Method read_teams

soccerdata/sofifa.py:170–219

Retrieve all teams for the selected leagues. Returns ------- pd.DataFrame

(self)

        return pd.DataFrame(versions).set_index("version_id").sort_index()

    def read_teams(self) -> pd.DataFrame:
        """Retrieve all teams for the selected leagues.

        Returns
        -------
        pd.DataFrame
        """
        # build url
        urlmask = SO_FIFA_API + "/teams?lg={}&r={}&set=true"
        filemask = "teams_{}_{}.html"

        # get league IDs
        leagues = self.read_leagues()

        # collect teams
        teams = []
        iterator = list(product(leagues.iterrows(), self.versions.iterrows()))
        for i, ((lkey, league), (version_id, version)) in enumerate(iterator):
            logger.info(
                "[%s/%s] Retrieving teams for %s in %s edition",
                i + 1,
                len(iterator),
                lkey,
                version["update"],
            )
            league_id = league["league_id"]
            # read html page (league overview)
            filepath = self.data_dir / filemask.format(league_id, version_id)
            url = urlmask.format(league_id, version_id)
            reader = self.get(url, filepath)

            # extract team links
            tree = html.parse(reader)
            pat_team = re.compile(r"\/team\/(\d+)\/[\w-]+\/")
            for node in tree.xpath("//table/tbody/tr"):
                # extract team IDs from links
                team_link = node.xpath(".//td[2]//a")[0]
                teams.append(
                    {
                        "team_id": int(
                            re.search(pat_team, team_link.get("href")).group(1)  # type: ignore
                        ),
                        "team": team_link.text,
                        "league": lkey,
                        **version.to_dict(),
                    }
                )

        # return data frame
        return pd.DataFrame(teams).replace({"team": TEAMNAME_REPLACEMENTS}).set_index(["team_id"])

    def read_players(self, team: Optional[Union[str, list[str]]] = None) -> pd.DataFrame:
        """Retrieve all players for the selected leagues.