter-form','share-links-header','teaser--wrapped', 'latest-updates-panel__container',
                    'latest-updates-panel__article-link','blog-post__section'
                ]
            }
        ),
        dict(attrs={
                'class': lambda x: x and 'blog-post__siblings-list-aside' in x.split()}),
        classes(
            'share-links-header teaser--wrapped latest-updates-panel__container'
            ' latest-updates-panel__article-link blog-post__section newsletter-form blog-post__bottom-panel'
        )
    ]
    keep_only_tags = [dict(name='article', id=lambda x: not x)]
    no_stylesheets = True
    remove_attributes = ['data-reactid', 'width', 'height']
    # economist.com has started throttling after about 60% of the total has
    # downloaded with connection reset by peer (104) errors.
    delay = 1

    needs_subscription = False

    def __init__(self, *args, **kwargs):
        BasicNewsRecipe.__init__(self, *args, **kwargs)
        if self.output_profile.short_name.startswith('kindle'):
            # Reduce image sizes to get file size below amazon's email
            # sending threshold
            self.web2disk_options.compress_news_images = True
            self.web2disk_options.compress_news_images_auto_size = 5
            self.log.warn('Kindle Output profile being used, reducing image quality to keep file size below amazon email threshold')

    def get_browser(self):
        br = BasicNewsRecipe.get_browser(self)
        # Add a cookie indicating we have accepted Economist's cookie
        # policy (needed when running from some European countries)
        ck = Cookie(
            version=0,
            name='notice_preferences',
            value='2:',
            port=None,
            port_specified=False,
            domain='.economist.com',
            domain_specified=False,
            domain_initial_dot=True,
            path='/',
            path_specified=False,
            secure=False,
            expires=None,
            discard=False,
            comment=None,
            comment_url=None,
            rest={'HttpOnly': None},
            rfc2109=False
        )
        br.cookiejar.set_cookie(ck)
        br.set_handle_gzip(True)
        return br

    def preprocess_raw_html(self, raw, url):
        # open('/t/raw.html', 'wb').write(raw.encode('utf-8'))
        root = parse(raw)
        script = root.xpath('//script[@id="__NEXT_DATA__"]')
        if script:
            try:
                load_article_from_json(script[0].text, root)
            except JSONHasNoContent:
                cleanup_html_article(root)
        for div in root.xpath('//div[@class="lazy-image"]'):
            noscript = list(div.iter('noscript'))
            if noscript and noscript[0].text:
                img = list(parse(noscript[0].text).iter('img'))
                if img:
                    p = noscript[0].getparent()
                    idx = p.index(noscript[0])
                    p.insert(idx, p.makeelement('img', src=img[0].get('src')))
                    p.remove(noscript[0])
        for x in root.xpath('//*[name()="script" or name()="style" or name()="source" or name()="meta"]'):
            x.getparent().remove(x)
        # the economist uses <small> for small caps with a custom font
        for x in root.xpath('//small'):
            if x.text and len(x) == 0:
                x.text = x.text.upper()
                x.tag = 'span'
                x.set('style', 'font-variant: small-caps')
        raw = etree.tostring(root, encoding='unicode')
        return raw
    
    needs_subscription = True
    
    def parse_index(self):
        url = 'https://www.economist.com/search?q={query}&sort={sort}&page={page}'
        if self.username is not None:
            search = self.username.replace(' ', '+')
            sort_type = 'relevance' # or 'date'
            self.title = 'The Economist - ' + search
            ans = []
            for num in range(1, 4): # increase range if you want more than 3 pages 
                ans.extend(self.economist_parse_index(url.format(query=search, sort=sort_type, page=num)))
                
            return [('Articles', ans)]

    def economist_parse_index(self, url):
        self.log(url)
        soup = self.index_to_soup(url) 
        results = soup.find('ol', id='search-results')
        for a in results.findAll('a', attrs={'class':'_search-result'}):
            url = a['href']
            title = self.tag_to_string(a.find(attrs={'class':'_headline'}))
            desc = self.tag_to_string(a.find(attrs={'class':'_snippet'}))
            self.log('\t', title, '\n\t', desc, '\n\t\t', url)
            yield {'title': title, 'url': url, 'description': desc}

    def eco_find_image_tables(self, soup):
        for x in soup.findAll('table', align=['right', 'center']):
            if len(x.findAll('font')) in (1, 2) and len(x.findAll('img')) == 1:
                yield x

    def postprocess_html(self, soup, first):
        for img in soup.findAll('img', srcset=True):
            del img['srcset']
        for table in list(self.eco_find_image_tables(soup)):
            caption = table.find('font')
            img = table.find('img')
            div = new_tag(soup, 'div')
            div['style'] = 'text-align:left;font-size:70%'
            ns = NavigableString(self.tag_to_string(caption))
            div.insert(0, ns)
            div.insert(1, new_tag(soup, 'br'))
            del img['width']
            del img['height']
            img.extract()
            div.insert(2, img)
            table.replaceWith(div)
        return soup

    def canonicalize_internal_url(self, url, is_link=True):
        if url.endswith('/print'):
            url = url.rpartition('/')[0]
        return BasicNewsRecipe.canonicalize_internal_url(self, url, is_link=is_link)


calibre_most_common_ua = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.87 Safari/537.36'