Add UTF-8 BOM to files written to S3/Minio

src/main/java/org/folio/dew/repository/BaseFilesStorage.java

@@ -41,6 +41,7 @@
 import java.io.BufferedReader;
 import java.io.BufferedWriter;
 import java.io.ByteArrayInputStream;
+import java.io.ByteArrayOutputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.InputStreamReader;
@@ -177,16 +178,23 @@ public String upload(String path, String filename) throws IOException {
    * @throws IOException - if an I/O error occurs
    */
   public String write(String path, byte[] bytes, Map<String, String> headers) throws IOException {
+    byte[] bom = {(byte)0xEF, (byte)0xBB, (byte)0xBF};
+
+    ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
+    outputStream.write(bom);
+    outputStream.write(bytes);
+    byte[] bytesWithBom = outputStream.toByteArray();
+
     path = getS3Path(path);
     if (isComposeWithAwsSdk) {
       log.info("Writing with using AWS SDK client");
       s3Client.putObject(PutObjectRequest.builder().bucket(bucket)
           .key(path).build(),
-        RequestBody.fromBytes(bytes));
+        RequestBody.fromBytes(bytesWithBom));
       return path;
     } else {
       log.info("Writing with using Minio client");
-      try(var is = new ByteArrayInputStream(bytes)) {
+      try (var is = new ByteArrayInputStream(bytesWithBom)) {
         return client.putObject(PutObjectArgs.builder()
             .bucket(bucket)
             .region(region)

src/test/resources/output/authority_control/auth_heading_update.csv

@@ -1,4 +1,4 @@
-Last updated,Original heading,New heading,Identifier,Original 1XX,New 1XX,Authority source file name,Number of bibliographic records linked,Updater
+Last updated,Original heading,New heading,Identifier,Original 1XX,New 1XX,Authority source file name,Number of bibliographic records linked,Updater
 2023-10-01 12:00:00.000Z,"Charles, Prince of Wales","Charles III, King",n1234567,150,110,LC Name Authority file (LCNAF),105,"Admin, Diku"
 2023-10-01 12:00:00.000Z,"Charles, Prince of Wales","Charles III, King",n1234567,150,110,LC Name Authority file (LCNAF),105,"Admin, Diku"
 2023-08-01 12:00:00.000Z,"Charles III, King",King Charles III,mo34056,100,100,Not specified,10,"Test, Folio"

src/test/resources/output/authority_control/auth_heading_update_empty.csv

@@ -1,2 +1,2 @@
-Last updated,Original heading,New heading,Identifier,Original 1XX,New 1XX,Authority source file name,Number of bibliographic records linked,Updater
+Last updated,Original heading,New heading,Identifier,Original 1XX,New 1XX,Authority source file name,Number of bibliographic records linked,Updater
 No records found

src/test/resources/output/authority_control/failed_linked_bib_updates.csv

@@ -1,3 +1,3 @@
-Failed,Bibliographic title,Bibliographic UUID,Failed bib field update,Linked authority identifier,Reason for error
+Failed,Bibliographic title,Bibliographic UUID,Failed bib field update,Linked authority identifier,Reason for error
 2019-08-24 14:15:22.000Z,First title,64d2028c-ae87-4069-a624-66089d957ef9,650,mo34056,Not found 404
 2019-08-24 14:15:22.000Z,Second title,65d2028c-ae87-4069-a624-66089d957ef9,150,mo34057,Invalid value

src/test/resources/output/bulk_edit_holdings_records_bad_reference_ids_errors_output.csv

@@ -1,4 +1,4 @@
-badHoldingsTypeId,Holdings type not found by id=1d51c4da-0d5f-4303-a922-3e9d7a9b22f3
+badHoldingsTypeId,Holdings type not found by id=1d51c4da-0d5f-4303-a922-3e9d7a9b22f3
 badHoldingsCallNumberTypeId,Call number type not found by id=1d51c4da-0d5f-4303-a922-3e9d7a9b22f3
 badHoldingsNoteTypeId,Note type not found by id=1d51c4da-0d5f-4303-a922-3e9d7a9b22f3
 badIllPolicyId,Ill policy not found by id=1d51c4da-0d5f-4303-a922-3e9d7a9b22f3

src/test/resources/output/bulk_edit_holdings_records_by_item_barcode.csv

@@ -1,3 +1,3 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 0b1e3760-f689-493e-a98e-9cc9dadb7e83,title,,ho13,FOLIO,d1670310-ceac-47d9-aaba-aaeeb890bc07,Physical,"Book, print (books)",Test administrative note,Annex,Main Library,shelving title,copy number,LC Modified,prefix,call number,suffix,10,statement;statement public note;statement staff note,statement for supplements;statement for supplements public note;statement for supplements staff note,statement for indexes;statement for indexes public note;statement for indexes staff note,Limited lending policy,digitisation policy,retention policy,Note;a note;false;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56|Note;a note;true;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56,"URL relationship;URI;Link text;Materials specified;URL public note
 Version of resource;www.someurl.com;link text;material;www.someurl.com",aquisition method,order format,receipt status,

src/test/resources/output/bulk_edit_holdings_records_empty_reference.csv

@@ -1,4 +1,4 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 855d4693-4087-4339-8c14-c25c350e5e59,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,,,,,,
 ae4288c6-4999-492a-82f4-eceb3a6e1ec6,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,,,,,,
 59b36165-fcf2-49d2-bf7f-25fedbc07e44,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,;a note;false;diku;null,,,,,

src/test/resources/output/bulk_edit_holdings_records_errors_output.csv

@@ -1,2 +1,2 @@
-789,No match found
+789,No match found
 123,Duplicate entry

src/test/resources/output/bulk_edit_holdings_records_errors_output_inst_hrid.csv

@@ -1,3 +1,3 @@
-456,Instance not found by hrid=456
+456,Instance not found by hrid=456
 789,Instance not found by hrid=789
 123,Duplicate entry

src/test/resources/output/bulk_edit_holdings_records_errors_output_item_barcode.csv

@@ -1,2 +1,2 @@
-789,Item not found by barcode=789
+789,Item not found by barcode=789
 123,Duplicate entry

src/test/resources/output/bulk_edit_holdings_records_output.csv

@@ -1,4 +1,4 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 0b1e3760-f689-493e-a98e-9cc9dadb7e83,title,,ho13,FOLIO,d1670310-ceac-47d9-aaba-aaeeb890bc07,Physical,"Book, print (books)",Test administrative note,Annex,Main Library,shelving title,copy number,LC Modified,prefix,call number,suffix,10,statement;statement public note;statement staff note,statement for supplements;statement for supplements public note;statement for supplements staff note,statement for indexes;statement for indexes public note;statement for indexes staff note,Limited lending policy,digitisation policy,retention policy,Note;a note;false;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56|Note;a note;true;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56,"URL relationship;URI;Link text;Materials specified;URL public note
 Version of resource;www.someurl.com;link text;material;www.someurl.com",aquisition method,order format,receipt status,
 855d4693-4087-4339-8c14-c25c350e5e59,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,,,,,,

src/test/resources/output/bulk_edit_holdings_records_output_instance_hrid.csv

@@ -1,4 +1,4 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 0b1e3760-f689-493e-a98e-9cc9dadb7e83,title,,ho13,FOLIO,d1670310-ceac-47d9-aaba-aaeeb890bc07,Physical,"Book, print (books)",Test administrative note,Annex,Main Library,shelving title,copy number,LC Modified,prefix,call number,suffix,10,statement;statement public note;statement staff note,statement for supplements;statement for supplements public note;statement for supplements staff note,statement for indexes;statement for indexes public note;statement for indexes staff note,Limited lending policy,digitisation policy,retention policy,Note;a note;false;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56|Note;a note;true;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56,"URL relationship;URI;Link text;Materials specified;URL public note
 Version of resource;www.someurl.com;link text;material;www.someurl.com",aquisition method,order format,receipt status,
 855d4693-4087-4339-8c14-c25c350e5e59,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,,,,,,

src/test/resources/output/bulk_edit_holdings_records_output_item_barcode.csv

@@ -1,4 +1,4 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 0b1e3760-f689-493e-a98e-9cc9dadb7e83,title,,ho13,FOLIO,d1670310-ceac-47d9-aaba-aaeeb890bc07,Physical,"Book, print (books)",Test administrative note,Annex,Main Library,shelving title,copy number,LC Modified,prefix,call number,suffix,10,statement;statement public note;statement staff note,statement for supplements;statement for supplements public note;statement for supplements staff note,statement for indexes;statement for indexes public note;statement for indexes staff note,Limited lending policy,digitisation policy,retention policy,Note;a note;false;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56|Note;a note;true;diku;b160f13a-ddba-4053-b9c4-60ec5ea45d56,"URL relationship;URI;Link text;Materials specified;URL public note
 Version of resource;www.someurl.com;link text;material;www.someurl.com",aquisition method,order format,receipt status,
 855d4693-4087-4339-8c14-c25c350e5e59,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,,,,,,

src/test/resources/output/bulk_edit_holdings_records_output_marc.csv

@@ -1,4 +1,4 @@
-Holdings record id,Version,HRID,Holdings type,Former ids,"Instance (Title, Publisher, Publication date)",Permanent location,Temporary location,Effective location,Electronic access,Call number type,Call number prefix,Call number,Call number suffix,Shelving title,Acquisition format,Acquisition method,Receipt status,Administrative note,Notes,Ill policy,Retention policy,Digitization policy,Holdings statements,Holdings statements for indexes,Holdings statements for supplements,Copy number,Number of items,Receiving history,Discovery suppress,Statistical codes,Tags,Source
+Holdings record id,Version,HRID,Holdings type,Former ids,"Instance (Title, Publisher, Publication date)",Permanent location,Temporary location,Effective location,Electronic access,Call number type,Call number prefix,Call number,Call number suffix,Shelving title,Acquisition format,Acquisition method,Receipt status,Administrative note,Notes,Ill policy,Retention policy,Digitization policy,Holdings statements,Holdings statements for indexes,Holdings statements for supplements,Copy number,Number of items,Receiving history,Discovery suppress,Statistical codes,Tags,Source
 0b1e3760-f689-493e-a98e-9cc9dadb7e83,1,ho13,Physical,d1670310-ceac-47d9-aaba-aaeeb890bc07,Sample instance;5bf370e0-8cca-4d9c-82e4-5170ab2a0a39,Annex,Main Library,Main Library,www.someurl.com;link text;material;www.someurl.com;Version of resource,LC Modified,prefix,call number,suffix,shelving title,order format,aquisition method,receipt status,Test administrative note,Note;a note;false|Note;a note;true,Limited lending policy,retention policy,digitisation policy,statement;statement public note;statement staff note,statement for indexes;statement for indexes public note;statement for indexes staff note,statement for supplements;statement for supplements public note;statement for supplements staff note,copy number,10,|true;enum;chronology|;enum2;chronology2,,"Book, print (books)",,FOLIO
 d29b9fa5-c7f5-4b77-9f99-79e5c3a9ae75,1,ho14,,,Sample instance;5bf370e0-8cca-4d9c-82e4-5170ab2a0a39,Main Library,Annex,Annex,,,,,,,,,,,,,,,,,,,,,,,,FOLIO
 855d4693-4087-4339-8c14-c25c350e5e59,1,ho15,,,Sample instance;5bf370e0-8cca-4d9c-82e4-5170ab2a0a39,Main Library,Main Library,Main Library,,,,,,,,,,,,,,,,,,,,,,,,MARC

src/test/resources/output/bulk_edit_holdings_records_reference_not_found.csv

@@ -1,4 +1,4 @@
-Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
+Holdings UUID,"Instance (Title, Publisher, Publication date)",Suppress from discovery,Holdings HRID,Source,Former holdings Id,Holdings type,Statistical codes,Administrative note,Holdings permanent location,Holdings temporary location,Shelving title,Holdings copy number,Holdings level call number type,Holdings level call number prefix,Holdings level call number,Holdings level call number suffix,Number of items,Holdings statement,Holdings statement for supplements,Holdings statement for indexes,ILL policy,Digitization policy,Retention policy,Notes,Electronic access,Acquisition method,Order format,Receipt status,Tags
 855d4693-4087-4339-8c14-c25c350e5e59,title,,ho14,FOLIO,,1d51c4da-0d5f-4303-a922-3e9d7a9b22f3,,,Main Library,,,,,,,,,,,,,,,,,,,,
 f836ca1d-c55a-4689-9b57-bb0de0c4d43d,title,,ho14,FOLIO,,,,,Main Library,,,,1d51c4da-0d5f-4303-a922-3e9d7a9b22f3,,,,,,,,,,,,,,,,
 c10bb70f-b0fd-4623-a8cf-25e5198087ad,title,,ho14,FOLIO,,,,,Main Library,,,,,,,,,,,,,,,1d51c4da-0d5f-4303-a922-3e9d7a9b22f3;a note;false;diku;1d51c4da-0d5f-4303-a922-3e9d7a9b22f3,,,,,

src/test/resources/output/bulk_edit_instance_by_issn_isbn_output.csv

@@ -1,2 +1,2 @@
-Instance UUID,Suppress from discovery,Staff suppress,Previously held,Instance HRID,Source,Cataloged date,Instance status term,Mode of issuance,Administrative note,Resource title,Index title,Series statements,Contributors,Edition,Physical description,Resource type,Nature of content,Formats,Languages,Publication frequency,Publication range,Notes
+Instance UUID,Suppress from discovery,Staff suppress,Previously held,Instance HRID,Source,Cataloged date,Instance status term,Mode of issuance,Administrative note,Resource title,Index title,Series statements,Contributors,Edition,Physical description,Resource type,Nature of content,Formats,Languages,Publication frequency,Publication range,Notes
 7fbd5d84-62d1-44c6-9c45-6cb173998bbd,false,false,true,inst000000000006,FOLIO,2024-01-26,,,Cataloging data,Bridget Jones's Baby: the diaries,,,"Fielding, Helen",First American Edition,219 pages ; 20 cm.,6312d172-f0cf-40f6-b27d-9fa8feaf332f,,,eng,A frequency description,A publication range,

src/test/resources/output/bulk_edit_instance_identifiers_errors_output.csv

@@ -1,4 +1,4 @@
-inst000000000022,Instance type was not found by id: [6312d172-f0cf-40f6-b27d-9fa8feaf332f]
+inst000000000022,Instance type was not found by id: [6312d172-f0cf-40f6-b27d-9fa8feaf332f]
 inst000000000022,Nature of content term was not found by id: [921e6d93-bafb-4a02-b62f-dcd027c45406]
 inst000000000022,Instance format was not found by id: [5cb91d15-96b1-4b8a-bf60-ec310538da66]
 inst000000000003,No match found