def main(argv):
reader = GomiReader()
reader.municipality_id = "232114"
reader.municipality_name = "愛知県豊田市"
reader.datasource_url = "https://manage.delight-system.com/threeR/web/bunbetsu?menu=bunbetsu&jichitaiId=toyotashi&lang=ja"
reader.target_url_base = "https://manage.delight-system.com/threeR/web/bunbetsu?menu=bunbetsu&jichitaiId=toyotashi&lang=ja"
reader.target_pages = [ "" ]
reader.datetime_selector = None
reader.datetime_pattern = None
reader.article_row_selector = "div.panel-heading"
reader.article_column_selector = "a.panel-title, div.panel-body"
reader.category_to_category_id = {
PatternValuePair("燃やすごみ", "burnable"),
PatternValuePair("埋めるごみ", "unburnable"),
PatternValuePair("粗大ごみ", "oversized"),
PatternValuePair("資源（ガラスびん・飲料缶・ペットボトル・有害ごみ・危険ごみ）", "recyclable"),
PatternValuePair("プラスチック製容器包装", "plasticpackaging"),
PatternValuePair("金属ごみ", "metal"),
PatternValuePair("古布類", "pointcollection.cloth"),
PatternValuePair("古紙類", "pointcollection.paper"),
PatternValuePair("家電４品目（市では収集しないもの）", "legalrecycling"),
PatternValuePair("禁止（市では収集しないもの）", "uncollectible")
}
# reader.note_to_category_id
reader.category_definitions = {
"burnable": { "name": "燃やすごみ" },
"unburnable": { "name": "埋めるごみ" },
"oversized": { "name": "粗大ごみ" },
"plasticpackaging": { "name": "プラ容器包装" },
"metal": { "name": "金属ごみ" },
"recyclable": { "name": "資源" },
"pointcollection.cloth": { "name": "古布類" },
"pointcollection.paper": { "name": "古紙類" },
"legalrecycling": { "name": "家電リサイクル法対象" },
"uncollectible": { "name": "回収できません" }
}
print(reader.to_json())
def main(argv):
reader = GomiReader()
reader.municipality_id = "232076"
reader.municipality_name = "愛知県豊川市"
reader.datasource_url = "https://www.city.toyokawa.lg.jp/smph/kurashi/gomirecycle/gomihayamihyo/"
reader.target_url_base = "https://www.city.toyokawa.lg.jp/smph/kurashi/gomirecycle/gomihayamihyo/"
reader.target_pages = [
"agyo.html", "kagyo.html", "sagyo.html", "tagyo.html", "nagyo.html",
"hagyo.html", "magyo.html", "yagyo.html", "ragyo.html", "wagyo.html"
]
reader.datetime_selector = "p.update"
reader.datetime_pattern = "更新日:%Y年%m月%d日"
reader.article_row_selector = "caption ~ tr"
reader.article_column_selector = "td"
reader.category_to_category_id = [
PatternValuePair("可燃ごみ", "burnable"),
PatternValuePair("不燃ごみ", "unburnable"),
PatternValuePair("資源", "recyclable"),
PatternValuePair("危険ごみ", "hazardous"),
PatternValuePair("粗大ごみ", "oversized"),
PatternValuePair("家電リサイクル法対象品目", "legalrecycling"),
PatternValuePair("市で処理できません。", "uncollectible")
]
# reader.note_to_category_id
reader.category_definitions = {
"burnable": {
"name": "可燃ごみ"
},
"unburnable": {
"name": "不燃ごみ"
},
"recyclable": {
"name": "資源"
},
"hazardous": {
"name": "危険ごみ"
},
"oversized": {
"name": "粗大ごみ"
},
"legalrecycling": {
"name": "家電リサイクル法対象"
},
"uncollectible": {
"name": "回収できません"
}
}
print(reader.to_json())
def main(args):
reader = AichiNishioGomiReader()
reader.municipality_id = "232131"
reader.municipality_name = "愛知県西尾市"
reader.datasource_url = None
reader.target_url_base = "file:data/nishio/"
reader.target_pages = [ "20191202-092059.pdf" ]
# reader.datetime_selector = None
# reader.datetime_pattern = None
# reader.article_row_selector = None
# reader.article_column_selector = None
reader.category_to_category_id = {
PatternValuePair("/.*/", "burnable")
}
# reader.note_to_category_id
reader.category_definitions = {
"burnable": { "name": "もやすごみ" },
}
print(reader.to_json())
def main(argv):
# メモ:
# 市のホームページでは「ビールびん」の分類が「空きびん」と誤記されているので
# 空きびん・生きびんについては分類ではなく備考により分別するようにした。
reader = GomiReader()
reader.municipality_id = "232025"
reader.municipality_name = "愛知県岡崎市"
reader.datasource_url = "https://www.city.okazaki.lg.jp/1100/1108/1151/p003039.html"
reader.target_url_base = "https://www.city.okazaki.lg.jp/1100/1108/1151/"
reader.target_pages = [
"p003041.html", "p003042.html", "p003043.html", "p003044.html",
"p003045.html", "p003046.html", "p003047.html", "p003048.html",
"p003049.html", "p003050.html"
]
reader.datetime_selector = "span.date"
reader.datetime_pattern = "最終更新日%Y年%m月%d日"
reader.article_row_selector = "tbody > tr"
reader.article_column_selector = "td"
reader.category_to_category_id = {
PatternValuePair(r"/（?可燃ごみ）?/", "burnable"),
PatternValuePair("不燃ごみ", "unburnable"),
PatternValuePair("粗大ごみ", "oversized"),
PatternValuePair("有害ごみ", "hazardous.harmful"),
PatternValuePair("発火性危険ごみ", "hazardous.ignitable"),
PatternValuePair("紙製容器包装", "paperpackaging"),
PatternValuePair(r"/プラスチック製容器包?装/", "plasticpackaging"),
PatternValuePair("ペットボトル", "petbottle"),
PatternValuePair("空き缶", "can"),
PatternValuePair("処理困難物", "uncollectible.difficult"),
PatternValuePair("リサイクル料金が必要。製造メーカーの電話受付窓口へ",
"uncollectible.makercollection"),
PatternValuePair(r"/家電4品目\sリサイクル料金が必要/", "legalrecycling")
}
reader.note_to_category_id = [
PatternValuePair("資源回収または拠点回収へ", "localcollection"),
PatternValuePair(r"/.*回収協力店へ.*/", "pointcollection"),
PatternValuePair(r"/.*販売店へ.*/", "uncollectible.sellercollection"),
PatternValuePair(r"/.*自主回収へ.*/", "uncollectible.sellercollection"),
PatternValuePair(r"/.*青色のコンテナへ/", "can"),
PatternValuePair(r"/.*茶色のコンテナへ/", "grassbottle"),
PatternValuePair(r"/.*白色のコンテナへ/", "reusebottle")
]
reader.category_definitions = {
"burnable": {
"name": "可燃ごみ"
},
"unburnable": {
"name": "不燃ごみ"
},
"oversized": {
"name": "粗大ごみ"
},
"hazardous": {
"name": "危険/有害ごみ"
},
"hazardous.ignitable": {
"name": "発火性危険ごみ"
},
"hazardous.harmful": {
"name": "有害ごみ"
},
"paperpackaging": {
"name": "紙製容器包装"
},
"plasticpackaging": {
"name": "プラ容器包装"
},
"petbottle": {
"name": "ペットボトル"
},
"reusebottle": {
"name": "生きびん"
},
"grassbottle": {
"name": "空きびん"
},
"can": {
"name": "空き缶"
},
"localcollection": {
"name": "資源回収/拠点回収"
},
"pointcollection": {
"name": "回収ボックス"
},
"legalrecycling": {
"name": "家電リサイクル法対象"
},
"uncollectible": {
"name": "回収できません"
},
"uncollectible.difficult": {
"name": "回収できません(処理困難)"
},
"uncollectible.sellercollection": {
"name": "回収できません(販売店回収)"
},
"uncollectible.makercollection": {
"name": "回収できません(メーカー回収)"
}
}
print(reader.to_json())
def main(argv):
reader = GomiReader()
reader.municipality_id = "232033"
reader.municipality_name = "愛知県一宮市"
reader.datasource_url = "https://www.city.ichinomiya.aichi.jp/kurashi/gomi/1000388/1000167/index.html"
reader.target_url_base = "https://www.city.ichinomiya.aichi.jp/kurashi/gomi/1000388/1000167/"
reader.target_pages = [
"1001702.html", "1001703.html", "1001700.html", "1001701.html",
"1001698.html", "1001699.html", "1001696.html", "1001697.html",
"1001694.html", "1001695.html"
]
reader.datetime_selector = "p.update"
reader.datetime_pattern = "更新日\r\n%Y年%m月%d日"
reader.article_row_selector = "tbody > tr"
reader.article_column_selector = "td"
reader.category_to_category_id = {
PatternValuePair(r"/可燃ごみ.*/", "burnable"),
PatternValuePair(r"/不燃ごみ.*/", "unburnable"),
PatternValuePair(r"/粗大ごみ.*/", "oversized"),
PatternValuePair(r"/空き缶・金属類.*/", "metal"),
PatternValuePair(r"/プラスチック製容器包装.*/", "plasticpackaging"),
PatternValuePair("ペットボトル", "petbottle"),
PatternValuePair(r"/町内回収資源.*/", "localcollection"),
PatternValuePair("戸別収集", "housecollection"),
PatternValuePair(r"/市では収集.*できません.*/", "uncollectible")
}
reader.note_to_category_id = [
PatternValuePair(r"/家電リサイクル法対象品目.*/", "legalrecycling")
]
reader.category_definitions = {
"burnable": {
"name": "可燃ごみ"
},
"unburnable": {
"name": "不燃ごみ"
},
"oversized": {
"name": "粗大ごみ"
},
"metal": {
"name": "金属ごみ"
},
"plasticpackaging": {
"name": "プラ容器包装"
},
"petbottle": {
"name": "ペットボトル"
},
"localcollection": {
"name": "町内回収資源"
},
"housecollection": {
"name": "戸別回収"
},
"legalrecycling": {
"name": "家電リサイクル法対象品"
},
"uncollectible": {
"name": "回収できません"
}
}
print(reader.to_json())
def main(argv):
reader = GomiReader()
reader.municipality_id = "232122"
reader.municipality_name = "愛知県安城市"
reader.datasource_url = "https://www.city.anjo.aichi.jp/kurasu/gomi/gomidashi/gomidasu/index.html"
reader.target_url_base = "https://www.city.anjo.aichi.jp/kurasu/gomi/gomidashi/gomidasu/"
reader.target_pages = [
"a.html", "i.html", "u.html", "e.html", "o.html", "ka.html", "ki.html",
"ku.html", "ke.html", "ko.html", "sa.html", "shi.html", "su.html",
"se.html", "so.html", "ta.html", "chi.html", "tsu.html", "te.html",
"to.html", "na.html", "ni.html", "nu.html", "ne.html", "no.html",
"ha.html", "hi.html", "fu.html", "he.html", "ho.html", "ma.html",
"mi.html", "mu.html", "me.html", "mo.html", "ya.html", "yu.html",
"yo.html", "ra.html", "ri.html", "ru.html", "re.html", "ro.html",
"wa.html"
]
reader.datetime_selector = "p#tmp_update"
reader.datetime_pattern = "更新日：%Y年%m月%d日"
reader.article_row_selector = "p ~ table > tbody > tr"
reader.article_column_selector = "td"
reader.category_to_category_id = [
PatternValuePair("燃やせるごみ", "burnable"),
PatternValuePair("燃やせないごみ", "unburnable"),
PatternValuePair("プラスチック製容器包装", "plasticpackaging"),
PatternValuePair(r"/拠点.*品目/", "pointcollection"),
PatternValuePair(r"/資源ごみ（(びん|缶類|びん・缶類)）/", "grassbottle"),
PatternValuePair("資源ごみ（破砕困難ごみ・危険ごみ）", "hazardous"),
PatternValuePair("資源ごみ（古着）", "cloth"),
PatternValuePair("資源ごみ（古紙）", "paper"),
PatternValuePair("粗大ごみ", "oversized"),
PatternValuePair("持ち込みごみ", "oversized")
]
reader.note_to_category_id = {
PatternValuePair(r"/家電リサイクル法対象品目です.*/", "legalrecycling"),
PatternValuePair(r"/.*/", "uncollectible")
}
reader.category_definitions = {
"burnable": {
"name": "燃やせるごみ"
},
"unburnable": {
"name": "燃やせないごみ"
},
"plasticpackaging": {
"name": "プラスチック製容器包装"
},
"pointcollection": {
"name": "拠点回収品目"
},
"grassbottle": {
"name": "びん・缶類"
},
"hazardous": {
"name": "破砕困難ごみ・危険ごみ"
},
"cloth": {
"name": "古着"
},
"paper": {
"name": "古紙"
},
"oversized": {
"name": "粗大ごみ・持ち込みごみ"
},
"legalrecycling": {
"name": "家電リサイクル法対象品目"
},
"uncollectible": {
"name": "回収できません"
}
}
print(reader.to_json())
def main(argv):
reader = GomiReader()
reader.municipality_id = "231002"
reader.municipality_name = "愛知県名古屋市"
reader.datasource_url = "http://www.city.nagoya.jp/kurashi/category/5-6-22-0-0-0-0-0-0-0.html"
reader.target_url_base = "http://www.city.nagoya.jp/kankyo/page/"
reader.target_pages = [
"0000066278.html", "0000066280.html", "0000066282.html",
"0000066283.html", "0000066302.html", "0000066303.html",
"0000066305.html", "0000066307.html", "0000066308.html",
"0000066309.html"
]
reader.datetime_selector = "span.syosai_hiduke"
reader.datetime_pattern = "最終更新日：%Y年%m月%d日"
reader.article_row_selector = "tbody > tr"
reader.article_column_selector = "td"
reader.category_to_category_id = [
PatternValuePair("可燃ごみ", "burnable"),
PatternValuePair("不燃ごみ", "unburnable"),
PatternValuePair("粗大ごみ", "oversized"),
PatternValuePair("発火性危険物", "hazardous"),
PatternValuePair("紙製容器包装", "paperpackaging"),
PatternValuePair("プラ容器包装", "plasticpackaging"),
PatternValuePair("ペットボトル", "petbottle"),
PatternValuePair("空きびん", "grassbottle"),
PatternValuePair("空き缶", "can"),
PatternValuePair("紙パック", "beveragepack"),
PatternValuePair("食用油", "pointcollection.edibleoil"),
PatternValuePair("小型家電", "pointcollection.smallappliances")
]
reader.note_to_category_id = [
PatternValuePair(r"/.*処理して可燃ごみへ/", "burnable"),
PatternValuePair(r"/.*家電リサイクル法対象.*/", "legalrecycling"),
PatternValuePair(r"/.*集団資源回収.*/", "localcollection"),
PatternValuePair(r"/.*小型家電回収ボックス.*/",
"pointcollection.smallappliances"),
PatternValuePair(r"/.*(環境事業所|協力店|販売店|消火器|病院・診療所|ご相談ください).*/",
"uncollectible")
]
reader.category_definitions = {
"burnable": {
"name": "可燃ごみ"
},
"unburnable": {
"name": "不燃ごみ"
},
"oversized": {
"name": "粗大ごみ"
},
"hazardous": {
"name": "発火性危険物"
},
"paperpackaging": {
"name": "紙製容器包装"
},
"plasticpackaging": {
"name": "プラ容器包装"
},
"beveragepack": {
"name": "紙パック"
},
"petbottle": {
"name": "ペットボトル"
},
"grassbottle": {
"name": "空きびん"
},
"can": {
"name": "空き缶"
},
"legalrecycling": {
"name": "集団資源回収"
},
"pointcollection": {
"name": "回収ボックス"
},
"pointcollection.edibleoil": {
"name": "回収ボックス(食用油)"
},
"pointcollection.smallappliances": {
"name": "回収ボックス(小型家電)"
},
"localcollection": {
"name": "集団資源回収"
},
"legalrecycling": {
"name": "家電リサイクル法対象"
},
"uncollectible": {
"name": "回収できません"
}
}
print(reader.to_json())
def main(argv):
reader = AichiToyohashiGomiReader()
reader.municipality_id = "232017"
reader.municipality_name = "愛知県豊橋市"
reader.datasource_url = "http://www2.city.toyohashi.aichi.jp/kankyo/dictionary/index.html"
reader.target_url_base = "http://www2.city.toyohashi.aichi.jp/kankyo/dictionary/search.cgi?mode=index&key="
reader.target_pages = ["1", "2", "3", "4", "5", "6", "7", "8", "9"]
reader.datetime_selector = None
reader.datetime_pattern = None
reader.article_row_selector = "form > table > tr > td > table > tr > td > table > tr"
reader.article_column_selector = "td"
reader.category_to_category_id = {
PatternValuePair("生ごみ", "foodscraps"),
PatternValuePair("もやすごみ", "burnable"),
PatternValuePair("こわすごみ", "crushable"),
PatternValuePair("危険ごみ", "hazardous"),
PatternValuePair("うめるごみ", "unburnable"),
PatternValuePair("大きなごみ", "oversized"),
PatternValuePair("びん・カン", "grassbottle"),
PatternValuePair("ペットボトル", "petbottle"),
PatternValuePair("プラマークごみ", "plasticpackaging"),
PatternValuePair("布類", "cloth"),
PatternValuePair("古紙", "localcollection.paper"),
PatternValuePair("家電４品目", "legalrecycling"),
PatternValuePair("―", "uncollectible"),
PatternValuePair("パソコン", "uncollectible"),
PatternValuePair("収集しません", "uncollectible")
}
# reader.note_to_category_id
reader.category_definitions = {
"burnable": {
"name": "もやすごみ",
"note": "週2回／指定ごみ袋"
},
"foodscraps": {
"name": "生ごみ",
"note": "週2回／指定ごみ袋"
},
"plasticpackaging": {
"name": "プラマークごみ",
"note": "水曜日／透明又は半透明の袋"
},
"crushable": {
"name": "こわすごみ",
"note": "月曜日又は火曜日（４週に１回）／指定ごみ袋"
},
"unburnable": {
"name": "うめるごみ",
"note": "月曜日又は火曜日（８週に１回）／透明又は半透明の袋"
},
"grassbottle": {
"name": "びん・カン",
"note": "木曜日又は金曜日／透明又は半透明の袋"
},
"petbottle": {
"name": "ペットボトル",
"note": "水曜日／透明又は半透明の袋"
},
"localcollection": {
"name": "地域資源回収"
},
"localcollection.paper": {
"name": "地域資源回収(古紙)"
},
"cloth": {
"name": "布類",
"note": "月曜日又は火曜日（８週に１回）／透明又は半透明の袋"
},
"hazardous": {
"name": "危険ごみ",
"note": "水曜日（４週に１回）／透明又は半透明の袋"
},
"oversized": {
"name": "大きなごみ",
"note": "戸別有料収集又は資源化センターへの自己搬入"
},
"legalrecycling": {
"name": "家電４品目"
},
"uncollectible": {
"name": "収集しません"
}
}
print(reader.to_json())