ggs production data

GGS/ggsLnew.csv

@@ -1,4 +1,4 @@
-;;
+name;vorname;
 Backhaus;Tina;
 Beck;Christopher;
 Beck;Marina;
@@ -95,4 +95,4 @@ Voigt;Finja;
 Weiß;Lea Pauline Katharina Laetitia;
 Welder;Jennifer;
 Wolter;Michael;
-Wulf;Alexander;
+Wulf;Alexander;

GGS/ggsSnew.csv

@@ -1,4 +1,4 @@
-Nachname;Vorname;Klasse
+name;vorname;klasse
 Ahmad Rahman;Elaya;01a
 Al-Ahmar;Mazen;01a
 Becirovic;Dejan;01a

GGS/ggsSold2.cvs

@@ -394,7 +394,7 @@ Clasen;Isger;HL0707104-9b
 Petrovic;Iwana;HL0707104-DaZSEK
 Prassler;Jaale;HL0707104-7d
 Olszak;Jack;HL0707104-5d
-Bréchôt;Jakob;HL0707104-6a
+Bréchot;Jakob;HL0707104-6a
 Bolte;Jakob;HL0707104-6c
 Barucker;Jakob Jan;HL0707104-8e
 Schmähl;Jakob;HL0707104-7b
@@ -804,7 +804,7 @@ Hattenbach;Marvin;HL0707104-6c
 Kerkhoff;Marvin;HL0707104-9b
 Küpper-Fahrenberg;Marvin;HL0707104-13c
 Petersen;Marvin Luca;HL0707104-13a
-Mihajlovic';Mateja;HL0707104-10a
+Mihajlovic;Mateja;HL0707104-10a
 Danowski;Mateusz;HL0707104-5b
 Schwertfeger;Mathilda;HL0707104-5e
 Gohl;Mathis;HL0707104-4c
@@ -840,7 +840,7 @@ Gußmann;Melina;HL0707104-13c
 Hagemann;Melina;HL0707104-9d
 Kollmorgen;Melina;HL0707104-6b
 Wegner;Melina;HL0707104-5d
-Döring;Merle-Marlen;HL0707104-3b
+Döhring;Merle-Marlen;HL0707104-3b
 Neethen;Merle;HL0707104-6e
 Zeitz;Merle;HL0707104-8d
 Bozuklu;Mert;HL0707104-4b
@@ -1098,8 +1098,9 @@ Asseme;Zeineb;HL0707104-6c
 Ertugrul;Zeyneb;HL0707104-12b
 Alofi;Ziyad;HL0707104-4a
 Berger;Zoe;HL0707104-6d
-Bohnhoff;Zoé Gertrud;HL0707104-6e
+Bohnhoff;Zoe Gertrud;HL0707104-6e
 Grimm;Zoe Lynn;HL0707104-7e
-Lorenz;Zoé;HL0707104-5b
+Lorenz;Zoe;HL0707104-5b
 Walter;Zoe;HL0707104-3b
 Niemczewska;Zuzanna Maria;HL0707104-13b
+Schüler;Der GGS-SJ Test;HL0707104-testgruppe

GGS/outputLehrer.csv

@@ -0,0 +1,102 @@
+name,vorname,klasse,schuelerid,mailUserQuota,oxUserQuota,oxContext
+Backhaus,Tina,,e3de9cf9-b0f4-49f5-8a94-840815d4d85a,2048,20480,16
+Beck,Christopher,,f0f2ce5b-6452-4596-9821-56a1ebb82188,2048,20480,16
+Beck,Marina,,8afe5ead-ed43-4253-8909-a7ae886715d3,2048,20480,16
+Berger,Charlyn,,86880eb3-9210-42e2-be40-8d2a4265324a,2048,20480,16
+Bischoff,Meike,,d6cf8df5-51a9-43c0-8ffd-ff5480b3a4da,2048,20480,16
+Blöß,Anne-Kristin,,ecf88a00-b838-438b-b1c5-d3119b1d2027,2048,20480,16
+Böhm,Susanne,,e14c5538-4ecc-4a2c-975f-628c3dbba6ec,2048,20480,16
+Boles,Simone Hildegard,,4327c5b6-a434-4e58-b863-f5832ec38d03,2048,20480,16
+Brabant,Nadja,,909e8802-7a8f-4043-8c4d-4b5cc163c935,2048,20480,16
+Bremer,Heinz-Dieter,,5224606f-9de3-4b1e-bb6d-2cf31dd546fc,2048,20480,16
+Bremer,Timo,,f82d00d2-7d05-4fe4-967a-03ae72f54c7d,2048,20480,16
+Buchbinder,Jürgen,,8138abcc-942f-42ba-9910-69536e26eed6,2048,20480,16
+Bücker,Meike Charlotte,,bea33288-a0c6-4bea-868b-cedae41b43a8,2048,20480,16
+Bünger,Birte,,627f357e-169c-4efc-a544-ed2f14b5dff9,2048,20480,16
+Delfs,Dorothee,,c7ef3d84-b8a9-458a-976a-571813ddce0d,2048,20480,16
+Didt,Marlen,,f055faf7-fa21-44ec-b6d1-cc957e1de0e1,2048,20480,16
+Diercks,Berit-Deike,,455baf87-49d7-4e00-af70-09d88b0c55a7,2048,20480,16
+"Lehnau, Dr.",Frank,,6c4e8feb-6702-433a-9531-98a2b9a8e73e,2048,20480,16
+Ehler,Patricia,,1097df38-d0c4-45c1-baf0-0968d6330a2b,2048,20480,16
+Elgert,Dorit,,4eb689b2-b535-4cd6-9642-ba38871a4c48,2048,20480,16
+Fischer,Rieke Flavia,,683cef23-bea2-4e7c-a09d-e401ffc3eca2,2048,20480,16
+Freyer,Jennifer,,7cc6650c-e5f5-4c36-9157-07d10d3aa17c,2048,20480,16
+Gilbert,Amke,,64cc443d-882b-4dc1-a49e-19b8e5522bad,2048,20480,16
+Glinzner,Sylke,,72bc43ff-064b-41f6-993f-317dfc749783,2048,20480,16
+Healy-Kloppenburg,Insa,,e7f93db5-bf1b-419f-8b93-8c824debd850,2048,20480,16
+Herrig,Swantje,,07004384-5856-43b6-9e32-467f27a63cea,2048,20480,16
+Holst,Patrick,,15780d68-62c1-4651-bb6e-67c5056e7344,2048,20480,16
+Iven,Meike,,0802930d-5b72-4f68-ac1f-3a94be3e9070,2048,20480,16
+Jacke,Friederike-Lisette,,5bf01e41-fc4d-41aa-be7d-1d34d33e599d,2048,20480,16
+Janshen,Friederike,,e7570384-8dee-4f0d-ad14-c2208e6d6f1a,2048,20480,16
+Kaiser,Yvonne,,91431b7f-3c25-4c8f-a996-de648e0f9f4c,2048,20480,16
+Klingelhöfer,Stefanie,,854193b8-6fa8-476a-94d3-4d7a702ce222,2048,20480,16
+Krohn,Iwanka,,5e01c5a3-7d21-43c7-b9aa-6b5aa26845af,2048,20480,16
+Lang,Alisa,,30ad1551-1724-47c7-a61d-42341ff143f3,2048,20480,16
+Lippitsch,Annette,,f4ee88bc-8d0d-43cf-bc2b-aaa82ce4b5f5,2048,20480,16
+Llerena,Susanne,,4254a59a-e0a2-4d63-b221-abd72cd713d2,2048,20480,16
+Marohn,Kai,,0c5a1343-8b3a-4e93-b1a7-33f6bd2dcf65,2048,20480,16
+Martens,Annika,,1931fc8e-d626-4a5d-a339-cd4194507efc,2048,20480,16
+Maßmann,Jillian,,8bc7e49e-892a-463a-a1fa-eb69c12bc66d,2048,20480,16
+Möller,Alexandra,,f44935df-81b5-4196-b3a2-68c96ac5afeb,2048,20480,16
+Möllerherm,Antje,,8cd5774d-03ad-4f38-b8c4-496997773322,2048,20480,16
+Müller,Thomas,,739cbc79-096e-498a-8e38-be9bbbaf3e3e,2048,20480,16
+Neumann,Lisa Marie,,953d2ebd-de8e-4c61-a827-dde777a65212,2048,20480,16
+Nicoloff,Alexandra,,d5438535-9c7a-4d7a-adef-92d5ee58ecc1,2048,20480,16
+Niemann,Lara,,c0564030-539d-4218-a9f8-e1475ccdb357,2048,20480,16
+Ochse,Olaf,,c65a070f-e525-4b2f-aabe-c14c4d2edec5,2048,20480,16
+Pabst,Stefan,,39a78afb-3802-48c1-9626-0176b84d5573,2048,20480,16
+Peters,Nicole,,1a94837f-9b1f-4fea-8899-4801275c328d,2048,20480,16
+Piehl,Jan-Philip,,569aaa43-9727-469f-a050-84a0caf9e019,2048,20480,16
+Prüfert,Birte,,191a4a36-d1c0-4f59-90dd-f931027fab2e,2048,20480,16
+Riedel,Dietmar,,c7020213-2d76-463b-8275-b3d312bc39f3,2048,20480,16
+Ries,Claudia,,57d4677d-a213-4af2-948d-59c555bd76fb,2048,20480,16
+Rudolf,Julia,,e3aa8d5c-5015-47bf-a086-3251a2c5b6a1,2048,20480,16
+Scharfe,Sarah Marie,,d26b2121-8442-4a20-9b1c-2ddd079d4359,2048,20480,16
+Schrader,Lea Ann,,6b64819b-329c-41d9-ac92-3b777604b721,2048,20480,16
+Schröder,Sandra,,811aa921-9326-4273-9848-8a1b03623c82,2048,20480,16
+Schwarz,Anneli-Dorothea,,2a6dd23c-f310-4bd8-ba1b-28a39c929f5e,2048,20480,16
+Seebauer,Astrid,,921902ff-fc4d-428d-9120-d3b2c10d43bc,2048,20480,16
+Seebauer,Benjamin,,c7bce821-fd69-43ea-8ab2-78fc8b6f0526,2048,20480,16
+Siedtmann,Katja,,d42c9b41-8d15-4b26-884c-4e6fde1b0659,2048,20480,16
+Sommer,Alexander,,692357fb-6083-4ea1-90cb-98b93ed66ff8,2048,20480,16
+Steinfatt,Jan Hendrik,,764148a7-a37b-4aaf-8db0-3a282cc22d0b,2048,20480,16
+Stukenbrock,Taira,,4a9d4044-7bed-4952-8dc6-55de9d01f488,2048,20480,16
+Teichmann,Sascha,,e3428db4-0005-4498-84c6-3c3b967427cd,2048,20480,16
+Thiele,Jörg,,697e7b38-381c-4ced-a80f-efb9bb7fc8f4,2048,20480,16
+Thiele,Ulrike,,e030fc2d-97a8-4ce3-913c-4a940b20dbbb,2048,20480,16
+Villwock,Hanno,,e953e7e5-ea76-47b6-baf7-7bd76d317ca2,2048,20480,16
+Weiß,Lea Pauline Katharina Laetitia,,18882825-9771-46f2-8fe1-c220504d7453,2048,20480,16
+Wolter,Michael,,53607793-36bc-4f7e-b80a-3723d0171ed6,2048,20480,16
+Wulf,Alexander,,c2bd5c52-3028-4bf4-a92c-ba425606aa5f,2048,20480,16
+Lehrer,Der GGS-SJ Test,HL0707104-testgruppe,1a9da9b5-8a73-40a4-b6a7-2cf68c6a398d,2048,20480,16
+Schuladmin,Der GGS-SJ Test,HL0707104-testgruppe,a9dffd3f-24a2-4ba2-940f-864b75085020,2048,20480,16
+Apitzsch,Birgit,,646f9c9c-2aa2-47a0-834f-2bc3f2470b18,2048,20480,16
+Ervis,Anastasia,,e35bf36a-2aac-4150-9064-02840a543756,2048,20480,16
+Bernhardi,Antonia,,d5b09f4c-e154-4f75-95e7-ebbf0b07cdd2,2048,20480,16
+Dohrendorf,Hella,,78981ba5-26f8-45b5-9362-030231a00b39,2048,20480,16
+Ehrenforth,Maike,,54a7e278-17f4-4a06-9958-dd5315207147,2048,20480,16
+Gosch,Christina,,7b038ed4-cdae-4a31-ac91-398b09a686d6,2048,20480,16
+Grimm,Antonia,,420c63cf-d908-4c11-8c63-77c4741ebaff,2048,20480,16
+Hartmann,Marisa Sophie,,ab94a0ba-b198-44b2-a681-9f29020d06b5,2048,20480,16
+Knobbe,Svenja Elena,,4c4aaec4-0abd-42a6-97cf-3e7cc0a0c276,2048,20480,16
+Krieglstein,Lena,,6cf5c7e1-3fae-43dd-8332-c519820a328d,2048,20480,16
+Markus-Schnabel,Karsten,,f33b6708-4f91-410d-ac59-fc91b12749e7,2048,20480,16
+Möbius,Christina,,6194e315-181c-4c37-af47-c35649fb030c,2048,20480,16
+Noguera Abreu,Kristina,,9a4c0591-4006-45ed-996f-9ee5d54972d6,2048,20480,16
+Posselt,Janett,,f4abc2fc-58f0-4bc0-acc8-a33fe33a46f4,2048,20480,16
+Radloff,Jannika,,66ada98a-e0e5-4504-85a4-44d9eaaf239e,2048,20480,16
+Resch,Christina Carola,,fad8ac82-9319-4def-9a34-8f57dfddd5d3,2048,20480,16
+Richter-Conrad,Barbara,,e3655bde-e2e7-4cfd-b2eb-3c7ad9081354,2048,20480,16
+Ritter,Ewa Joanna,,4dc8bcfc-9c5a-4a2d-9f20-c4fc680c7ef1,2048,20480,16
+Rohmer,Thomas,,9969c680-1a15-4670-8934-174ef75e1663,2048,20480,16
+Schacht,Dörthe,,c7e9014e-73c5-424b-85af-262c4b0ad027,2048,20480,16
+Schwarz,Berit,,1a0aaf4b-fc65-43bf-8f25-f81757769f5b,2048,20480,16
+Steiger,Lisa,,390880f4-3c18-4a50-8409-d7234dcf4761,2048,20480,16
+Tereshchenko,Oksana,,9ef4f253-472f-4198-8ad7-a955c535beb5,2048,20480,16
+Topal,Sebnem,,6d2e205f-8055-4cf8-a8c4-de5832a2de83,2048,20480,16
+Torkuhl,Clara,,0076b459-2ff6-42da-8ed5-13d5cfee8b5e,2048,20480,16
+Treichel,Hannah Lena,,bbba962f-b95b-40b4-b428-fc396a156f20,2048,20480,16
+Vehrs,Marian,,0b8162d5-0ca5-49ca-864d-bc4b3c2cae9b,2048,20480,16
+Voigt,Finja,,a3d30f75-8ef1-4c67-88e5-b4352679980b,2048,20480,16
+Welder,Jennifer,,823feb09-9311-4ed4-a86e-de989869eb67,2048,20480,16

main.py

@@ -4,14 +4,14 @@ if __name__ == "__main__":
     oldlist = 'Data/alte-liste-utf.csv'
     newlist = 'Data/neue-liste-utf.csv'
 
-    test = 'GGS/export_schueler.csv'
-    new_test = 'GGS/ggsSnew.csv'
+    test = 'GGS/downloadS.csv'
+    new_test = 'GGS/ggsSold.csv'
     clean = 'GGS/ggsSold2.cvs'
 
     # Liste nach Fehler prüfen und Zeichen ersetzen
-    check_file(test)
-    format_csv(test, new_test)
-    # clean_data(new_test, clean)
+    # check_file(test)
+    # format_csv(test, new_test)
+    clean_data(new_test, clean)
 
     # path_old_csv = input('Pfad zur alten Liste eingeben eingeben: ')
     # format_csv(path_old_csv, oldlist)

src/compare.py

@@ -1,5 +1,8 @@
 import csv
 from Levenshtein import distance
+import pandas as pd
+import uuid
+
 
 # TODO Filter für Spalten, ggfs. Klasse benötigt
 
@@ -42,21 +45,87 @@ def find_similar_pairs(pair, other_data):
     return similar_pairs
 
 
+def create_uuid():
+    return str(uuid.uuid4())
+
+
+def add_hl_tag(row):
+    return 'HL0707104-' + row['klasse']
+
+
+def create_import_list(path, path_new, old_pairs, new_pairs, common_pairs):
+    df = pd.read_csv(path, sep=';', encoding='utf-8')
+    df = df[~df[['name', 'vorname']].apply(tuple, axis=1).isin(old_pairs)]
+    df = df.drop('username', axis=1, errors='ignore')
+    df = df.drop('klasse', axis=1, errors='ignore')
+    # print(len(df))
+    # print(df)
+    new_data = pd.read_csv(path_new, sep=';', encoding='utf-8')
+
+    # Bei Schüler: alte Klassen gelöscht, mit neuen Klassen aus new-data auffüllen
+
+    matches = new_data[~new_data[['name', 'vorname']].apply(tuple, axis=1).isin(new_pairs)]
+    matches.loc[:, 'klasse'] = matches.apply(add_hl_tag, axis=1)
+    # print(len(matches))
+    # print(matches)
+    df = pd.merge(df, matches, how='outer', left_on=['name', 'vorname'], right_on=['name', 'vorname'])
+    df = df[['name', 'vorname', 'klasse', 'schuelerid', 'mailUserQuota', 'oxUserQuota', 'oxContext']]
+    # print(df)
+    print(len(df))
+
+
+    new_data = new_data[~new_data[['name', 'vorname']].apply(tuple, axis=1).isin(common_pairs)]
+    # new_data = new_data.drop('Unnamed: 2', axis=1, errors='ignore')
+    new_uuids = []
+    for row in range(len(new_data)):
+        new_uuids.append(create_uuid())
+    # Klasse?? Unterschied zwischen Lehrer und Schüler
+    #new_data['klasse'] = None
+    new_data.loc[:, 'klasse'] = new_data.apply(add_hl_tag, axis=1)
+    new_data.insert(loc=2, column='schuelerid', value=new_uuids)
+    mailUserQuota = 1024
+    oxUserQuota = 5120
+    oxContext = 16
+    new_data['mailUserQuota'] = mailUserQuota
+    new_data['oxUserQuota'] = oxUserQuota
+    new_data['oxContext'] = oxContext
+    # print(new_data)
+    print(len(new_data))
+
+    # vor dem merge daten ergänzen
+    # import_df = pd.merge(df, new_data, how='outer', left_on=['name', 'vorname', 'mailUserQuota', 'oxUserQuota', 'oxContext'], right_on=['name', 'vorname', 'mailUserQuota', 'oxUserQuota', 'oxContext'])
+    import_df = pd.merge(df, new_data, how='outer', left_on=['name', 'vorname', 'klasse', 'schuelerid', 'mailUserQuota', 'oxUserQuota', 'oxContext'], right_on=['name', 'vorname', 'klasse', 'schuelerid', 'mailUserQuota', 'oxUserQuota', 'oxContext'])
+    # pd.set_option('display.max_rows', None)
+    # pd.set_option('display.max_columns', None)
+    # print(import_df)
+    # print(len(import_df))
+    # pd.reset_option('display.max_rows')
+    # pd.reset_option('display.max_columns')
+    out_path = '../GGS/outputSchueler.csv'
+    import_df.to_csv(out_path, index=False)
+    print(len(import_df))
+    print('Schüler Import Liste erzeugt')
+    print('Testuser manuell nachtragen!!')
+
+
 def main():
     file1_path = '../GGS/ggsSnew.csv'  # Pfad zur ersten CSV-Datei
     file2_path = '../GGS/ggsSold2.cvs'  # Pfad zur zweiten CSV-Datei
+    file3_path = '../GGS/downloadS.csv'
 
-    common_pairs, unique_pairs1, unique_pairs2, data1, data2 = compare_csv(file1_path, file2_path)
+    common_pairs, new_pairs, old_pairs, data1, data2 = compare_csv(file1_path, file2_path)
 
     print(f"Anzahl der übereinstimmenden Paare: {len(common_pairs)}")
-    print(f"Anzahl der neuen Einträge: {len(unique_pairs1)}")
-    print(f"Anzahl der veralteten Einträge: {len(unique_pairs2)}")
+    print(f"Anzahl der neuen Einträge: {len(new_pairs)}")
+    print(f"Anzahl der veralteten Einträge: {len(old_pairs)}")
 
-    for pair in data1:
-        similar_pairs_list2 = find_similar_pairs(pair, set(data2) - {pair})
+    # for pair in data1:
+    #     similar_pairs_list2 = find_similar_pairs(pair, set(data2) - {pair})
+    #
+    #     if similar_pairs_list2:
+    #         print(f"Ähnliche Paare zu {pair} in Liste 2: {similar_pairs_list2}")
 
-        if similar_pairs_list2:
-            print(f"Ähnliche Paare zu {pair} in Liste 2: {similar_pairs_list2}")
+    create_import_list(file3_path, file1_path, old_pairs, new_pairs, common_pairs)
 
 
 if __name__ == "__main__":