/android/import-contacts

To get this branch, use:
bzr branch http://bzr.ed.am/android/import-contacts

« back to all changes in this revision

Viewing changes to src/org/waxworlds/edam/importcontacts/VcardImporter.java

  • Committer: edam
  • Date: 2011-06-04 16:52:28 UTC
  • Revision ID: edam@waxworlds.org-20110604165228-jam230oo29u1m06u
- properly handle multiple TYPE= params in one entry in a v3.0 vCard
- when deciding which phone number to use as the pimary number, a voice number takes precedence over a non-voice (fax or pager) number of the same standing in terms of being preferred or not.

Show diffs side-by-side

added added

removed removed

src/org/waxworlds/edam/importcontacts/VcardImporter.java
1
1
/*
2
2
 * VCFImporter.java
3
3
 *
4
 
 * Copyright (C) 2009 to 2011 Tim Marston <tim@ed.am>
 
4
 * Copyright (C) 2009 to 2011 Tim Marston <edam@waxworlds.org>
5
5
 *
6
6
 * This file is part of the Import Contacts program (hereafter referred
7
7
 * to as "this program"). For more information, see
8
 
 * http://ed.am/dev/android/import-contacts
 
8
 * http://www.waxworlds.org/edam/software/android/import-contacts
9
9
 *
10
10
 * This program is free software: you can redistribute it and/or modify
11
11
 * it under the terms of the GNU General Public License as published by
21
21
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
22
22
 */
23
23
 
24
 
package am.ed.importcontacts;
 
24
package org.waxworlds.edam.importcontacts;
25
25
 
26
26
import java.io.BufferedReader;
27
27
import java.io.File;
38
38
import java.util.HashSet;
39
39
import java.util.Iterator;
40
40
import java.util.List;
41
 
import java.util.Locale;
42
41
import java.util.NoSuchElementException;
43
42
import java.util.Set;
44
43
import java.util.Vector;
45
44
import java.util.regex.Matcher;
46
45
import java.util.regex.Pattern;
47
46
 
48
 
import android.annotation.SuppressLint;
49
47
import android.content.SharedPreferences;
 
48
import android.provider.Contacts;
 
49
import android.provider.Contacts.PhonesColumns;
50
50
 
51
51
public class VcardImporter extends Importer
52
52
{
58
58
                super( doit );
59
59
        }
60
60
 
61
 
        @SuppressLint( "SdCardPath" )
62
61
        @Override
63
62
        protected void onImport() throws AbortImportException
64
63
        {
83
82
                                // get files
84
83
                                class VCardFilter implements FilenameFilter {
85
84
                                        public boolean accept( File dir, String name ) {
86
 
                                                return name.toLowerCase( Locale.US ).endsWith( ".vcf" );
 
85
                                                return name.toLowerCase().endsWith( ".vcf" );
87
86
                                        }
88
87
                                }
89
88
                                files = file.listFiles( new VCardFilter() );
202
201
                                        buffer.limit() - buffer.position(), "US-ASCII" );
203
202
                        }
204
203
                        catch( UnsupportedEncodingException e ) {
205
 
                                // we know US-ASCII *is* supported, so appease the compiler...
 
204
                                // we know US-ASCII is supported, so appease the compiler...
206
205
                                line = "";
207
206
                        }
208
207
 
209
208
                        if( vcard == null ) {
210
209
                                // look for vcard beginning
211
 
                                if( line.matches( "^BEGIN[ \t]*:[ \t]*VCARD" ) ) {
 
210
                                if( line.matches( "^BEGIN:VCARD" ) ) {
212
211
                                        setProgress( _progress++ );
213
212
                                        vcard = new Vcard();
214
213
                                        vcard_start_line = cli.getLineNumber();
216
215
                        }
217
216
                        else {
218
217
                                // look for vcard content or ending
219
 
                                if( line.matches( "^END[ \t]*:[ \t]*VCARD" ) )
 
218
                                if( line.matches( "^END:VCARD" ) )
220
219
                                {
221
220
                                        // finalise the vcard/contact
222
221
                                        try {
351
350
                public boolean doesNextLineLookFolded()
352
351
                {
353
352
                        return _pos > 0 && _pos < _content.length &&
354
 
                                _content[ _pos - 1 ] == '\n' &&
355
 
                                ( _content[ _pos ] == ' ' || _content[ _pos ] == '\t' );
 
353
                                _content[ _pos - 1 ] == '\n' && _content[ _pos ] == ' ';
356
354
                }
357
355
 
358
356
                public int getLineNumber()
370
368
                private final static int MULTILINE_NONE = 0;
371
369
                private final static int MULTILINE_ENCODED = 1; // v2.1 quoted-printable
372
370
                private final static int MULTILINE_ESCAPED = 2; // v2.1 \\CRLF
373
 
                private final static int MULTILINE_FOLDED = 3;  // MIME-DIR folding
 
371
                private final static int MULTILINE_FOLDED = 3;  // v3.0 folding
374
372
 
375
373
                private String _version = null;
376
374
                private Vector< ByteBuffer > _buffers = null;
474
472
 
475
473
                                // is it a version line?
476
474
                                if( name_and_params != null &&
477
 
                                        name_and_params.equalsIgnoreCase( "VERSION" ) )
 
475
                                        name_and_params.equals( "VERSION" ) )
478
476
                                {
479
477
                                        // yes, get it!
480
478
                                        String value = extractValueFromLine( buffer, line );
504
502
                        else
505
503
                        {
506
504
                                // name and params and the position in the buffer where the
507
 
                                // "value" part of the line starts
 
505
                                // "value" part of the line start
508
506
                                String name_and_params;
509
507
                                int pos;
510
508
 
540
538
                                }
541
539
                                else
542
540
                                {
543
 
                                        // skip empty lines
544
 
                                        if( line.trim().length() == 0 ) return;
545
 
 
546
541
                                        // get name and params from line, and since we're not
547
542
                                        // parsing a subsequent line in a multi-line, this should
548
543
                                        // not fail, or it's an error
573
568
 
574
569
                                // determine whether we care about this entry
575
570
                                final HashSet< String > interesting_fields =
576
 
                                        new HashSet< String >( Arrays.asList( new String[] { "N",
577
 
                                                "FN", "ORG", "TITLE", "TEL", "EMAIL", "ADR", "LABEL" }
 
571
                                        new HashSet< String >( Arrays.asList( new String[]
 
572
                                                { "N", "FN", "ORG", "TITLE", "TEL", "EMAIL", "ADR" }
578
573
                                ) );
579
574
                                boolean is_interesting_field =
580
 
                                        interesting_fields.contains(
581
 
                                                name_param_parts[ 0 ].toUpperCase( Locale.US ) );
 
575
                                        interesting_fields.contains( name_param_parts[ 0 ] );
582
576
 
583
577
                                // parse encoding parameter
584
578
                                String encoding = checkParam( name_param_parts, "ENCODING" );
585
 
                                if( encoding != null )
586
 
                                        encoding = encoding.toUpperCase( Locale.US );
 
579
                                if( encoding != null ) encoding = encoding.toUpperCase();
587
580
                                if( is_interesting_field && encoding != null &&
588
 
                                        !encoding.equalsIgnoreCase( "8BIT" ) &&
589
 
                                        !encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
590
 
                                        //&& !encoding.equalsIgnoreCase( "BASE64" ) )
 
581
                                        !encoding.equals( "8BIT" ) &&
 
582
                                        !encoding.equals( "QUOTED-PRINTABLE" ) )
 
583
                                        //&& !encoding.equals( "BASE64" ) )
591
584
                                {
592
585
                                        throw new ParseException( R.string.error_vcf_encoding );
593
586
                                }
594
587
 
595
588
                                // parse charset parameter
596
589
                                String charset = checkParam( name_param_parts, "CHARSET" );
597
 
                                if( charset != null )
598
 
                                        charset = charset.toUpperCase( Locale.US );
 
590
                                if( charset != null ) charset = charset.toUpperCase();
599
591
                                if( charset != null &&
600
 
                                        !charset.equalsIgnoreCase( "US-ASCII" ) &&
601
 
                                        !charset.equalsIgnoreCase( "ASCII" ) &&
602
 
                                        !charset.equalsIgnoreCase( "UTF-8" ) )
 
592
                                        !charset.equals( "US-ASCII" ) &&
 
593
                                        !charset.equals( "ASCII" ) &&
 
594
                                        !charset.equals( "UTF-8" ) )
603
595
                                {
604
596
                                        throw new ParseException( R.string.error_vcf_charset );
605
597
                                }
607
599
                                // do unencoding (or default to a fake unencoding result with
608
600
                                // the raw string)
609
601
                                UnencodeResult unencoding_result = null;
610
 
                                if( encoding != null &&
611
 
                                        encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
612
 
                                {
 
602
                                if( encoding != null && encoding.equals( "QUOTED-PRINTABLE" ) )
613
603
                                        unencoding_result = unencodeQuotedPrintable( value );
614
 
                                }
615
 
//                              else if( encoding != null &&
616
 
//                                      encoding.equalsIgnoreCase( "BASE64" ) )
617
 
//                              {
 
604
//                              else if( encoding != null && encoding.equals( "BASE64" ) )
618
605
//                                      unencoding_result = unencodeBase64( props[ 1 ], charset );
619
 
//                              }
620
606
                                if( unencoding_result != null ) {
621
607
                                        value = unencoding_result.getBuffer();
622
608
                                        if( unencoding_result.isAnotherLineRequired() )
627
613
                                // specified for a v2.1 vcard entry, we assume it's US-ASCII)
628
614
                                if( ( charset == null && _version.equals( "2.1" ) ) ||
629
615
                                        ( charset != null && (
630
 
                                                charset.equalsIgnoreCase( "ASCII" ) ||
631
 
                                                charset.equalsIgnoreCase( "US-ASCII" ) ) ) )
 
616
                                                charset.equals( "ASCII" ) ||
 
617
                                                charset.equals( "US-ASCII" ) ) ) )
632
618
                                {
633
619
                                        value = transcodeAsciiToUtf8( value );
634
620
                                }
645
631
                                // for some entries that have semicolon-separated value parts,
646
632
                                // check to see if the value ends in an escape character, which
647
633
                                // indicates that we have a multi-line value
648
 
                                if( ( name_param_parts[ 0 ].equalsIgnoreCase( "N" ) ||
649
 
                                        name_param_parts[ 0 ].equalsIgnoreCase( "ORG" ) ||
650
 
                                        name_param_parts[ 0 ].equalsIgnoreCase( "ADR" ) ) &&
 
634
                                if( ( name_param_parts[ 0 ].equals( "N" ) ||
 
635
                                        name_param_parts[ 0 ].equals( "ORG" ) ||
 
636
                                        name_param_parts[ 0 ].equals( "ADR" ) ) &&
651
637
                                        doesStringEndInAnEscapeChar( string_value ) )
652
638
                                {
653
639
                                        _parser_multiline_state = MULTILINE_ESCAPED;
655
641
                                                string_value.length() - 1 );
656
642
                                }
657
643
 
658
 
                                // if we know we're not in an encoding-based multi-line, check
659
 
                                // to see if we're in a folded multi-line
 
644
                                // now we know whether we're in an encoding multi-line,
 
645
                                // determine if we're in a v3 folded multi-line or not
660
646
                                if( _parser_multiline_state == MULTILINE_NONE &&
661
 
                                        next_line_looks_folded )
 
647
                                        _version.equals( "3.0" ) && next_line_looks_folded )
662
648
                                {
663
649
                                        _parser_multiline_state = MULTILINE_FOLDED;
664
650
                                }
676
662
                                if( complete_value.length() < 1 ) return;
677
663
 
678
664
                                // parse some properties
679
 
                                if( name_param_parts[ 0 ].equalsIgnoreCase( "N" ) )
 
665
                                if( name_param_parts[ 0 ].equals( "N" ) )
680
666
                                        parseN( name_param_parts, complete_value );
681
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "FN" ) )
 
667
                                else if( name_param_parts[ 0 ].equals( "FN" ) )
682
668
                                        parseFN( name_param_parts, complete_value );
683
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ORG" ) )
 
669
                                else if( name_param_parts[ 0 ].equals( "ORG" ) )
684
670
                                        parseORG( name_param_parts, complete_value );
685
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TITLE" ) )
 
671
                                else if( name_param_parts[ 0 ].equals( "TITLE" ) )
686
672
                                        parseTITLE( name_param_parts, complete_value );
687
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TEL" ) )
 
673
                                else if( name_param_parts[ 0 ].equals( "TEL" ) )
688
674
                                        parseTEL( name_param_parts, complete_value );
689
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "EMAIL" ) )
 
675
                                else if( name_param_parts[ 0 ].equals( "EMAIL" ) )
690
676
                                        parseEMAIL( name_param_parts, complete_value );
691
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ADR" ) )
 
677
                                else if( name_param_parts[ 0 ].equals( "ADR" ) )
692
678
                                        parseADR( name_param_parts, complete_value );
693
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "LABEL" ) )
694
 
                                        parseLABEL( name_param_parts, complete_value );
695
 
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "NOTE" ) )
696
 
                                        parseNOTE( name_param_parts, complete_value );
697
679
                        }
698
680
                }
699
681
 
771
753
                                in_escape = false;
772
754
                                switch( c )
773
755
                                {
774
 
                                case 'T':
775
 
                                case 't':
776
 
                                        // add tab (invalid/non-standard, but accepted)
777
 
                                        ret.append( '\t' );
778
 
                                        break;
779
756
                                case 'N':
780
757
                                case 'n':
781
758
                                        // add newline
789
766
                                        break;
790
767
                                default:
791
768
                                        // unknown escape sequence, so add it unescaped
792
 
                                        // (invalid/non-standard, but accepted)
793
769
                                        ret.append( "\\" );
794
770
                                        ret.append( Character.toChars( c ) );
795
771
                                        break;
901
877
                        int type;
902
878
                        if( types.contains( "FAX" ) )
903
879
                                if( types.contains( "HOME" ) )
904
 
                                        type = TYPE_FAX_HOME;
 
880
                                        type = PhonesColumns.TYPE_FAX_HOME;
905
881
                                else
906
 
                                        type = TYPE_FAX_WORK;
 
882
                                        type = PhonesColumns.TYPE_FAX_WORK;
907
883
                        else if( types.contains( "CELL" ) || types.contains( "VIDEO" ) )
908
 
                                type = TYPE_MOBILE;
 
884
                                type = PhonesColumns.TYPE_MOBILE;
909
885
                        else if( types.contains( "PAGER" ) )
910
 
                                type = TYPE_PAGER;
 
886
                                type = PhonesColumns.TYPE_PAGER;
911
887
                        else if( types.contains( "WORK" ) )
912
 
                                type = TYPE_WORK;
 
888
                                type = PhonesColumns.TYPE_WORK;
913
889
                        else
914
 
                                type = TYPE_HOME;
 
890
                                type = PhonesColumns.TYPE_HOME;
915
891
 
916
892
                        // add phone number
917
893
                        addNumber( value, type, is_preferred );
928
904
                        boolean is_preferred = types.contains( "PREF" );
929
905
                        int type;
930
906
                        if( types.contains( "WORK" ) )
931
 
                                type = TYPE_WORK;
 
907
                                type = Contacts.ContactMethods.TYPE_WORK;
932
908
                        else
933
 
                                type = TYPE_HOME;
 
909
                                type = Contacts.ContactMethods.TYPE_HOME;
934
910
 
935
911
                        addEmail( unescapeValue( value ), type, is_preferred );
936
912
                }
945
921
                        for( int a = 0; a < adr_parts.length; a++ )
946
922
                                if( adr_parts[ a ].length() > 0 )
947
923
                                {
948
 
                                        // version 3.0 vCards allow further splitting by comma
949
 
                                        if( _version.equals( "3.0" ) )
950
 
                                        {
951
 
                                                // split this part in to it's comma-separated bits and
952
 
                                                // add them on individual lines
953
 
                                                String[] adr_part_parts =
954
 
                                                        splitValueByCharacter( adr_parts[ a ], ',' );
955
 
                                                for( int b = 0; b < adr_part_parts.length; b++ )
956
 
                                                        if( adr_part_parts[ b ].length() > 0 )
957
 
                                                        {
958
 
                                                                if( value.length() > 0 ) value += "\n";
959
 
                                                                value += adr_part_parts[ b ];
960
 
                                                        }
961
 
                                        }
962
 
                                        else
963
 
                                        {
964
 
                                                // add this part on an individual line
965
 
                                                if( value.length() > 0 ) value += "\n";
966
 
                                                value += adr_parts[ a ];
967
 
                                        }
 
924
                                        // split this part in to it's comma-separated bits
 
925
                                        String[] adr_part_parts =
 
926
                                                splitValueByCharacter( adr_parts[ a ], ',' );
 
927
                                        for( int b = 0; b < adr_part_parts.length; b++ )
 
928
                                                if( adr_part_parts[ b ].length() > 0 )
 
929
                                                {
 
930
                                                        if( value.length() > 0 ) value += "\n";
 
931
                                                        value += adr_part_parts[ b ];
 
932
                                                }
968
933
                                }
969
934
 
970
935
                        Set< String > types = extractTypes( params, Arrays.asList(
971
 
                                "PREF", "WORK", "HOME" ) );
972
 
 
973
 
                        // add address
974
 
                        int type;
975
 
                        if( types.contains( "WORK" ) )
976
 
                                type = TYPE_WORK;
977
 
                        else
978
 
                                type = TYPE_HOME;
979
 
 
980
 
                        addAddress( unescapeValue( value ), type );
981
 
                }
982
 
 
983
 
                private void parseLABEL( String[] params, String value )
984
 
                {
985
 
                        Set< String > types = extractTypes( params, Arrays.asList(
986
 
                                "PREF", "WORK", "HOME" ) );
987
 
 
988
 
                        // add address
989
 
                        int type;
990
 
                        if( types.contains( "WORK" ) )
991
 
                                type = TYPE_WORK;
992
 
                        else
993
 
                                type = TYPE_HOME;
994
 
 
995
 
                        addAddress( unescapeValue( value ), type );
996
 
                }
997
 
 
998
 
                private void parseNOTE( String[] params, String value )
999
 
                {
1000
 
                        addNote( unescapeValue( value ) );
 
936
                                "PREF", "WORK", "HOME", "INTERNET" ) );
 
937
 
 
938
                        // add address
 
939
                        int type;
 
940
                        if( types.contains( "WORK" ) )
 
941
                                type = Contacts.ContactMethods.TYPE_WORK;
 
942
                        else
 
943
                                type = Contacts.ContactMethods.TYPE_HOME;
 
944
 
 
945
                        addAddress( unescapeValue( value ), type );
1001
946
                }
1002
947
 
1003
948
                public void finaliseVcard()
1035
980
                        HashSet< String > ret = new HashSet< String >();
1036
981
 
1037
982
                        Pattern p = Pattern.compile(
1038
 
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$",
1039
 
                                Pattern.CASE_INSENSITIVE );
 
983
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$" );
1040
984
                        for( int i = 0; i < params.length; i++ ) {
1041
985
                                Matcher m = p.matcher( params[ i ] );
1042
986
                                if( m.matches() )
1050
994
                 * Amongst the params, return any type values present. For v2.1 vCards,
1051
995
                 * those types are just parameters. For v3.0, they are prefixed with
1052
996
                 * "TYPE=". There may also be multiple type parameters.
1053
 
                 * @param params an array of params to look for types in
1054
 
                 * @param valid_types an list of upper-case type values to look for
 
997
                 * @param params
 
998
                 * @param a list of type values to look for
1055
999
                 * @return a set of present type values
1056
1000
                 */
1057
1001
                private Set< String > extractTypes( String[] params,
1063
1007
                        String type_params[] = checkParams( params, "TYPE" );
1064
1008
                        for( int a = 0; a < type_params.length; a++ )
1065
1009
                        {
1066
 
                                // check for a comma-separated list of types (why? I don't think
1067
 
                                // this is in the specs!)
 
1010
                                // check for a comma-separated list of types (why? this isn't in
 
1011
                                // the specs!)
1068
1012
                                String[] parts = type_params[ a ].split( "," );
1069
 
                                for( int i = 0; i < parts.length; i++ ) {
1070
 
                                        String ucpart = parts[ i ].toUpperCase( Locale.US );
1071
 
                                        if( valid_types.contains( ucpart ) )
1072
 
                                                types.add( ucpart );
1073
 
                                }
 
1013
                                for( int i = 0; i < parts.length; i++ )
 
1014
                                        if( valid_types.contains( parts[ i ] ) )
 
1015
                                                types.add( parts[ i ] );
1074
1016
                        }
1075
1017
 
1076
1018
                        // get 2.1-style type param
1077
1019
                        if( _version.equals( "2.1" ) ) {
1078
 
                                for( int i = 1; i < params.length; i++ ) {
1079
 
                                        String ucparam = params[ i ].toUpperCase( Locale.US );
1080
 
                                        if( valid_types.contains( ucparam ) )
1081
 
                                                types.add( ucparam );
1082
 
                                }
 
1020
                                for( int i = 1; i < params.length; i++ )
 
1021
                                        if( valid_types.contains( params[ i ] ) )
 
1022
                                                types.add( params[ i ] );
1083
1023
                        }
1084
1024
 
1085
1025
                        return types;