output.clj

{:starters
(["Marcus" 17]
 ["Baba" 16]
 ["Sami" 14]
 ["Tuomas" 14]
 ["Kabu" 14]
 ["Jesse" 11]
 ["Lydä" 9]
 ["Jonde" 9]
 ["Jebu" 9]
 ["Matias" 9]
 ["Eetu" 8]
 ["Pete" 8]
 ["Ripa" 8]
 ["Miku" 8]
 ["Sintti" 8]
 ["Nuppis" 7]
 ["Micke" 6]
 ["Kunnas" 6]
 ["Mige" 3]
 ["Jussi" 3]
 ["Milton" 3]
 ["Tero" 3]
 ["Faba" 3]
 ["Michael" 2]
 ["Gusse" 2]
 ["Krisu" 2]
 ["Kimmo" 2]
 ["JW" 2]
 ["Nemo" 1]
 ["Maukka" 1]
 ["Wille" 1]),
:bench
(["Max" 15]
 ["Milton" 11]
 ["Sintti" 7]
 ["Jesse" 7]
 ["Pete" 5]
 ["Michael" 5]
 ["Jussi" 5]
 ["Micke" 5]
 ["Jonde" 5]
 ["Jebu" 5]
 ["Kunnas" 5]
 ["Eetu" 4]
 ["Maukka" 4]
 ["Eki" 4]
 ["Tero" 4]
 ["Sami" 3]
 ["Lydä" 2]
 ["Nemo" 2]
 ["Nuppis" 2]
 ["Pasi" 2]
 ["Matias" 2]
 ["Tuomas" 2]
 ["Wille" 2]
 ["Baba" 1]
 ["Mige" 1]
 ["Krisu" 1]
 ["Ripa" 1]
 ["Faba" 1]
 ["Kimmo" 1]
 ["Kabu" 1]),
:mom
(["Baba" 3]
 ["Pete" 2]
 ["Lydä" 2]
 ["Tuomas" 2]
 ["Micke" 1]
 ["Milton" 1]
 ["Sami" 1]
 ["Marcus" 1]
 ["Jebu" 1]
 ["Kunnas" 1]
 ["Kimmo" 1]
 ["Matias" 1]
 ["Miku" 1]
 ["Jesse" 1])}

ppj_stats.clj

(ns ppj-stats
  (:import [java.io BufferedReader StringReader])
  (:require [clojure.string :refer (join)]
            [clojure.pprint :refer (pprint)]))

(def data-string
  "22.4. HJK/Kantsu - PPJ 1 - 3 (1 - 1)
   A Pete, Tuomas - Marcus - Kabu - Babar, Ripatti - Jebu, Jesse, Faba - Gusse - Kunnas
   P Milton, Tero, Jonde, Sintti, Max, Maukka
   M Faba 27' 1-1 (rp), Milton 63' 1- 2, Sintti 89' 1-3 (Jonde)
   O Marcus
   28.4. PPJ - FC ALPC 3 - 0 (2 - 0)
   A Eetu, Mige - Marcus - Kabu - Tuomas, Sami - Micke, Jebu, Faba - Babar - Sintti
   P Milton, Kunnas, Jonde, Jesse, Max, Maukka
   M Babar 15' (Sintti), Jebu 33' (Sintti), Kunnas 82' (Babar)
   O Babar
   7.5. LPS/Kuninkaat - PPJ 1 - 3 (1-2)
   A Eetu, Jonde - Marcus - Kabu - Tuomas, Sami - Micke, Milton, Faba - Babar - Sintti
   P Matias, Jebu, Kunnas, Jebu, Jessu, Max, Maukka
   M Faba 19' 0-1 (Babar), Milton 34' 0-2 (Sintti), Jesse 79' 1-3 (Matias) 
   O Tuomas
   12.5. PPJ - PPV 2 -1 (1-0)
   A Pete, Mige - Kabu - Matias - Jonde, Ripatti - Jebu, Jesse, Sintti - Babar - Lydä
   P Faba, Sami, Micke, Wille, Eetu
   M Sintti 4' 1-0, Jebu 81' 2-0
   O Pete
   21.5. HIFK/3 - PPJ 0 - 2 (0-2)
   A Eetu, Matias - Marcus - Kabu - Tuomas, Ripatti - Jebu, Jesse, Lydä - Kunnas - Babar
   P Sintti, Micke, Max, Jonde, Maukka
   M Kunnas 14' (Babar), Jebu 29' (Marcus)
   O Jebu
   26.5. SaPa/3 - PPJ 2 - 2 (2-1)
   K Maukka, Jonde - Marcus - Matias - Tuomas, Micke - Sami, Jesse, Baba - Wille - Kunnas
   P Sintti, Milton, Max, Michael
   M Jesse 44' 2-1 (Micke), Babar 85' 2-2 (Max)
   O Jesse
   2.6. PPJ - HPS/2 1 - 3 (0-1)
   K Eetu, Jonde-Marcus-Kabu-Matias, Micke-Sami, Milton, Kunnas-Babar- Sintti 
   P Michael, Jesse, Kimmo, Jebu, Mige, Wille, Max
   M Kunnas 73' 1-1 (Jesse)
   O Kunnas
   9.6. PPJ - Vesa 3 - 1 (1-1)
   K Michael, Jonde-Marcus-Kabu-Matias, Ripatti-Sami, Jesper, Kunnas - Lydä - Miku
   P Babar, Milton, Tuomas, Micke, Tero, Max, Pete
   M Marcus 32' 1-0 (vp), Babar 46' 2-1 (Miku), Lydä 73' 3-1 (Miku)
   O Lydä
   16.6. Gnistan/2 - PPJ 2 - 1 (1-1)
   K Eetu, Jonde - Marcus - Matias - Mige, Micke - Ripatti, Jesse, Lydä - Babar - Krisu
   P Milton, Pasi, Sami, Max, Jussi, Pete
   M Babar 45' 1-1 (Jesse)
   O Babar
   23.6. PPJ - Fc Degis 1 - 2 (1-1)
   K Pete, Jonde - Jussi - Matias - Tuomas, Jesse - Sami, Jebu, Ripatti - Kimmo - Tero
   P Milton, Lydä, Micke, Kunnas, Nuppis, Eetu, Kabu
   M Ripatti 45' 1-1 (Jebu)
   O Matias
   30.6. Ponnistus - PPJ 0 - 2 (0-0)
   K Eetu, Matias - Marcus - Kabu - Tuomas, Jebu - Jesse, Nuppis, Kimmo - Lydä - Miku
   P Jussi, Jonde, Sami, Milton, Pasi, Tero
   M Nuppis 7' (Jebu), Marcus 70'
   O Kimmo
   28.7. PPJ - HJK/Kantsu 2 - 0 (1-0)
   K Eetu, Babar - Marcus - Jussi - Tuomas, Sami - Antti, JW, Sintti - Lydä - Miku
   P Nemo, Jesse, Tero, Ripa, Eki, Max, Pete
   M Lydä 10' (Tuomas), JW 52' (Tuomas)
   O Tuomas
   4.8. PPJ - SaPa/3 0 - 2 (0-2)
   K Eetu, Baba - Marcus - Kabu - Tuomas, Ripa - Sami, Johannes, Lydä - Jesse - Miku 
   P Nuppis, Sintti, Milton, Jussi, Max, Eki, Pete
   O Sami
   11.8. PPJ - LPS/Kuninkaat 6 - 0 (1-0)
   K Pete - Tuomas, Marcus, Kabu, Miku - Ripa, Sami - Nuppis - Lydä, Baba, Sintti
   P Milton, Jesse, Jussi, Max, Eki; Eetu
   M Sami 13' (Lydä), Baba 48', Kabu 51' (Marcus), Lydä 65' (Baba), Milton 68' (Baba), Jesse 74' (Marcus)
   O Miku 
   18.8. PPV - PPJ 0 - 3 (0-2)
   K Pete - Tuomas, Marcus, Kasper, Miku - Jesse, Sami - Nuppis - Lydä, Baba, Sintti
   P Milton, Jebu, Nemo, Jussi, Michael
   M Nuppis 4' (Baba), Lydä 32', Lydä 51' (Miku)
   O Lydä
   27.8. PPJ - HIFK/3 2 - 1 (1-1)
   K Michael, Tuomas - Marcus - Kabu - Miku, Sami - Jesse, Nuppis, Nemo - Babar - Sintti
   P Matias, Lydä, Milton, Jebu, Krisu, Max, Pete
   M Babar 26' 1-1 (Tuomas), Babar 46' 2-1 (Sami)
   O Babar
   6.9. Fc Alp - PPJ 1 - 3 (1-0)
   A Pete - Jonde Jussi Marcus Miku - Jebu Nuppis Sami - Milton Babs Tero
   P Tuomas, Jesse, Sintti, Kunnas, Max, Michael
   M Milton 46' 1-1, Babar 52' 1-2 (Jebu), Babar 56' 1-3 (Jussi)
   O Milton
   9.9. HPS/2 - PPJ 2 - 1 (2-0) 
   A Pete - Jonde, Marcus, Kabu, Tuomas - Sami, Jebu - Nuppis - Tero, Babs, Krisu
   P Jesse, Micke, Max, Sintti, Kunnas, Eetu
   M Kunnas 89' 2-1
   O Pete
   16.9. Vesa - PPJ 2 - 2 (0-1)
   A Pete - Tuomas, Marcus, Kabu, Matias - Micke, Sami - Jesse - Kunnas, Babs, Gusse
   P Sintti, Eki, Max, Michael, Jonde
   M Babar 25' 0-1 (Marcus), Babar 61' 0-2 (Sintti)
   O Micke")

(defn partition-many
  [pred coll]
  (letfn [(partition-many' [coll]
            (when-let [s (seq coll)]
              (let [[x & xs] s]
                (let [start (cons x (take-while (complement not pred) xs))
                      end (drop-while (complement pred) xs)]
                  (cons start (lazy-seq (partition-many' end)))))))]
    (partition-many' coll)))

(def data
  (with-open [rdr (BufferedReader. (StringReader. data-string))]
    (doall (line-seq rdr))))

(def mapping
  {\A :starters
   \K :starters
   \P :bench
   \M :goals
   \O :mom})

(def name-fixes {"Johannes" "JW"
                 "Babs" "Baba"
                 "Babar" "Baba"
                 "Antti" "Nuppis"
                 "Ripatti" "Ripa"
                 "Kasper" "Kabu"
                 "Jesper" "Jebu"
                 "Jessu" "Jesse"})

(defn fix-name
  [s]
  (or (name-fixes s) s))

(defn parse-row
  [row]
  (let [[c & cs] row]
    {(mapping c) (map fix-name (re-seq #"[äöüÄÖÜß\w]+" (join cs)))}))

(defn group-data
  [data]
  (->> data
       (map #(.trim %) ,,,)
       (partition-many #(.matches % "^\\d.+") ,,,)
       (map (fn [[match & xs]]
              (into {} (map parse-row xs))) ,,,)
       ))

(defn sorted-frequency-data-for-key
  [grouped-data k]
  (->> grouped-data
       (mapcat k)
       frequencies
       (sort-by second >)))

(let [grouped-data (group-data data)
      starters (sorted-frequency-data-for-key grouped-data :starters)
      bench (sorted-frequency-data-for-key grouped-data :bench)
      mom (sorted-frequency-data-for-key grouped-data :mom)]
  (pprint {:starters starters 
           :bench bench
           :mom mom}))