improve KKTIX guess answer feature.

master
Max 2023-02-05 00:23:01 +08:00
parent b8667e56f5
commit 107e814c8a
1 changed files with 156 additions and 32 deletions

View File

@ -473,38 +473,101 @@ def convert_string_to_pattern(my_str, dynamic_length=True):
return my_formated
def guess_answer_list_from_multi_options(tmp_text):
show_debug_message = True # debug.
show_debug_message = False # online
options_list = None
matched_pattern = ""
if options_list is None:
if u'' in tmp_text and u'' in tmp_text:
options_list = re.findall(u'【.*?】', tmp_text)
pattern = '【.*?】'
options_list = re.findall(pattern, tmp_text)
if len(options_list) <= 2:
options_list = None
else:
matched_pattern = pattern
if options_list is None:
if u'(' in tmp_text and u')' in tmp_text:
options_list = re.findall(u'\(.*?\)', tmp_text)
pattern = '\(.*?\)'
options_list = re.findall(pattern, tmp_text)
if len(options_list) <= 2:
options_list = None
else:
matched_pattern = pattern
if options_list is None:
if u'[' in tmp_text and u']' in tmp_text:
options_list = re.findall(u'[.*?]', tmp_text)
pattern = '\[.*?\]'
options_list = re.findall(pattern, tmp_text)
if len(options_list) <= 2:
options_list = None
else:
matched_pattern = pattern
if show_debug_message:
print("matched pattern:", matched_pattern)
# default remove quota
is_trim_quota = not check_answer_keep_symbol(tmp_text)
if show_debug_message:
print("is_trim_quota:", is_trim_quota)
return_list = None
if not options_list is None:
options_list_length = len(options_list)
if show_debug_message:
print("options_list_length:", options_list_length)
print("options_list:", options_list)
if options_list_length > 2:
is_all_options_same_length = True
options_length_count = {}
for i in range(options_list_length-1):
if len(options_list[i]) != len(options_list[i]):
current_option_length = len(options_list[i])
next_option_length = len(options_list[i+1])
if current_option_length != next_option_length:
is_all_options_same_length = False
if current_option_length in options_length_count:
options_length_count[current_option_length] += 1
else:
options_length_count[current_option_length] = 1
if show_debug_message:
print("is_all_options_same_length:", is_all_options_same_length)
if is_all_options_same_length:
return_list = []
for each_option in options_list:
return_list.append(each_option[1:-1])
if is_trim_quota:
return_list.append(each_option[1:-1])
else:
return_list.append(each_option)
else:
#print("options_length_count:", options_length_count)
if len(options_length_count) > 0:
target_option_length = 0
most_length_count = 0
for k in options_length_count.keys():
if options_length_count[k] > most_length_count:
most_length_count = options_length_count[k]
target_option_length = k
#print("most_length_count:", most_length_count)
#print("target_option_length:", target_option_length)
if target_option_length > 0:
return_list = []
for each_option in options_list:
current_option_length = len(each_option)
if current_option_length == target_option_length:
if is_trim_quota:
return_list.append(each_option[1:-1])
else:
return_list.append(each_option)
# something is wrong, give up those options.
if not return_list is None:
if len(return_list) <= 2:
return_list = None
return return_list
#PS: this may get a wrong answer list. XD
@ -610,23 +673,13 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
# ps: hint_list is not options list
if offical_hint_string == "":
offical_hint_string = get_offical_hint_string_from_symbol(CONST_EXAMPLE_SYMBOL, tmp_text)
if len(offical_hint_string) > 0:
right_part = offical_hint_string.split(CONST_EXAMPLE_SYMBOL)[1]
if len(offical_hint_string) == len(tmp_text):
offical_hint_string = right_part
new_hint = find_continuous_text(right_part)
if len(new_hint) > 0:
offical_hint_string_anwser = new_hint
if offical_hint_string == "":
# for: 若你覺得答案為 a請輸入 a
if '答案' in tmp_text and CONST_INPUT_SYMBOL in tmp_text:
offical_hint_string = get_offical_hint_string_from_symbol(CONST_INPUT_SYMBOL, tmp_text)
if len(offical_hint_string) > 0:
right_part = offical_hint_string.split(CONST_INPUT_SYMBOL)[1]
#print("right_part:", right_part)
if len(offical_hint_string) == len(tmp_text):
offical_hint_string = right_part
@ -636,6 +689,20 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
#if u'答案' in offical_hint_string and CONST_INPUT_SYMBOL in offical_hint_string:
offical_hint_string_anwser = new_hint
if offical_hint_string == "":
offical_hint_string = get_offical_hint_string_from_symbol(CONST_EXAMPLE_SYMBOL, tmp_text)
if len(offical_hint_string) > 0:
right_part = offical_hint_string.split(CONST_EXAMPLE_SYMBOL)[1]
if len(offical_hint_string) == len(tmp_text):
offical_hint_string = right_part
# PS: find first text will only get B char in this case: 答案為B需填入Bb)
new_hint = find_continuous_text(right_part)
if len(new_hint) > 0:
offical_hint_string_anwser = new_hint
if show_debug_message:
print("offical_hint_string:",offical_hint_string)
@ -650,7 +717,7 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
# is need to merge next block
if len(offical_hint_string) > 0:
target_symbol = offical_hint_string + u" "
target_symbol = offical_hint_string + " "
if target_symbol in tmp_text :
star_index = tmp_text.find(target_symbol)
next_block_index = star_index + len(target_symbol)
@ -723,16 +790,14 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
offical_hint_string = tmp_text[star_index: space_index]
if len(offical_hint_string) > 0:
#print("offical_hint_string_anwser:", offical_hint_string_anwser)
if show_debug_message:
print("offical_hint_string_anwser:", offical_hint_string_anwser)
my_anwser_formated = convert_string_to_pattern(offical_hint_string_anwser)
my_options = tmp_text
if len(my_question) < len(tmp_text):
my_options = my_options.replace(my_question,u"")
my_options = my_options.replace(offical_hint_string,u"")
#print("tmp_text:", tmp_text)
#print("my_options:", my_options)
#print("offical_hint_string:", offical_hint_string)
# try rule7:
# check is chinese/english in question, if match, apply my_options rule.
@ -745,7 +810,9 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
star_index = tmp_text_org.find(target_symbol)
my_options = tmp_text_org[star_index-1:]
#print(u"my_options:", my_options)
if show_debug_message:
print("tmp_text:", tmp_text)
print("my_options:", my_options)
if len(my_anwser_formated) > 0:
allow_delimitor_symbols = ")].: }"
@ -757,20 +824,41 @@ def guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
maybe_delimitor = my_options[span_end+0:span_end+1]
if maybe_delimitor in allow_delimitor_symbols:
my_answer_delimitor = maybe_delimitor
#print(u"my_answer_delimitor:", my_answer_delimitor)
if show_debug_message:
print(u"my_answer_delimitor:", my_answer_delimitor)
# default remove quota
is_trim_quota = not check_answer_keep_symbol(tmp_text)
if show_debug_message:
print("is_trim_quota:", is_trim_quota)
if len(my_anwser_formated) > 0:
#print("text:" , re.findall('\([\w]+\)', tmp_text))
new_pattern = my_anwser_formated
if len(my_answer_delimitor) > 0:
new_pattern = my_anwser_formated + u'\\' + my_answer_delimitor
return_list = re.findall(new_pattern, my_options)
if show_debug_message:
print("my_anwser_formated:", my_anwser_formated)
print("new_pattern:", new_pattern)
print("return_list:" , return_list)
if not return_list is None:
if len(return_list) == 1:
# re-sample for this case.
return_list = re.findall(my_anwser_formated, my_options)
# clean delimitor
if is_trim_quota:
return_list_length = len(return_list)
if return_list_length >= 1:
if len(my_answer_delimitor) > 0:
for idx in range(return_list_length):
return_list[idx]=return_list[idx].replace(my_answer_delimitor,'')
if show_debug_message:
print("cleaned return_list:" , return_list)
return return_list
def format_question_string(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captcha_text_div_text):
@ -815,6 +903,9 @@ def format_question_string(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captcha_tex
return tmp_text
def get_answer_list_by_question(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captcha_text_div_text):
show_debug_message = True # debug.
show_debug_message = False # online
return_list = None
tmp_text = format_question_string(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captcha_text_div_text)
@ -822,14 +913,21 @@ def get_answer_list_by_question(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captch
# guess answer list from multi-options: 【】() []
if return_list is None:
return_list = guess_answer_list_from_multi_options(tmp_text)
pass
if show_debug_message:
if not return_list is None:
print("guess_answer_list_from_multi_options:", return_list)
if return_list is None:
return_list = guess_answer_list_from_hint(CONST_EXAMPLE_SYMBOL, CONST_INPUT_SYMBOL, captcha_text_div_text)
if show_debug_message:
if not return_list is None:
print("guess_answer_list_from_hint:", return_list)
# try rule8:
if return_list is None:
return_list = guess_answer_list_from_symbols(captcha_text_div_text)
if show_debug_message:
if not return_list is None:
print("guess_answer_list_from_symbols:", return_list)
return return_list
@ -2614,6 +2712,22 @@ def check_answer_keep_symbol(captcha_text_div_text):
if u'符號須相同' in keep_symbol_tmp:
is_need_keep_symbol = True
# for: 大小寫含括號需一模一樣
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'還有', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'以及', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'必須', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'而且', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'', '')
keep_symbol_tmp = keep_symbol_tmp.replace(u'一模', '')
#print("keep_symbol_tmp:", keep_symbol_tmp)
if '大小寫括號相同' in keep_symbol_tmp:
is_need_keep_symbol = True
return is_need_keep_symbol
def get_answer_list_from_question_string(registrationsNewApp_div, captcha_text_div_text):
@ -7200,18 +7314,28 @@ if __name__ == "__main__":
main()
else:
#for test kktix infer answer.
#captcha_text_div_text = u"請回答下列問題,請在下方空格輸入DELIGHT請以半形輸入法作答大小寫需要一模一樣"
captcha_text_div_text = u"請回答下列問題,請在下方空格輸入DELIGHT請以半形輸入法作答大小寫需要一模一樣"
#captcha_text_div_text = u"請在下方空白處輸入引號內文字「abc」"
#captcha_text_div_text = u"請在下方空白處輸入引號內文字「0118eveconcert」請以半形小寫作答。"
#captcha_text_div_text = "在《DEEP AWAKENING見過深淵的人》專輯中哪一首為合唱曲目 【V6】深淵 、【Z5】浮木、【J8】無聲、【C1】以上皆非 (請以半形輸入法作答,大小寫/阿拉伯數字需要一模一樣範例A2"
captcha_text_div_text = "Super Junior 的隊長是以下哪位? 【v】神童 【w】藝聲 【x】利特 【y】始源 若你覺得答案為 a請輸入 a (英文為半形小寫)"
#captcha_text_div_text = "Super Junior 的隊長是以下哪位? 【v】神童 【w】藝聲 【x】利特 【y】始源 若你覺得答案為 a請輸入 a (英文為半形小寫)"
#captcha_text_div_text = "請問XXX, 請以英文為半形小寫(例如a) a. 1月5日 b. 2月5日 c. 3月5日 d. 4月5日"
#captcha_text_div_text = "以下哪個「不是」正確的林俊傑與其他藝人合唱的歌曲組合?(選項為歌名/合作藝人 ,請以半形輸入法作答選項,大小寫需要一模一樣,範例:jju 選項: (jja)小酒窩/A-Sa蔡卓妍 (jjb)被風吹過的夏天/金莎 (jjc)友人說/張懷秋 (jjd)全面開戰/五月天阿信 (jje)小說/阿杜"
#captcha_text_div_text = "請問《龍的傳人2060》演唱會是以下哪位藝人的演出請以半形輸入法作答大小寫需要一模一樣範例B2A1.周杰倫 B2.林俊傑 C3.張學友 D4.王力宏"
#captcha_text_div_text = "王力宏何時發行第一張專輯?(請以半形輸入法作答,大小寫需要一模一樣,範例:B2 A1.1985 B2.2005 C3.2015 D4.1995"
#captcha_text_div_text = "朴寶劍三月以歌手出道的日期和單曲名為? Answer the singles name & the debut date. *以半形輸入,大小寫/符號須都相同。例:(E1) Please use the same format given in the options.ex:(E1) (A1)20/Bloomin'(B1)2/Blossom(C1)2/Bloomin'(D1)20/Blossom"
#captcha_text_div_text = "以下哪位不是LOVELYZ成員? (請以半形輸入選項內的英文及數字,大小寫須符合),範例:E5e。 (A1a)智愛 (B2b)美珠 (C3c)JON (D4d)叡仁"
#captcha_text_div_text = "題請問此次 RAVI的SOLO專輯名稱為?(請以半形輸入法作答,大小寫需要一模一樣,範例:Tt Aa [ BOOK] 、 Bb [OOK BOOK.R] 、 Cc [R.OOK BOOK] 、 Dd [OOK R. BOOK]"
captcha_text_div_text = "請問下列哪個選項皆為河成雲的創作歌曲? Aa) Dont Forget、Candle Bb) Dont Forget、Forever+1 Cc) Dont Forget、Flowerbomb Dd) Dont Forget、One Love 請以半形輸入,大小寫含括號需一模一樣 【範例:答案為B需填入Bb)】"
inferred_answer_string, answer_list = get_answer_list_from_question_string(None, captcha_text_div_text)
print("inferred_answer_string:", inferred_answer_string)
print("answer_list:", answer_list)
ocr = ddddocr.DdddOcr(show_ad=False, beta=True)
with open('captcha-xxxx.png', 'rb') as f:
image_bytes = f.read()
res = ocr.classification(image_bytes)
print(res)
image_file = 'captcha-xxxx.png'
if os.path.exists(image_file):
with open(image_file, 'rb') as f:
image_bytes = f.read()
res = ocr.classification(image_bytes)
print(res)