/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 14:47:57 UTC
  • Revision ID: tim@ed.am-20121221144757-5cb1lgsp7fdt7p2n
updated TODO

Show diffs side-by-side

added added

removed removed

202
202
                                        buffer.limit() - buffer.position(), "US-ASCII" );
203
203
                        }
204
204
                        catch( UnsupportedEncodingException e ) {
205
 
                                // we know US-ASCII is supported, so appease the compiler...
 
205
                                // we know US-ASCII *is* supported, so appease the compiler...
206
206
                                line = "";
207
207
                        }
208
208
 
209
209
                        if( vcard == null ) {
210
210
                                // look for vcard beginning
211
 
                                if( line.matches( "^BEGIN:VCARD" ) ) {
 
211
                                if( line.matches( "^BEGIN[ \t]*:[ \t]*VCARD" ) ) {
212
212
                                        setProgress( _progress++ );
213
213
                                        vcard = new Vcard();
214
214
                                        vcard_start_line = cli.getLineNumber();
216
216
                        }
217
217
                        else {
218
218
                                // look for vcard content or ending
219
 
                                if( line.matches( "^END:VCARD" ) )
 
219
                                if( line.matches( "^END[ \t]*:[ \t]*VCARD" ) )
220
220
                                {
221
221
                                        // finalise the vcard/contact
222
222
                                        try {
351
351
                public boolean doesNextLineLookFolded()
352
352
                {
353
353
                        return _pos > 0 && _pos < _content.length &&
354
 
                                _content[ _pos - 1 ] == '\n' && _content[ _pos ] == ' ';
 
354
                                _content[ _pos - 1 ] == '\n' &&
 
355
                                ( _content[ _pos ] == ' ' || _content[ _pos ] == '\t' );
355
356
                }
356
357
 
357
358
                public int getLineNumber()
369
370
                private final static int MULTILINE_NONE = 0;
370
371
                private final static int MULTILINE_ENCODED = 1; // v2.1 quoted-printable
371
372
                private final static int MULTILINE_ESCAPED = 2; // v2.1 \\CRLF
372
 
                private final static int MULTILINE_FOLDED = 3;  // v3.0 folding
 
373
                private final static int MULTILINE_FOLDED = 3;  // MIME-DIR folding
373
374
 
374
375
                private String _version = null;
375
376
                private Vector< ByteBuffer > _buffers = null;
473
474
 
474
475
                                // is it a version line?
475
476
                                if( name_and_params != null &&
476
 
                                        name_and_params.equals( "VERSION" ) )
 
477
                                        name_and_params.equalsIgnoreCase( "VERSION" ) )
477
478
                                {
478
479
                                        // yes, get it!
479
480
                                        String value = extractValueFromLine( buffer, line );
503
504
                        else
504
505
                        {
505
506
                                // name and params and the position in the buffer where the
506
 
                                // "value" part of the line start
 
507
                                // "value" part of the line starts
507
508
                                String name_and_params;
508
509
                                int pos;
509
510
 
539
540
                                }
540
541
                                else
541
542
                                {
 
543
                                        // skip empty lines
 
544
                                        if( line.trim().length() == 0 ) return;
 
545
 
542
546
                                        // get name and params from line, and since we're not
543
547
                                        // parsing a subsequent line in a multi-line, this should
544
548
                                        // not fail, or it's an error
573
577
                                                "FN", "ORG", "TITLE", "TEL", "EMAIL", "ADR", "LABEL" }
574
578
                                ) );
575
579
                                boolean is_interesting_field =
576
 
                                        interesting_fields.contains( name_param_parts[ 0 ] );
 
580
                                        interesting_fields.contains(
 
581
                                                name_param_parts[ 0 ].toUpperCase( Locale.US ) );
577
582
 
578
583
                                // parse encoding parameter
579
584
                                String encoding = checkParam( name_param_parts, "ENCODING" );
580
585
                                if( encoding != null )
581
586
                                        encoding = encoding.toUpperCase( Locale.US );
582
587
                                if( is_interesting_field && encoding != null &&
583
 
                                        !encoding.equals( "8BIT" ) &&
584
 
                                        !encoding.equals( "QUOTED-PRINTABLE" ) )
585
 
                                        //&& !encoding.equals( "BASE64" ) )
 
588
                                        !encoding.equalsIgnoreCase( "8BIT" ) &&
 
589
                                        !encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
 
590
                                        //&& !encoding.equalsIgnoreCase( "BASE64" ) )
586
591
                                {
587
592
                                        throw new ParseException( R.string.error_vcf_encoding );
588
593
                                }
592
597
                                if( charset != null )
593
598
                                        charset = charset.toUpperCase( Locale.US );
594
599
                                if( charset != null &&
595
 
                                        !charset.equals( "US-ASCII" ) &&
596
 
                                        !charset.equals( "ASCII" ) &&
597
 
                                        !charset.equals( "UTF-8" ) )
 
600
                                        !charset.equalsIgnoreCase( "US-ASCII" ) &&
 
601
                                        !charset.equalsIgnoreCase( "ASCII" ) &&
 
602
                                        !charset.equalsIgnoreCase( "UTF-8" ) )
598
603
                                {
599
604
                                        throw new ParseException( R.string.error_vcf_charset );
600
605
                                }
602
607
                                // do unencoding (or default to a fake unencoding result with
603
608
                                // the raw string)
604
609
                                UnencodeResult unencoding_result = null;
605
 
                                if( encoding != null && encoding.equals( "QUOTED-PRINTABLE" ) )
 
610
                                if( encoding != null &&
 
611
                                        encoding.equalsIgnoreCase( "QUOTED-PRINTABLE" ) )
 
612
                                {
606
613
                                        unencoding_result = unencodeQuotedPrintable( value );
607
 
//                              else if( encoding != null && encoding.equals( "BASE64" ) )
 
614
                                }
 
615
//                              else if( encoding != null &&
 
616
//                                      encoding.equalsIgnoreCase( "BASE64" ) )
 
617
//                              {
608
618
//                                      unencoding_result = unencodeBase64( props[ 1 ], charset );
 
619
//                              }
609
620
                                if( unencoding_result != null ) {
610
621
                                        value = unencoding_result.getBuffer();
611
622
                                        if( unencoding_result.isAnotherLineRequired() )
616
627
                                // specified for a v2.1 vcard entry, we assume it's US-ASCII)
617
628
                                if( ( charset == null && _version.equals( "2.1" ) ) ||
618
629
                                        ( charset != null && (
619
 
                                                charset.equals( "ASCII" ) ||
620
 
                                                charset.equals( "US-ASCII" ) ) ) )
 
630
                                                charset.equalsIgnoreCase( "ASCII" ) ||
 
631
                                                charset.equalsIgnoreCase( "US-ASCII" ) ) ) )
621
632
                                {
622
633
                                        value = transcodeAsciiToUtf8( value );
623
634
                                }
634
645
                                // for some entries that have semicolon-separated value parts,
635
646
                                // check to see if the value ends in an escape character, which
636
647
                                // indicates that we have a multi-line value
637
 
                                if( ( name_param_parts[ 0 ].equals( "N" ) ||
638
 
                                        name_param_parts[ 0 ].equals( "ORG" ) ||
639
 
                                        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" ) ) &&
640
651
                                        doesStringEndInAnEscapeChar( string_value ) )
641
652
                                {
642
653
                                        _parser_multiline_state = MULTILINE_ESCAPED;
644
655
                                                string_value.length() - 1 );
645
656
                                }
646
657
 
647
 
                                // now we know whether we're in an encoding multi-line,
648
 
                                // 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
649
660
                                if( _parser_multiline_state == MULTILINE_NONE &&
650
 
                                        _version.equals( "3.0" ) && next_line_looks_folded )
 
661
                                        next_line_looks_folded )
651
662
                                {
652
663
                                        _parser_multiline_state = MULTILINE_FOLDED;
653
664
                                }
665
676
                                if( complete_value.length() < 1 ) return;
666
677
 
667
678
                                // parse some properties
668
 
                                if( name_param_parts[ 0 ].equals( "N" ) )
 
679
                                if( name_param_parts[ 0 ].equalsIgnoreCase( "N" ) )
669
680
                                        parseN( name_param_parts, complete_value );
670
 
                                else if( name_param_parts[ 0 ].equals( "FN" ) )
 
681
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "FN" ) )
671
682
                                        parseFN( name_param_parts, complete_value );
672
 
                                else if( name_param_parts[ 0 ].equals( "ORG" ) )
 
683
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ORG" ) )
673
684
                                        parseORG( name_param_parts, complete_value );
674
 
                                else if( name_param_parts[ 0 ].equals( "TITLE" ) )
 
685
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TITLE" ) )
675
686
                                        parseTITLE( name_param_parts, complete_value );
676
 
                                else if( name_param_parts[ 0 ].equals( "TEL" ) )
 
687
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "TEL" ) )
677
688
                                        parseTEL( name_param_parts, complete_value );
678
 
                                else if( name_param_parts[ 0 ].equals( "EMAIL" ) )
 
689
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "EMAIL" ) )
679
690
                                        parseEMAIL( name_param_parts, complete_value );
680
 
                                else if( name_param_parts[ 0 ].equals( "ADR" ) )
 
691
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "ADR" ) )
681
692
                                        parseADR( name_param_parts, complete_value );
682
 
                                else if( name_param_parts[ 0 ].equals( "LABEL" ) )
 
693
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "LABEL" ) )
683
694
                                        parseLABEL( name_param_parts, complete_value );
 
695
                                else if( name_param_parts[ 0 ].equalsIgnoreCase( "NOTE" ) )
 
696
                                        parseNOTE( name_param_parts, complete_value );
684
697
                        }
685
698
                }
686
699
 
758
771
                                in_escape = false;
759
772
                                switch( c )
760
773
                                {
 
774
                                case 'T':
 
775
                                case 't':
 
776
                                        // add tab (invalid/non-standard, but accepted)
 
777
                                        ret.append( '\t' );
 
778
                                        break;
761
779
                                case 'N':
762
780
                                case 'n':
763
781
                                        // add newline
771
789
                                        break;
772
790
                                default:
773
791
                                        // unknown escape sequence, so add it unescaped
 
792
                                        // (invalid/non-standard, but accepted)
774
793
                                        ret.append( "\\" );
775
794
                                        ret.append( Character.toChars( c ) );
776
795
                                        break;
926
945
                        for( int a = 0; a < adr_parts.length; a++ )
927
946
                                if( adr_parts[ a ].length() > 0 )
928
947
                                {
929
 
                                        // split this part in to it's comma-separated bits
930
 
                                        String[] adr_part_parts =
931
 
                                                splitValueByCharacter( adr_parts[ a ], ',' );
932
 
                                        for( int b = 0; b < adr_part_parts.length; b++ )
933
 
                                                if( adr_part_parts[ b ].length() > 0 )
934
 
                                                {
935
 
                                                        if( value.length() > 0 ) value += "\n";
936
 
                                                        value += adr_part_parts[ b ];
937
 
                                                }
 
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
                                        }
938
968
                                }
939
969
 
940
970
                        Set< String > types = extractTypes( params, Arrays.asList(
965
995
                        addAddress( unescapeValue( value ), type );
966
996
                }
967
997
 
 
998
                private void parseNOTE( String[] params, String value )
 
999
                {
 
1000
                        addNote( unescapeValue( value ) );
 
1001
                }
 
1002
 
968
1003
                public void finaliseVcard()
969
1004
                        throws ParseException, ContactNotIdentifiableException
970
1005
                {
1000
1035
                        HashSet< String > ret = new HashSet< String >();
1001
1036
 
1002
1037
                        Pattern p = Pattern.compile(
1003
 
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$" );
 
1038
                                "^" + name + "[ \\t]*=[ \\t]*(\"?)(.*)\\1$",
 
1039
                                Pattern.CASE_INSENSITIVE );
1004
1040
                        for( int i = 0; i < params.length; i++ ) {
1005
1041
                                Matcher m = p.matcher( params[ i ] );
1006
1042
                                if( m.matches() )
1014
1050
                 * Amongst the params, return any type values present. For v2.1 vCards,
1015
1051
                 * those types are just parameters. For v3.0, they are prefixed with
1016
1052
                 * "TYPE=". There may also be multiple type parameters.
1017
 
                 * @param params
1018
 
                 * @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
1019
1055
                 * @return a set of present type values
1020
1056
                 */
1021
1057
                private Set< String > extractTypes( String[] params,
1027
1063
                        String type_params[] = checkParams( params, "TYPE" );
1028
1064
                        for( int a = 0; a < type_params.length; a++ )
1029
1065
                        {
1030
 
                                // check for a comma-separated list of types (why? this isn't in
1031
 
                                // the specs!)
 
1066
                                // check for a comma-separated list of types (why? I don't think
 
1067
                                // this is in the specs!)
1032
1068
                                String[] parts = type_params[ a ].split( "," );
1033
 
                                for( int i = 0; i < parts.length; i++ )
1034
 
                                        if( valid_types.contains( parts[ i ] ) )
1035
 
                                                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
                                }
1036
1074
                        }
1037
1075
 
1038
1076
                        // get 2.1-style type param
1039
1077
                        if( _version.equals( "2.1" ) ) {
1040
 
                                for( int i = 1; i < params.length; i++ )
1041
 
                                        if( valid_types.contains( params[ i ] ) )
1042
 
                                                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
                                }
1043
1083
                        }
1044
1084
 
1045
1085
                        return types;