【无标题】

前言

爬取数据见上一篇博客：《Python3 爬取OpenStreetMap平台的城市道路交通网数据》

代码

"""
osm 是xml文件 因此 利用 sax解析xml
将读取的结果保存至 csv 文件
"""
import xml.sax
import openpyxl
import re
import os

"""
* func checkIsExist
* desc 检查文件是否存在，若是存在则删除
* para path String
* retu 
"""
def checkIsExist(path:str):
    if os.path.exists(path):
        #删除它
        os.remove(path)

# 将数据 转换为 表格
def roadDataToXlsx(data:dict,path:str):
    wbk = openpyxl.Workbook()
    sd_sheet = wbk.active
    sd_sheet.cell(1,1,"id")
    sd_sheet.cell(1,2,"name")
    sd_sheet.cell(1,3,"lanes")
    sd_sheet.cell(1,4,"highway")
    sd_sheet.cell(1,5,"oneway")
    # sd_sheet.cell(1,6,"surface")
    # sd_sheet.cell(1,7,"layer")
    sd_sheet.cell(1,6,"city")
    global_index = 1
    for k,v in data.items():
        global_index+=1
        sd_sheet.cell(global_index,1,v[0])
        sd_sheet.cell(global_index,2,v[1])
        sd_sheet.cell(global_index,3,v[2])
        sd_sheet.cell(global_index,4,v[3])
        sd_sheet.cell(global_index,5,v[4])
        # sd_sheet.cell(global_index,6,v[5])
        # sd_sheet.cell(global_index,7,v[6])
        sd_sheet.cell(global_index,6,v[5])
    checkIsExist(path)#删除原文件
    wbk.save(path)

# 自定义解析 node 和 way 标签的 类
class WayHandler(xml.sax.ContentHandler):
    def __init__(self,city_name):
        self.rootName = "" #保存 根标签名称
        # way 标签里面的内容
        self.way_id = ""
        self.tag_name = ""
        self.tag_lanes = ""
        self.tag_highway = ""
        self.tag_oneway = ""
        # self.tag_surface = ""
        # self.tag_layer = ""
        self.tag_waterway = ""
        self.tag_power = ""
        self.tag_railway = ""
        self.city = city_name
        self.road_map = {}
        
    # 元素开始事件处理
    def startElement(self, tag, attributes):
        if tag == "way":
            #记录父标签
            self.rootName = tag
            #当前处理 way标签
            self.way_id = attributes.get("id","")
        elif tag == "tag":
            if self.rootName == "way":
                # 父节点是 way
                k = attributes.get("k","")
                v = attributes.get("v","")
                if k == "name":
                    self.tag_name = v
                elif k == "lanes":
                    self.tag_lanes = v
                elif k == "highway":
                    self.tag_highway = v
                elif k == "oneway":
                    self.tag_oneway = v
                # elif k == "surface":
                #     self.tag_surface = v
                # elif k == "layer":
                #     self.tag_layer = v
                elif k == "waterway":
                    self.tag_waterway = v
                elif k == "power":
                    self.tag_power = v
                elif k == "railway":
                    self.tag_railway = v
            
    # 元素结束事件处理    
    def endElement(self, tag):
        if tag == "way":
            self.rootName=""#释放根节点
            # 排除 隧道 无名路 无级别路 住宅路 桥 
            if not re.match(r".*(路|道|街)$",self.tag_name):
                self.clear()
                return
            # if self.tag_highway == "footway":
            #     self.clear()
            #     return
            if re.match(r".*隧道$",self.tag_name):
                self.clear()
                return
            if self.tag_power=="plant":
                self.clear()
                return
            if self.tag_highway=="":
                self.clear()
                return
            if self.tag_power=="line":
                self.clear()
                return
            if self.tag_power=="substation":
                self.clear()
                return
            if self.tag_railway=="subway":
                self.clear()
                return
            if self.tag_railway=="rail":
                self.clear()
                return
            # 保存至 self.road_map
            item = []
            item.append(self.way_id)
            item.append(self.tag_name)
            item.append(self.tag_lanes)
            item.append(self.tag_highway)
            item.append(self.tag_oneway)
            # item.append(self.tag_surface)
            # item.append(self.tag_layer)
            item.append(self.city)
            self.road_map[self.way_id] = item
            self.clear()

    # 内容事件处理
    def characters(self, content):
        # print(content)
        pass
    
    def clear(self):
        self.way_id = ""
        self.tag_name = ""
        self.tag_lanes = ""
        self.tag_highway = ""
        self.tag_oneway = ""
        # self.tag_surface = ""
        # self.tag_layer =""
        self.tag_waterway = ""
        self.tag_power = ""
        self.tag_railway = ""
    
    def saveToXlsx(self):
        road_map = {}
        city_name = self.city
        for k,v in self.road_map.items():
            name = v[1]
            lanes = v[2]
            # 如果没有 放进去
            if not road_map.get(name,None):
                road_map[name] = v
            # 替换没有车道信息的记录
            elif lanes and lanes.isalnum() and road_map[name][2] and road_map[name][2].isalnum()==False:
                road_map[name][2] = lanes
            # 替换车道最少的信息
            elif lanes and lanes.isalnum() and road_map[name][2] and road_map[name][2].isalnum() and lanes<road_map[name][2]:
                road_map[name][2] = lanes
        roadDataToXlsx(road_map,"./static/road/"+city_name+"道路信息表.xlsx")   

# 将JSOM文件 转换为 表格 入口函数
def josmToXlsx(city_name)->int:
    path = "./static/osm/"+city_name+".osm"# osm文件路径
    if not os.path.exists(path):
        return 700
    # 创建一个 XMLReader 
    parser = xml.sax.make_parser()
    parser.setFeature(xml.sax.handler.feature_namespaces, 0)
    # 重写 ContextHandler
    nodeHandler = WayHandler(city_name)
    parser.setContentHandler(nodeHandler)
    parser.parse(path)
    nodeHandler.saveToXlsx()
    return 200