BIDI_ML_INFO_EXTRACTION/BiddingKG/dl/interface/html_2_kvtree.py

#coding:utf8
from bs4 import BeautifulSoup
import json
import re
import traceback

import logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')

logger = logging.getLogger(__name__)
logger.setLevel(logging.INFO)
from BiddingKG.dl.interface.Preprocessing import tableToText
from uuid import uuid4

def log(msg):
    '''
    @summary:打印信息
    '''
    logger.info(msg)

class DotDict(dict):

    def __getattr__(self,name):
        try:
            return self[name]
        except KeyError:
            raise AttributeError("No attribute '%s'" % name)

    def __setattr__(self,name,value):
        self[name] = value



def get_tables(soup,dict_table = None):
    is_first = False
    if dict_table is None:
        dict_table = {"children":[]}
        is_first = True
    if soup and soup.name:
        childs = soup.contents

    else:
        childs = []

    # tr+tbody
    _flag = False
    if len(childs)>=2:
        if childs[0].name=="tr" and childs[1].name=="tbody":
            childs[1].insert(0,copy.copy(childs[0]))
            childs[0].decompose()
            _flag = True

    childs_bak = childs
    # tbody+tbody
    _flag = False
    if soup and soup.name:
        childs = soup.find_all("tbody",recursive=False)
        if len(childs)>=2:
            if childs[0].name=="tbody" and childs[1].name=="tbody":
                child0_tr = childs[0].find_all("tr",recursive=False)
                has_td_count = 0
                tr_line = None
                for tr in child0_tr:
                    if len(tr.find_all("td",recursive=False))>0:
                        has_td_count += 1
                        tr_line = tr
                if has_td_count==1:
                    childs[1].insert(0,copy.copy(tr_line))
                    childs[0].decompose()
                    _flag = True

    childs = childs_bak
    for child in childs:
        _d = {"children":[]}
        if child.name in ("table","tbody"):
            if len(child.find_all("tr",recursive=False))>0:
                # _d["table"] = str(child)
                _d["table"] = child
        dict_table["children"].append(_d)
        child_dict_table = get_tables(child,_d)

    if is_first:
        if soup.name in ("table","tbody"):
            if not _flag:
                if len(soup.find_all("tr",recursive=False))>0:
                    # dict_table["table"] = str(soup)
                    dict_table["table"] = soup

        dict_table = squeeze_tables(dict_table)

    return dict_table
def squeeze_tables(dict_table):
    _i = -1
    new_children = []
    for child in dict_table["children"]:
        _i += 1
        child_table = squeeze_tables(child)

        if child_table is not None:
            new_children.append(child_table)

    if dict_table.get("table") is not None:
        if len(new_children)>0:
            dict_table["children"] = new_children
        else:
            del dict_table["children"]
        return dict_table
    if len(new_children)==1:
        return new_children[0]
    if len(new_children)>1:
        dict_table["children"] = new_children
        return dict_table

    return None





def table_to_tree(soup,json_obj=None):


    if json_obj is None:
        json_obj = DotDict({"tag": "table","children":[]})


    dict_table = get_tables(soup)

    children = dict_table.get("children",[])
    for child in children:
        _d = DotDict({"tag": "table","children":[]})
        json_obj["children"].append(_d)
        table = child.get("table")
        if table is not None:
            table_id = str(uuid4())
            table_to_tree(table,_d)

    table = dict_table.get("table")
    if table is not None:
        table_id = str(uuid4())
        json_obj["table_id"] = table_id
        soup, kv_list, text = tableToText(table,return_kv=True)
        _flag = False
        if soup and soup.name:
            if soup.contents:
                _flag = True
                soup.contents[0].insert_before(table_id)
        if not _flag:
            soup.insert_before(table_id)
        json_obj["text"] = text
        json_obj["kv"] = kv_list
        for _d in kv_list:
            _d["position"] = {"key_begin_sentence":0,
                              "key_begin_sentence_start":_d.get("key_sen_index",0),
                              "key_end_sentence":0,
                              "key_end_sentence_end":_d.get("key_sen_index",0)+len(_d.get("key","")),
                              "value_begin_sentence":0,
                              "value_begin_sentence_start":_d.get("value_sen_index",0),
                              "value_end_sentence":0,
                              "value_end_sentence_end":_d.get("value_sen_index",0)+len(_d.get("value",""))
                              }
            if "key_sen_index" in _d:
                _d.pop("key_sen_index")
            if "value_sen_index" in _d:
                _d.pop("value_sen_index")
    return json_obj


def update_table_position(table,sentence_index):


    def get_table_idx_lengths(list_table_id,index):
        _length = 0
        for _d in list_table_id:
            table_id = _d.get("table_id")
            idx = _d.get("idx",-1)

            if idx>=0 and _idx<=index:
                _length += len(table_id)
        return _length

    def get_sentence_index(list_sent_span,idx):
        list_sent_span.sort(key=lambda x:x[0])
        for _i in range(len(list_sent_span)):
            if list_sent_span[_i][0]<=idx and idx<=list_sent_span[_i][1]:
                return _i
        return 0

    def get_list_tables(table,list_table=[]):

        table_id = table.get("table_id")
        if table_id:
            list_table.append(table)
        childs = table.get("children",[])
        for child in childs:
            get_list_tables(child,list_table)
        return list_table

    tables = get_list_tables(table)
    if tables:

        list_table_id = []
        text = tables[0].get("text","")

        for table in tables:
            table_id = table.get("table_id")

            if table_id:
                _idx = text.find(table_id)
                list_table_id.append({"table_id":table_id,"idx":_idx})
                if _idx>=0:
                    kv_list = table.get("kv",[])
                    for _d in kv_list:
                        _d["position"]["key_begin_sentence_start"] += _idx
                        _d["position"]["key_end_sentence_end"] += _idx
                        _d["position"]["value_begin_sentence_start"] += _idx
                        _d["position"]["value_end_sentence_end"] += _idx

        # remove table_id
        for table in tables:
            table_id = table.get("table_id")

            if table_id:
                kv_list = table.get("kv",[])
                for _d in kv_list:
                    _length = get_table_idx_lengths(list_table_id,_d["position"]["key_begin_sentence_start"])
                    _d["position"]["key_begin_sentence_start"] -= _length
                    _length = get_table_idx_lengths(list_table_id,_d["position"]["key_end_sentence_end"])
                    _d["position"]["key_end_sentence_end"] -= _length
                    _length = get_table_idx_lengths(list_table_id,_d["position"]["value_begin_sentence_start"])
                    _d["position"]["value_begin_sentence_start"] -= _length
                    _length = get_table_idx_lengths(list_table_id,_d["position"]["value_end_sentence_end"])
                    _d["position"]["value_end_sentence_end"] -= _length

        for table in tables:
            if table.get("table_id"):
                text = table.get("text","")
                for _d in list_table_id:
                    table_id = _d.get("table_id")
                    text = text.replace(table_id,"")
                table["text"] = text

        # split sentence
        text = tables[0].get("text","")
        list_sentence = str(text).split("。")
        list_sent_span = []
        _begin = 0
        for _i in range(len(list_sentence)):
            list_sentence[_i] += "。"
            _end = _begin+len(list_sentence[_i])
            list_sent_span.append([_begin,_end])
            _begin = _end
        tables[0]["sentences"] = list_sentence

        for table in tables:

            kv_list = table.get("kv",[])
            for _d in kv_list:
                key_begin_sentence = get_sentence_index(list_sent_span,_d["position"]["key_begin_sentence_start"])
                _d["position"]["key_begin_sentence"] = key_begin_sentence+sentence_index
                key_end_sentence = get_sentence_index(list_sent_span,_d["position"]["key_end_sentence_end"])
                _d["position"]["key_end_sentence"] = key_end_sentence+sentence_index
                value_begin_sentence = get_sentence_index(list_sent_span,_d["position"]["value_begin_sentence_start"])
                _d["position"]["value_begin_sentence"] = value_begin_sentence+sentence_index
                value_end_sentence = get_sentence_index(list_sent_span,_d["position"]["value_end_sentence_end"])
                _d["position"]["value_end_sentence"] = value_end_sentence+sentence_index

        return sentence_index + len(list_sentence)
    return sentence_index












def tree_reposition(tree,sentence_index=None):
    if sentence_index is None:
        sentence_index = 0

    wordOffset_begin = 0
    wordOffset_end = 0
    for obj in tree:
        is_table = True if obj.get("tag","")=="table" else False
        if not is_table:
            sentence_index += 1
            obj["sentence_index"] = sentence_index
            obj["sentences"] = [obj.get("text","")]
            for _t in obj["sentences"]:
                wordOffset_end += len(_t)
            obj["wordOffset_begin"] = wordOffset_begin
            obj["wordOffset_end"] = wordOffset_end
            wordOffset_begin = wordOffset_end

            

            list_kv = obj.get("kv",[])
            for _d in list_kv:
                _d["position"]["key_begin_sentence"] = sentence_index
                _d["position"]["key_end_sentence"] = sentence_index
                _d["position"]["value_begin_sentence"] = sentence_index
                _d["position"]["value_end_sentence"] = sentence_index

        else:
            sentence_index += 1
            obj["sentence_index"] = sentence_index
            obj["sentence_index_start"] = sentence_index
            obj["sentences"] = [obj.get("text","")]
            sentence_index_end = update_table_position(obj,sentence_index)
            obj["sentence_index_end"] = sentence_index_end
            sentence_index = sentence_index_end

            for _t in obj["sentences"]:
                wordOffset_end += len(_t)
            obj["wordOffset_begin"] = wordOffset_begin
            obj["wordOffset_end"] = wordOffset_end
            wordOffset_begin = wordOffset_end




# 递归地将 DOM 转换为 JSON
def dom_to_tree(node):
    if node.name:  # 如果是标签节点
        json_obj = DotDict({"tag": node.name})
        if node.attrs:
            json_obj["attributes"] = node.attrs

        is_table = False
        if node.name in ("table","tbody"):
            json_obj = table_to_tree(node)
            is_table = True

        if not is_table:
            children = []
            for child in node.contents:
                _child = dom_to_tree(child)
                if _child is not None:
                    children.append(_child)
            if children:
                json_obj["children"] = children
        json_obj["name"] = json_obj.get("tag")
        return json_obj
    elif node.string and node.string.strip():  # 如果是纯文本节点
        _text = node.string.strip()
        _text = re.sub('\xa0','',_text)
        list_text = re.split("\s",_text)
        _text = ""
        for _t in list_text:
            if len(_t)<3:
                if len(_t)>0:
                    _text += _t
            else:
                _text += _t+" "
        _text = _text.strip()
        return DotDict({"tag":"text","name":"text","text": _text})
    return None  # 忽略空白字符

def tree_pop_parent(tree):

    if isinstance(tree,list):
        for child in tree:
            tree_pop_parent(child)
    if isinstance(tree,dict):
        if "parent" in tree:
            del tree["parent"]
        for child in tree.get("children",[]):
            tree_pop_parent(child)


def html_to_tree(html_content):
    # 使用 BeautifulSoup 解析 HTML
    soup = BeautifulSoup(html_content, "lxml")
    dom_tree = dom_to_tree(soup)
    extract_kv_from_tree(dom_tree)
    list_objs = get_outobjs_from_tree(dom_tree)
    tree_reposition(list_objs)
    return dom_tree

def print_tree(dom_tree):
    # 转换为 JSON 格式
    tree_pop_parent(dom_tree)
    json_output = json.dumps(dom_tree,ensure_ascii=False, indent=2)

# kv_pattern = "\s*(?P<key>.{,10})[:：]\s*(?P<value>[^:：。，()]+?)(\s+|$|;|；)(?![\u4e00-\u9fa5]+：)"
kv_pattern = r"(?P<key>[\u4e00-\u9fa5]+)：\s*(?P<value>[^\s，。();；]+)"

def get_kv_pattern():
    import re

    text = """
    name: John age: 30 note: invalid;
    """

    # 正则模式
    kv_pattern = r"(?P<key>[a-zA-Z]+)[:：](?P<value>.+(?!.*[:：]))"

    # 提取匹配
    matches = re.findall(kv_pattern, text)

    # 打印结果
    for match in matches:
        key, value = match
        print("{%s}: {%s}"%(key,value))

def extract_kv_from_sentence(sentence):
    list_kv = []
    _iter = re.finditer("[:：]", sentence)
    if _iter:
        list_span = []
        for iter in _iter:
            list_span.append(iter.span())
        if len(list_span)==1:
            _begin,_end = list_span[0]
            if _begin<20 and _end<len(sentence)-1:
                _d = DotDict({"key":sentence[0:_begin],"value":sentence[_end:]})
                _d["position"] = {"key_begin_sentence":0,
                              "key_begin_sentence_start":0,
                              "key_end_sentence":0,
                              "key_end_sentence_end":_begin,
                              "value_begin_sentence":0,
                              "value_begin_sentence_start":_end,
                              "value_end_sentence":0,
                              "value_end_sentence_end":len(sentence)
                              }
                list_kv.append(_d)
            else:
                _begin = 0
                _end = len(sentence)-1
                iter = re.search(kv_pattern,sentence[_begin:_end])
                if iter is not None:
                    _d = DotDict({})

                    _d["key"] = iter.group("key")
                    _d["value"] = iter.group("value")

                    _d["position"] = {"key_begin_sentence":0,
                              "key_begin_sentence_start":iter.span("key")[0],
                              "key_end_sentence":0,
                              "key_end_sentence_end":iter.span("key")[0]+len(_d.get("key","")),
                              "value_begin_sentence":0,
                              "value_begin_sentence_start":iter.span("value")[0],
                              "value_end_sentence":0,
                              "value_end_sentence_end":iter.span("value")[0]+len(_d.get("value",""))
                              }
                    list_kv.append(_d)

        elif len(list_span)>1:

            _begin,_end = list_span[0]
            if _begin<20 and len(sentence)>100:
                _d = DotDict({"key":sentence[0:_begin],"value":sentence[_end:]})
                _d["position"] = {"key_begin_sentence":0,
                              "key_begin_sentence_start":0,
                              "key_end_sentence":0,
                              "key_end_sentence_end":_begin,
                              "value_begin_sentence":0,
                              "value_begin_sentence_start":_end,
                              "value_end_sentence":0,
                              "value_end_sentence_end":len(sentence)
                              }
                list_kv.append(_d)

            else:
                _begin = 0

                for _i in range(len(list_span)-1):
                    _end = list_span[_i+1][0]
                    iter = re.search(kv_pattern,sentence[_begin:_end])
                    _begin = list_span[_i][1]
                    if iter is not None:
                        _d = DotDict({})
                        _d["key"] = iter.group("key")
                        _d["value"] = iter.group("value")

                        _d["position"] = {"key_begin_sentence":0,
                                  "key_begin_sentence_start":iter.span("key")[0],
                                  "key_end_sentence":0,
                                  "key_end_sentence_end":iter.span("key")[0]+len(_d.get("key","")),
                                  "value_begin_sentence":0,
                                  "value_begin_sentence_start":iter.span("value")[0],
                                  "value_end_sentence":0,
                                  "value_end_sentence_end":iter.span("value")[0]+len(_d.get("value",""))
                                  }
                        list_kv.append(_d)

                _begin = list_span[-2][1]
                _end = len(sentence)
                iter = re.search(kv_pattern,sentence[_begin:_end])
                if iter is not None:
                    _d = DotDict({})
                    _d["key"] = iter.group("key")
                    _d["value"] = iter.group("value")

                    _d["position"] = {"key_begin_sentence":0,
                                  "key_begin_sentence_start":iter.span("key")[0],
                                  "key_end_sentence":0,
                                  "key_end_sentence_end":iter.span("key")[0]+len(_d.get("key","")),
                                  "value_begin_sentence":0,
                                  "value_begin_sentence_start":iter.span("value")[0],
                                  "value_end_sentence":0,
                                  "value_end_sentence_end":iter.span("value")[0]+len(_d.get("value",""))
                                  }
                    list_kv.append(_d)




        # for iter in _iter:
        #     _d = DotDict({})
        #     _d["key"] = iter.group("key")
        #     _d["value"] = iter.group("value")
        #     _d["key_span"] = iter.span("key")
        #     _d["value_span"] = iter.span("value")
        #     list_kv.append(_d)
    return list_kv


def extract_kv_from_node(node):
    list_kv = []
    list_text = []
    childs = node.get("children",[])
    _text = ""

    has_br = False
    if childs:
        for child in childs:
            node_name = child.get("tag","")
            child_text = child.get("text")
            if node_name=="br":
                list_text.append([])
                has_br = True
            if child_text:
                if len(list_text)==0:
                    list_text.append([])
                list_text[-1].append(child)

        node["kv"] = []
        if has_br:
            new_children = []
            for texts in list_text:
                if texts:
                    _text = "".join([a.get("text") for a in texts])
                    tag = texts[0]
                    list_kv = extract_kv_from_sentence(_text)
                    _n = DotDict({"tag":tag,"name":tag,"text":_text,"children":[],"kv":list_kv})
                    new_children.append(_n)
            node["children"] = new_children
        else:
            for texts in list_text:
                _text = "".join([a.get("text") for a in texts])
                if _text:
                    list_kv = extract_kv_from_sentence(_text)
                node["kv"].extend(list_kv)
    else:
        _text = node.get("text")
        if _text:
            list_kv = extract_kv_from_sentence(_text)
        node["kv"] = list_kv
    return list_kv


def get_child_text(node):
    _text = node.get("text","")
    for child in node.get("children",[]):
        _text += get_child_text(child)
    return _text
def extract_kv_from_tree(tree):
    if isinstance(tree,list):
        _count = 0
        has_table = False
        for child in tree:
            _c,_t = extract_kv_from_tree(child)
            _count += _c
            if _t:
                has_table = _t
        return _count,has_table
    if isinstance(tree,dict):
        if tree.get("tag","")!="table":
            childs = tree.get("children",[])

            if len(childs)>0:
                _count = 0
                has_table = False
                child_has_p_div = False
                child_has_br = False
                for child in childs:
                    _c,_t = extract_kv_from_tree(child)
                    _count += _c
                    if _t:
                        has_table = _t
                    if child.get("tag","") in ("p","div","li"):
                        child_has_p_div = True
                    if child.get("tag","")=="br":
                        child_has_br = True
                if _count==0:
                    if not has_table and not child_has_p_div and not child_has_br:
                        _text = get_child_text(tree)
                        if "children" in tree:
                            del tree["children"]
                        tree["text"] = _text
                        list_kv = extract_kv_from_node(tree)
                        _count = len(list_kv)
                    return _count,has_table
                if tree.get("tag","") in ("p","div","li") and not has_table and not child_has_p_div:
                    if not child_has_br:
                        _text = get_child_text(tree)
                        tree["text"] = _text
                        if "children" in tree:
                            del tree["children"]
                    p_list_kv = extract_kv_from_node(tree)
                    return len(p_list_kv),has_table

                return _count,has_table
            else:
                list_kv = extract_kv_from_node(tree)
                return len(list_kv),False
        else:
            return len(tree.get("kv",[])),True
    return 0,False

def update_kv_span(list_kv,append_length):
    for _d in list_kv:
        _d["position"] = {"key_begin_sentence":0,
                              "key_begin_sentence_start":_d.get("key_sen_index",0),
                              "key_end_sentence":0,
                              "key_end_sentence_end":_d.get("key_sen_index",0)+len(_d.get("key","")),
                              "value_begin_sentence":0,
                              "value_begin_sentence_start":_d.get("value_sen_index",0),
                              "value_end_sentence":0,
                              "value_end_sentence_end":_d.get("value_sen_index",0)+len(_d.get("value",""))
                              }
        _d["position"]["key_begin_sentence_start"] += append_length
        _d["position"]["key_end_sentence_end"] += append_length
        _d["position"]["value_begin_sentence_start"] += append_length
        _d["position"]["value_end_sentence_end"] += append_length

def get_outobjs_from_tree(tree,list_outobjs=None):

    is_first = False
    if list_outobjs is None:
        list_outobjs = []
        is_first = True
    if isinstance(tree,list):
        for child in tree:
            get_outobjs_from_tree(child,list_outobjs)
    if isinstance(tree,dict):
        childs = tree.get("children",[])
        _text = tree.get("text","")
        is_table = True if tree.get("tag","")=="table" else False
        if is_table:
            list_outobjs.append(tree)
        else:
            if _text!="":
                tree.name = tree.tag
                list_outobjs.append(tree)
            for child in childs:
                get_outobjs_from_tree(child,list_outobjs)

    return list_outobjs


def standard_title_context(_title_context):
    return _title_context.replace("（","(").replace("）",")").replace("：",":").replace("：",";").replace("，",".").replace(",",".").replace("、",".").replace("．",".")

def standard_product(sentence):
    return sentence.replace("（","(").replace("）",")")


import Levenshtein
import copy
def jaccard_score(source,target):
    source_set = set([s for s in source])
    target_set = set([s for s in target])
    if len(source_set)==0 or len(target_set)==0:
        return 0
    return max(len(source_set&target_set)/len(source_set),len(source_set&target_set)/len(target_set))


def judge_pur_chinese(keyword):
    """
    中文字符的编码范围为： u'\u4e00' -- u'\u9fff：只要在此范围内就可以判断为中文字符串
    @param keyword:
    @return:
    """
    # 定义一个需要删除的标点符号字符串列表
    remove_chars = '[·’!"\#$%&\'()＃！（）*+,-./:;<=>?\@，：?￥★、…．＞【】［］《》？“”‘’\[\\]^_`{|}~]+'
    # 利用re.sub来删除中文字符串中的标点符号
    strings = re.sub(remove_chars, "", keyword)  # 将keyword中文字符串中remove_chars中包含的标点符号替换为空字符串
    for ch in strings:
        if u'\u4e00' <= ch <= u'\u9fff':
            pass
        else:
            return False
    return True
def is_similar(source,target,_radio=None):
    source = str(source).lower()
    target = str(target).lower()
    max_len = max(len(source),len(target))
    min_len = min(len(source),len(target))

    min_ratio = 90
    if min_len>=3:
        min_ratio = 87
    if min_len>=5:
        min_ratio = 85
    if _radio is not None:
        min_ratio = _radio
    # dis_len = abs(len(source)-len(target))
    # min_dis = min(max_len*0.2,4)
    if min_len==0 and max_len>0:
        return False
    if max_len<=2:
        if source==target:
            return True
    if min_len<2:
        return False
    #判断相似度
    similar = Levenshtein.ratio(source,target)*100
    if similar>=min_ratio:
        log("%s and %s similar_jaro %d"%(source,target,similar))
        return True
    similar_jaro = Levenshtein.jaro(source,target)
    if similar_jaro*100>=min_ratio:
        log("%s and %s similar_jaro %d"%(source,target,similar_jaro*100))
        return True
    similar_jarow = Levenshtein.jaro_winkler(source,target)
    if similar_jarow*100>=min_ratio:
        log("%s and %s similar_jaro %d"%(source,target,similar_jarow*100))
        return True

    if min_len>=5:
        if len(source)==max_len and str(source).find(target)>=0:
                return True
        elif len(target)==max_len and target.find(source)>=0:
                return True
        elif jaccard_score(source, target)==1 and judge_pur_chinese(source) and judge_pur_chinese(target):
            return True
    return False


end_pattern = "商务要求|评分标准|商务条件|商务条件"
_param_pattern = "(产品|技术|清单|配置|参数|具体|明细|项目|招标|货物|服务|规格|工作|具体)[及和与]?(指标|配置|条件|要求|参数|需求|规格|条款|名称及要求)|配置清单|(质量|技术).{,10}要求|验收标准|^(参数|功能)$"
meter_pattern = "[><≤≥±]\d+|\d+(?:[μucmkK微毫千]?[米升LlgGmMΩ]|摄氏度|英寸|度|天|VA|dB|bpm|rpm|kPa|mol|cmH20|%|°|Mpa|Hz|K?HZ|℃|W|min|[*×xX])|[*×xX]\d+|/min|\ds[^a-zA-Z]|GB.{,20}标准|PVC|PP|角度|容积|色彩|自动|流量|外径|轴位|折射率|帧率|柱镜|振幅|磁场|镜片|防漏|强度|允差|心率|倍数|瞳距|底座|色泽|噪音|间距|材质|材料|表面|频率|阻抗|浓度|兼容|防尘|防水|内径|实时|一次性|误差|性能|距离|精确|温度|超温|范围|跟踪|对比度|亮度|[横纵]向|均压|负压|正压|可调|设定值|功能|检测|高度|厚度|宽度|深度|[单双多]通道|效果|指数|模式|尺寸|重量|峰值|谷值|容量|寿命|稳定性|高温|信号|电源|电流|转换率|效率|释放量|转速|离心力|向心力|弯曲|电压|功率|气量|国标|标准协议|灵敏度|最大值|最小值|耐磨|波形|高压|性强|工艺|光源|低压|压力|压强|速度|湿度|重量|毛重|[MLX大中小]+码|净重|颜色|[红橙黄绿青蓝紫]色|不锈钢|输入|输出|噪声|认证|配置"
not_meter_pattern = "投标报价|中标金额|商务部分|公章|分值构成|业绩|详见|联系人|联系电话|合同价|金额|采购预算|资金来源|费用|质疑|评审因素|评审标准|商务资信|商务评分|专家论证意见|评标方法|代理服务费|售后服务|评分类型|评分项目|预算金额|得\d+分|项目金额|详见招标文件|乙方"


def getTrs(tbody):
    #获取所有的tr
    trs = []
    if tbody.name=="table":
        body = tbody.find("tbody",recursive=False)
        if body is not None:
            tbody = body
    objs = tbody.find_all(recursive=False)
    for obj in objs:
        if obj.name=="tr":
            trs.append(obj)
        if obj.name=="tbody" or obj.name=="table":
            for tr in obj.find_all("tr",recursive=False):
                trs.append(tr)
    return trs

def fixSpan(tbody):
    # 处理colspan, rowspan信息补全问题
    #trs = tbody.findChildren('tr', recursive=False)

    trs = getTrs(tbody)
    ths_len = 0
    ths = list()
    trs_set = set()
    #修改为先进行列补全再进行行补全，否则可能会出现表格解析混乱
    # 遍历每一个tr

    for indtr, tr in enumerate(trs):
        ths_tmp = tr.findChildren('th', recursive=False)
        #不补全含有表格的tr
        if len(tr.findChildren('table'))>0:
            continue
        if len(ths_tmp) > 0:
            ths_len = ths_len + len(ths_tmp)
            for th in ths_tmp:
                ths.append(th)
            trs_set.add(tr)
        # 遍历每行中的element
        tds = tr.findChildren(recursive=False)
        for indtd, td in enumerate(tds):
            # 若有colspan 则补全同一行下一个位置
            if 'colspan' in td.attrs:
                if str(re.sub("[^0-9]","",str(td['colspan'])))!="":
                    col = int(re.sub("[^0-9]","",str(td['colspan'])))
                    if col<100 and len(td.get_text())<1000:
                        td['colspan'] = 1
                        for i in range(1, col, 1):
                            td.insert_after(copy.copy(td))

    for indtr, tr in enumerate(trs):
        ths_tmp = tr.findChildren('th', recursive=False)
        #不补全含有表格的tr
        if len(tr.findChildren('table'))>0:
            continue
        if len(ths_tmp) > 0:
            ths_len = ths_len + len(ths_tmp)
            for th in ths_tmp:
                ths.append(th)
            trs_set.add(tr)
        # 遍历每行中的element
        tds = tr.findChildren(recursive=False)
        for indtd, td in enumerate(tds):
            # 若有rowspan 则补全下一行同样位置
            if 'rowspan' in td.attrs:
                if str(re.sub("[^0-9]","",str(td['rowspan'])))!="":
                    row = int(re.sub("[^0-9]","",str(td['rowspan'])))
                    td['rowspan'] = 1
                    for i in range(1, row, 1):
                        # 获取下一行的所有td， 在对应的位置插入
                        if indtr+i<len(trs):
                            tds1 = trs[indtr + i].findChildren(['td','th'], recursive=False)
                            if len(tds1) >= (indtd) and len(tds1)>0:
                                if indtd > 0:
                                    tds1[indtd - 1].insert_after(copy.copy(td))
                                else:
                                    tds1[0].insert_before(copy.copy(td))
                            elif indtd-2>0 and len(tds1) > 0 and len(tds1) == indtd - 1:  # 修正某些表格最后一列没补全
                                tds1[indtd-2].insert_after(copy.copy(td))
def getTable(tbody):
    #trs = tbody.findChildren('tr', recursive=False)
    fixSpan(tbody)
    trs = getTrs(tbody)
    inner_table = []
    for tr in trs:
        tr_line = []
        tds = tr.findChildren(['td','th'], recursive=False)
        if len(tds)==0:
            tr_line.append([re.sub('\xa0','',tr.get_text()),0]) # 2021/12/21 修复部分表格没有td 造成数据丢失
        for td in tds:
            tr_line.append([re.sub('\xa0','',td.get_text()),0])
            #tr_line.append([td.get_text(),0])
        inner_table.append(tr_line)
    return inner_table

def extract_products(list_data,_product,_param_pattern = "产品名称|设备材料|采购内存|标的名称|采购内容|(标的|维修|系统|报价构成|商品|产品|物料|物资|货物|设备|采购品|采购条目|物品|材料|印刷品?|采购|物装|配件|资产|耗材|清单|器材|仪器|器械|备件|拍卖物|标的物|物件|药品|药材|药械|货品|食品|食材|品目|^品名|气体|标项|分项|项目|计划|包组|标段|[分子]?包|子目|服务|招标|中标|成交|工程|招标内容)[\)）的]?([、\w]{,4}名称|内容|描述)|标的|标项|项目$|商品|产品|物料|物资|货物|设备|采购品|采购条目|物品|材料|印刷品|物装|配件|资产|招标内容|耗材|清单|器材|仪器|器械|备件|拍卖物|标的物|物件|药品|药材|药械|货品|食品|食材|菜名|^品目$|^品名$|^名称|^内容$"):
    _product = standard_product(_product)
    list_result = []
    list_table_products = []
    for _data_i in range(len(list_data)):
        _data = list_data[_data_i]
        _type = _data["type"]
        _text = _data["text"]

        if _type=="table":
            list_table = _data["list_table"]
            if list_table is None:
                continue
            _check = True
            max_length = max([len(a) for a in list_table])
            min_length = min([len(a) for a in list_table])
            if min_length<max_length/2:
                continue
            list_head_index = []
            _begin_index = 0
            head_cell_text = ""
            for line_i in range(len(list_table[:2])):
                line = list_table[line_i]
                line_text = ",".join([cell[0] for cell in line])
                for cell_i in range(len(line)):
                    cell = line[cell_i]
                    cell_text = cell[0]
                    if len(cell_text)<10 and re.search(_param_pattern,cell_text) is not None and re.search("单价|数量|预算|限价|总价|品牌|规格|型号|用途|要求|采购量",line_text) is not None:
                        _begin_index = line_i+1
                        list_head_index.append(cell_i)

            for line_i in range(len(list_table)):
                line = list_table[line_i]
                for cell_i in list_head_index:
                    if cell_i>=len(line):
                        continue
                    cell = line[cell_i]
                    cell_text = cell[0]
                    head_cell_text += cell_text

            # print("===head_cell_text",head_cell_text)
            if re.search("招标人|采购人|项目编号|项目名称|金额|^\d+$",head_cell_text) is not None:
                list_head_index = []

            for line in list_table:
                line_text = ",".join([cell[0] for cell in line])
                for cell_i in range(len(line)):
                    cell = line[cell_i]
                    cell_text = cell[0]
                    if cell_text is not None and _product is not None and len(cell_text)<len(_product)*10 and cell_text.find(_product)>=0 and re.search("单价|数量|总价|规格|品牌|型号|用途|要求|采购量",line_text) is not None:
                        list_head_index.append(cell_i)

            list_head_index = list(set(list_head_index))
            if len(list_head_index)>0:
                has_number = False
                for cell_i in list_head_index:
                    table_products = []

                    for line_i in range(_begin_index,len(list_table)):
                        line = list_table[line_i]

                        for _i in range(len(line)):
                            cell = line[_i]
                            cell_text = cell[0]
                            if re.search("^\d+$",cell_text) is not None:
                                has_number = True

                        if cell_i>=len(line):
                            continue
                        cell = line[cell_i]
                        cell_text = cell[0]
                        if re.search(_param_pattern,cell_text) is None or has_number:
                            if re.search("^[\da-zA-Z]+$",cell_text) is None:
                                table_products.append(cell_text)

                    if len(table_products)>0:
                        logger.debug("table products %s"%(str(table_products)))
                        if min([len(x) for x in table_products])>0 and max([len(x) for x in table_products])<=30:
                            if re.search("招标人|代理人|预算|数量|交货期|品牌|产地","".join(table_products)) is None:
                                list_table_products.append(table_products)
    _find = False
    for table_products in list_table_products:
        for _p in table_products:
            if is_similar(_product,_p,90):
                _find = True
                logger.debug("similar table_products %s"%(str(table_products)))
                list_result = list(set([a for a in table_products if len(a)>1 and len(a)<20 and re.search("费用|预算|合计|金额|万元|运费|^其他$",a) is None]))
                break
    if not _find:
        for table_products in list_table_products:
            list_result.extend(table_products)
        list_result = list(set([a for a in list_result if len(a)>1 and len(a)<30 and re.search("费用|预算|合计|金额|万元|运费",a) is None]))
    return list_result

def get_childs(childs, max_depth=None):
    list_data = []
    for _child in childs:
        list_data.append(_child)
        childs2 = _child.get("child_title",[])

        if len(childs2)>0 and (max_depth==None or max_depth>0):
            for _child2 in childs2:
                if max_depth != None:
                    list_data.extend(get_childs([_child2], max_depth-1))
                else:
                    list_data.extend(get_childs([_child2], None))
    return list_data

class Html2KVTree():

    def __init__(self,_html,auto_merge_table=True,list_obj = []):
        if _html is None:
            _html = ""
        self.html = _html
        self.auto_merge_table = auto_merge_table

        if list_obj:
            self.list_obj = list_obj
        else:

            _tree = html_to_tree(html_content)
            self.list_obj = get_outobjs_from_tree(_tree)



        # for obj in self.list_obj:
        #     print("obj",obj.get_text()[:20])

        self.tree = self.buildParsetree(self.list_obj,[],auto_merge_table)


        # #识别目录树
        # self.print_tree(self.tree,"-|")

    def get_soup_objs(self,soup,list_obj=None):
        if list_obj is None:
            list_obj = []
        childs = soup.find_all(recursive=False)
        for _obj in childs:
            childs1 = _obj.find_all(recursive=False)
            if len(childs1)==0 or len(_obj.get_text())<40 or _obj.name=="table":
                list_obj.append(_obj)
            elif _obj.name=="p":
                list_obj.append(_obj)
            else:
                self.get_soup_objs(_obj,list_obj)
        return list_obj

    def fix_tree(self,_product):
        products = extract_products(self.tree,_product)
        if len(products)>0:
            self.tree = self.buildParsetree(self.list_obj,products,self.auto_merge_table)

    def print_tree(self,tree,append="",set_tree_id=None):
        if set_tree_id is None:
            set_tree_id = set()
        if append=="":
            for t in tree:
                logger.debug("%s text:%s title:%s title_text:%s before:%s after%s product:%s"%("==>",t["text"][:50],t["sentence_title"],t["sentence_title_text"],t["title_before"],t["title_after"],t["has_product"]))

        for t in tree:
            _id = id(t)
            if _id in set_tree_id:
                continue
            set_tree_id.add(_id)
            logger.info("%s text:%s title:%s title_text:%s before:%s after%s product:%s kv:%s"%(append,t["text"][:50],t["sentence_title"],t["sentence_title_text"],t["title_before"],t["title_after"],t["has_product"],str(t["kv"])))
            childs = t["child_title"]
            self.print_tree(childs,append=append+"-|",set_tree_id=set_tree_id)

    def is_title_first(self,title):
        if title in ("一","1","Ⅰ","a","A"):
            return True
        return False

    def find_title_by_pattern(self,_text,_pattern="(^|★|▲|:|：|\s+)(?P<title_1>(?P<title_1_index_0_0>第?)(?P<title_1_index_1_1>[一二三四五六七八九十ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]+)(?P<title_1_index_2_0>[、章册包标部.:：]+))|" \
                                             "([\s★▲\*]*)(?P<title_3>(?P<title_3_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?)(?P<title_3_index_0_1>[ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]+)(?P<title_3_index_0_2>[、章册包标部.:：]+))|" \
                                             "([\s★▲\*]*)(?P<title_4>(?P<title_4_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?第?)(?P<title_4_index_1_1>[一二三四五六七八九十]+)(?P<title_4_index_2_0>[节章册部\.:：、、]+))|" \
                                             "([\s★▲\*]*)(?P<title_5>(?P<title_5_index_0_0>^)(?P<title_5_index_1_1>[一二三四五六七八九十]+)(?P<title_5_index_2_0>)[^一二三四五六七八九十节章册部\.:：、])|" \
                                             "([\s★▲\*]*)(?P<title_12>(?P<title_12_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-])(?P<title_12_index_1_1>\d{1,2})(?P<title_12_index_2_0>[\.．、\s\-]?))|"\
                                             "([\s★▲\*]*)(?P<title_11>(?P<title_11_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-])(?P<title_11_index_1_1>\d{1,2})(?P<title_11_index_2_0>[\.．、\s\-]?))|" \
                                             "([\s★▲\*]*)(?P<title_10>(?P<title_10_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?\d{1,2}[\.．、\s\-]\d{1,2}[\.．、\s\-])(?P<title_10_index_1_1>\d{1,2})(?P<title_10_index_2_0>[\.．、\s\-]?))|" \
                                             "([\s★▲\*]*)(?P<title_7>(?P<title_7_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?\d{1,2}[\.．\s\-])(?P<title_7_index_1_1>\d{1,2})(?P<title_7_index_2_0>[\.．包标:：、\s\-]*))|" \
                                             "(^[\s★▲\*]*)(?P<title_6>(?P<title_6_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?包?)(?P<title_6_index_0_1>\d{1,2})(?P<title_6_index_2_0>[\.．、\s\-包标]*))|" \
                                             "([\s★▲\*]*)(?P<title_15>(?P<title_15_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?[(（]?)(?P<title_15_index_1_1>\d{1,2})(?P<title_15_index_2_0>[)）包标\.．:：、]+))|" \
                                             "([\s★▲\*]+)(?P<title_17>(?P<title_17_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?[(（]?)(?P<title_17_index_1_1>[a-zA-Z]+)(?P<title_17_index_2_0>[)）包标\.．:：、]+))|" \
                                             "([\s★▲\*]*)(?P<title_19>(?P<title_19_index_0_0>[^一二三四五六七八九十\dⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]{,3}?[(（]?)(?P<title_19_index_1_1>[一二三四五六七八九十ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]+)(?P<title_19_index_2_0>[)）]))"
                              ):
        _se = re.search(_pattern,_text)
        groups = []
        if _se is not None:
            e = _se.end()
            if re.search('(时间|日期|编号|账号|号码|手机|价格|\w价|人民币|金额|得分|分值|总分|满分|最高得|扣|减|数量|评委)[:：]?\d', _se.group(0)) or (re.search('\d[.:：]?$', _se.group(0)) and re.search('^[\d年月日万元天个分秒台条A-Za-z]|^(小时)', _text[e:])):
                return None
            elif re.match('[二三四五六七八九十]\w{1,2}[市区县]|五金|四川|八疆|九龙|[一二三四五六七八九十][层天标包]', _text) and re.match('[一二三四五六七八九十]', _se.group(0)): # 289765335 排除三明市等开头作为大纲
                return None
            elif re.search('^[\u4e00-\u9fa5]+[:：]', _text[:e]):
                return None
            _gd = _se.groupdict()
            for k,v in _gd.items():
                if v is not None:
                    groups.append((k,v))
        if len(groups):
            groups.sort(key=lambda x:x[0])
            return groups
        return None

    def make_increase(self,_sort,_title,_add=1):
        if len(_title)==0 and _add==0:
            return ""
        if len(_title)==0 and _add==1:
            return _sort[0]
        _index = _sort.index(_title[-1])
        next_index = (_index+_add)%len(_sort)
        next_chr = _sort[next_index]
        if _index==len(_sort)-1:
            _add = 1
        else:
            _add = 0
        return next_chr+self.make_increase(_sort,_title[:-1],_add)


    def get_next_title(self,_title):
        if re.search("^\d+$",_title) is not None:
            return str(int(_title)+1)
        if re.search("^[一二三四五六七八九十百]+$",_title) is not None:
            if _title[-1]=="十":
                return _title+"一"
            if _title[-1]=="百":
                return _title+"零一"

            if _title[-1]=="九":
                if len(_title)==1:
                    return "十"
                if len(_title)==2:
                    if _title[0]=="十":
                        return "二十"
                if len(_title)==3:
                    if _title[0]=="九":
                        return "一百"
                    else:
                        _next_title = self.make_increase(['一','二','三','四','五','六','七','八','九','十'],re.sub("[十百]",'',_title[0]))
                        return _next_title+"十"

            _next_title = self.make_increase(['一','二','三','四','五','六','七','八','九','十'],re.sub("[十百]",'',_title))
            _next_title = list(_next_title)
            _next_title.reverse()
            if _next_title[-1]!="十":
                if len(_next_title)>=2:
                    _next_title.insert(-1,'十')
            if len(_next_title)>=4:
                _next_title.insert(-3,'百')
            if _title[0]=="十":
                if _next_title=="十":
                    _next_title = ["二","十"]
                _next_title.insert(0,"十")
            _next_title = "".join(_next_title)
            return _next_title
        if re.search("^[a-z]+$",_title) is not None:
            _next_title = self.make_increase([chr(i+ord('a')) for i in range(26)],_title)
            _next_title = list(_next_title)
            _next_title.reverse()
            return "".join(_next_title)
        if re.search("^[A-Z]+$",_title) is not None:
            _next_title = self.make_increase([chr(i+ord('A')) for i in range(26)],_title)
            _next_title = list(_next_title)
            _next_title.reverse()
            return "".join(_next_title)
        if re.search("^[ⅠⅡⅢⅣⅤⅥⅦⅧⅨⅩⅪⅫ]$",_title) is not None:
            _sort = ["Ⅰ","Ⅱ","Ⅲ","Ⅳ","Ⅴ","Ⅵ","Ⅶ","Ⅷ","Ⅸ","Ⅹ","Ⅺ","Ⅻ"]
            _index = _sort.index(_title)
            if _index<len(_sort)-1:
                return _sort[_index+1]
            return None

    def count_title_before(self,list_obj):
        dict_before = {}
        dict_sentence_count = {}
        illegal_sentence = set()
        for obj_i in range(len(list_obj)):
            obj = list_obj[obj_i]
            _type = "sentence"
            _text = obj.text.strip()
            if obj.name=="table":
                _type = "table"
                _text = str(obj)
            _append = False


            if _type=="sentence":
                if len(_text)>10 and len(_text)<100:
                    if _text not in dict_sentence_count:
                        dict_sentence_count[_text] = 0
                    dict_sentence_count[_text] += 1
                    if re.search("\d+页",_text) is not None:
                        illegal_sentence.add(_text)
                elif len(_text)<10:
                    if re.search("第\d+页",_text) is not None:
                        illegal_sentence.add(_text)

                sentence_groups = self.find_title_by_pattern(_text[:10])
                if sentence_groups:
                    # c062f53cf83401e671822003d63c1828print("sentence_groups",sentence_groups)
                    sentence_title = sentence_groups[0][0]
                    sentence_title_text = sentence_groups[0][1]
                    title_index = sentence_groups[-2][1]
                    title_before = sentence_groups[1][1].replace("（","(").replace("：",":").replace("：",";").replace("，",".").replace(",",".").replace("、",".")
                    title_after = sentence_groups[-1][1].replace("）",")").replace("：",":").replace("：",";").replace("，",".").replace(",",".").replace("、",".")
                    next_index = self.get_next_title(title_index)
                    if title_before not in dict_before:
                        dict_before[title_before] = 0
                    dict_before[title_before] += 1

        for k,v in dict_sentence_count.items():
            if v>10:
                illegal_sentence.add(k)
        return dict_before,illegal_sentence

    def is_page_no(self,sentence):
        if len(sentence)<10:
            if re.search("\d+页|^\-\d+\-$",sentence) is not None:
                return True

    def block_tree(self,childs):
        for child in childs:

            if not child["block"]:
                child["block"] = True
                childs2 = child["child_title"]
                self.block_tree(childs2)


    def buildParsetree(self,list_obj,products=[],auto_merge_table=True,auto_append=False):

        self.parseTree = None
        trees = []
        list_length = []
        for obj in list_obj[:200]:
            if obj.name!="table":
                list_length.append(len(obj.text))
        if len(list_length)>0:
            max_length = max(list_length)
        else:
            max_length = 40
        max_length = min(max_length,40)

        logger.debug("%s:%d"%("max_length",max_length))


        list_data = []
        last_table_index = None
        last_table_columns = None
        last_table = None
        dict_before,illegal_sentence = self.count_title_before(list_obj)
        for obj_i in range(len(list_obj)):
            obj = list_obj[obj_i]

            # logger.debug("==obj %s"%obj.text[:20])

            _type = "sentence"
            _text = standard_product(obj.text)
            if obj.name=="table":
                _type = "table"
                _text = standard_product(str(obj))
            _append = False
            sentence_title = None
            sentence_title_text = None
            sentence_groups = None
            title_index = None
            next_index = None
            parent_title = None
            title_before = None
            title_after = None
            title_next = None
            childs = []
            # new
            sentence_index = obj.sentence_index
            wordOffset_begin = obj.wordOffset_begin
            wordOffset_end = obj.wordOffset_end
            sentences = obj.sentences

            list_kv = obj.get("kv",[])

            table_id = obj.get("table_id")

            list_table = None
            block = False

            has_product = False
            position = obj.get("position",{})

            if _type=="sentence":
                if _text in illegal_sentence:
                    continue


                sentence_groups = self.find_title_by_pattern(_text[:10])
                if sentence_groups:
                    title_before = standard_title_context(sentence_groups[1][1])
                    title_after = sentence_groups[-1][1]
                    sentence_title_text = sentence_groups[0][1]
                    other_text = _text.replace(sentence_title_text,"")
                    if (title_before in dict_before and dict_before[title_before]>1) or title_after!="":
                        sentence_title = sentence_groups[0][0]

                        title_index = sentence_groups[-2][1]
                        next_index = self.get_next_title(title_index)

                        other_text = _text.replace(sentence_title_text,"")

                        for p in products:
                            if other_text.strip()==p.strip():
                                has_product = True

                    else:
                        _fix = False

                        for p in products:
                            if other_text.strip()==p.strip():
                                title_before = "=产品"
                                sentence_title = "title_0"
                                sentence_title_text = p
                                title_index = "0"
                                title_after = "产品="
                                next_index = "0"
                                _fix = True
                                has_product = True
                                break
                        if not _fix:
                            title_before = None
                            title_after = None
                            sentence_title_text = None
                else:
                    if len(_text)<40 and re.search(_param_pattern,_text) is not None:
                        for p in products:
                            if _text.find(p)>=0:
                                title_before = "=产品"
                                sentence_title = "title_0"
                                sentence_title_text = p
                                title_index = "0"
                                title_after = "产品="
                                next_index = "0"
                                _fix = True
                                has_product = True
                                break

            # 合并两个非标题句子 20241106 注销，由于 485441521 招标内容结束位置不对
            if auto_append:
                if _type=="sentence":
                    if sentence_title is None and len(list_data)>0 and list_data[-1]["sentence_title"] is not None and list_data[-1]["line_width"]>=max_length*0.6:
                        list_data[-1]["text"] += _text
                        list_data[-1]["line_width"] = len(_text)
                        update_kv_span(list_kv,len(_text))
                        list_data[-1]["kv"].extend(list_kv)
                        list_data[-1]["sentences"].extend(sentences)
                        _append = True
                    elif sentence_title is None and len(list_data)>0 and _type==list_data[-1]["type"]:
                        if list_data[-1]["line_width"]>=max_length*0.7:
                            list_data[-1]["text"] += _text
                            list_data[-1]["line_width"] = len(_text)
                            update_kv_span(list_kv,len(_text))
                            list_data[-1]["kv"].extend(list_kv)
                            list_data[-1]["sentences"].extend(sentences)
                            _append = True


            if not _append:
                _data = {"type":_type,"tag":obj.get("tag"),"table_id":table_id, "text":_text,"sentences":sentences,"list_table":list_table,
                         "line_width":len(_text),"sentence_title":sentence_title,"title_index":title_index,
                         "sentence_title_text":sentence_title_text,"sentence_groups":sentence_groups,"parent_title":parent_title,
                         "child_title":childs,"title_before":title_before,"title_after":title_after,"title_next":title_next,"next_index":next_index,
                         "block":block,"has_product":has_product,
                         "sentence_index":sentence_index,"wordOffset_begin":wordOffset_begin,"wordOffset_end":wordOffset_end,
                         "kv":list_kv,"position":position
                        }


                if sentence_title is not None:
                    if len(list_data)>0:
                        if self.is_title_first(title_index):
                            for i in range(1,len(list_data)+1):
                                _d = list_data[-i]
                                if _d["sentence_title"] is not None:
                                    _data["parent_title"] = _d
                                    _d["child_title"].append(_data)
                                    break
                        else:
                            _find = False
                            for i in range(1,len(list_data)+1):
                                if _find:
                                    break
                                _d = list_data[-i]
                                if _d.get("sentence_title")==sentence_title and title_before==_d["title_before"] and title_after==_d["title_after"]:
                                    if _d["next_index"]==title_index and _d["title_next"] is None and not _d["block"]:
                                        _data["parent_title"] = _d["parent_title"]
                                        _d["title_next"] = _data
                                        if len(_d["child_title"])>0:
                                            _d["child_title"][-1]["title_next"] = ""
                                            self.block_tree(_d["child_title"])
                                        if _d["parent_title"] is not None:
                                            _d["parent_title"]["child_title"].append(_data)
                                        _find = True
                                        break
                            for i in range(1,len(list_data)+1):
                                if _find:
                                    break
                                _d = list_data[-i]
                                if i==1 and not _d["block"] and _d.get("sentence_title")==sentence_title and title_before==_d["title_before"] and title_after==_d["title_after"]:
                                    _data["parent_title"] = _d["parent_title"]
                                    _d["title_next"] = _data
                                    if len(_d["child_title"])>0:
                                        _d["child_title"][-1]["title_next"] = ""
                                        self.block_tree(_d["child_title"])
                                    if _d["parent_title"] is not None:
                                        _d["parent_title"]["child_title"].append(_data)
                                    _find = True
                                    break
                            title_before = standard_title_context(title_before)
                            title_after = standard_title_context(title_after)
                            for i in range(1,len(list_data)+1):
                                if _find:
                                    break
                                _d = list_data[-i]
                                if _d.get("sentence_title")==sentence_title and title_before==standard_title_context(_d["title_before"]) and title_after==standard_title_context(_d["title_after"]):
                                    if _d["next_index"]==title_index and _d["title_next"] is None and not _d["block"]:
                                        _data["parent_title"] = _d["parent_title"]
                                        _d["title_next"] = _data
                                        if len(_d["child_title"])>0:
                                            _d["child_title"][-1]["title_next"] = ""
                                            self.block_tree(_d["child_title"])
                                        if _d["parent_title"] is not None:
                                            _d["parent_title"]["child_title"].append(_data)
                                        _find = True
                                        break
                            for i in range(1,len(list_data)+1):
                                if _find:
                                    break
                                _d = list_data[-i]
                                if not _d["block"] and _d.get("sentence_title")==sentence_title and title_before==standard_title_context(_d["title_before"]) and title_after==standard_title_context(_d["title_after"]):
                                    _data["parent_title"] = _d["parent_title"]
                                    _d["title_next"] = _data
                                    if len(_d["child_title"])>0:
                                        _d["child_title"][-1]["title_next"] = ""
                                        # self.block_tree(_d["child_title"])
                                    if _d["parent_title"] is not None:
                                        _d["parent_title"]["child_title"].append(_data)
                                    _find = True
                                    break
                            for i in range(1,min(len(list_data)+1,20)):
                                if _find:
                                    break
                                _d = list_data[-i]
                                if not _d["block"] and _d.get("sentence_title")==sentence_title and title_before==standard_title_context(_d["title_before"]):
                                    _data["parent_title"] = _d["parent_title"]
                                    _d["title_next"] = _data
                                    if len(_d["child_title"])>0:
                                        _d["child_title"][-1]["title_next"] = ""
                                        # self.block_tree(_d["child_title"])
                                    if _d["parent_title"] is not None:
                                        _d["parent_title"]["child_title"].append(_data)
                                    _find = True
                                    break

                            if not _find:
                                if len(list_data)>0:
                                    for i in range(1,len(list_data)+1):
                                        _d = list_data[-i]
                                        if _d.get("sentence_title") is not None:
                                            _data["parent_title"] = _d
                                            _d["child_title"].append(_data)
                                            break


                else:
                    if len(list_data)>0:
                        for i in range(1,len(list_data)+1):
                            _d = list_data[-i]
                            if _d.get("sentence_title") is not None:
                                _data["parent_title"] = _d
                                _d["child_title"].append(_data)
                                break

                list_data.append(_data)

        for _data in list_data:

            childs = _data["child_title"]

            for c_i in range(len(childs)):
                cdata = childs[c_i]
                if cdata["has_product"]:
                    continue
                else:
                    if c_i>0:
                        last_cdata = childs[c_i-1]
                        if cdata["sentence_title"] is not None and last_cdata["sentence_title"] is not None and last_cdata["title_before"]==cdata["title_before"] and last_cdata["title_after"]==cdata["title_after"] and last_cdata["has_product"]:
                            cdata["has_product"] = True
                    if c_i<len(childs)-1:
                        last_cdata = childs[c_i+1]
                        if cdata["sentence_title"] is not None and last_cdata["sentence_title"] is not None and last_cdata["title_before"]==cdata["title_before"] and last_cdata["title_after"]==cdata["title_after"] and last_cdata["has_product"]:
                            cdata["has_product"] = True
            for c_i in range(len(childs)):
                cdata = childs[len(childs)-1-c_i]
                if cdata["has_product"]:
                    continue
                else:
                    if c_i>0:
                        last_cdata = childs[c_i-1]
                        if cdata["sentence_title"] is not None and last_cdata["sentence_title"] is not None and last_cdata["title_before"]==cdata["title_before"] and last_cdata["title_after"]==cdata["title_after"] and last_cdata["has_product"]:
                            cdata["has_product"] = True
                    if c_i<len(childs)-1:
                        last_cdata = childs[c_i+1]
                        if cdata["sentence_title"] is not None and last_cdata["sentence_title"] is not None and last_cdata["title_before"]==cdata["title_before"] and last_cdata["title_after"]==cdata["title_after"] and last_cdata["has_product"]:
                            cdata["has_product"] = True


        return list_data

    def get_tree_sentence(self):
        list_sentence = []
        for obj in self.tree:
            list_sentence.extend(obj.get("sentences",[]))

        return list_sentence


    def extract_kvs_from_table(self,list_pattern,tree=None,result_kv=None):
        if result_kv is None:
            result_kv = [[] for i in list_pattern]
            try:
                for pattern in list_pattern:
                    re.compile(pattern)
            except Exception as e:
                log("list_pattern error: "+str(e))
                return result_kv
        if tree is None:
            tree = self.tree
        for obj in tree:
            is_table = True if obj.get("tag","")=="table" else False
            if is_table:

                table_id = obj.get("table_id")
                list_kv = obj.get("kv")
                for _pi in range(len(list_pattern)):
                    table_kvs = []
                    for _d0 in list_kv:
                        _k = _d0.get("key","")
                        _v = _d0.get("value","")
                        _d = {"key":_k,"value":_v,"position":_d0.get("position",{})}
                        if re.search(list_pattern[_pi],_k) is not None:
                            table_kvs.append(_d)
                    if table_kvs:
                        result_kv[_pi].append({"table_id":table_id,"kv":table_kvs})
                childs = obj.get("children",[])
                for child in childs:
                    self.extract_kvs_from_table(list_pattern,child,result_kv)
        return result_kv

    def extract_kvs_from_sentence(self,list_pattern,tree=None,result_kv=None):
        if result_kv is None:
            result_kv = [[] for i in list_pattern]
            try:
                for pattern in list_pattern:
                    re.compile(pattern)
            except Exception as e:
                log("list_pattern error: "+str(e))
                return result_kv
        if tree is None:
            tree = self.tree
        for obj in tree:
            is_table = True if obj.get("tag","")=="table" else False
            if not is_table:
                list_kv = obj.get("kv",[])
                for _pi in range(len(list_pattern)):
                    for _d in list_kv:
                        _k = _d.get("key","")
                        _v = _d.get("value","")
                        if re.search(list_pattern[_pi],_k) is not None:
                            result_kv[_pi].append(_d)
        return result_kv

    def extract_kvs_from_outline(self,list_pattern,tree=None,result_kv=None):
        if result_kv is None:
            result_kv = [[] for i in list_pattern]
            try:
                for pattern in list_pattern:
                    re.compile(pattern)
            except Exception as e:
                log("list_pattern error: "+str(e))
                return result_kv
        if tree is None:
            tree = self.tree
        for obj in tree:
            is_table = True if obj.get("tag","")=="table" else False
            if not is_table:

                _text = obj["text"]

                for _pi in range(len(list_pattern)):

                    sentence_index_from = obj["sentence_index"]
                    sentence_index_to = sentence_index_from

                    if re.search(list_pattern[_pi],_text) is not None and obj.get("sentence_title") is not None:

                        childs = get_childs([obj])
                        _child_text = ""
                        for _child in childs:
                            sentence_index_to = _child["sentence_index"]
                            _child_text+=_child["text"]+"\n"
                        result_kv[_pi].append({"key":_text,"value":_child_text,"from_outline":True,"key_sentence_index_from":sentence_index_from,
                                          "key_sentence_index_to":sentence_index_from,"value_sentence_index_from":sentence_index_from,
                                          "value_sentence_index_to":sentence_index_to,})
        return result_kv


    def extract_kv(self,k_pattern,from_sentence=True,from_outline=True,from_table=True):
        result_kv = []
        try:
            re.compile(k_pattern)
        except Exception as e:
            log("k_pattern error: "+str(e))
            traceback.print_exc()
            return result_kv
        result_kv = []
        if from_table:
            result_kv_table = self.extract_kvs_from_table([k_pattern])
            for table_d in result_kv_table[0]:
                table_id = table_d.get("table_id")
                table_kvs = table_d.get("kv",[])
                for _d in table_kvs:
                    _d["from_table"] = True
                result_kv.extend(table_kvs)
        if from_sentence:
            result_kv_sentence = self.extract_kvs_from_sentence([k_pattern])
            for _d in result_kv_sentence[0]:
                _d["from_sentence"] = True
            result_kv.extend(result_kv_sentence[0])
        if from_outline:
            result_kv_outline = self.extract_kvs_from_outline([k_pattern])
            for _d in result_kv_outline[0]:
                _d["from_outline"] = True
            result_kv.extend(result_kv_outline[0])

        return result_kv

    # def extract_kvs_from_table(self,list_pattern):




if __name__ == '__main__':
    # HTML 文本
    html_content = """
<div>
<div>
   工程造价咨询 
 </div>
<div>
<div>
    关于为【意溪镇四宁村农村人居环境及村道整治提升项目-前期费用结算审核】公开选取【工程造价咨询】机构的公告 
  </div>
</div>
<div>
<p> 2024-12-24 17:30 ，在广东省网上中介服务超市为<a target="_blank" class="markBlue" href="/bdqyhx/216515788655505408.html" style="color: #3083EB !important;text-decoration: underline;">潮州市湘桥区意溪镇四宁村民委员会</a> 公开选取工程造价咨询中介服务机构，现将相关事项公告如下： </p>
<p>此项目采用多选一的直接选取方式，项目业主将在报名的若干家中介机构中，自主选定一家作为中选机构，未被选中的机构不应有任何异议。</p>
<ul>
<li> <b>项目业主</b>
<div data-purorgcode="G19187326" data-tyshxydm="54445102G191873262">
      <a target="_blank" class="markBlue" href="/bdqyhx/216515788655505408.html" style="color: #3083EB !important;text-decoration: underline;">潮州市湘桥区意溪镇四宁村民委员会</a> 
    </div> </li>
<li> <b>采购项目名称</b>
<div data-divisioncode="445100">
      意溪镇四宁村农村人居环境及村道整治提升项目-前期费用结算审核 
    </div> </li>
<li> <b>中介服务事项</b>
<div data-servicesubjectcode="">
      无（属于非行政管理的中介服务项目采购） 
    </div> </li>
<li> <b>投资审批项目</b>
<div>
      否 
    </div> </li>
<li> <b>采购项目编码</b>
<div>
      445100G191873262412200638 
    </div> </li>
<li> <b>项目规模</b>
<div data-restrictionsforehead="1500000.0" data-restrictionsforeheadtype="amountInvested">
<p>投资额（￥1,500,000.00元）</p>
</div> </li>
<li> <b>所需服务</b>
<div>
      工程造价咨询 
    </div> </li>
<li> <b>服务内容</b>
<div>
      本次工程为意溪镇四宁村农村人居环境及村道整治提升项目，位于潮州市湘桥区意溪镇四宁村，投资额为1500000.00元，建设内容包括路面便底化约800平方米，黑底化约2150平方米，新安装太阳能路灯14盏等，现已竣工验收完成，根据业主要求，对本项目进行前期费用结算审核（概算编制费2571.86元，施工图审查费5024.5元，设计费54700元，预算编制费5315.66元，工程监理费35275.46元，工程测绘费22634.89元，建设方案编制费10500元），并出具前期费用结算审核报告及定案表。 
    </div> </li>
<li> <b>中介机构要求</b>
<div>
      仅承诺服务即可 
    </div> </li>
<li> <b>其他要求说明：</b>
<div>
      无 
    </div> </li>
<li> <b>服务时限说明</b>
<div>
      无要求，按合同约定。 
    </div> </li>
<li> <b>服务金额</b>
<div data-biddingmode="" data-highprice="" data-lowprice="1600.0">
      ￥1,600.00元 
    </div> </li>
<li> <b>金额说明</b>
<div>
      按广东省物价局关于调整我省建设工程造价咨询服务收费的复函（粤价函[2011]742号）规定的造价咨询行业收费标准，最终价格以财政审核或第三方工程造价公司审核价为准。 
    </div> </li>
<li> <b>选取中介服务机构方式</b>
<div data-selectmodetype="DXYZJXQ">
      直接选取 
    </div> </li>
<li> <b>是否选取中介</b>
<div>
      否 
    </div> </li>
<li> <b>有无回避情况</b>
<div>
      否 
    </div> </li>
<li> <b> 截止报名时间 </b>
<div>
      2024-12-24 17:30 
    </div> </li>
<li> <b>业主单位咨询电话</b>
<div>
      <a target="_blank" class="markBlue" href="/bdqyhx/216515788655505408.html" style="color: #3083EB !important;text-decoration: underline;">潮州市湘桥区意溪镇四宁村民委员会</a> （登录后查看） 
    </div> </li>
<li> <b>采购需求书下载</b>
<div>
<a target="_blank" class="markBlue" filelink="277a2f13e4a4d41149766c82adfc8762" href="https://attachment-hub.oss-cn-hangzhou.aliyuncs.com/277a/20230710/2023-07-10/04733/1688982497468.jpg?Expires=1734688561&amp;OSSAccessKeyId=LTAI5tHoEUDSy6FnZjMKsNiZ&amp;Signature=G3SAJQuJlYZ5lOpHNc%2BWHspfDpE%3D" original="https://ygp.gdzwfw.gov.cn/zjfwcs/gd-zjcs-pub/file/downloadfile/PjAttachment/7618f614-ae9c-48cc-97f7-17ffd05d4200" rel="noreferrer">资金说明1.jpg</a>
<br>
<a target="_blank" class="markBlue" filelink="3e3795ea0244f1b4cc77123512edd30a" href="https://attachment-hub.oss-cn-hangzhou.aliyuncs.com/3e37/20230710/2023-07-10/04733/1688982505603.jpg?Expires=1734688561&amp;OSSAccessKeyId=LTAI5tHoEUDSy6FnZjMKsNiZ&amp;Signature=ZP8xdy%2F1a%2Blbb%2FOAhFyzjpadprg%3D" original="https://ygp.gdzwfw.gov.cn/zjfwcs/gd-zjcs-pub/file/downloadfile/PjAttachment/beb5d765-4ce3-471f-a290-14a12d9ad64e" rel="noreferrer">资金说明2.jpg</a>
<br>
</div> </li>
</ul>
</div>
<p> 广东省网上中介服务超市已经向符合资质条件的在库中介服务机构的业务授权人手机号码和中介专属网页发送通知，诚邀符合资质条件的在库中介服务机构登录中介专属网页进行报名。 </p>
<p><span>潮州市公共资源交易中心</span><br> <span>2024-12-20</span></p>
</div>
    """
    _tree = html_to_tree(html_content)


    _pd = Html2KVTree(html_content)
    _pd.print_tree(_pd.tree,"-|")

    list_kv = _pd.extract_kv("资质要求")
    print(list_kv)

    #获取预处理后的所有句子，该句子与kv值对应
    print(_pd.get_tree_sentence())

    # soup = BeautifulSoup(html_content,"lxml")
    # table_tree = table_to_tree(soup)
    # print(json.dumps(table_tree,ensure_ascii=False))