From 7f43f22faf57c308d5a2baf2e579323d2f84ac01 Mon Sep 17 00:00:00 2001 From: interlark Date: Tue, 25 Oct 2022 00:25:55 +0500 Subject: [PATCH] Update cities, rubtics. Disable headless parsing by default. --- CHANGELOG.md | 10 +++- parser_2gis/chrome/options.py | 2 +- parser_2gis/data/cities.json | 24 ++++----- parser_2gis/data/rubrics.json | 95 +++++++++++++++++++++++++++++------ parser_2gis/version.py | 2 +- 5 files changed, 102 insertions(+), 31 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index bfc7e2a..508c8d9 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,6 +2,13 @@ ## [Невошедшее] +## [0.1.10] - 25-10-2022 +### Добавлено +- Обновлён список рубрик и городов. + +### Исправлено +- Отключен скрытый режим парсинга по-умолчанию. + ## [0.1.9] - 18-08-2022 ### Добавлено - Новые рубрики: *Клубы настольного тенниса, Атрибутика для болельщиков, Полицейские станции*. @@ -57,7 +64,8 @@ - Первый релиз. -[Невошедшее]: https://github.com/interlark/parser-2gis/compare/v0.1.9...HEAD +[Невошедшее]: https://github.com/interlark/parser-2gis/compare/v0.1.10...HEAD +[0.1.10]: https://github.com/interlark/parser-2gis/compare/v0.1.9...v0.1.10 [0.1.9]: https://github.com/interlark/parser-2gis/compare/v0.1.8...v0.1.9 [0.1.8]: https://github.com/interlark/parser-2gis/compare/v0.1.7...v0.1.8 [0.1.7]: https://github.com/interlark/parser-2gis/compare/v0.1.6...v0.1.7 diff --git a/parser_2gis/chrome/options.py b/parser_2gis/chrome/options.py index d36a4f2..0e054e7 100644 --- a/parser_2gis/chrome/options.py +++ b/parser_2gis/chrome/options.py @@ -28,7 +28,7 @@ class ChromeOptions(BaseModel): """ binary_path: Optional[pathlib.Path] start_maximized: bool = False - headless: bool = True + headless: bool = False disable_images: bool = True silent_browser: bool = True memory_limit: PositiveInt = default_memory_limit() diff --git a/parser_2gis/data/cities.json b/parser_2gis/data/cities.json index 4cf5d8d..a3344ad 100644 --- a/parser_2gis/data/cities.json +++ b/parser_2gis/data/cities.json @@ -479,6 +479,12 @@ "domain": "ru", "country_code": "ru" }, + { + "name": "Кызылорда", + "code": "kyzylorda", + "domain": "ru", + "country_code": "kz" + }, { "name": "Ленинск-Кузнецкий", "code": "lenkuz", @@ -527,6 +533,12 @@ "domain": "ru", "country_code": "ru" }, + { + "name": "Московская область", + "code": "moscow_region", + "domain": "ru", + "country_code": "ru" + }, { "name": "Мурманская область", "code": "murmansk", @@ -767,18 +779,6 @@ "domain": "ru", "country_code": "ru" }, - { - "name": "Сергиев Посад", - "code": "s_posad", - "domain": "ru", - "country_code": "ru" - }, - { - "name": "Серпухов", - "code": "serpuhov", - "domain": "ru", - "country_code": "ru" - }, { "name": "Сланцы", "code": "slantsy", diff --git a/parser_2gis/data/rubrics.json b/parser_2gis/data/rubrics.json index d7d6d31..fd578bc 100644 --- a/parser_2gis/data/rubrics.json +++ b/parser_2gis/data/rubrics.json @@ -220,6 +220,7 @@ "50535", "116", "50642", + "112584", "70514", "112583", "71182", @@ -346,10 +347,10 @@ "9786", "67763", "162", + "112719", "16677", "111594", "51459", - "112683", "164", "52248", "166", @@ -368,6 +369,7 @@ "parentCode": "2", "children": [ "112455", + "110380", "537", "11648", "72370", @@ -1098,6 +1100,7 @@ "56161", "112582", "58618", + "112691", "108687", "110371", "95584", @@ -1334,12 +1337,12 @@ "code": "69", "parentCode": "1035", "children": [ + "112448", "1182", "14426", "354", "605", "606", - "112448", "112678", "355", "551", @@ -1396,6 +1399,7 @@ "561", "1273", "112674", + "112704", "110976", "366", "559", @@ -2036,6 +2040,7 @@ "20165", "652", "13726", + "112695", "110355", "67142", "112650", @@ -6337,7 +6342,7 @@ "879": { "isRussian": true, "isNonRussian": true, - "label": "Снэковая продукция", + "label": "Снековая продукция", "code": "879", "parentCode": "70", "children": [] @@ -6355,7 +6360,8 @@ "110370", "19288", "372", - "558" + "558", + "112683" ] }, "921": { @@ -9986,6 +9992,7 @@ "416", "517", "422", + "112697", "9505", "420" ] @@ -11392,7 +11399,7 @@ }, "52248": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Рюмочные", "code": "52248", "parentCode": "25", @@ -12217,7 +12224,7 @@ }, "56581": { "isRussian": true, - "isNonRussian": true, + "isNonRussian": false, "label": "Избирательные участки", "code": "56581", "parentCode": "44", @@ -13498,6 +13505,14 @@ "parentCode": "19519", "children": [] }, + "110380": { + "isRussian": true, + "isNonRussian": false, + "label": "Аквапарки", + "code": "110380", + "parentCode": "26", + "children": [] + }, "110381": { "isRussian": true, "isNonRussian": true, @@ -14020,7 +14035,7 @@ }, "110469": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Промышленный клининг", "code": "110469", "parentCode": "114", @@ -14100,7 +14115,7 @@ }, "110487": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Полигоны размещения отходов", "code": "110487", "parentCode": "18", @@ -14124,7 +14139,7 @@ }, "110490": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Лодочные станции", "code": "110490", "parentCode": "19519", @@ -14973,7 +14988,7 @@ "112448": { "isRussian": true, "isNonRussian": true, - "label": "Косплей товары", + "label": "Аниме / косплей товары", "code": "112448", "parentCode": "69", "children": [] @@ -14996,7 +15011,7 @@ }, "112455": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Авиатренажеры", "code": "112455", "parentCode": "26", @@ -15076,7 +15091,7 @@ }, "112525": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Кризисные центры", "code": "112525", "parentCode": "44", @@ -15170,6 +15185,14 @@ "parentCode": "16", "children": [] }, + "112584": { + "isRussian": true, + "isNonRussian": false, + "label": "Администрации районов города", + "code": "112584", + "parentCode": "16", + "children": [] + }, "112585": { "isRussian": true, "isNonRussian": false, @@ -15316,7 +15339,7 @@ }, "112617": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Стендап-клубы", "code": "112617", "parentCode": "26", @@ -15655,12 +15678,12 @@ "isNonRussian": true, "label": "Продажа чая", "code": "112683", - "parentCode": "25", + "parentCode": "913", "children": [] }, "112684": { "isRussian": true, - "isNonRussian": false, + "isNonRussian": true, "label": "Клубы настольного тенниса", "code": "112684", "parentCode": "19519", @@ -15674,6 +15697,46 @@ "parentCode": "48", "children": [] }, + "112691": { + "isRussian": true, + "isNonRussian": false, + "label": "Общественный душ", + "code": "112691", + "parentCode": "56", + "children": [] + }, + "112695": { + "isRussian": true, + "isNonRussian": true, + "label": "Услуги пирсинга", + "code": "112695", + "parentCode": "106", + "children": [] + }, + "112697": { + "isRussian": true, + "isNonRussian": false, + "label": "Залы ожидания повышенного комфорта", + "code": "112697", + "parentCode": "22191", + "children": [] + }, + "112704": { + "isRussian": true, + "isNonRussian": false, + "label": "Растениеводство", + "code": "112704", + "parentCode": "70", + "children": [] + }, + "112719": { + "isRussian": true, + "isNonRussian": true, + "label": "Кофейные автоматы", + "code": "112719", + "parentCode": "25", + "children": [] + }, "-1": { "isRussian": true, "isNonRussian": true, @@ -15717,4 +15780,4 @@ "969" ] } -} +} \ No newline at end of file diff --git a/parser_2gis/version.py b/parser_2gis/version.py index b654e16..81ab893 100644 --- a/parser_2gis/version.py +++ b/parser_2gis/version.py @@ -1,4 +1,4 @@ """Version info.""" -version = '0.1.9' +version = '0.1.10' config_version = '0.1'