assorted: add russian title filters (#13155)

src/Jackett.Common/Definitions/nntt.yml

@@ -654,6 +654,14 @@ caps:
     book-search: [q]
 
 settings:
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: false
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: false
   - name: sort
     type: select
     label: Sort requested from site
@@ -702,6 +710,35 @@ search:
           args: f
     title:
       selector: a.topictitle
+      filters:
+        - name: replace
+          args: ["Кураж-Бамбей", "kurazh"]
+        - name: replace
+          args: ["Кубик в Кубе", "Kubik"]
+        - name: replace
+          args: ["Кравец", "Kravec"]
+        - name: replace
+          args: ["Пифагор", "Pifagor"]
+        - name: replace
+          args: ["Невафильм", "Nevafilm"]
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
+        - name: re_replace
+          args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$1$2$3$4{{ end }}"]
+        - name: re_replace
+          args: ["[\\:\\-\\/\\|\\[\\]]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: [" Rip", "Rip"]
+        - name: replace
+          args: ["WEB DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: a.topictitle
       attribute: href

src/Jackett.Common/Definitions/piratbit.yml

@@ -630,6 +630,14 @@ caps:
     book-search: [q]
 
 settings:
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: true
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: true
   - name: sort
     type: select
     label: Sort requested from site
@@ -645,11 +653,6 @@ settings:
     options:
       2: desc
       1: asc
-  - name: striprussian
-    type: checkbox
-    label: Strip Russian Letters
-    default: true
-
 
 download:
   selectors:
@@ -706,15 +709,39 @@ search:
     title:
       selector: td a.genmed
       filters:
+        - name: replace
+          args: ["Кураж-Бамбей", "kurazh"]
+        - name: replace
+          args: ["Кубик в Кубе", "Kubik"]
+        - name: replace
+          args: ["Кравец", "Kravec"]
+        - name: replace
+          args: ["Пифагор", "Pifagor"]
+        - name: replace
+          args: ["Невафильм", "Nevafilm"]
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
         # normalize to SXXEYY format
         - name: re_replace
-          args: ["[\\:\\-\\/\\|]", " "]
+          args: ["([CСcс]езоны?:?\\s+((?:\\d+)(?:-\\d+)?).*[CСcс]ери[ия]:?\\s+((?:\\d+)(?:-\\d+)?).*?\\d+\\)?)", " - S$2E$3 - "]
         - name: re_replace
-          args: ["(.+)\\s+[CСcс]езон\\s+(\\d+)\\s+[CСcс]ерии\\s+(\\d+)\\s+(\\d+)\\s+из\\s+\\d+(.+)", "$1 S$2E$3-$4 rus $5"]
+          args: ["([CСcс]ери[ия]:?\\s+((?:\\d+)(?:-\\d+)?).*?[?\\d]+\\)?)", " - E$2 - "]
-        - name: re_replace
-          args: ["(.+)\\s+[CСcс]езон\\s+(\\d+)\\s+[CСcс]ери[ия]\\s+(\\d+)\\s+из\\s+\\d+(.+)", "$1 S$2E$3 rus $4"]
         - name: re_replace
           args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$0{{ end }}"]
+        - name: re_replace
+          args: ["[«»()=.,:|\\[\\]\\/]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: ["-Rip", "Rip"]
+        - name: replace
+          args: ["WEB-DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: td a.genmed
       attribute: href

src/Jackett.Common/Definitions/piratbitl.yml

@@ -630,6 +630,14 @@ caps:
     book-search: [q]
 
 settings:
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: true
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: true
   - name: username
     type: text
     label: Username
@@ -651,10 +659,7 @@ settings:
     options:
       2: desc
       1: asc
-  - name: striprussian
+
-    type: checkbox
-    label: Strip Russian Letters
-    default: true
 login:
   path: login.php
   method: post
@@ -720,15 +725,39 @@ search:
     title:
       selector: td a.genmed
       filters:
+        - name: replace
+          args: ["Кураж-Бамбей", "kurazh"]
+        - name: replace
+          args: ["Кубик в Кубе", "Kubik"]
+        - name: replace
+          args: ["Кравец", "Kravec"]
+        - name: replace
+          args: ["Пифагор", "Pifagor"]
+        - name: replace
+          args: ["Невафильм", "Nevafilm"]
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
         # normalize to SXXEYY format
         - name: re_replace
-          args: ["[\\:\\-\\/\\|]", " "]
+          args: ["([CСcс]езоны?:?\\s+((?:\\d+)(?:-\\d+)?).*[CСcс]ери[ия]:?\\s+((?:\\d+)(?:-\\d+)?).*?\\d+\\)?)", " - S$2E$3 - "]
         - name: re_replace
-          args: ["(.+)\\s+[CСcс]езон\\s+(\\d+)\\s+[CСcс]ерии\\s+(\\d+)\\s+(\\d+)\\s+из\\s+\\d+(.+)", "$1 S$2E$3-$4 rus $5"]
+          args: ["([CСcс]ери[ия]:?\\s+((?:\\d+)(?:-\\d+)?).*?[?\\d]+\\)?)", " - E$2 - "]
-        - name: re_replace
-          args: ["(.+)\\s+[CСcс]езон\\s+(\\d+)\\s+[CСcс]ери[ия]\\s+(\\d+)\\s+из\\s+\\d+(.+)", "$1 S$2E$3 rus $4"]
         - name: re_replace
           args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$0{{ end }}"]
+        - name: re_replace
+          args: ["[«»()=.,:|\\[\\]\\/]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: ["-Rip", "Rip"]
+        - name: replace
+          args: ["WEB-DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: td a.genmed
       attribute: href

src/Jackett.Common/Definitions/riperam.yml

@@ -843,6 +843,14 @@ settings:
   - name: password
     type: password
     label: Password
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: false
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: false
   - name: sort
     type: select
     label: Sort requested from site
@@ -855,12 +863,11 @@ settings:
   - name: type
     type: select
     label: Order requested from site
-    default: desc
+    default: d
     options:
       d: desc
       a: asc
 
-
 login:
   path: ucp.php?mode=login
   method: form
@@ -894,9 +901,16 @@ search:
     st: 0
     ch: 300
     t: 0
+  keywordsfilters:
+    - name: re_replace # S01 to 1
+      args: ["(?i)\\bS0*(\\d+)\\b", "$1"]
+    - name: re_replace # S01E01 to 1 1
+      args: ["(?i)\\bS0*(\\d+)E0*(\\d+)\\b", "$1 $2"]
 
   rows:
     selector: li.row:has(a[href^="./download/file.php?id="])
+    filters:
+      - name: andmatch
 
   fields:
     category:
@@ -910,6 +924,44 @@ search:
           args: f
     title:
       selector: a.topictitle
+      filters:
+        - name: replace
+          args: ["Кураж-Бамбей", "kurazh"]
+        - name: replace
+          args: ["Кубик в Кубе", "Kubik"]
+        - name: replace
+          args: ["Кравец", "Kravec"]
+        - name: replace
+          args: ["Пифагор", "Pifagor"]
+        - name: replace
+          args: ["Невафильм", "Nevafilm"]
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
+        # normalize to SXXEYY format
+        - name: re_replace
+          args: ["((?:\\d+)(?:-\\d+)?)\\s+[CСcс]езоны?[,:]?\\s+((?:\\d+)(?:-\\d+)?)\\s+[CСcс]ери[ия].*?\\d+(.*)\\[((?:[12][0-9]{3})(?:-(?:[12][0-9]{3}))?)(.*)", "$3 - S$1E$2 - $4 $5"]
+        - name: re_replace
+          args: ["(.*)?[CСcс]езоны?:?\\s+((?:\\d+)(?:-\\d+)?).*[CСcс]ери[ия]\\s+((?:\\d+)(?:-\\d+)?).*?\\d+(.*)\\[((?:[12][0-9]{3})(?:-(?:[12][0-9]{3}))?)(.*)", "$1 $4 - S$2E$3 - $5 $6"]
+        - name: re_replace
+          args: ["(((?:\\d+)(?:-\\d+)?)\\s+[CСcс]ери[ия].*?\\d+)", "E$2 - "]
+        - name: re_replace
+          args: ["([CСcс]ери[ия]\\s+((?:\\d+)(?:-\\d+)?).*?\\d+)", "E$2 - "]
+        - name: re_replace
+          args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$1$2$3$4{{ end }}"]
+        - name: re_replace
+          args: ["[«»()=.,:|\\[\\]\\/]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: ["-Rip", "Rip"]
+        - name: replace
+          args: ["WEB-DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: a.topictitle
       attribute: href

src/Jackett.Common/Definitions/rus-media.yml

@@ -279,6 +279,14 @@ caps:
     book-search: [q]
 
 settings:
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: false
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: false
   - name: sort
     type: select
     label: Sort requested from site
@@ -327,6 +335,35 @@ search:
           args: f
     title:
       selector: a.topictitle
+      filters:
+        - name: replace
+          args: ["Кураж-Бамбей", "kurazh"]
+        - name: replace
+          args: ["Кубик в Кубе", "Kubik"]
+        - name: replace
+          args: ["Кравец", "Kravec"]
+        - name: replace
+          args: ["Пифагор", "Pifagor"]
+        - name: replace
+          args: ["Невафильм", "Nevafilm"]
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
+        - name: re_replace
+          args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$1$2$3$4{{ end }}"]
+        - name: re_replace
+          args: ["[\\:\\-\\/\\|\\[\\]]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: [" Rip", "Rip"]
+        - name: replace
+          args: ["WEB DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: a.topictitle
       attribute: href

src/Jackett.Common/Definitions/selezen.yml

@@ -50,6 +50,14 @@ settings:
   - name: password
     type: password
     label: Password
+  - name: striprussian
+    type: checkbox
+    label: Strip Russian Letters
+    default: false
+  - name: addrussian
+    type: checkbox
+    label: Add RUSSIAN to end of all titles to improve language detection by Sonarr and Radarr. Will cause English-only results to be misidentified.
+    default: false
   - name: sort
     type: select
     label: Sort requested from site
@@ -133,6 +141,25 @@ search:
         a[href$="/3dblu-ray1080p/"]: 65
     title:
       selector: div.short-title > a
+      filters:
+        - name: replace
+          args: ["Лицензия", "Lic"]
+        - name: replace
+          args: ["селезень", "selezen"]
+        - name: re_replace
+          args: ["(\\([А-Яа-яЁё\\W]+\\))|(^[А-Яа-яЁё\\W\\d]+\\/ )|([а-яА-ЯЁё \\-]+,+)|([а-яА-ЯЁё]+)", "{{ if .Config.striprussian }}{{ else }}$1$2$3$4{{ end }}"]
+        - name: re_replace
+          args: ["[\\:\\-\\/\\|\\[\\]]", " "]
+        - name: append
+          args: "{{ if .Config.addrussian }} - RUSSIAN{{ else }}{{ end }}"
+        - name: replace
+          args: [" Rip", "Rip"]
+        - name: replace
+          args: ["WEB DL", "WEBDL"]
+        - name: replace
+          args: ["WEBDLRip", "WEBDL"]
+        - name: replace
+          args: ["HDTVRip", "HDTV"]
     details:
       selector: div.short-title > a
       attribute: href