From 55f88e28a5dc31724bd973e82d580ad43ea4ef59 Mon Sep 17 00:00:00 2001
From: Felipe Allegretti <felipeallh@gmail.com>
Date: Tue, 19 Sep 2023 15:23:56 -0300
Subject: [PATCH] Implement goalkeeper parser

---
 app/services/players/stats.py | 69 ++++++++++++++++++++++++++++++++++-
 app/utils/xpath.py            |  1 +
 2 files changed, 69 insertions(+), 1 deletion(-)

diff --git a/app/services/players/stats.py b/app/services/players/stats.py
index fa95457..d7971c5 100644
--- a/app/services/players/stats.py
+++ b/app/services/players/stats.py
@@ -8,6 +8,7 @@
     clean_response,
     extract_from_url,
     get_text_by_xpath,
+    remove_str,
     request_url_page,
     trim,
 )
@@ -22,8 +23,17 @@ class TransfermarktPlayerStats:
     def __post_init__(self):
         self._request_player_page()
         self._check_player_found()
+        self._check_player_goalkeeper()
 
     def get_player_stats(self) -> dict:
+        self.player_stats["id"] = self.player_id
+        self.player_stats["stats"] = (
+            self.parse_player_stats_goalkeeper() if self.goalkeeper else self.parse_player_stats_field()
+        )
+        self.player_stats["lastUpdate"] = datetime.now()
+        return clean_response(self.player_stats)
+
+    def parse_player_stats_field(self) -> dict:
         stats: list[ElementTree] = self.page.xpath(Players.Stats.PLAYER_STATS)
 
         seasons = [elem.text for elem in stats[::9]]
@@ -40,7 +50,7 @@ def get_player_stats(self) -> dict:
         yellow_cards = [card[0] for card in cards]
         second_yellow_cards = [card[1] for card in cards]
         red_cards = [card[2] for card in cards]
-        minutes_played = [trim(elem.text) for elem in stats[8::9]]
+        minutes_played = [remove_str(trim(elem.text), ["'", "."]) for elem in stats[8::9]]
 
         self.player_stats["id"] = self.player_id
         self.player_stats["stats"] = [
@@ -77,6 +87,60 @@ def get_player_stats(self) -> dict:
         self.player_stats["lastUpdate"] = datetime.now()
         return clean_response(self.player_stats)
 
+    def parse_player_stats_goalkeeper(self):
+        stats: list[ElementTree] = self.page.xpath(Players.Stats.PLAYER_STATS)
+
+        seasons = [elem.text for elem in stats[::10]]
+        competitions_ids = [
+            extract_from_url(get_text_by_xpath(elem, Players.Stats.COMPETITIONS_IDS)) for elem in stats[2::10]
+        ]
+        competitions_names = [get_text_by_xpath(elem, Players.Stats.COMPETITIONS_NAMES) for elem in stats[2::10]]
+        clubs_ids = [extract_from_url(get_text_by_xpath(elem, Players.Stats.CLUBS_IDS)) for elem in stats[3::10]]
+        clubs_names = [get_text_by_xpath(elem, Players.Stats.CLUBS_NAMES) for elem in stats[3::10]]
+        appearances = [get_text_by_xpath(elem, Players.Stats.APPEARANCES) for elem in stats[4::10]]
+
+        goals = [elem.text for elem in stats[5::10]]
+        cards = [trim(elem.text).split("/") for elem in stats[6::10]]
+        yellow_cards = [card[0] for card in cards]
+        second_yellow_cards = [card[1] for card in cards]
+        red_cards = [card[2] for card in cards]
+        goals_conceded = [elem.text for elem in stats[7::10]]
+        clean_sheets = [elem.text for elem in stats[8::10]]
+        minutes_played = [remove_str(trim(elem.text), ["'", "."]) for elem in stats[9::10]]
+
+        return [
+            {
+                "season": season,
+                "competitionId": competition_id,
+                "competitionName": competition_name,
+                "clubId": club_id,
+                "clubName": club_name,
+                "appearances": appearance,
+                "goals": goal,
+                "yellowCards": yellow_card,
+                "secondYellowCards": second_yellow_card,
+                "redCards": red_card,
+                "goalsConceded": red_card,
+                "cleanSheets": red_card,
+                "minutesPlayed": minute_played,
+            }
+            for season, competition_id, competition_name, club_id, club_name, appearance, goal, yellow_card, second_yellow_card, red_card, conceded, clean, minute_played in zip(  # noqa: E501
+                seasons,
+                competitions_ids,
+                competitions_names,
+                clubs_ids,
+                clubs_names,
+                appearances,
+                goals,
+                yellow_cards,
+                second_yellow_cards,
+                red_cards,
+                goals_conceded,
+                clean_sheets,
+                minutes_played,
+            )
+        ]
+
     def _request_player_page(self) -> None:
         player_url = f"https://www.transfermarkt.com/-/leistungsdatendetails/spieler/{self.player_id}"
         self.page = request_url_page(url=player_url)
@@ -84,3 +148,6 @@ def _request_player_page(self) -> None:
     def _check_player_found(self) -> None:
         if not get_text_by_xpath(self, Players.Profile.URL):
             raise HTTPException(status_code=404, detail=f"Player Stats not found for id: {self.player_id}")
+
+    def _check_player_goalkeeper(self):
+        self.goalkeeper = get_text_by_xpath(self, Players.Stats.GOALKEEPER) == "Goalkeeper"
diff --git a/app/utils/xpath.py b/app/utils/xpath.py
index 3f3224b..0411ce0 100644
--- a/app/utils/xpath.py
+++ b/app/utils/xpath.py
@@ -84,6 +84,7 @@ class Stats:
         CLUBS_IDS: str = ".//a/@href"
         CLUBS_NAMES: str = ".//a//@title"
         APPEARANCES: str = ".//a//text()"
+        GOALKEEPER: str = "//li[contains(text(), 'Position:')]//span//text()"
 
 
 class Clubs: