/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: 2012-12-20 12:33:54 UTC
  • Revision ID: tim@ed.am-20121220123354-y1ls9f4awasf53d5
updated TODO and NOTES

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