/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/am/ed/importcontacts/VcardImporter.java

  • Committer: edam
  • Date: 2012-12-21 14:48:44 UTC
  • Revision ID: tim@ed.am-20121221144844-licgvkppbjq4th4m
updated copyright dates and improved an error message

Show diffs side-by-side

added added

removed removed

38
38
import java.util.HashSet;
39
39
import java.util.Iterator;
40
40
import java.util.List;
 
41
import java.util.Locale;
41
42
import java.util.NoSuchElementException;
42
43
import java.util.Set;
43
44
import java.util.Vector;
44
45
import java.util.regex.Matcher;
45
46
import java.util.regex.Pattern;
46
47
 
 
48
import android.annotation.SuppressLint;
47
49
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" )
61
62
        @Override
62
63
        protected void onImport() throws AbortImportException
63
64
        {
82
83
                                // get files
83
84
                                class VCardFilter implements FilenameFilter {
84
85
                                        public boolean accept( File dir, String name ) {
85
 
                                                return name.toLowerCase().endsWith( ".vcf" );
 
86
                                                return name.toLowerCase( Locale.US ).endsWith( ".vcf" );
86
87
                                        }
87
88
                                }
88
89
                                files = file.listFiles( new VCardFilter() );
201
202
                                        buffer.limit() - buffer.position(), "US-ASCII" );
202
203
                        }
203
204
                        catch( UnsupportedEncodingException e ) {
204
 
                                // we know US-ASCII is supported, so appease the compiler...
 
205
                                // we know US-ASCII *is* supported, so appease the compiler...
205
206
                                line = "";
206
207
                        }
207
208
 
208
209
                        if( vcard == null ) {
209
210
                                // look for vcard beginning
210
 
                                if( line.matches( "^BEGIN:VCARD" ) ) {
 
211
                                if( line.matches( "^BEGIN[ \t]*:[ \t]*VCARD" ) ) {
211
212
                                        setProgress( _progress++ );
212
213
                                        vcard = new Vcard();
213
214
                                        vcard_start_line = cli.getLineNumber();
215
216
                        }
216
217
                        else {
217
218
                                // look for vcard content or ending
218
 
                                if( line.matches( "^END:VCARD" ) )
 
219
                                if( line.matches( "^END[ \t]*:[ \t]*VCARD" ) )
219
220
                                {
220
221
                                        // finalise the vcard/contact
221
222
                                        try {
350
351
                public boolean doesNextLineLookFolded()
351
352
                {
352
353
                        return _pos > 0 && _pos < _content.length &&
353
 
                                _content[ _pos - 1 ] == '\n' && _content[ _pos ] == ' ';
 
354
                                _content[ _pos - 1 ] == '\n' &&
 
355
                                ( _content[ _pos ] == ' ' || _content[ _pos ] == '\t' );
354
356
                }
355
357
 
356
358
                public int getLineNumber()
368
370
                private final static int MULTILINE_NONE = 0;
369
371
                private final static int MULTILINE_ENCODED = 1; // v2.1 quoted-printable
370
372
                private final static int MULTILINE_ESCAPED = 2; // v2.1 \\CRLF
371
 
                private final static int MULTILINE_FOLDED = 3;  // v3.0 folding
 
373
                private final static int MULTILINE_FOLDED = 3;  // MIME-DIR folding
372
374
 
373
375
                private String _version = null;
374
376
                private Vector< ByteBuffer > _buffers = null;
472
474
 
473
475
                                // is it a version line?
474
476
                                if( name_and_params != null &&
475
 
                                        name_and_params.equals( "VERSION" ) )
 
477
                                        name_and_params.equalsIgnoreCase( "VERSION" ) )
476
478
                                {
477
479
                                        // yes, get it!
478
480
                                        String value = extractValueFromLine( buffer, line );
502
504
                        else
503
505
                        {
504
506
                                // name and params and the position in the buffer where the
505
 
                                // "value" part of the line start
 
507
                                // "value" part of the line starts
506
508
                                String name_and_params;
507
509
                                int pos;
508
510
 
538
540
                                }
539
541
                                else
540
542
                                {
 
543
                                        // skip empty lines
 
544
                                        if( line.trim().length() == 0 ) return;
 
545
 
541
546
                                        // get name and params from line, and since we're not
542
547
                                        // parsing a subsequent line in a multi-line, this should
543
548
                                        // not fail, or it's an error
572
577
                                                "FN", "ORG", "TITLE", "TEL", "EMAIL", "ADR", "LABEL" }
573
578
                                ) );
574
579
                                boolean is_interesting_field =
575
 
                                        interesting_fields.contains( name_param_parts[ 0 ] );
 
580
                                        interesting_fields.contains(
 
581
                                                name_param_parts[ 0 ].toUpperCase( Locale.US ) );
576
582
 
577
583
                                // parse encoding parameter
578
584
                                String encoding = checkParam( name_param_parts, "ENCODING" );
579
 
                                if( encoding != null ) encoding = encoding.toUpperCase();
 
585
                                if( encoding != null )
 
586
                                        encoding = encoding.toUpperCase( Locale.US );
580
587
                                if( is_interesting_field && encoding != null &&
581
 
                                        !encoding.equals( "8BIT" ) &&
582
 
                                        !encoding.equals( "QUOTED-PRINTABLE" ) )
583
 
                                        //&& !encoding.equals( "BASE64" ) )
 
588
                                        !encoding.equalsIgnoreCase( "8BIT" ) &&
 
589
                                        !encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
 
590
                                        //&& !encoding.equalsIgnoreCase( "BASE64" ) )
584
591
                                {
585
592
                                        throw new ParseException( R.string.error_vcf_encoding );
586
593
                                }
587
594
 
588
595
                                // parse charset parameter
589
596
                                String charset = checkParam( name_param_parts, "CHARSET" );
590
 
                                if( charset != null ) charset = charset.toUpperCase();
 
597
                                if( charset != null )
 
598
                                        charset = charset.toUpperCase( Locale.US );
591
599
                                if( charset != null &&
592
 
                                        !charset.equals( "US-ASCII" ) &&
593
 
                                        !charset.equals( "ASCII" ) &&
594
 
                                        !charset.equals( "UTF-8" ) )
 
600
                                        !charset.equalsIgnoreCase( "US-ASCII" ) &&
 
601
                                        !charset.equalsIgnoreCase( "ASCII" ) &&
 
602
                                        !charset.equalsIgnoreCase( "UTF-8" ) )
595
603
                                {
596
604
                                        throw new ParseException( R.string.error_vcf_charset );
597
605
                                }
599
607
                                // do unencoding (or default to a fake unencoding result with
600
608
                                // the raw string)
601
609
                                UnencodeResult unencoding_result = null;
602
 
                                if( encoding != null && encoding.equals( "QUOTED-PRINTABLE" ) )
 
610
                                if( encoding != null &&
 
611
                                        encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
 
612
                                {
603
613
                                        unencoding_result = unencodeQuotedPrintable( value );
604
 
//                              else if( encoding != null && encoding.equals( "BASE64" ) )
 
614
                                }
 
615
//                              else if( encoding != null &&
 
616
//                                      encoding.equalsIgnoreCase( "BASE64" ) )
 
617
//                              {
605
618
//                                      unencoding_result = unencodeBase64( props[ 1 ], charset );
 
619
//                              }
606
620
                                if( unencoding_result != null ) {
607
621
                                        value = unencoding_result.getBuffer();
608
622
                                        if( unencoding_result.isAnotherLineRequired() )
613
627
                                // specified for a v2.1 vcard entry, we assume it's US-ASCII)
614
628
                                if( ( charset == null && _version.equals( "2.1" ) ) ||
615
629
                                        ( charset != null && (
616
 
                                                charset.equals( "ASCII" ) ||
617
 
                                                charset.equals( "US-ASCII" ) ) ) )
 
630
                                                charset.equalsIgnoreCase( "ASCII" ) ||
 
631
                                                charset.equalsIgnoreCase( "US-ASCII" ) ) ) )
618
632
                                {
619
633
                                        value = transcodeAsciiToUtf8( value );
620
634
                                }
631
645
                                // for some entries that have semicolon-separated value parts,
632
646
                                // check to see if the value ends in an escape character, which
633
647
                                // indicates that we have a multi-line value
634
 
                                if( ( name_param_parts[ 0 ].equals( "N" ) ||
635
 
                                        name_param_parts[ 0 ].equals( "ORG" ) ||
636
 
                                        name_param_parts[ 0 ].equals( "ADR" ) ) &&
 
648
                                if( ( name_param_parts[ 0 ].equalsIgnoreCase( "N" ) ||
 
649
                                        name_param_parts[ 0 ].equalsIgnoreCase( "ORG" ) ||
 
650
                                        name_param_parts[ 0 ].equalsIgnoreCase( "ADR" ) ) &&
637
651
                                        doesStringEndInAnEscapeChar( string_value ) )
638
652
                                {
639
653
                                        _parser_multiline_state = MULTILINE_ESCAPED;
641
655
                                                string_value.length() - 1 );
642
656
                                }
643
657
 
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
 
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
646
660
                                if( _parser_multiline_state == MULTILINE_NONE &&
647
 
                                        _version.equals( "3.0" ) && next_line_looks_folded )
 
661
                                        next_line_looks_folded )
648
662
                                {
649
663
                                        _parser_multiline_state = MULTILINE_FOLDED;
650
664
                                }
662
676
                                if( complete_value.length() < 1 ) return;
663
677
 
664
678
                                // parse some properties
665
 
                                if( name_param_parts[ 0 ].equals( "N" ) )
 
679
                                if( name_param_parts[ 0 ].equalsIgnoreCase( "N" ) )
666
680
                                        parseN( name_param_parts, complete_value );
667
 
                                else if( name_param_parts[ 0 ].equals( "FN" ) )
 
681
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "FN" ) )
668
682
                                        parseFN( name_param_parts, complete_value );
669
 
                                else if( name_param_parts[ 0 ].equals( "ORG" ) )
 
683
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ORG" ) )
670
684
                                        parseORG( name_param_parts, complete_value );
671
 
                                else if( name_param_parts[ 0 ].equals( "TITLE" ) )
 
685
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TITLE" ) )
672
686
                                        parseTITLE( name_param_parts, complete_value );
673
 
                                else if( name_param_parts[ 0 ].equals( "TEL" ) )
 
687
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TEL" ) )
674
688
                                        parseTEL( name_param_parts, complete_value );
675
 
                                else if( name_param_parts[ 0 ].equals( "EMAIL" ) )
 
689
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "EMAIL" ) )
676
690
                                        parseEMAIL( name_param_parts, complete_value );
677
 
                                else if( name_param_parts[ 0 ].equals( "ADR" ) )
 
691
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ADR" ) )
678
692
                                        parseADR( name_param_parts, complete_value );
679
 
                                else if( name_param_parts[ 0 ].equals( "LABEL" ) )
 
693
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "LABEL" ) )
680
694
                                        parseLABEL( name_param_parts, complete_value );
 
695
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "NOTE" ) )
 
696
                                        parseNOTE( name_param_parts, complete_value );
681
697
                        }
682
698
                }
683
699
 
755
771
                                in_escape = false;
756
772
                                switch( c )
757
773
                                {
 
774
                                case 'T':
 
775
                                case 't':
 
776
                                        // add tab (invalid/non-standard, but accepted)
 
777
                                        ret.append( '\t' );
 
778
                                        break;
758
779
                                case 'N':
759
780
                                case 'n':
760
781
                                        // add newline
768
789
                                        break;
769
790
                                default:
770
791
                                        // unknown escape sequence, so add it unescaped
 
792
                                        // (invalid/non-standard, but accepted)
771
793
                                        ret.append( "\\" );
772
794
                                        ret.append( Character.toChars( c ) );
773
795
                                        break;
879
901
                        int type;
880
902
                        if( types.contains( "FAX" ) )
881
903
                                if( types.contains( "HOME" ) )
882
 
                                        type = PhonesColumns.TYPE_FAX_HOME;
 
904
                                        type = TYPE_FAX_HOME;
883
905
                                else
884
 
                                        type = PhonesColumns.TYPE_FAX_WORK;
 
906
                                        type = TYPE_FAX_WORK;
885
907
                        else if( types.contains( "CELL" ) || types.contains( "VIDEO" ) )
886
 
                                type = PhonesColumns.TYPE_MOBILE;
 
908
                                type = TYPE_MOBILE;
887
909
                        else if( types.contains( "PAGER" ) )
888
 
                                type = PhonesColumns.TYPE_PAGER;
 
910
                                type = TYPE_PAGER;
889
911
                        else if( types.contains( "WORK" ) )
890
 
                                type = PhonesColumns.TYPE_WORK;
 
912
                                type = TYPE_WORK;
891
913
                        else
892
 
                                type = PhonesColumns.TYPE_HOME;
 
914
                                type = TYPE_HOME;
893
915
 
894
916
                        // add phone number
895
917
                        addNumber( value, type, is_preferred );
906
928
                        boolean is_preferred = types.contains( "PREF" );
907
929
                        int type;
908
930
                        if( types.contains( "WORK" ) )
909
 
                                type = Contacts.ContactMethods.TYPE_WORK;
 
931
                                type = TYPE_WORK;
910
932
                        else
911
 
                                type = Contacts.ContactMethods.TYPE_HOME;
 
933
                                type = TYPE_HOME;
912
934
 
913
935
                        addEmail( unescapeValue( value ), type, is_preferred );
914
936
                }
923
945
                        for( int a = 0; a < adr_parts.length; a++ )
924
946
                                if( adr_parts[ a ].length() > 0 )
925
947
                                {
926
 
                                        // split this part in to it's comma-separated bits
927
 
                                        String[] adr_part_parts =
928
 
                                                splitValueByCharacter( adr_parts[ a ], ',' );
929
 
                                        for( int b = 0; b < adr_part_parts.length; b++ )
930
 
                                                if( adr_part_parts[ b ].length() > 0 )
931
 
                                                {
932
 
                                                        if( value.length() > 0 ) value += "\n";
933
 
                                                        value += adr_part_parts[ b ];
934
 
                                                }
 
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
                                        }
935
968
                                }
936
969
 
937
970
                        Set< String > types = extractTypes( params, Arrays.asList(
940
973
                        // add address
941
974
                        int type;
942
975
                        if( types.contains( "WORK" ) )
943
 
                                type = Contacts.ContactMethods.TYPE_WORK;
 
976
                                type = TYPE_WORK;
944
977
                        else
945
 
                                type = Contacts.ContactMethods.TYPE_HOME;
 
978
                                type = TYPE_HOME;
946
979
 
947
980
                        addAddress( unescapeValue( value ), type );
948
981
                }
955
988
                        // add address
956
989
                        int type;
957
990
                        if( types.contains( "WORK" ) )
958
 
                                type = Contacts.ContactMethods.TYPE_WORK;
 
991
                                type = TYPE_WORK;
959
992
                        else
960
 
                                type = Contacts.ContactMethods.TYPE_HOME;
 
993
                                type = TYPE_HOME;
961
994
 
962
995
                        addAddress( unescapeValue( value ), type );
963
996
                }
964
997
 
 
998
                private void parseNOTE( String[] params, String value )
 
999
                {
 
1000
                        addNote( unescapeValue( value ) );
 
1001
                }
 
1002
 
965
1003
                public void finaliseVcard()
966
1004
                        throws ParseException, ContactNotIdentifiableException
967
1005
                {
997
1035
                        HashSet< String > ret = new HashSet< String >();
998
1036
 
999
1037
                        Pattern p = Pattern.compile(
1000
 
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$" );
 
1038
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$",
 
1039
                                Pattern.CASE_INSENSITIVE );
1001
1040
                        for( int i = 0; i < params.length; i++ ) {
1002
1041
                                Matcher m = p.matcher( params[ i ] );
1003
1042
                                if( m.matches() )
1011
1050
                 * Amongst the params, return any type values present. For v2.1 vCards,
1012
1051
                 * those types are just parameters. For v3.0, they are prefixed with
1013
1052
                 * "TYPE=". There may also be multiple type parameters.
1014
 
                 * @param params
1015
 
                 * @param a list of type values to look for
 
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
1016
1055
                 * @return a set of present type values
1017
1056
                 */
1018
1057
                private Set< String > extractTypes( String[] params,
1024
1063
                        String type_params[] = checkParams( params, "TYPE" );
1025
1064
                        for( int a = 0; a < type_params.length; a++ )
1026
1065
                        {
1027
 
                                // check for a comma-separated list of types (why? this isn't in
1028
 
                                // the specs!)
 
1066
                                // check for a comma-separated list of types (why? I don't think
 
1067
                                // this is in the specs!)
1029
1068
                                String[] parts = type_params[ a ].split( "," );
1030
 
                                for( int i = 0; i < parts.length; i++ )
1031
 
                                        if( valid_types.contains( parts[ i ] ) )
1032
 
                                                types.add( parts[ i ] );
 
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
                                }
1033
1074
                        }
1034
1075
 
1035
1076
                        // get 2.1-style type param
1036
1077
                        if( _version.equals( "2.1" ) ) {
1037
 
                                for( int i = 1; i < params.length; i++ )
1038
 
                                        if( valid_types.contains( params[ i ] ) )
1039
 
                                                types.add( params[ i ] );
 
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
                                }
1040
1083
                        }
1041
1084
 
1042
1085
                        return types;