diff options
| author | unwox <me@unwox.com> | 2024-10-22 10:54:33 +0600 |
|---|---|---|
| committer | unwox <me@unwox.com> | 2024-10-22 10:54:33 +0600 |
| commit | 477dd002b008db50361d9a87a12c6731cccb6b78 (patch) | |
| tree | 82d5ee8bf9022d41964bc17873b6a2f5669cbae7 /parser | |
| parent | a24b2b3de7995d28420f09c988fbb7ef31d0e15c (diff) | |
improve categories a bit
Diffstat (limited to 'parser')
| -rw-r--r-- | parser/artoftea.fnl | 10 | ||||
| -rw-r--r-- | parser/chaekshop.fnl | 4 | ||||
| -rw-r--r-- | parser/clubcha.fnl | 14 | ||||
| -rw-r--r-- | parser/parser.fnl | 3 |
4 files changed, 16 insertions, 15 deletions
diff --git a/parser/artoftea.fnl b/parser/artoftea.fnl index 7ad4e66..a62cd34 100644 --- a/parser/artoftea.fnl +++ b/parser/artoftea.fnl @@ -71,12 +71,12 @@ {:path "xej-cha-chernyj-chaj" :tags ["Хэй ча"]} {:path "posuda" :tags ["Посуда"]} {:path "tipoty-lightking" :tags ["Посуда"]} - {:path "accesories" :tags ["Аксессуары"]} - {:path "matcha-i-aksessuary"} - {:path "upakovka"} + {:path "accesories" :tags ["Посуда"]} + {:path "matcha-i-aksessuary" :tags ["Матча"]} + {:path "upakovka" :tags ["Упаковка"]} {:path "eksklyuzivny-chay"} - {:path "taiwan-tea"} - {:path "sinij-chaj"}] + {:path "taiwan-tea" :tags ["Тайвань"]} + {:path "sinij-chaj" :tags ["Синий чай"]}] format-url product-peg normalize)) diff --git a/parser/chaekshop.fnl b/parser/chaekshop.fnl index b7699e9..80e6cd5 100644 --- a/parser/chaekshop.fnl +++ b/parser/chaekshop.fnl @@ -57,8 +57,8 @@ {:path "chay/zhyeltyy_chay" :tags ["Желтый чай"]} {:path "chay/zelyenyy_chay" :tags ["Зеленый чай"]} {:path "chay/krasnyy_chay" :tags ["Красный чай"]} - {:path "chay/severo_futszyanskie_uluny" :tags ["Улун"]} - {:path "chay/tayvanskie_uluny" :tags ["Улун"]} + {:path "chay/severo_futszyanskie_uluny" :tags ["Улун" "Фудзянь"]} + {:path "chay/tayvanskie_uluny" :tags ["Улун" "Тайвань"]} {:path "chay/khey_cha" :tags ["Хэй ча"]} {:path "chay/shen_puer" :tags ["Шен пуэр"]} {:path "chay/shu_puery" :tags ["Шу пуэр"]} diff --git a/parser/clubcha.fnl b/parser/clubcha.fnl index 71fbc0f..57334a7 100644 --- a/parser/clubcha.fnl +++ b/parser/clubcha.fnl @@ -80,18 +80,18 @@ (fetcher.from-html [{:path "shu-puer" :tags ["Шу пуэр"]} {:path "shen-puer" :tags ["Шен пуэр"]} - {:path "bai-hao-puer"} + {:path "bai-hao-puer" :tags ["Белый пуэр"]} {:path "hei-cha" :tags ["Хэй ча"]} - {:path "yuzhnofutszyanskij-ulun" :tags ["Улун"]} - {:path "guandunskij-ulun" :tags ["Улун"]} - {:path "severofujianskiy-ulun" :tags ["Улун"]} - {:path "taiwanskiy-ulun" :tags ["Улун"]} - {:path "gaba-ulun-gaba-alishan-i-dr" :tags ["Улун"]} + {:path "yuzhnofutszyanskij-ulun" :tags ["Улун" "Фудзянь"]} + {:path "guandunskij-ulun" :tags ["Улун" "Гуандун"]} + {:path "severofujianskiy-ulun" :tags ["Улун" "Фудзянь"]} + {:path "taiwanskiy-ulun" :tags ["Улун" "Тайвань"]} + {:path "gaba-ulun-gaba-alishan-i-dr" :tags ["Улун" "Габа"]} {:path "zelenyi-chay" :tags ["Зеленый чай"]} {:path "belyi-chay" :tags ["Белый чай"]} {:path "zheltyi-chay" :tags ["Желтый чай"]} {:path "kitajskij-krasnyj-chaj" :tags ["Красный чай"]} - {:path "tajvanskij-krasnyj-chaj" :tags ["Красный чай"]} + {:path "tajvanskij-krasnyj-chaj" :tags ["Красный чай" "Тайвань"]} {:path "Nechaynyye_chai_i_tsvetochnyye_dobavki"} {:path "upakovka-dlya-puera"} {:path "upakovka-dlya-posudy"} diff --git a/parser/parser.fnl b/parser/parser.fnl index acc8a19..bf8feed 100644 --- a/parser/parser.fnl +++ b/parser/parser.fnl @@ -105,7 +105,6 @@ (: (peg.Ct (^ (peg.Ct tag) 1)) :match html)) -;; FIXME: make guessing case insensitive (fn guess-tags [text] (local text (if text (must (luna.utf8.lower text)) "")) @@ -127,6 +126,8 @@ ["Шен пуэр"] (: (anywhere (+ (peg.P "матча") "маття")) :match text) ["Матча"] + (: (anywhere (peg.P "габа")) :match text) + ["Габа"] [])) (fn guess-year [text] |
