feat(重构代码): 重构代码，适配新版jar包的方法

1. 新增 Geocoding 自定义
2. addRegionEntry 方法新增 replace 参数
3. 新增分词
segments
4. 分离 similarityWithResult 和 similarity 两个方法

GeocodingCHN/__init__.py

@@ -0,0 +1 @@
+from GeocodingCHN.Geocoding import Geocoding

GeocodingCHN/model/__init__.py

@@ -0,0 +1,12 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:07
+# @File     : __init__.py.py
+# @Author  : CasuallyName
+# @Email   : fjkl@vip.qq.com
+# @Software : Python 3.7
+# @About    :
+from .address import Address
+from .region_type import RegionType
+from .version import Version
+from .document import Document
+from .matched import MatchedResult,MatchedTerm

GeocodingCHN/model/address.py

@@ -0,0 +1,85 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:08
+# @File     : address.py
+# @Author   : Zhou Hang
+# @Email    : zhouhang@idataway.com
+# @Software : Python 3.7
+# @About    :
+import jpype
+class Address(object):
+    def __init__(self, provinceId=None, province=None, cityId=None, city=None, districtId=None, district=None,
+                 streetId=None, street=None, townId=None, town=None, villageId=None, village=None, road=None,
+                 roadNum=None, buildingNum=None, text=None, java=None):
+        self.provinceId = int(provinceId) if provinceId else provinceId
+        self.province = province
+        self.cityId = int(cityId) if cityId else cityId
+        self.city = city
+        self.districtId = int(districtId) if districtId else districtId
+        self.district = district
+        self.streetId = int(streetId) if streetId else streetId
+        self.street = street
+        self.townId = townId
+        self.town = town
+        self.villageId = villageId if villageId is not None else None
+        self.village = village
+        self.road = road
+        self.roadNum = roadNum
+        self.buildingNum = buildingNum
+        self.text = text
+        self._AddressClass = jpype.JClass('org.bitlap.geocoding.model.Address')
+        self._java = java if java is not None else self._AddressClass(self.provinceId, self.province, self.cityId,
+                                                                      self.city, self.districtId, self.district,
+                                                                      self.streetId, self.street, self.townId,
+                                                                      self.town,
+                                                                      self.villageId, self.village, self.road,
+                                                                      self.roadNum, self.buildingNum, self.text)
+
+    def __repr__(self):
+        return (f"Address(provinceId={self.provinceId}, province={self.province}, " +
+                f"cityId={self.cityId}, city={self.city}, " +
+                f"districtId={self.districtId}, district={self.district}, " +
+                f"streetId={self.streetId}, street={self.street}, " +
+                f"townId={self.townId}, town={self.town}, " +
+                f"villageId={self.villageId}, village={self.village}, " +
+                f"road={self.road}, " +
+                f"roadNum={self.roadNum}, " +
+                f"buildingNum={self.buildingNum}, " +
+                f"text={self.text})")
+
+    def __str__(self):
+        return (f"Address(\n\tprovinceId={self.provinceId}, province={self.province}, " +
+                f"\n\tcityId={self.cityId}, city={self.city}, " +
+                f"\n\tdistrictId={self.districtId}, district={self.district}, " +
+                f"\n\tstreetId={self.streetId}, street={self.street}, " +
+                f"\n\ttownId={self.townId}, town={self.town}, " +
+                f"\n\tvillageId={self.villageId}, village={self.village}, " +
+                f"\n\troad={self.road}, " +
+                f"\n\troadNum={self.roadNum}, " +
+                f"\n\tbuildingNum={self.buildingNum}, " +
+                f"\n\ttext={self.text}\n)")
+
+    @property
+    def __dict__(self):
+        return {
+            'provinceId': self.provinceId,
+            'province': self.province,
+            'cityId': self.cityId,
+            'city': self.city,
+            'districtId': self.districtId,
+            'district': self.district,
+            'streetId': self.streetId,
+            'street': self.street,
+            'townId': self.townId,
+            'town': self.town,
+            'villageId': self.villageId,
+            'village': self.village,
+            'road': self.road,
+            'roadNum': self.roadNum,
+            'buildingNum': self.buildingNum,
+            'text': self.text
+        }
+
+    @property
+    def __java__(self):
+        return self._java
+

GeocodingCHN/model/document.py

@@ -0,0 +1,23 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:12
+# @File     : document.py
+# @Author  : CasuallyName
+# @Email   : fjkl@vip.qq.com
+# @Software : Python 3.7
+# @About    :
+
+class Document():
+    def __init__(self, terms=None, termsMap=None, town=None, village=None, road=None, roadNum=None, roadNumValue=None):
+        self.terms = terms
+        self.termsMap = termsMap
+        self.town = town
+        self.village = village
+        self.road = road
+        self.roadNum = roadNum
+        self.roadNumValue = roadNumValue
+
+    def __str__(self):
+        return "Document(terms={terms}, town={town}, village={village}, road={road}, roadNum={roadNum}, roadNumValue={roadNumValue})".format(
+            terms=self.terms, town=self.town, village=self.village, road=self.road, roadNum=self.roadNum,
+            roadNumValue=self.roadNumValue
+        )

GeocodingCHN/model/matched.py

@@ -0,0 +1,51 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:33
+# @File     : matched.py
+# @Author   : Zhou Hang
+# @Email    : zhouhang@idataway.com
+# @Software : Python 3.7
+# @About    :
+
+
+class MatchedTerm:
+    def __init__(self, term=None, coord=0, density=0, boost=0, tfidf=0):
+        # 匹配的词条
+        self.term = term
+        # 匹配率
+        self.coord = coord
+        # 稠密度
+        self.density = density
+        # 权重
+        self.boost = boost
+        # 特征值 TF - IDF
+        self.tfidf = tfidf
+
+    def __str__(self):
+        return "MatchedTerm({}, coord={}, density={}, boost={}, tfidf={})".format(
+            self.term, self.coord, self.density, self.boost, self.tfidf)
+
+
+class MatchedResult:
+    def __init__(self, doc1=None, doc2=None, terms=None, similarity=0,java=None):
+        self.doc1 = doc1
+        self.doc2 = doc2
+        self.terms = terms
+        self.terms = terms if terms is None else [MatchedTerm(term.getTerm(),
+                                                              term.getCoord(),
+                                                              term.getDensity(),
+                                                              term.getBoost(),
+                                                              term.getTfidf(),
+                                                              )
+                                                  for term in terms]
+        self.similarity = similarity
+        self.java = java
+
+    def __str__(self):
+        return "MatchedResult(\n\tdoc1={doc1}, \n\tdoc2={doc2}, \n\tterms={terms}, \n\tsimilarity={similarity}\n)".format(
+            doc1=self.doc1, doc2=self.doc2, terms=str([str(i) for i in self.terms]), similarity=self.similarity
+        )
+
+    def __repr__(self):
+        return "MatchedResult(doc1={doc1}, doc2={doc2}, terms={terms}, similarity={similarity})".format(
+            doc1=self.doc1, doc2=self.doc2, terms=str([str(i) for i in self.terms]), similarity=self.similarity
+        )

GeocodingCHN/model/region_type.py

@@ -0,0 +1,41 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:09
+# @File     : region_type.py
+# @Author  : CasuallyName
+# @Email   : fjkl@vip.qq.com
+# @Software : Python 3.7
+# @About    :
+
+class RegionType(object):
+    def __init__(self, RegionTypeClass):
+        self.Undefined = RegionTypeClass.Undefined  # 未定义区域类型
+        self.Country = RegionTypeClass.Country  # 国家
+        self.Province = RegionTypeClass.Province  # 省份
+        self.ProvinceLevelCity1 = RegionTypeClass.ProvinceLevelCity1  # 直辖市 - 与省份并行的一级
+        self.ProvinceLevelCity2 = RegionTypeClass.ProvinceLevelCity2  # 直辖市 - 与城市并行的一级
+        self.City = RegionTypeClass.City  # 地级市
+        self.CityLevelDistrict = RegionTypeClass.CityLevelDistrict  # 省直辖县级市
+        self.District = RegionTypeClass.District  # 县、区
+        self.Street = RegionTypeClass.Street  # 街道乡镇一级
+        self.PlatformL4 = RegionTypeClass.PlatformL4  # 特定平台的4级地址
+        self.Town = RegionTypeClass.Town  # 附加：乡镇
+        self.Village = RegionTypeClass.Village  # 附加：村
+
+    @staticmethod
+    def help():
+        print('\n'.join([
+            'RegionType 说明:',
+            '               Country : 国家',
+            '              Province : 省份',
+            '    ProvinceLevelCity1 : 直辖市(与省份并行的一级)',
+            '    ProvinceLevelCity2 : 直辖市(与城市并行的一级)',
+            '                  City : 地级市',
+            '     CityLevelDistrict : 省直辖县级市',
+            '              District : 县、区',
+            '                Street : 街道乡镇一级',
+            '            PlatformL4 : 特定平台的4级地址',
+            '                  Town : 乡镇(附加)',
+            '               Village : 村(附加)',
+            '             Undefined : 未定义区域类型',
+        ]))
+

GeocodingCHN/model/version.py

@@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+# @Time     : 2022/5/26 11:10
+# @File     : version.py
+# @Author  : CasuallyName
+# @Email   : fjkl@vip.qq.com
+# @Software : Python 3.7
+# @About    :
+
+class Version:
+    def __init__(self, package, jar):
+        self.package = package
+        self.jar = jar
+
+    def __repr__(self):
+        return f'Package(GeocodingCHN) version: {self.package}\nSource(geocoding.jar) version: {self.jar}'
+
+    def __str__(self):
+        return f'Package(GeocodingCHN) version: {self.package}, Source(geocoding.jar) version: {self.jar}'

tests/test.py

@@ -0,0 +1,26 @@
+#!/usr/bin/python3
+# -*- coding: utf-8 -*-
+# @Time    : 2021/1/18 13:13
+# @Author  : CasuallyName
+# @Email   : fjkl@vip.qq.com
+# @File    : test.py
+# @Software: PyCharm
+from GeocodingCHN import Geocoding
+
+if __name__ == '__main__':
+    geocoding = Geocoding()
+    geocoding = Geocoding(data_class_path='core/region.dat')
+    print(geocoding.__version__)
+    text1 = '山东青岛李沧区延川路116号绿城城园东区7号楼2单元802户'
+    text2 = '山东青岛李沧区延川路绿城城园东区7-2-802'
+    Address_1 = geocoding.normalizing(text1)
+    print(Address_1)
+    Address_2 = geocoding.normalizing(text2)
+    similar = geocoding.similarityWithResult(Address_1, Address_2)
+    print(similar)
+    print(similar.similarity)
+    print(geocoding.similarity(Address_1, Address_2))
+
+    geocoding.addRegionEntry(1, 321200000000, "A街道", geocoding.RegionType.Street)
+    print(geocoding.normalizing("江苏泰州A街道"))
+    print(geocoding.segment(text2))