BIDI_ML_INFO_EXTRACTION/BiddingKG/dl/interface/html_2_kv.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)

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



# 递归地将 DOM 转换为 JSON
# 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
        children = []
        for child in node.contents:
            _child = dom_to_tree(child)
            if _child is not None:
                _child["parent"] = json_obj
                children.append(_child)
        if children:
            json_obj["children"] = children
        return json_obj
    elif node.string and node.string.strip():  # 如果是纯文本节点
        return DotDict({"tag":"text","text": node.string.strip()})
    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)
    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)
    print(json_output)

# 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:],"key_span":[0,_begin],"value_span":[_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["key_span"] = iter.span("key")
                    _d["value_span"] = iter.span("value")
                    list_kv.append(_d)

        elif len(list_span)>1:
            _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["key_span"] = iter.span("key")
                    _d["value_span"] = iter.span("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["key_span"] = iter.span("key")
                _d["value_span"] = iter.span("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 = []
    _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
        for child in tree:
            _count += extract_kv_from_tree(child)
        return _count
    if isinstance(tree,dict):
        childs = tree.get("children",[])
        if len(childs)>0:
            _count = 0
            for child in childs:
                _count += extract_kv_from_tree(child)
            if _count==0:
                _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
            if tree.get("tag","")=="p":
                _text = get_child_text(tree)
                tree["text"] = _text
                p_list_kv = extract_kv_from_node(tree)
                if len(p_list_kv)>=_count:
                    if "children" in tree:
                        del tree["children"]
                else:
                    tree["text"] = ""
                return len(p_list_kv)

            return _count
        else:
            list_kv = extract_kv_from_node(tree)
            return len(list_kv)


def update_kv_span(list_kv,append_length):
    for _d in list_kv:
        _d["key_span"][0] += append_length
        _d["key_span"][1] += append_length
        _d["value_span"][0] += append_length
        _d["value_span"][1] += append_length

def get_sentence_from_tree(tree,list_sentences=None):

    is_first = False
    if list_sentences is None:
        list_sentences = []
        is_first = True
    if isinstance(tree,list):
        for child in tree:
            get_sentence_from_tree(child,list_sentences)
    if isinstance(tree,dict):
        childs = tree.get("children",[])
        _text = tree.get("text","")
        if _text!="":
            tree.name = tree.tag
            list_sentences.append(tree)
        for child in childs:
            get_sentence_from_tree(child,list_sentences)
    if is_first:
        wordOffset_begin = 0
        wordOffset_end = 0
        for _i in range(len(list_sentences)):
            list_sentences[_i].sentence_index = _i
            list_sentences[_i].wordOffset_begin = wordOffset_begin
            wordOffset_end = wordOffset_begin+len(list_sentences[_i].text)
            list_sentences[_i].wordOffset_end = wordOffset_end
            wordOffset_begin = wordOffset_end

    return list_sentences


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)
            extract_kv_from_tree(_tree)
            self.list_obj = get_sentence_from_tree(_tree)

            print(len(self.list_obj))


        # 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

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

            list_table = None
            block = False

            has_product = False

            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)
                        _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)
                            _append = True

            if _type=="table":
                _soup = BeautifulSoup(_text,"lxml")
                _table = _soup.find("table")
                if _table is not None:
                    list_table = getTable(_table)
                    if len(list_table)==0:
                        continue
                    table_columns = len(list_table[0])

                    if auto_merge_table:
                        if last_table_index is not None and abs(obj_i-last_table_index)<=2 and last_table_columns is not None and last_table_columns==table_columns:
                            if last_table is not None:
                                trs = getTrs(_table)
                                last_tbody = BeautifulSoup(last_table["text"],"lxml")
                                _table = last_tbody.find("table")
                                last_trs = getTrs(_table)
                                _append = True

                                for _line in list_table:
                                    last_table["list_table"].append(_line)
                                if len(last_trs)>0:
                                    for _tr in trs:
                                        last_trs[-1].insert_after(copy.copy(_tr))
                                    last_table["text"] = re.sub("</?html>|</?body>","",str(last_tbody))

                                last_table_index = obj_i
                                last_table_columns = len(list_table[-1])


            if not _append:
                _data = {"type":_type, "text":_text,"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
                        }

                if _type=="table":
                    last_table = _data
                    last_table_index = obj_i
                    if list_table:
                        last_table_columns = last_table_columns = len(list_table[-1])

                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_sentence_tree(self):
        return self.tree

    def extract_kv(self,k_pattern,from_kv=True,from_outline=True):
        result_kv = []
        try:
            re.compile(k_pattern)
        except Exception as e:
            log("k_pattern error: "+str(e))
            traceback.print_exc()
            return result_kv
        for sentence in self.tree:
            _text = sentence["text"]
            list_kv = sentence.get("kv",[])
            sentence_index_from = sentence["sentence_index"]
            sentence_index_to = sentence_index_from
            if from_kv:
                for _d in list_kv:
                    _k = _d.get("key","")
                    _v = _d.get("value","")
                    _k_span = _d.get("key_span",[])
                    _v_span = _d.get("value_span",[])
                    if re.search(k_pattern,_k) is not None:
                        result_kv.append({"key":_k,"value":_v,"from_kv":True,"key_sentence_index_from":sentence_index_from,
                                          "key_sentence_index_to":sentence_index_to,"value_sentence_index_from":sentence_index_from,
                                          "value_sentence_index_to":sentence_index_to,
                                          "key_span":_k_span,
                                          "value_span":_v_span})
            if from_outline:
                if re.search(k_pattern,_text) is not None and sentence.get("sentence_title") is not None:

                    childs = get_childs([sentence])
                    _child_text = ""
                    for _child in childs:
                        sentence_index_to = _child["sentence_index"]
                        _child_text+=_child["text"]
                    result_kv.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_kvs_from_table(self,list_pattern):




if __name__ == '__main__':
    # HTML 文本
    html_content = """
<div>
 <div>
   中介服务交易|中介服务交易；中选公示；广东省中介超市交易系统 
 </div> 
 <div>
   采购项目编码： 4451024573035462411180923 
 </div> 
 <div>
   采购项目名称： 意溪中学办公楼一楼卫生间、保密室、运动场围墙及零碎修缮工程（结算审核） 
 </div> 
 <div>
   项目业主名称： <a target="_blank" class="markBlue" href="/bdqyhx/212724763956953088.html" style="color: #3083EB !important;text-decoration: underline;">潮州市湘桥区意溪中学</a> 
 </div> 
 <div>
   中介服务事项： 无（属于非行政管理的中介服务项目采购） 
 </div> 
 <div>
   投资审批项目编码： 
 </div> 
 <div>
   服务金额： 暂不做评估与测算 
 </div> 
 <div>
   金额说明： 按粤价函【2011】742号文及潮财建【2019】19号文的收费标准执行 （行业收费标准）计算。 
 </div> 
 <div>
   选取中介机构方式： 直接选取 
 </div> 
 <div>
   业务单位咨询电话： 13715781262 
 </div> 
 <div>
   监督举报： 
 </div> 
 <div>
   中选中介机构名称： <a target="_blank" class="markBlue" href="/bdqyhx/216479817075470337.html" style="color: #3083EB !important;text-decoration: underline;">中庭国际设计有限公司</a> 
 </div> 
 <div>
   中介机构联系地址： 广州市天河区大观南路26号C203A 
 </div>
</div>
    """
    _tree = html_to_tree(html_content)
    extract_kv_from_tree(_tree)

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


    list_kv = _pd.extract_kv("交货地点")
    print(list_kv)