Create a new version of paste:

import os
import re
from cudatext import *
import cudax_lib as appx

FN_CONFIG = os.path.join(app_path(APP_DIR_SETTINGS), 'plugins.ini')
SECTION = 'complete_from_text'
NONWORDS_DEFAULT = '''-+*=/\()[]{}<>"'.,:;~?!@#$%^&|`…''' # without space/tab
nonwords = ''

def bool_to_str(v): return '1' if v else '0'
def str_to_bool(s): return s=='1'

# '-' here is none-lexer
option_lexers = ini_read(FN_CONFIG, SECTION, 'lexers', '-,ini files ^,markdown,restructuredtext,properties')
option_min_len = int(ini_read(FN_CONFIG, SECTION, 'min_len', '3'))
option_case_sens = str_to_bool(ini_read(FN_CONFIG, SECTION, 'case_sens', '1'))
#option_no_cmt = str_to_bool(ini_read(FN_CONFIG, SECTION, 'no_comments', '1'))
#option_no_str = str_to_bool(ini_read(FN_CONFIG, SECTION, 'no_strings', '0'))
option_what_editors = int(ini_read(FN_CONFIG, SECTION, 'what_editors', '0'))
option_max_lines = int(ini_read(FN_CONFIG, SECTION, 'max_lines', '10000'))
option_use_acp = str_to_bool(ini_read(FN_CONFIG, SECTION, 'use_acp', '1'))
option_show_acp_first = str_to_bool(ini_read(FN_CONFIG, SECTION, 'show_acp_first', '0'))
option_case_split = str_to_bool(ini_read(FN_CONFIG, SECTION, 'case_split', '0'))
option_underscore_split = str_to_bool(ini_read(FN_CONFIG, SECTION, 'underscore_split', '0'))
option_fuzzy = str_to_bool(ini_read(FN_CONFIG, SECTION, 'fuzzy_search', '1'))


def get_editors(ed, lexer):

    if option_what_editors==0:
        return [ed]
    elif option_what_editors==1:
        return [Editor(h) for h in ed_handles()]
    elif option_what_editors==2:
        res = []
        for h in ed_handles():
            e = Editor(h)
            if e.get_prop(PROP_LEXER_FILE)==lexer:
                res += [e]
        return res


def is_lexer_allowed(lex):

    return (',*,' in ','+option_lexers+',') or (','+(lex or '-').lower()+',' in ','+option_lexers.lower()+',')


def isword(s):

    return s not in ' \t'+nonwords


def is_plain_match(stext, sfind):

    if option_case_sens:
        return stext.startswith(sfind)
    else:
        return stext.upper().startswith(sfind.upper())


def is_fuzzy_match(stext, sfind):
    '''Ported from CudaText's Pascal code''' 
    stext = stext.upper()
    sfind = sfind.upper()
    n = -1
    for ch in sfind:
        n = stext.find(ch, n+1)
        if n<0:
            return False
    return True


def is_text_with_begin(s, begin):

    if option_fuzzy:
        return is_fuzzy_match(s, begin)

    if option_case_split or option_underscore_split:
        if option_case_split  and  s[0] == begin[0]:
            searchwords = re.findall('.[^A-Z]*', begin)
            wordwords = re.findall('.[^A-Z]*', s)

            swl = len(searchwords)
            wwl = len(wordwords)
            if swl <= wwl:
                for i in range(swl):
                    if not wordwords[i].startswith(searchwords[i]):
                        break
                else:
                    return True

        if option_underscore_split:
            ss = s.strip('_') # ignore starting,ending underscores
            if '_' in ss:

                if not option_case_sens:
                    begin = begin.lower()
                    ss = ss.lower()

                wordwords = re.findall('[^_]+', ss) # PROP_LEX => [PROP, LEX]
                wwl = len(wordwords)
                bl = len(begin)

                if bl <= wwl: # most common case, first chars of each word: PLF|PL|P => PROP_LEXER_FILE
                    for i in range(bl):
                        if begin[i] != wordwords[i][0]:
                            break
                    else:
                        return True

                if spl_match(begin, wordwords):
                    return True

    if option_case_sens:
        return s.startswith(begin)
    else:
        return s.upper().startswith(begin.upper())


def spl_match(begin, words):
    '''returns True if 'begin' fits consecutively into 'words' (prefixes of 'words')'''

    word = words[0]
    common_len = 0
    for i in range(min(len(begin), len(word))):
        if begin[i] == word[i]:
            common_len = i+1
        else:
            break

    if common_len > 0:
        if len(begin) == common_len: # last match success
            return True
        elif len(words) == 1: # last word - should be full prefix
            return False

        for i in range(common_len): # i: 0 and 1 for common_len=2 ->
            # ... on calling spl_match() 'begin' will always be non-empty - less than full match
            res = spl_match(begin[common_len-i:], words[1:])
            if res:
                return True

    return False


def escape_regex(s):

    res = ''
    for ch in s:
        if ch in '$^#-+|.?*()[]{}':
            res += '\\'
        res += ch
    return res

def get_regex(begin, nonwords):

    w_content = r'\w'
    for ch in NONWORDS_DEFAULT:
        if ch not in nonwords:
            w_content += escape_regex(ch)

    repeats = max(1, option_min_len-len(begin))
    regex = r'\$?\b' + escape_regex(begin) + '[' + w_content + ']{' + str(repeats) + ',}'

    #print('begin:', repr(begin))
    #print('regex:', repr(regex))
    return regex

def get_words_list(ed, regex):

    if ed.get_line_count() > option_max_lines:
        return []

    '''
    if option_no_cmt and option_no_str:
        ops = 'T6'
    elif option_no_cmt:
        ops = 'T4'
    elif option_no_str:
        ops = 'T5'
    else:
        ops = ''

    res = ed.action(EDACTION_FIND_ALL, regex, 'r'+ops) or []
    l = [ed.get_text_substr(*r) for r in res]
    # slower than via python regex, 3 times
    '''

    flags = re.I if not option_case_sens else 0
    l = re.findall(regex, ed.get_text_all(), flags)
    l = list(set(l))

    return l


def get_word(ed, x, y):

    if not 0<=y<ed.get_line_count():
        return
    s = ed.get_text_line(y)
    if not 0<x<=len(s):
        return

    x0 = x
    while (x0>0) and isword(s[x0-1]):
        x0-=1
    text1 = s[x0:x]

    x0 = x
    while (x0<len(s)) and isword(s[x0]):
        x0+=1
    text2 = s[x:x0]

    return (text1, text2)


def _read_acp_file(sfile):

    with open(sfile, 'r', encoding='utf-8', errors='ignore') as f:
        lines = f.readlines()
    return lines
    
def get_acp_words(ed, word1, word2):

    lex = ed.get_prop(PROP_LEXER_CARET, '')
    sfile = os.path.join(app_path(APP_DIR_DATA), 'autocomplete', lex+'.acp')
    if os.path.isfile(sfile):
        acp_lines = _read_acp_file(sfile)
    else:
        sfile = os.path.join(app_path(APP_DIR_DATA), 'autocomplete', lex+'_.acp')
        if os.path.isfile(sfile):
            acp_lines = _read_acp_file(sfile)
        else:
            return [], set()

    target_word = word1+word2

    if len(acp_lines) > 0  and  acp_lines[0].startswith('#chars'):
        del acp_lines[0]

    acp_words_plain = []
    acp_words_fuzzy = []
    words = set()
    fstr = '{}|{}|{}'+chr(9)+'{}'
    for line in acp_lines:
        line = line.rstrip()
        if not line:
            continue

        # using '#chars'
        #m = re.match('^([^\s]+)\s([\w'+ word_chars +']+)([^|]*)\|?(.*)?$', line) # ^Prefix Word Args Descr?
        m = re.match('^([^\s]+)\s([^(| ]+)([^|]*)\|?(.*)?$', line) # ^Prefix Word Args Descr?
        if not m:
            continue

        # avoid 'm[index]' to support old Python 3.4
        pre, word, args, descr = m.group(1), m.group(2).rstrip(), m.group(3), m.group(4) or ''

        if len(word) < option_min_len:
            continue
        if is_text_with_begin(word, word1)  and  word != word1  and  word != target_word:
            word = word.replace('%20', ' ')
            s = fstr.format(pre, word, args, descr)
            if is_plain_match(word, word1):
                acp_words_plain.append(s)
            else:
                acp_words_fuzzy.append(s)
            words.add(word)

    acp_words = acp_words_plain + acp_words_fuzzy
    return acp_words, words


def get_completions(ed_self, x0, y0, with_acp, ignore_lexer=False):

    lex = ed_self.get_prop(PROP_LEXER_FILE, '')
    if lex is None: return
    if not ignore_lexer and not is_lexer_allowed(lex): return

    global nonwords
    nonwords = appx.get_opt(
        'nonword_chars',
        NONWORDS_DEFAULT,
        appx.CONFIG_LEV_ALL,
        ed_self,
        lex)

    word = get_word(ed_self, x0, y0)

    if not word: return
    word1, word2 = word
    if not word1: return # to fix https://github.com/Alexey-T/CudaText/issues/3175
    if word1[0].isdigit(): return

    # filter by word1[0], not by entire word1, because fuzzy must be supported
    regex = get_regex(word1[0], nonwords)

    words_by_tabs = []
    tab_titles = []
    for e in get_editors(ed_self, lex):
        words_by_tabs.append(get_words_list(e, regex))
        title = e.get_prop(PROP_TAB_TITLE).replace('|', '/')
        tab_titles.append(title)
    #print('words_by_tabs:', words_by_tabs)

    words = []
    for w in words_by_tabs:
        words += w

    def search_tab(w):
        if option_what_editors in [1, 2]:
            tabs_found = []
            for i, words_by_tabs_ in enumerate(words_by_tabs):
                if w in words_by_tabs_:
                    tabs_found.append(i)
            if tabs_found:
                return '|' + '; '.join([tab_titles[tf] for tf in tabs_found])
            return ''
        else:
            return ''

    #exclude numbers
    #words = [w for w in words if not w.isdigit()]

    if words:
        words = sorted(list(set(words)))

    acp_words, acp_set = get_acp_words(ed_self, word1, word2) if with_acp else ([], set())

    if not words and not acp_words:
        return

    def get_prefix(w):
        return 'var' if w.startswith('$') else 'text'

    word1and2 = word1+word2
    words = [w for w in words
             if w not in acp_set # do not repeat words from .acp
             and w!=word1
             and w!=word1and2
             ]
    words = [w for w in words
             if is_text_with_begin(w, word1)
             ]

    words_plain = []
    words_fuzzy = []
    for w in words:
        if is_plain_match(w, word1):
            words_plain.append(w)
        else:
            words_fuzzy.append(w)

    words = words_plain + words_fuzzy

    words_decorated = [get_prefix(w)+'|'+w+search_tab(w) for w in words]
    return (words_decorated, acp_words, word1, word2)


class Command:
    
    def on_complete(self, ed_self):

        carets = ed_self.get_carets()
        if len(carets)!=1: return # don't allow multi-carets
        x0, y0, x1, y1 = carets[0]
        if y1>=0: return # don't allow selection
        
        res = get_completions(ed_self, x0, y0, option_use_acp)
        if res is None: return
        words, acp_words, word1, word2 = res
        word_list = acp_words+words if option_show_acp_first else words+acp_words 
        ed_self.complete('\n'.join(word_list), len(word1), len(word2))
        return True

    def config(self):

        ini_write(FN_CONFIG, SECTION, 'lexers', option_lexers)
        ini_write(FN_CONFIG, SECTION, 'min_len', str(option_min_len))
        ini_write(FN_CONFIG, SECTION, 'case_sens', bool_to_str(option_case_sens))
        #ini_write(FN_CONFIG, SECTION, 'no_comments', bool_to_str(option_no_cmt))
        #ini_write(FN_CONFIG, SECTION, 'no_strings', bool_to_str(option_no_str))
        ini_write(FN_CONFIG, SECTION, 'what_editors', str(option_what_editors))
        ini_write(FN_CONFIG, SECTION, 'max_lines', str(option_max_lines))
        ini_write(FN_CONFIG, SECTION, 'use_acp', bool_to_str(option_use_acp))
        ini_write(FN_CONFIG, SECTION, 'show_acp_first', bool_to_str(option_show_acp_first))
        ini_write(FN_CONFIG, SECTION, 'fuzzy_search', bool_to_str(option_fuzzy))
        ini_write(FN_CONFIG, SECTION, 'case_split', bool_to_str(option_case_split))
        ini_write(FN_CONFIG, SECTION, 'underscore_split', bool_to_str(option_underscore_split))
        file_open(FN_CONFIG)

        lines = [ed.get_text_line(i) for i in range(ed.get_line_count())]
        try:
            index = lines.index('['+SECTION+']')
            ed.set_caret(0, index)
        except:
            pass