epicyon/content.py

__filename__ = "content.py"
__author__ = "Bob Mottram"
__license__ = "AGPL3+"
__version__ = "1.5.0"
__maintainer__ = "Bob Mottram"
__email__ = "bob@libreserver.org"
__status__ = "Production"
__module_group__ = "Core"

import difflib
import math
import html
import os
import email.parser
import urllib.parse
from shutil import copyfile
from dateutil.parser import parse
from utils import remove_link_tracking
from utils import string_contains
from utils import string_ends_with
from utils import is_account_dir
from utils import get_url_from_post
from utils import is_right_to_left_text
from utils import language_right_to_left
from utils import binary_is_image
from utils import get_content_from_post
from utils import get_full_domain
from utils import get_user_paths
from utils import convert_published_to_local_timezone
from utils import has_object_dict
from utils import valid_hash_tag
from utils import dangerous_svg
from utils import remove_domain_port
from utils import get_image_extensions
from utils import load_json
from utils import save_json
from utils import file_last_modified
from utils import get_link_prefixes
from utils import dangerous_markup
from utils import is_pgp_encrypted
from utils import contains_pgp_public_key
from utils import acct_dir
from utils import is_float
from utils import get_currencies
from utils import remove_html
from utils import remove_eol
from petnames import get_pet_name
from session import download_image

MUSIC_SITES = ('soundcloud.com', 'bandcamp.com', 'resonate.coop')

MAX_LINK_LENGTH = 40

REMOVE_MARKUP = (
    'b', 'i', 'ul', 'ol', 'li', 'em', 'strong',
    'blockquote', 'h1', 'h2', 'h3', 'h4', 'h5'
)

INVALID_CONTENT_STRINGS = (
    'mute', 'unmute', 'editeventpost', 'notifypost',
    'delete', 'options', 'page', 'repeat',
    'bm', 'tl', 'actor', 'unrepeat', 'eventid',
    'unannounce', 'like', 'unlike', 'bookmark',
    'unbookmark', 'likedBy', 'time',
    'year', 'month', 'day', 'editnewpost',
    'graph', 'showshare', 'category', 'showwanted',
    'rmshare', 'rmwanted', 'repeatprivate',
    'unrepeatprivate', 'replyto',
    'replyfollowers', 'replydm', 'replychat', 'editblogpost',
    'handle', 'blockdomain'
)


def valid_url_lengths(content: str, max_url_length: int) -> bool:
    """Returns true if the given content contains urls which are too long
    """
    if '://' not in content:
        return True
    sections = content.split('://')
    ctr = 0
    for text in sections:
        if ctr == 0:
            ctr += 1
            continue
        if '"' in text:
            url = text.split('"')[0]
            if '<' not in url and '>' not in url:
                if len(url) > max_url_length:
                    return False
    return True


def remove_html_tag(html_str: str, tag: str) -> str:
    """Removes a given tag from a html string
    """
    tag_found = True
    while tag_found:
        match_str = ' ' + tag + '="'
        if match_str not in html_str:
            tag_found = False
            break
        sections = html_str.split(match_str, 1)
        if '"' not in sections[1]:
            tag_found = False
            break
        html_str = sections[0] + sections[1].split('"', 1)[1]
    return html_str


def _remove_quotes_within_quotes(content: str) -> str:
    """Removes any blockquote inside blockquote
    """
    if '<blockquote>' not in content:
        return content
    if '</blockquote>' not in content:
        return content
    ctr = 1
    found = True
    while found:
        prefix = content.split('<blockquote>', ctr)[0] + '<blockquote>'
        quoted_str = content.split('<blockquote>', ctr)[1]
        if '</blockquote>' not in quoted_str:
            found = False
        else:
            end_str = quoted_str.split('</blockquote>')[1]
            quoted_str = quoted_str.split('</blockquote>')[0]
            if '<blockquote>' not in end_str:
                found = False
            if '<blockquote>' in quoted_str:
                quoted_str = quoted_str.replace('<blockquote>', '')
                content = prefix + quoted_str + '</blockquote>' + end_str
        ctr += 1
    return content


def html_replace_email_quote(content: str) -> str:
    """Replaces an email style quote "> Some quote" with html blockquote
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content
    # replace quote paragraph
    if '<p>&quot;' in content:
        if '&quot;</p>' in content:
            if content.count('<p>&quot;') == content.count('&quot;</p>'):
                content = content.replace('<p>&quot;', '<p><blockquote>')
                content = content.replace('&quot;</p>', '</blockquote></p>')
    if '>\u201c' in content:
        if '\u201d<' in content:
            if content.count('>\u201c') == content.count('\u201d<'):
                content = content.replace('>\u201c', '><blockquote>')
                content = content.replace('\u201d<', '</blockquote><')
    # replace email style quote
    if '>&gt; ' not in content:
        return content
    content_str = content.replace('<p>', '')
    content_lines = content_str.split('</p>')
    new_content = ''
    for line_str in content_lines:
        if not line_str:
            continue
        if '>&gt; ' not in line_str:
            if line_str.startswith('&gt; '):
                line_str = line_str.replace('&gt; ', '<blockquote>')
                line_str = line_str.replace('&gt;', '<br>')
                new_content += '<p>' + line_str + '</blockquote></p>'
            else:
                new_content += '<p>' + line_str + '</p>'
        else:
            line_str = line_str.replace('>&gt; ', '><blockquote>')
            if line_str.startswith('&gt;'):
                line_str = line_str.replace('&gt;', '<blockquote>', 1)
            else:
                line_str = line_str.replace('&gt;', '<br>')
            new_content += '<p>' + line_str + '</blockquote></p>'
    return _remove_quotes_within_quotes(new_content)


def html_replace_quote_marks(content: str) -> str:
    """Replaces quotes with html formatting
    "hello" becomes <q>hello</q>
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content
    if '"' not in content:
        if '&quot;' not in content:
            return content

    # only if there are a few quote marks
    if content.count('"') > 4:
        return content
    if content.count('&quot;') > 4:
        return content

    new_content = content
    if '"' in content:
        sections = content.split('"')
        if len(sections) > 1:
            new_content = ''
            open_quote = True
            markup = False
            for char in content:
                curr_char = char
                if char == '<':
                    markup = True
                elif char == '>':
                    markup = False
                elif char == '"' and not markup:
                    if open_quote:
                        curr_char = '“'
                    else:
                        curr_char = '”'
                    open_quote = not open_quote
                new_content += curr_char

    if '&quot;' in new_content:
        open_quote = True
        content = new_content
        new_content = ''
        ctr = 0
        sections = content.split('&quot;')
        no_of_sections = len(sections)
        for sec in sections:
            new_content += sec
            if ctr < no_of_sections - 1:
                if open_quote:
                    new_content += '“'
                else:
                    new_content += '”'
                open_quote = not open_quote
            ctr += 1
    return new_content


def dangerous_css(filename: str, allow_local_network_access: bool) -> bool:
    """Returns true is the css file contains code which
    can create security problems
    """
    if not os.path.isfile(filename):
        return False

    content = None
    try:
        with open(filename, 'r', encoding='utf-8') as css_file:
            content = css_file.read().lower()
    except OSError:
        print('EX: unable to read css file ' + filename)

    if not content:
        return False

    css_matches = (
        'behavior:', ':expression', '?php', '.php',
        'google', 'regexp', 'localhost',
        '127.0.', '192.168', '10.0.', '@import'
    )
    for cssmatch in css_matches:
        if cssmatch in content:
            return True

    # search for non-local web links
    if 'url(' in content:
        url_list = content.split('url(')
        ctr = 0
        for url_str in url_list:
            if ctr > 0:
                if ')' in url_str:
                    url_str = url_str.split(')')[0]
                    if string_contains(url_str, ('http', 'ipfs', 'ipns')):
                        print('ERROR: non-local web link in CSS ' +
                              filename)
                        return True
            ctr += 1

    # an attacker can include html inside of the css
    # file as a comment and this may then be run from the html
    if dangerous_markup(content, allow_local_network_access, []):
        return True
    return False


def switch_words(base_dir: str, nickname: str, domain: str, content: str,
                 rules: [] = []) -> str:
    """Performs word replacements. eg. Trump -> The Orange Menace
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content

    if not rules:
        switch_words_filename = \
            acct_dir(base_dir, nickname, domain) + '/replacewords.txt'
        if not os.path.isfile(switch_words_filename):
            return content
        try:
            with open(switch_words_filename, 'r',
                      encoding='utf-8') as words_file:
                rules = words_file.readlines()
        except OSError:
            print('EX: unable to read switches ' + switch_words_filename)

    for line in rules:
        replace_str = remove_eol(line)
        splitters = ('->', ':', ',', ';', '-')
        word_transform = None
        for split_str in splitters:
            if split_str in replace_str:
                word_transform = replace_str.split(split_str)
                break
        if not word_transform:
            continue
        if len(word_transform) == 2:
            replace_str1 = word_transform[0].strip().replace('"', '')
            replace_str2 = word_transform[1].strip().replace('"', '')
            content = content.replace(replace_str1, replace_str2)
    return content


def _save_custom_emoji(session, base_dir: str, emojiName: str, url: str,
                       debug: bool) -> None:
    """Saves custom emoji to file
    """
    if not session:
        if debug:
            print('EX: _save_custom_emoji no session')
        return
    if '.' not in url:
        return
    ext = url.split('.')[-1]
    if ext != 'png':
        if debug:
            print('EX: Custom emoji is wrong format ' + url)
        return
    emojiName = emojiName.replace(':', '').strip().lower()
    custom_emoji_dir = base_dir + '/emojicustom'
    if not os.path.isdir(custom_emoji_dir):
        os.mkdir(custom_emoji_dir)
    emoji_image_filename = custom_emoji_dir + '/' + emojiName + '.' + ext
    if not download_image(session, url,
                          emoji_image_filename, debug, False):
        if debug:
            print('EX: custom emoji not downloaded ' + url)
        return
    emoji_json_filename = custom_emoji_dir + '/emoji.json'
    emoji_json = {}
    if os.path.isfile(emoji_json_filename):
        emoji_json = load_json(emoji_json_filename, 0, 1)
        if not emoji_json:
            emoji_json = {}
    if not emoji_json.get(emojiName):
        emoji_json[emojiName] = emojiName
        save_json(emoji_json, emoji_json_filename)
        if debug:
            print('EX: Saved custom emoji ' + emoji_json_filename)
    elif debug:
        print('EX: cusom emoji already saved')


def _get_emoji_name_from_code(base_dir: str, emoji_code: str) -> str:
    """Returns the emoji name from its code
    """
    emojis_filename = base_dir + '/emoji/emoji.json'
    if not os.path.isfile(emojis_filename):
        emojis_filename = base_dir + '/emoji/default_emoji.json'
        if not os.path.isfile(emojis_filename):
            return None
    emojis_json = load_json(emojis_filename)
    if not emojis_json:
        return None
    for emoji_name, code in emojis_json.items():
        if code == emoji_code:
            return emoji_name
    return None


def _update_common_emoji(base_dir: str, emoji_content: str) -> None:
    """Updates the list of commonly used emoji
    """
    if '.' in emoji_content:
        emoji_content = emoji_content.split('.')[0]
    emoji_content = emoji_content.replace(':', '')
    if emoji_content.startswith('0x'):
        # lookup the name for an emoji code
        emoji_code = emoji_content[2:]
        emoji_content = _get_emoji_name_from_code(base_dir, emoji_code)
        if not emoji_content:
            return
    common_emoji_filename = base_dir + '/accounts/common_emoji.txt'
    common_emoji = None
    if os.path.isfile(common_emoji_filename):
        try:
            with open(common_emoji_filename, 'r',
                      encoding='utf-8') as fp_emoji:
                common_emoji = fp_emoji.readlines()
        except OSError:
            print('EX: unable to load common emoji file')
    if common_emoji:
        new_common_emoji = []
        emoji_found = False
        for line in common_emoji:
            if ' ' + emoji_content in line:
                if not emoji_found:
                    emoji_found = True
                    counter = 1
                    count_str = line.split(' ')[0]
                    if count_str.isdigit():
                        counter = int(count_str) + 1
                    count_str = str(counter).zfill(16)
                    line = count_str + ' ' + emoji_content
                    new_common_emoji.append(line)
            else:
                line1 = remove_eol(line)
                new_common_emoji.append(line1)
        if not emoji_found:
            new_common_emoji.append(str(1).zfill(16) + ' ' + emoji_content)
        new_common_emoji.sort(reverse=True)
        try:
            with open(common_emoji_filename, 'w+',
                      encoding='utf-8') as fp_emoji:
                for line in new_common_emoji:
                    fp_emoji.write(line + '\n')
        except OSError:
            print('EX: error writing common emoji 1')
            return
    else:
        line = str(1).zfill(16) + ' ' + emoji_content + '\n'
        try:
            with open(common_emoji_filename, 'w+',
                      encoding='utf-8') as fp_emoji:
                fp_emoji.write(line)
        except OSError:
            print('EX: error writing common emoji 2')
            return


def replace_emoji_from_tags(session, base_dir: str,
                            content: str, tag: [], message_type: str,
                            debug: bool, screen_readable: bool) -> str:
    """Uses the tags to replace :emoji: with html image markup
    """
    for tag_item in tag:
        if not isinstance(tag_item, dict):
            continue
        if not tag_item.get('type'):
            continue
        if tag_item['type'] != 'Emoji':
            continue
        if not tag_item.get('name'):
            continue
        if not tag_item.get('icon'):
            continue
        if not tag_item['icon'].get('url'):
            continue
        url_str = get_url_from_post(tag_item['icon']['url'])
        if '/' not in url_str:
            continue
        if tag_item['name'] not in content:
            continue
        tag_url = remove_html(url_str)
        if not tag_url:
            continue
        icon_name = tag_url.split('/')[-1]
        if icon_name:
            if len(icon_name) > 1:
                if icon_name[0].isdigit():
                    if '.' in icon_name:
                        icon_name = icon_name.split('.')[0]
                        # see https://unicode.org/
                        # emoji/charts/full-emoji-list.html
                        if '-' not in icon_name:
                            # a single code
                            replaced = False
                            try:
                                replace_char = chr(int("0x" + icon_name, 16))
                                if not screen_readable:
                                    replace_char = \
                                        '<span aria-hidden="true">' + \
                                        replace_char + '</span>'
                                content = \
                                    content.replace(tag_item['name'],
                                                    replace_char)
                                replaced = True
                            except BaseException:
                                if debug:
                                    print('EX: replace_emoji_from_tags 1 ' +
                                          'no conversion of ' +
                                          str(icon_name) + ' to chr ' +
                                          tag_item['name'] + ' ' +
                                          tag_url)
                            if not replaced:
                                _save_custom_emoji(session, base_dir,
                                                   tag_item['name'],
                                                   tag_url, debug)
                                _update_common_emoji(base_dir, icon_name)
                            else:
                                _update_common_emoji(base_dir,
                                                     "0x" + icon_name)
                        else:
                            # sequence of codes
                            icon_codes = icon_name.split('-')
                            icon_code_sequence = ''
                            for icode in icon_codes:
                                replaced = False
                                try:
                                    icon_code_sequence += chr(int("0x" +
                                                                  icode, 16))
                                    replaced = True
                                except BaseException:
                                    icon_code_sequence = ''
                                    if debug:
                                        print('EX: ' +
                                              'replace_emoji_from_tags 2 ' +
                                              'no conversion of ' +
                                              str(icode) + ' to chr ' +
                                              tag_item['name'] + ' ' +
                                              tag_url)
                                if not replaced:
                                    _save_custom_emoji(session, base_dir,
                                                       tag_item['name'],
                                                       tag_url, debug)
                                    _update_common_emoji(base_dir,
                                                         icon_name)
                                else:
                                    _update_common_emoji(base_dir,
                                                         "0x" + icon_name)
                            if icon_code_sequence:
                                if not screen_readable:
                                    icon_code_sequence = \
                                        '<span aria-hidden="true">' + \
                                        icon_code_sequence + '</span>'
                                content = content.replace(tag_item['name'],
                                                          icon_code_sequence)

        html_class = 'emoji'
        if message_type == 'post header':
            html_class = 'emojiheader'
        if message_type == 'profile':
            html_class = 'emojiprofile'
        if screen_readable:
            emoji_tag_name = tag_item['name'].replace(':', '')
        else:
            emoji_tag_name = ''
        url_str = get_url_from_post(tag_item['icon']['url'])
        tag_url = remove_html(url_str)
        emoji_html = "<img src=\"" + tag_url + "\" alt=\"" + \
            emoji_tag_name + \
            "\" align=\"middle\" class=\"" + html_class + "\"/>"
        content = content.replace(tag_item['name'], emoji_html)
    return content


def _add_music_tag(content: str, tag: str) -> str:
    """If a music link is found then ensure that the post is
    tagged appropriately
    """
    if '#podcast' in content or '#documentary' in content:
        return content
    if '#' not in tag:
        tag = '#' + tag
    if tag in content:
        return content
    music_site_found = False
    for site in MUSIC_SITES:
        if site + '/' in content:
            music_site_found = True
            break
    if not music_site_found:
        return content
    return ':music: ' + content + ' ' + tag + ' '


def _shorten_linked_urls(content: str) -> str:
    """If content comes with a web link included then make sure
    that it is short enough
    """
    if 'href=' not in content:
        return content
    if '>' not in content:
        return content
    if '<' not in content:
        return content
    sections = content.split('>')
    ctr = 0
    for section_text in sections:
        if ctr == 0:
            ctr += 1
            continue
        if '<' not in section_text:
            ctr += 1
            continue
        section_text = section_text.split('<')[0]
        if ' ' in section_text:
            continue
        if len(section_text) > MAX_LINK_LENGTH:
            content = content.replace('>' + section_text + '<',
                                      '>' +
                                      section_text[:MAX_LINK_LENGTH-1] + '<')
        ctr += 1
    return content


def _contains_doi_reference(wrd: str, replace_dict: {}) -> bool:
    """Handle DOI scientific references
    """
    if not wrd.startswith('doi:') and \
       not wrd.startswith('DOI:'):
        return False

    doi_ref_str = wrd.split(':', 1)[1]
    doi_site = 'https://sci-hub.ru'
    markup = '<a href="' + doi_site + '/' + \
        doi_ref_str + '" tabindex="10" ' + \
        'rel="nofollow noopener noreferrer" ' + \
        'target="_blank">' + \
        '<span class="ellipsis">doi:' + doi_ref_str + \
        '</span></a>'
    replace_dict[wrd] = markup
    return True


def _contains_arxiv_reference(wrd: str, replace_dict: {}) -> bool:
    """Handle arxiv scientific references
    """
    if not wrd.startswith('arXiv:') and \
       not wrd.startswith('arx:') and \
       not wrd.startswith('arxiv:'):
        return False

    arxiv_ref_str = wrd.split(':', 1)[1].lower()
    if '.' in arxiv_ref_str:
        arxiv_ref = arxiv_ref_str.split('.')
    elif ':' in arxiv_ref_str:
        arxiv_ref = arxiv_ref_str.split(':')
    else:
        return False
    if len(arxiv_ref) != 2:
        return False
    if not arxiv_ref[0].isdigit():
        return False
    arxiv_day = arxiv_ref[1]
    if 'v' in arxiv_day:
        arxiv_day = arxiv_day.split('v')[0]
    if not arxiv_day.isdigit():
        return False
    ref_str = arxiv_ref[0] + '.' + arxiv_ref[1]
    markup = '<a href="https://arxiv.org/abs/' + \
        ref_str + '" tabindex="10" ' + \
        'rel="nofollow noopener noreferrer" ' + \
        'target="_blank">' + \
        '<span class="ellipsis">arXiv:' + ref_str + \
        '</span></a>'
    replace_dict[wrd] = markup
    return True


def _contains_academic_references(content: str) -> bool:
    """Does the given content contain academic references
    """
    prefixes = (
        'arXiv:', 'arx:', 'arxiv:',
        'doi:', 'DOI:'
    )
    for reference in prefixes:
        if reference in content:
            return True
    return False


def remove_link_trackers_from_content(content: str) -> str:
    """ Removes any link trackers from urls within the content
    """
    if '?utm_' not in content:
        return content
    sections = content.split('?utm_')
    ctr = 0
    new_content = ''
    for section_str in sections:
        if ctr == 0:
            new_content = section_str
            ctr = 1
            continue
        if '"' in section_str:
            new_content += '"' + section_str.split('"', 1)[1]
        else:
            new_content += section_str
        ctr += 1
    return new_content


def add_web_links(content: str) -> str:
    """Adds markup for web links
    """
    content = _shorten_linked_urls(content)

    if ':' not in content:
        return content

    prefixes = get_link_prefixes()

    # do any of these prefixes exist within the content?
    prefix_found = False
    for prefix in prefixes:
        if prefix in content:
            prefix_found = True
            break

    # if there are no prefixes then just keep the content we have
    if not prefix_found:
        if _contains_academic_references(content):
            prefix_found = True
        else:
            return content

    content = content.replace('\r', '')
    words = content.replace('\n', ' --linebreak-- ').split(' ')
    replace_dict = {}
    for wrd in words:
        if ':' not in wrd:
            continue
        if _contains_arxiv_reference(wrd, replace_dict):
            continue
        if _contains_doi_reference(wrd, replace_dict):
            continue
        # does the word begin with a link prefix?
        prefix_found = False
        for prefix in prefixes:
            if wrd.startswith(prefix):
                prefix_found = True
                break
        if not prefix_found:
            continue
        # the word contains a link prefix
        url = wrd
        if url.endswith('.') or wrd.endswith(';'):
            url = url[:-1]
        url = remove_link_tracking(url)
        markup = '<a href="' + url + '" tabindex="10" ' + \
            'rel="nofollow noopener noreferrer" target="_blank">'
        for prefix in prefixes:
            if url.startswith(prefix):
                markup += '<span class="invisible">' + prefix + '</span>'
                break
        link_text = url
        for prefix in prefixes:
            link_text = link_text.replace(prefix, '')
        # prevent links from becoming too long
        if len(link_text) > MAX_LINK_LENGTH:
            markup += '<span class="ellipsis">' + \
                link_text[:MAX_LINK_LENGTH] + '</span>'
            markup += '<span class="invisible">' + \
                link_text[MAX_LINK_LENGTH:] + '</span></a>'
        else:
            markup += '<span class="ellipsis">' + link_text + '</span></a>'
        replace_dict[url] = markup

    # do the replacements
    for url, markup in replace_dict.items():
        content = content.replace(url, markup)

    # replace any line breaks
    content = content.replace(' --linebreak-- ', '<br>')

    return content


def safe_web_text(arbitrary_html: str) -> str:
    """Turns arbitrary html into something safe.
    So if the arbitrary html contains attack scripts those will be removed
    """
    # first remove the markup, so that we have something safe
    safe_text = remove_html(arbitrary_html)
    if not safe_text:
        return ''
    # remove any spurious characters found in podcast descriptions
    remove_chars = ('Œ', 'â€', 'ğŸ', '<EFBFBD>', ']]', '__')
    for remchar in remove_chars:
        safe_text = safe_text.replace(remchar, '')
    # recreate any url links safely
    return add_web_links(safe_text)


def _add_hash_tags(word_str: str, http_prefix: str, domain: str,
                   replace_hashtags: {}, post_hashtags: {}) -> bool:
    """Detects hashtags and adds them to the replacements dict
    Also updates the hashtags list to be added to the post
    """
    if replace_hashtags.get(word_str):
        return True
    hashtag = word_str[1:]
    if not valid_hash_tag(hashtag):
        return False
    hashtag_url = http_prefix + "://" + domain + "/tags/" + hashtag
    post_hashtags[hashtag] = {
        'href': hashtag_url,
        'name': '#' + hashtag,
        'type': 'Hashtag'
    }
    replace_hashtags[word_str] = "<a href=\"" + hashtag_url + \
        "\" class=\"mention hashtag\" rel=\"tag\" tabindex=\"10\">" + \
        "<span aria-hidden=\"true\">#</span><span>" + \
        hashtag + "</span></a>"
    return True


def replace_remote_hashtags(content: str,
                            nickname: str, domain: str) -> str:
    """Replaces remote hashtags with a local version
    """
    if not domain:
        return content

    if ' href="' not in content:
        return content

    sections = content.split(' href="')
    ctr = 0
    replacements = {}
    for section in sections:
        if ctr == 0:
            ctr += 1
            continue
        if '"' not in section:
            ctr += 1
            continue
        link = section.split('"')[0]
        if '://' not in link:
            continue
        if '?remotetag=' in link:
            ctr += 1
            continue
        if '/tags/' not in link:
            ctr += 1
            continue
        if '/' + domain not in link:
            new_link = '/users/' + nickname + \
                '?remotetag=' + link.replace('/', '--')
            replacements[link] = new_link
        ctr += 1
    if not replacements:
        return content
    for old_link, new_link in replacements.items():
        content = content.replace('"' + old_link + '"',
                                  '"' + new_link + '"')
    return content


def _add_emoji(base_dir: str, word_str: str,
               http_prefix: str, domain: str,
               replace_emoji: {}, post_tags: {},
               emoji_dict: {}) -> bool:
    """Detects Emoji and adds them to the replacements dict
    Also updates the tags list to be added to the post
    """
    if not word_str.startswith(':'):
        return False
    if not word_str.endswith(':'):
        return False
    if len(word_str) < 3:
        return False
    if replace_emoji.get(word_str):
        return True
    # remove leading and trailing : characters
    emoji = word_str[1:]
    emoji = emoji[:-1]
    # is the text of the emoji valid?
    if not valid_hash_tag(emoji):
        return False
    if not emoji_dict.get(emoji):
        return False
    emoji_filename = base_dir + '/emoji/' + emoji_dict[emoji] + '.png'
    if not os.path.isfile(emoji_filename):
        emoji_filename = \
            base_dir + '/emojicustom/' + emoji_dict[emoji] + '.png'
        if not os.path.isfile(emoji_filename):
            return False
    emoji_url = http_prefix + "://" + domain + \
        "/emoji/" + emoji_dict[emoji] + '.png'
    post_tags[emoji] = {
        'icon': {
            'mediaType': 'image/png',
            'type': 'Image',
            'url': emoji_url
        },
        'name': ':' + emoji + ':',
        "updated": file_last_modified(emoji_filename),
        "id": emoji_url.replace('.png', ''),
        'type': 'Emoji'
    }
    return True


def post_tag_exists(tag_type: str, tag_name: str, tags: {}) -> bool:
    """Returns true if a tag exists in the given dict
    """
    for tag in tags:
        if tag['name'] == tag_name and tag['type'] == tag_type:
            return True
    return False


def _mention_to_url(base_dir: str, http_prefix: str,
                    domain: str, nickname: str) -> str:
    """Convert https://somedomain/@somenick to
    https://somedomain/users/somenick
    This uses the hack of trying the cache directory to see if
    there is a matching actor
    """
    possible_paths = get_user_paths()
    cache_dir = base_dir + '/cache/actors'
    cache_path_start = cache_dir + '/' + http_prefix + ':##' + domain
    for users_path in possible_paths:
        users_path = users_path.replace('/', '#')
        possible_cache_entry = \
            cache_path_start + users_path + nickname + '.json'
        if os.path.isfile(possible_cache_entry):
            return http_prefix + '://' + \
                domain + users_path.replace('#', '/') + nickname
    possible_cache_entry = \
        cache_path_start + '#' + nickname + '.json'
    if os.path.isfile(possible_cache_entry):
        return http_prefix + '://' + domain + '/' + nickname
    return http_prefix + '://' + domain + '/users/' + nickname


def _add_mention(base_dir: str, word_str: str, http_prefix: str,
                 following: [], petnames: [], replace_mentions: {},
                 recipients: [], tags: {}) -> bool:
    """Detects mentions and adds them to the replacements dict and
    recipients list
    """
    possible_handle = word_str[1:]
    # @nick
    if following and '@' not in possible_handle:
        # fall back to a best effort match against the following list
        # if no domain was specified. eg. @nick
        possible_nickname = possible_handle
        for follow in following:
            if '@' not in follow:
                continue
            follow_nick = follow.split('@')[0]
            if possible_nickname == follow_nick:
                follow_str = remove_eol(follow)
                replace_domain = follow_str.split('@')[1]
                recipient_actor = \
                    _mention_to_url(base_dir, http_prefix,
                                    replace_domain, possible_nickname)
                if recipient_actor not in recipients:
                    recipients.append(recipient_actor)
                tags[word_str] = {
                    'href': recipient_actor,
                    'name': word_str,
                    'type': 'Mention'
                }
                replace_mentions[word_str] = \
                    "<span class=\"h-card\"><a href=\"" + recipient_actor + \
                    "\" tabindex=\"10\" class=\"u-url mention\">@<span>" + \
                    possible_nickname + "</span></a></span>"
                return True
        # try replacing petnames with mentions
        follow_ctr = 0
        for follow in following:
            if '@' not in follow:
                follow_ctr += 1
                continue
            pet = remove_eol(petnames[follow_ctr])
            if pet:
                if possible_nickname == pet:
                    follow_str = remove_eol(follow)
                    replace_nickname = follow_str.split('@')[0]
                    replace_domain = follow_str.split('@')[1]
                    recipient_actor = \
                        _mention_to_url(base_dir, http_prefix,
                                        replace_domain, replace_nickname)
                    if recipient_actor not in recipients:
                        recipients.append(recipient_actor)
                    tags[word_str] = {
                        'href': recipient_actor,
                        'name': word_str,
                        'type': 'Mention'
                    }
                    replace_mentions[word_str] = \
                        "<span class=\"h-card\"><a href=\"" + \
                        recipient_actor + "\" tabindex=\"10\" " + \
                        "class=\"u-url mention\">@<span>" + \
                        replace_nickname + "</span></a></span>"
                    return True
            follow_ctr += 1
        return False
    possible_nickname = None
    possible_domain = None
    if '@' not in possible_handle:
        return False
    possible_nickname = possible_handle.split('@')[0]
    if not possible_nickname:
        return False
    possible_domain = \
        possible_handle.split('@')[1].strip('\n').strip('\r')
    if not possible_domain:
        return False
    if following:
        for follow in following:
            if remove_eol(follow) != possible_handle:
                continue
            recipient_actor = \
                _mention_to_url(base_dir, http_prefix,
                                possible_domain, possible_nickname)
            if recipient_actor not in recipients:
                recipients.append(recipient_actor)
            tags[word_str] = {
                'href': recipient_actor,
                'name': word_str,
                'type': 'Mention'
            }
            replace_mentions[word_str] = \
                "<span class=\"h-card\"><a href=\"" + recipient_actor + \
                "\" tabindex=\"10\" class=\"u-url mention\">@<span>" + \
                possible_nickname + "</span></a></span>"
            return True
    # @nick@domain
    if not (possible_domain == 'localhost' or '.' in possible_domain):
        return False
    recipient_actor = \
        _mention_to_url(base_dir, http_prefix,
                        possible_domain, possible_nickname)
    if recipient_actor not in recipients:
        recipients.append(recipient_actor)
    tags[word_str] = {
        'href': recipient_actor,
        'name': word_str,
        'type': 'Mention'
    }
    replace_mentions[word_str] = \
        "<span class=\"h-card\"><a href=\"" + recipient_actor + \
        "\" tabindex=\"10\" class=\"u-url mention\">@<span>" + \
        possible_nickname + "</span></a></span>"
    return True


def replace_content_duplicates(content: str) -> str:
    """Replaces invalid duplicates within content
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content
    while '<<' in content:
        content = content.replace('<<', '<')
    while '>>' in content:
        content = content.replace('>>', '>')
    content = content.replace('<\\p>', '')
    return content


def remove_text_formatting(content: str, bold_reading: bool) -> str:
    """Removes markup for bold, italics, etc
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content
    if '<' not in content:
        return content
    for markup in REMOVE_MARKUP:
        if bold_reading:
            if markup == 'b':
                continue
        content = content.replace('<' + markup + '>', '')
        content = content.replace('</' + markup + '>', '')
        content = content.replace('<' + markup.upper() + '>', '')
        content = content.replace('</' + markup.upper() + '>', '')
    return content


def remove_long_words(content: str, max_word_length: int,
                      long_words_list: []) -> str:
    """Breaks up long words so that on mobile screens this doesn't
    disrupt the layout
    """
    if is_pgp_encrypted(content) or contains_pgp_public_key(content):
        return content
    content = replace_content_duplicates(content)
    if ' ' not in content and '</p><p>' not in content:
        # handle a single very long string with no spaces
        content_str = content.replace('<p>', '').replace(r'<\p>', '')
        if '://' not in content_str:
            if len(content_str) > max_word_length:
                if '<p>' in content:
                    content = '<p>' + content_str[:max_word_length] + r'<\p>'
                else:
                    content = content[:max_word_length]
                return content
    content = content.replace('<p></p>', '<p> </p>')
    words = content.split(' ')
    if not long_words_list:
        long_words_list = []
        for word_str in words:
            if len(word_str) > max_word_length:
                if word_str not in long_words_list:
                    long_words_list.append(word_str)
    for word_str in long_words_list:
        original_word_str = word_str
        if word_str.startswith('<p>'):
            word_str = word_str.replace('<p>', '')
        if word_str.startswith('<'):
            continue
        if len(word_str) == 76:
            if word_str.upper() == word_str:
                # tox address
                continue
        if '=\"' in word_str:
            continue
        if '@' in word_str:
            if '@@' not in word_str:
                continue
        if '=.ed25519' in word_str:
            continue
        if '.onion' in word_str:
            continue
        if '.i2p' in word_str:
            continue
        if 'https:' in word_str:
            continue
        if 'http:' in word_str:
            continue
        if 'i2p:' in word_str:
            continue
        if 'gnunet:' in word_str:
            continue
        if 'dat:' in word_str:
            continue
        if 'rad:' in word_str:
            continue
        if 'hyper:' in word_str:
            continue
        if 'briar:' in word_str:
            continue
        if '<' in word_str:
            replace_word = word_str.split('<', 1)[0]
            # if len(replace_word) > max_word_length:
            #     replace_word = replace_word[:max_word_length]
            content = content.replace(word_str, replace_word)
            word_str = replace_word
        if '/' in word_str:
            continue
        if len(word_str[max_word_length:]) < max_word_length:
            end_of_line_char = '\n'
            if '<br>' in original_word_str:
                end_of_line_char = ''
            content = content.replace(word_str,
                                      word_str[:max_word_length] +
                                      end_of_line_char +
                                      word_str[max_word_length:])
        else:
            content = content.replace(word_str,
                                      word_str[:max_word_length])
    if content.startswith('<p>'):
        if not content.endswith('</p>'):
            content = content.strip() + '</p>'
    content = content.replace('<p> </p>', '<p></p>')
    return content


def _load_auto_tags(base_dir: str, nickname: str, domain: str) -> []:
    """Loads automatic tags file and returns a list containing
    the lines of the file
    """
    filename = acct_dir(base_dir, nickname, domain) + '/autotags.txt'
    if not os.path.isfile(filename):
        return []
    try:
        with open(filename, 'r', encoding='utf-8') as tags_file:
            return tags_file.readlines()
    except OSError:
        print('EX: unable to read auto tags ' + filename)
    return []


def _auto_tag(word_str: str, auto_tag_list: [], append_tags: []) -> None:
    """Generates a list of tags to be automatically appended to the content
    """
    for tag_rule in auto_tag_list:
        if word_str not in tag_rule:
            continue
        if '->' not in tag_rule:
            continue
        rulematch = tag_rule.split('->')[0].strip()
        if rulematch != word_str:
            continue
        tag_name = tag_rule.split('->')[1].strip()
        if tag_name.startswith('#'):
            if tag_name not in append_tags:
                append_tags.append(tag_name)
        else:
            if '#' + tag_name not in append_tags:
                append_tags.append('#' + tag_name)


def _get_simplified_content(content: str) -> str:
    """Returns a simplified version of the content suitable for
    splitting up into individual words
    """
    content_simplified = \
        content.replace(',', ' ').replace(';', ' ').replace('- ', ' ')
    content_simplified = content_simplified.replace('. ', ' ').strip()
    if content_simplified.endswith('.'):
        content_simplified = content_simplified[:len(content_simplified)-1]
    return content_simplified


def detect_dogwhistles(content: str, dogwhistles: {}) -> {}:
    """Returns a dict containing any detected dogwhistle words
    """
    content = remove_html(content).lower()
    result = {}
    words = _get_simplified_content(content).split(' ')
    for whistle, category in dogwhistles.items():
        if not category:
            continue
        ending = False
        starting = False
        whistle = whistle.lower()

        if whistle.startswith('x-'):
            whistle = whistle[2:]
            ending = True
        elif string_ends_with(whistle, ('*', '~', '-')):
            whistle = whistle[1:]
            ending = True

        if ending:
            prev_wrd = ''
            for wrd in words:
                wrd2 = (prev_wrd + ' ' + wrd).strip()
                if wrd.endswith(whistle) or wrd2.endswith(whistle):
                    if not result.get(whistle):
                        result[whistle] = {
                            "count": 1,
                            "category": category
                        }
                    else:
                        result[whistle]['count'] += 1
                prev_wrd = wrd
            continue

        if whistle.lower().endswith('-x'):
            whistle = whistle[:len(whistle)-2]
            starting = True
        elif string_ends_with(whistle, ('*', '~', '-')):
            whistle = whistle[:len(whistle)-1]
            starting = True

        if starting:
            prev_wrd = ''
            for wrd in words:
                wrd2 = (prev_wrd + ' ' + wrd).strip()
                if wrd.startswith(whistle) or wrd2.startswith(whistle):
                    if not result.get(whistle):
                        result[whistle] = {
                            "count": 1,
                            "category": category
                        }
                    else:
                        result[whistle]['count'] += 1
                prev_wrd = wrd
            continue

        if '*' in whistle:
            whistle_start = whistle.split('*', 1)[0]
            whistle_end = whistle.split('*', 1)[1]
            prev_wrd = ''
            for wrd in words:
                wrd2 = (prev_wrd + ' ' + wrd).strip()
                if ((wrd.startswith(whistle_start) and
                     wrd.endswith(whistle_end)) or
                    (wrd2.startswith(whistle_start) and
                     wrd2.endswith(whistle_end))):
                    if not result.get(whistle):
                        result[whistle] = {
                            "count": 1,
                            "category": category
                        }
                    else:
                        result[whistle]['count'] += 1
                prev_wrd = wrd
            continue

        prev_wrd = ''
        for wrd in words:
            wrd2 = (prev_wrd + ' ' + wrd).strip()
            if whistle in (wrd, wrd2):
                if not result.get(whistle):
                    result[whistle] = {
                        "count": 1,
                        "category": category
                    }
                else:
                    result[whistle]['count'] += 1
            prev_wrd = wrd
    return result


def load_dogwhistles(filename: str) -> {}:
    """Loads a list of dogwhistles from file
    """
    if not os.path.isfile(filename):
        return {}
    dogwhistle_lines = []
    try:
        with open(filename, 'r', encoding='utf-8') as fp_dogwhistles:
            dogwhistle_lines = fp_dogwhistles.readlines()
    except OSError:
        print('EX: unable to load dogwhistles from ' + filename)
        return {}
    separators = ('->', '=>', ',', ';', '|', '=')
    dogwhistles = {}
    for line in dogwhistle_lines:
        line = remove_eol(line).strip()
        if not line:
            continue
        if line.startswith('#'):
            continue
        whistle = None
        category = None
        for sep in separators:
            if sep in line:
                whistle = line.split(sep, 1)[0].strip()
                category = line.split(sep, 1)[1].strip()
                break
        if not whistle:
            whistle = line
        dogwhistles[whistle] = category
    return dogwhistles


def add_html_tags(base_dir: str, http_prefix: str,
                  nickname: str, domain: str, content: str,
                  recipients: [], hashtags: {}, translate: {},
                  is_json_content: bool = False) -> str:
    """ Replaces plaintext mentions such as @nick@domain into html
    by matching against known following accounts
    """
    if content.startswith('<p>'):
        content = html_replace_email_quote(content)
        return html_replace_quote_marks(content)
    max_word_length = 40
    content = content.replace('\r', '')
    content = content.replace('\n', ' --linebreak-- ')
    now_playing_str = 'NowPlaying'
    if translate.get(now_playing_str):
        now_playing_str = translate[now_playing_str]
    now_playing_lower_str = 'nowplaying'
    if translate.get(now_playing_lower_str):
        now_playing_lower_str = translate[now_playing_lower_str]
    if '#' + now_playing_lower_str in content:
        content = content.replace('#' + now_playing_lower_str,
                                  '#' + now_playing_str)
    content = _add_music_tag(content, now_playing_str)
    words = _get_simplified_content(content).split(' ')

    # remove . for words which are not mentions
    new_words = []
    for _, word_str in enumerate(words):
        if word_str.endswith('.'):
            if not word_str.startswith('@'):
                word_str = word_str[:-1]
        if word_str.startswith('.'):
            word_str = word_str[1:]
        new_words.append(word_str)
    words = new_words

    replace_mentions = {}
    replace_hashtags = {}
    replace_emoji = {}
    emoji_dict = {}
    original_domain = domain
    domain = remove_domain_port(domain)
    following_filename = \
        acct_dir(base_dir, nickname, domain) + '/following.txt'

    # read the following list so that we can detect just @nick
    # in addition to @nick@domain
    following = None
    petnames = None
    if '@' in words:
        if os.path.isfile(following_filename):
            following = []
            try:
                with open(following_filename, 'r',
                          encoding='utf-8') as foll_file:
                    following = foll_file.readlines()
            except OSError:
                print('EX: unable to read ' + following_filename)
            for handle in following:
                pet = get_pet_name(base_dir, nickname, domain, handle)
                if pet:
                    petnames.append(pet + '\n')

    # extract mentions and tags from words
    long_words_list = []
    prev_word_str = ''
    auto_tags_list = _load_auto_tags(base_dir, nickname, domain)
    append_tags = []
    for word_str in words:
        word_len = len(word_str)
        if word_len <= 2:
            continue
        if word_len > max_word_length:
            long_words_list.append(word_str)
        first_char = word_str[0]
        if first_char == '@':
            if _add_mention(base_dir, word_str, http_prefix, following,
                            petnames, replace_mentions, recipients,
                            hashtags):
                prev_word_str = ''
                continue
        elif first_char == '#':
            # remove any endings from the hashtag
            hash_tag_endings = ('.', ':', ';', '-', '\n')
            for ending in hash_tag_endings:
                if word_str.endswith(ending):
                    word_str = word_str[:len(word_str) - 1]
                    break

            if _add_hash_tags(word_str, http_prefix, original_domain,
                              replace_hashtags, hashtags):
                prev_word_str = ''
                continue
        elif ':' in word_str:
            word_str2 = word_str.split(':')[1]
            if not emoji_dict:
                # emoji.json is generated so that it can be customized and
                # the changes will be retained even if default_emoji.json
                # is subsequently updated
                if not os.path.isfile(base_dir + '/emoji/emoji.json'):
                    copyfile(base_dir + '/emoji/default_emoji.json',
                             base_dir + '/emoji/emoji.json')
            emoji_dict = load_json(base_dir + '/emoji/emoji.json')

            # append custom emoji to the dict
            custom_emoji_filename = base_dir + '/emojicustom/emoji.json'
            if os.path.isfile(custom_emoji_filename):
                custom_emoji_dict = load_json(custom_emoji_filename)
                if custom_emoji_dict:
                    # combine emoji dicts one by one
                    for ename, eitem in custom_emoji_dict.items():
                        if ename and eitem:
                            if not emoji_dict.get(ename):
                                emoji_dict[ename] = eitem

            _add_emoji(base_dir, ':' + word_str2 + ':', http_prefix,
                       original_domain, replace_emoji, hashtags,
                       emoji_dict)
        else:
            if _auto_tag(word_str, auto_tags_list, append_tags):
                prev_word_str = ''
                continue
            if prev_word_str:
                if _auto_tag(prev_word_str + ' ' + word_str,
                             auto_tags_list, append_tags):
                    prev_word_str = ''
                    continue
        prev_word_str = word_str

    # add any auto generated tags
    for appended in append_tags:
        content = content + ' ' + appended
        _add_hash_tags(appended, http_prefix, original_domain,
                       replace_hashtags, hashtags)

    # replace words with their html versions
    for word_str, replace_str in replace_mentions.items():
        content = content.replace(word_str, replace_str)
    for word_str, replace_str in replace_hashtags.items():
        content = content.replace(word_str, replace_str)
    if not is_json_content:
        for word_str, replace_str in replace_emoji.items():
            content = content.replace(word_str, replace_str)

    content = add_web_links(content)
    if long_words_list:
        content = remove_long_words(content, max_word_length, long_words_list)
    content = limit_repeated_words(content, 6)
    content = content.replace(' --linebreak-- ', '</p><p>')
    content = html_replace_email_quote(content)
    return '<p>' + html_replace_quote_marks(content) + '</p>'


def _string_starts_with_url_prefix(text: str) -> bool:
    """ Does the given text begin with one of the url prefixes?
    """
    url_prefixes = ('http', 'gnunet', 'i2p', 'ipfs', 'ipns', 'hyper', 'dat:')
    for possible_prefix in url_prefixes:
        if text.startswith(possible_prefix):
            return True
    return False


def get_mentions_from_html(html_text: str, match_str: str) -> []:
    """Extracts mentioned actors from the given html content string
    """
    mentions = []
    if match_str not in html_text:
        return mentions
    mentions_list = html_text.split(match_str)
    for mention_str in mentions_list:
        if '"' not in mention_str:
            continue
        actor_str = mention_str.split('"')[0]
        if _string_starts_with_url_prefix(actor_str):
            if actor_str not in mentions:
                mentions.append(actor_str)
    return mentions


def extract_media_in_form_post(post_bytes, boundary, name: str):
    """Extracts the binary encoding for image/video/audio within a http
    form POST
    Returns the media bytes and the remaining bytes
    """
    image_start_boundary = b'Content-Disposition: form-data; name="' + \
        name.encode('utf8', 'ignore') + b'";'
    image_start_location = post_bytes.find(image_start_boundary)
    if image_start_location == -1:
        return None, post_bytes

    # bytes after the start boundary appears
    media_bytes = post_bytes[image_start_location:]

    # look for the next boundary
    image_end_boundary = boundary.encode('utf8', 'ignore')
    image_end_location = media_bytes.find(image_end_boundary)
    if image_end_location == -1:
        # no ending boundary
        return media_bytes, post_bytes[:image_start_location]

    # remaining bytes after the end of the image
    remainder = media_bytes[image_end_location:]

    # remove bytes after the end boundary
    media_bytes = media_bytes[:image_end_location]

    # return the media and the before+after bytes
    return media_bytes, post_bytes[:image_start_location] + remainder


def _valid_follows_csv(content: str) -> bool:
    """is the given content a valid csv file containing imported follows?
    """
    if ',' not in content:
        return False
    if 'Account address,' not in content:
        return False
    return True


def save_media_in_form_post(media_bytes, debug: bool,
                            filename_base: str) -> (str, str):
    """Saves the given media bytes extracted from http form POST
    Returns the filename and attachment type
    """
    if not media_bytes:
        if filename_base:
            # remove any existing files
            extension_types = get_image_extensions()
            for ex in extension_types:
                possible_other_format = filename_base + '.' + ex
                if os.path.isfile(possible_other_format):
                    try:
                        os.remove(possible_other_format)
                    except OSError:
                        if debug:
                            print('EX: save_media_in_form_post ' +
                                  'unable to delete other ' +
                                  str(possible_other_format))
            if os.path.isfile(filename_base):
                try:
                    os.remove(filename_base)
                except OSError:
                    if debug:
                        print('EX: save_media_in_form_post ' +
                              'unable to delete ' +
                              str(filename_base))

        if debug:
            print('DEBUG: No media found within POST')
        return None, None

    media_location = -1
    search_str = ''
    filename = None

    # directly search the binary array for the beginning
    # of an image, zip or csv
    extension_list = {
        'png': 'image/png',
        'jpeg': 'image/jpeg',
        'jxl': 'image/jxl',
        'gif': 'image/gif',
        'svg': 'image/svg+xml',
        'webp': 'image/webp',
        'avif': 'image/avif',
        'heic': 'image/heic',
        'mp4': 'video/mp4',
        'ogv': 'video/ogv',
        'mp3': 'audio/mpeg',
        'ogg': 'audio/ogg',
        'wav': 'audio/vnd.wave',
        'wav2': 'audio/wav',
        'wav3': 'audio/x-wav',
        'wav4': 'audio/x-pn-wave',
        'opus': 'audio/opus',
        'spx': 'audio/speex',
        'flac': 'audio/flac',
        'zip': 'application/zip',
        'csv': 'text/csv',
        'csv2': 'text/plain'
    }
    detected_extension = None
    for extension, content_type in extension_list.items():
        search_str = b'Content-Type: ' + content_type.encode('utf8', 'ignore')
        media_location = media_bytes.find(search_str)
        if media_location > -1:
            # image/video/audio binaries
            if extension == 'jpeg':
                extension = 'jpg'
            elif extension == 'mpeg':
                extension = 'mp3'
            elif extension == 'csv2':
                extension = 'csv'
            elif extension == 'wav2':
                extension = 'wav'
            elif extension == 'wav3':
                extension = 'wav'
            elif extension == 'wav4':
                extension = 'wav'
            if filename_base:
                if not filename_base.endswith('.' + extension):
                    filename = filename_base + '.' + extension
                else:
                    # already has the extension
                    filename = filename_base
            search_lst = search_str.decode().split('/', maxsplit=1)
            attachment_media_type = \
                search_lst[0].replace('Content-Type: ', '')
            detected_extension = extension
            break

    if not filename:
        return None, None

    # locate the beginning of the image, after any
    # carriage returns
    start_pos = media_location + len(search_str)
    for offset in range(1, 8):
        if media_bytes[start_pos+offset] != 10:
            if media_bytes[start_pos+offset] != 13:
                start_pos += offset
                break

    # remove any existing image files with a different format
    if detected_extension != 'zip':
        extension_types = get_image_extensions()
        for ex in extension_types:
            if ex == detected_extension:
                continue
            possible_other_format = \
                filename.replace('.temp', '').replace('.' +
                                                      detected_extension, '.' +
                                                      ex)
            if os.path.isfile(possible_other_format):
                try:
                    os.remove(possible_other_format)
                except OSError:
                    if debug:
                        print('EX: save_media_in_form_post ' +
                              'unable to delete other 2 ' +
                              str(possible_other_format))

    # don't allow scripts within svg files
    if detected_extension == 'svg':
        svg_str = media_bytes[start_pos:]
        svg_str = svg_str.decode()
        if dangerous_svg(svg_str, False):
            return None, None
    elif detected_extension == 'csv':
        csv_str = media_bytes[start_pos:]
        csv_str = csv_str.decode()
        if not _valid_follows_csv(csv_str):
            return None, None

    # if this is an image then check that the binary looks like an image
    image_extension_types = get_image_extensions()
    if detected_extension in image_extension_types:
        if not binary_is_image(filename, media_bytes[start_pos:]):
            print('WARN: save_media_in_form_post ' +
                  'image binary not recognized ' + filename)
            return None, None

    try:
        with open(filename, 'wb') as fp_media:
            fp_media.write(media_bytes[start_pos:])
    except OSError:
        print('EX: unable to write media')

    if not os.path.isfile(filename):
        print('WARN: Media file could not be written to file: ' + filename)
        return None, None
    print('Uploaded media file written: ' + filename)

    return filename, attachment_media_type


def combine_textarea_lines(text: str) -> str:
    """Combines separate lines
    """
    result = ''
    ctr = 0
    paragraphs = text.split('\n\n')
    for para in paragraphs:
        para = para.replace('\n* ', '***BULLET POINT*** ')
        para = para.replace('\n * ', '***BULLET POINT*** ')
        para = para.replace('\n- ', '***DASH POINT*** ')
        para = para.replace('\n - ', '***DASH POINT*** ')
        para = para.replace('\n', ' ')
        para = para.replace('  ', ' ')
        para = para.replace('***BULLET POINT*** ', '\n* ')
        para = para.replace('***DASH POINT*** ', '\n- ')
        if ctr > 0:
            result += '</p><p>'
        result += para
        ctr += 1
    return result


def extract_text_fields_in_post(post_bytes, boundary: str, debug: bool,
                                unit_test_data: str) -> {}:
    """Returns a dictionary containing the text fields of a http form POST
    The boundary argument comes from the http header
    """
    if boundary == 'LYNX':
        if debug:
            print('POST from lynx browser')
        boundary = '--LYNX'

    if not unit_test_data:
        msg_bytes = email.parser.BytesParser().parsebytes(post_bytes)
        message_fields = msg_bytes.get_payload(decode=True).decode('utf-8')
    else:
        message_fields = unit_test_data

    if debug:
        if 'password' not in message_fields:
            print('DEBUG: POST arriving ' + message_fields)

    message_fields = message_fields.split(boundary)
    fields = {}
    fields_with_semicolon_allowed = (
        'message', 'bio', 'autoCW', 'password', 'passwordconfirm',
        'instanceDescription', 'instanceDescriptionShort',
        'subject', 'location', 'imageDescription', 'importBlocks',
        'importFollows', 'importTheme'
    )
    if debug:
        if 'password' not in message_fields:
            print('DEBUG: POST message_fields: ' + str(message_fields))
    lynx_content_type = 'Content-Type: text/plain; charset=utf-8\r\n'
    # examine each section of the POST, separated by the boundary
    for fld in message_fields:
        if fld == '--':
            continue
        if ' name="' not in fld:
            continue
        post_str = fld.split(' name="', 1)[1]
        if '"' not in post_str:
            continue
        post_key = post_str.split('"', 1)[0]
        if debug:
            print('post_key: ' + post_key)
        post_value_str = post_str.split('"', 1)[1]
        if boundary == '--LYNX':
            post_value_str = \
                post_value_str.replace(lynx_content_type, '')
        if debug and 'password' not in post_key:
            print('boundary: ' + boundary)
            print('post_value_str1: ' + post_value_str)
        if ';' in post_value_str:
            if post_key not in fields_with_semicolon_allowed and \
               not post_key.startswith('edited'):
                if debug:
                    print('extract_text_fields_in_post exit 1')
                continue
        if debug and 'password' not in post_key:
            print('post_value_str2: ' + post_value_str)
        if '\r\n' not in post_value_str:
            if debug:
                print('extract_text_fields_in_post exit 2')
            continue
        post_lines = post_value_str.split('\r\n')
        if debug and 'password' not in post_key:
            print('post_lines: ' + str(post_lines))
        post_value = ''
        if len(post_lines) > 2:
            for line in range(2, len(post_lines)-1):
                if line > 2:
                    post_value += '\n'
                post_value += post_lines[line]
        fields[post_key] = urllib.parse.unquote(post_value)
        if boundary == '--LYNX' and post_key in ('message', 'bio'):
            fields[post_key] = combine_textarea_lines(fields[post_key])
    return fields


def limit_repeated_words(text: str, max_repeats: int) -> str:
    """Removes words which are repeated many times
    """
    words = text.replace('\n', ' ').split(' ')
    repeat_ctr = 0
    repeated_text = ''
    replacements = {}
    prev_word = ''
    for word in words:
        if word == prev_word:
            repeat_ctr += 1
            if repeated_text:
                repeated_text += ' ' + word
            else:
                repeated_text = word + ' ' + word
        else:
            if repeat_ctr > max_repeats:
                new_text = ((prev_word + ' ') * max_repeats).strip()
                replacements[prev_word] = [repeated_text, new_text]
            repeat_ctr = 0
            repeated_text = ''
        prev_word = word

    if repeat_ctr > max_repeats:
        new_text = ((prev_word + ' ') * max_repeats).strip()
        replacements[prev_word] = [repeated_text, new_text]

    for word, item in replacements.items():
        text = text.replace(item[0], item[1])
    return text


def get_price_from_string(price_str: str) -> (str, str):
    """Returns the item price and currency
    """
    currencies = get_currencies()
    for symbol, name in currencies.items():
        if symbol in price_str:
            price = price_str.replace(symbol, '')
            if is_float(price):
                return price, name
        elif name in price_str:
            price = price_str.replace(name, '')
            if is_float(price):
                return price, name
    if is_float(price_str):
        return price_str, "EUR"
    return "0.00", "EUR"


def _words_similarity_histogram(words: []) -> {}:
    """Returns a histogram for word combinations
    """
    histogram = {}
    for index in range(1, len(words)):
        combined_words = words[index - 1] + words[index]
        if histogram.get(combined_words):
            histogram[combined_words] += 1
        else:
            histogram[combined_words] = 1
    return histogram


def _words_similarity_words_list(content: str) -> []:
    """Returns a list of words for the given content
    """
    remove_punctuation = ('.', ',', ';', '-', ':', '"')
    content = remove_html(content).lower()
    for punc in remove_punctuation:
        content = content.replace(punc, ' ')
        content = content.replace('  ', ' ')
    return content.split(' ')


def words_similarity(content1: str, content2: str, min_words: int) -> int:
    """Returns percentage similarity
    """
    if content1 == content2:
        return 100

    words1 = _words_similarity_words_list(content1)
    if len(words1) < min_words:
        return 0

    words2 = _words_similarity_words_list(content2)
    if len(words2) < min_words:
        return 0

    histogram1 = _words_similarity_histogram(words1)
    histogram2 = _words_similarity_histogram(words2)

    diff = 0
    for combined_words, _ in histogram1.items():
        if not histogram2.get(combined_words):
            diff += 1
        else:
            diff += \
                abs(histogram2[combined_words] - histogram1[combined_words])
    return 100 - int(diff * 100 / len(histogram1.items()))


def contains_invalid_local_links(content: str) -> bool:
    """Returns true if the given content has invalid links
    """
    for inv_str in INVALID_CONTENT_STRINGS:
        if '?' + inv_str + '=' in content:
            return True
    return False


def bold_reading_string(text: str) -> str:
    """Returns bold reading formatted text
    """
    text = html.unescape(text)
    add_paragraph_markup = False
    if '<p>' in text:
        text = text.replace('</p>', '\n').replace('<p>', '')
        add_paragraph_markup = True
    paragraphs = text.split('\n')
    parag_ctr = 0
    new_text = ''
    for parag in paragraphs:
        words = parag.split(' ')
        new_parag = ''
        reading_markup = False
        for wrd in words:
            if '<' in wrd:
                reading_markup = True
            if reading_markup and '>' in wrd:
                reading_markup = False
            wrd_len = len(wrd)
            if not reading_markup and wrd_len > 1 and \
               '<' not in wrd and '>' not in wrd and \
               '&' not in wrd and '=' not in wrd and \
               not wrd.startswith(':'):

                prefix = ''
                postfix = ''
                if wrd.startswith('"'):
                    prefix = '"'
                    wrd = wrd[1:]
                if wrd.endswith('"'):
                    postfix = '"'
                    wrd = wrd[:wrd_len - 1]

                initial_chars = int(math.ceil(wrd_len / 2.0))
                new_parag += \
                    prefix + '<b>' + wrd[:initial_chars] + '</b>' + \
                    wrd[initial_chars:] + postfix + ' '
            else:
                new_parag += wrd + ' '
        parag_ctr += 1
        new_parag = new_parag.strip()
        if not new_parag:
            continue
        if parag_ctr < len(paragraphs):
            if not add_paragraph_markup:
                new_text += new_parag + '\n'
            else:
                new_text += '<p>' + new_parag + '</p>'
        else:
            if not add_paragraph_markup:
                new_text += new_parag
            else:
                new_text += '<p>' + new_parag + '</p>'

    return new_text


def import_emoji(base_dir: str, import_filename: str, session) -> None:
    """Imports emoji from the given filename
    Each line should be [emoji url], :emojiname:
    """
    if not os.path.isfile(import_filename):
        return
    emoji_dict = load_json(base_dir + '/emoji/default_emoji.json', 0, 1)
    added = 0
    with open(import_filename, "r", encoding='utf-8') as fp_emoji:
        lines = fp_emoji.readlines()
        for line in lines:
            if ', ' not in line:
                continue
            url = line.split(', ')[0]
            tag = line.split(', ')[1].strip()
            if ':' not in tag:
                continue
            tag = tag.split(':')[1]
            if emoji_dict.get(tag):
                continue
            emoji_image_filename = base_dir + '/emoji/' + tag + '.png'
            if os.path.isfile(emoji_image_filename):
                continue
            if download_image(session, url,
                              emoji_image_filename, True, False):
                emoji_dict[tag] = tag
                added += 1
    save_json(emoji_dict, base_dir + '/emoji/default_emoji.json')
    print(str(added) + ' custom emoji added')


def content_diff(content: str, prev_content: str) -> str:
    """Returns a diff for the given content
    """
    cdiff = difflib.Differ()
    text1_lines = content.splitlines()
    text1_sentences = []
    for line in text1_lines:
        sentences = line.split('.')
        for sentence in sentences:
            text1_sentences.append(sentence.strip())

    text2_lines = prev_content.splitlines()
    text2_sentences = []
    for line in text2_lines:
        sentences = line.split('.')
        for sentence in sentences:
            text2_sentences.append(sentence.strip())

    diff = cdiff.compare(text1_sentences, text2_sentences)

    diff_text = ''
    for line in diff:
        if line.startswith('- '):
            if not diff_text:
                diff_text = '<p>'
            else:
                diff_text += '<br>'
            diff_text += '<label class="diff_add">+ ' + line[2:] + '</label>'
        elif line.startswith('+ '):
            if not diff_text:
                diff_text = '<p>'
            else:
                diff_text += '<br>'
            diff_text += \
                '<label class="diff_remove">- ' + line[2:] + '</label>'
    if diff_text:
        diff_text += '</p>'
    return diff_text


def create_edits_html(edits_json: {}, post_json_object: {},
                      translate: {}, timezone: str,
                      system_language: str,
                      languages_understood: []) -> str:
    """ Creates html showing historical edits made to a post
    """
    if not edits_json:
        return ''
    if not has_object_dict(post_json_object):
        return ''
    if 'content' not in post_json_object['object']:
        if 'contentMap' not in post_json_object['object']:
            return ''
    edit_dates_list = []
    for modified, _ in edits_json.items():
        edit_dates_list.append(modified)
    edit_dates_list.sort(reverse=True)
    edits_str = ''
    content = get_content_from_post(post_json_object, system_language,
                                    languages_understood, "content")
    if not content:
        return ''
    content = remove_html(content)
    for modified in edit_dates_list:
        prev_json = edits_json[modified]
        if not has_object_dict(prev_json):
            continue
        prev_content = get_content_from_post(prev_json, system_language,
                                             languages_understood, "content")
        if not prev_content:
            continue
        prev_content = remove_html(prev_content)
        if content == prev_content:
            continue
        diff = content_diff(content, prev_content)
        if not diff:
            continue
        diff = diff.replace('\n', '</p><p>')
        # convert to local time
        datetime_object = parse(modified)
        datetime_object = \
            convert_published_to_local_timezone(datetime_object, timezone)
        modified_str = datetime_object.strftime("%a %b %d, %H:%M")
        diff = '<p><b>' + modified_str + '</b></p>' + diff
        edits_str += diff
        content = prev_content
    if not edits_str:
        return ''
    return '<details><summary class="cw" tabindex="10">' + \
        translate['SHOW EDITS'] + '</summary>' + \
        edits_str + '</details>'


def remove_script(content: str, log_filename: str,
                  actor: str, url: str) -> str:
    """Removes <script> from some content
    """
    separators = [['<', '>'], ['&lt;', '&gt;']]
    for sep in separators:
        prefix = sep[0] + 'script'
        ending = '/script' + sep[1]
        if prefix not in content:
            continue
        sections = content.split(prefix)
        ctr = 0
        for text in sections:
            if ctr == 0:
                ctr += 1
                continue
            if ending not in text:
                if '/' + sep[1] not in text:
                    continue
            if ending in text:
                text = prefix + text.split(ending)[0] + ending
            else:
                text = prefix + text.split('/' + sep[1])[0] + '/' + sep[1]
                if log_filename and actor:
                    # write the detected script to a log file
                    log_str = actor + ' ' + url + ' ' + text + '\n'
                    write_type = 'a+'
                    if os.path.isfile(log_filename):
                        write_type = 'w+'
                    try:
                        with open(log_filename, write_type,
                                  encoding='utf-8') as fp_log:
                            fp_log.write(log_str)
                    except OSError:
                        print('EX: cannot append to svg script log')
            content = content.replace(text, '')
    return content


def reject_twitter_summary(base_dir: str, nickname: str, domain: str,
                           summary: str) -> bool:
    """Returns true if the post should be rejected due to twitter
    existing within the summary
    """
    if not summary:
        return False
    remove_twitter = \
        acct_dir(base_dir, nickname, domain) + '/.removeTwitter'
    if not os.path.isfile(remove_twitter):
        return False
    summary_lower = summary.lower()
    if 'twitter' in summary_lower or \
       'birdsite' in summary_lower:
        return True
    return False


def add_name_emojis_to_tags(base_dir: str, http_prefix: str,
                            domain: str, port: int,
                            actor_json: {}) -> None:
    """Add any custom emojis within the name of an actor to
    the tag list
    """
    if not actor_json.get('name'):
        return
    name = actor_json['name']

    # does the name contain an emoji?
    if ':' not in name:
        return
    if ':' not in name.split(':', 1)[1]:
        return

    # get emojis from the actor name
    words = name.split(' ')
    emojis = []
    for wrd in words:
        if wrd.startswith(':') and wrd.endswith(':'):
            if wrd not in emojis:
                emojis.append(wrd)
    if not emojis:
        return

    actor_tags = []
    if actor_json.get('tag'):
        actor_tags = actor_json['tag']

    # is the emoji already in the tag list?
    for tag_dict in actor_tags:
        if not tag_dict.get('type'):
            continue
        if tag_dict['type'] != 'Emoji':
            continue
        if not tag_dict.get('name'):
            continue
        if not tag_dict['name'].startswith(':'):
            continue
        if not tag_dict['name'].endswith(':'):
            continue
        if tag_dict['name'] in emojis:
            emojis.remove(tag_dict['name'])
    if not emojis:
        return

    domain_full = get_full_domain(domain, port)
    for emoji_tag_name in emojis:
        emoji_name = emoji_tag_name.replace(':', '')
        emoji_id = \
            http_prefix + '://' + domain_full + '/emoji/' + \
            emoji_name
        url = emoji_id + '.png'
        emoji_filename = base_dir + '/emoji/' + emoji_name + '.png'
        updated = None
        if os.path.isfile(emoji_filename):
            updated = file_last_modified(emoji_filename)
        new_tag = {
            'icon': {
                'mediaType': 'image/png',
                'type': 'Image',
                'url': url
            },
            'id': emoji_id,
            'name': emoji_tag_name,
            'type': 'Emoji',
            'updated': '2022-11-15T23:45:42Z'
        }
        if updated:
            new_tag['updated'] = updated
        actor_json['tag'].append(new_tag)


def format_mixed_right_to_left(content: str,
                               language: str) -> str:
    """Adds RTL direction formatting for non-RTL language
    eg. where some paragraphs are English and others are Arabic
    """
    # not a RTL language
    if language_right_to_left(language):
        return content
    result = ''
    changed = False
    paragraphs = content.split('<p>')
    for text_html in paragraphs:
        if '</p>' not in text_html:
            continue
        text_html = '<p>' + text_html
        text_plain = remove_html(text_html)
        if is_right_to_left_text(text_plain):
            text_html = text_html.replace('<p>', '<p><div dir="rtl">', 1)
            text_html = text_html.replace('</p>', '</div></p>', 1)
            changed = True
        result += text_html
    if not changed:
        result = ''
        prev_distilled = ''
        distilled = content
        while prev_distilled != distilled:
            prev_distilled = distilled
            distilled = distilled.replace('<br><br><br>', '<br><br>')
        paragraphs = distilled.split('<br><br>')
        ctr = 0
        for text_html in paragraphs:
            ctr += 1
            if ctr < len(paragraphs):
                text_html += '<br><br>'
            text_plain = remove_html(text_html)
            if is_right_to_left_text(text_plain):
                text_html = '<div dir="rtl">' + text_html
                if ctr < len(paragraphs):
                    text_html = \
                        text_html.replace('<br><br>', '</div><br><br>', 1)
                else:
                    text_html += '</div>'
                changed = True
            result += text_html
    if not changed:
        return content
    return result


def _load_auto_cw(base_dir: str, nickname: str, domain: str) -> []:
    """Loads automatic CWs file and returns a list containing
    the lines of the file
    """
    auto_cw_filename = acct_dir(base_dir, nickname, domain) + '/autocw.txt'
    if not os.path.isfile(auto_cw_filename):
        return []
    try:
        with open(auto_cw_filename, 'r', encoding='utf-8') as fp_auto:
            return fp_auto.read().split('\n')
    except OSError:
        print('EX: unable to load auto cw file ' + auto_cw_filename)
    return []


def load_auto_cw_cache(base_dir: str) -> {}:
    """Returns a dictionary containing the automatic content warning lists
    for each account
    """
    auto_cw_cache = {}
    for _, dirs, _ in os.walk(base_dir + '/accounts'):
        for handle in dirs:
            if not is_account_dir(handle):
                continue
            nickname = handle.split('@')[0]
            domain = handle.split('@')[1]
            auto_cw_cache[nickname] = _load_auto_cw(base_dir, nickname, domain)
        break
    return auto_cw_cache


def add_auto_cw(base_dir: str, nickname: str, domain: str,
                subject: str, content: str,
                auto_cw_cache: {}) -> str:
    """Appends any automatic content warnings to the subject line
    and returns the new subject line
    """
    new_subject = subject
    if auto_cw_cache.get(nickname):
        auto_cw_list = auto_cw_cache[nickname]
    else:
        auto_cw_list = _load_auto_cw(base_dir, nickname, domain)
        auto_cw_cache[nickname] = auto_cw_list
    for cw_rule in auto_cw_list:
        if '->' not in cw_rule:
            continue
        sections = cw_rule.split('->')
        rulematch = sections[0].strip()
        if rulematch not in content:
            continue
        cw_str = sections[1].strip()
        if not cw_str:
            continue
        if new_subject:
            if cw_str not in new_subject:
                new_subject += ', ' + cw_str
        else:
            new_subject = cw_str
    return new_subject