… | |
… | |
7 | and $self->selection (map { y/A-Za-z/N-ZA-Mn-za-m/; $_ } $self->selection); |
7 | and $self->selection (map { y/A-Za-z/N-ZA-Mn-za-m/; $_ } $self->selection); |
8 | |
8 | |
9 | () |
9 | () |
10 | } |
10 | } |
11 | |
11 | |
|
|
12 | sub on_init { |
|
|
13 | my ($self) = @_; |
|
|
14 | |
|
|
15 | if (defined (my $res = $self->resource ("cutchars"))) { |
|
|
16 | $res = $self->locale_decode ($res); |
|
|
17 | push @{ $self->{patterns} }, qr{\G [\Q$res\E[:space:]]* ([^\Q$res\E[:space:]]+) }x; |
|
|
18 | } |
|
|
19 | |
|
|
20 | for (my $idx = 0; defined (my $res = $self->x_resource ("selection.pattern-$idx")); $idx++) { |
|
|
21 | $res = $self->locale_decode ($res); |
|
|
22 | utf8::encode $res; |
|
|
23 | push @{ $self->{patterns} }, qr/$res/; |
|
|
24 | } |
|
|
25 | |
|
|
26 | () |
|
|
27 | } |
|
|
28 | |
12 | # "find interetsing things"-patterns |
29 | # "find interesting things"-patterns |
13 | my @mark_patterns = ( |
30 | my @mark_patterns = ( |
14 | qr{([[:alnum:]]+)}, |
|
|
15 | |
|
|
16 | # common "parentheses" |
31 | # common types of "parentheses" |
17 | qr{(?:^|\s) ‘ ([^‘’]+?) ’ (?:\s|\)|$)}x, |
32 | qr{ (?<![^[:space:]]) ‘ ([^‘’]+) ’ (?![^[:space]]) }x, |
18 | qr{(?:^|\s) ` ([^`']+?) ' (?:\s|\)|$)}x, |
33 | qr{ (?<![^[:space:]]) ` ([^`']+) ' (?![^[:space]]) }x, |
19 | qr{ \{ ([^{}]+?) \} }x, |
34 | qr{ (?<![^[:space:]]) (" [^[:space:]] [^"]* ") }x, |
20 | qr{ \[ ([^{}]+?) \] }x, |
35 | qr{ (" [^"]* [^[:space:]] ") (?![^[:space]]) }x, |
|
|
36 | qr{ \< ([^<>[:space:]]+) \> }x, |
|
|
37 | qr{ \{ ([^{}[:space:]]+) \} }x, |
|
|
38 | qr{ \[ ([^{}[:space:]]+) \] }x, |
|
|
39 | qr{ \( ([^()[:space:]]+) \) }x, |
21 | |
40 | |
22 | # urls, just a heuristic |
41 | # urls, just a heuristic |
23 | qr{( |
42 | qr{( |
24 | (?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
43 | (?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
25 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~] # exclude some trailing characters (heuristic) |
44 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~] # exclude some trailing characters (heuristic) |
26 | )}x, |
45 | )}x, |
27 | |
46 | |
28 | # shell-like argument quoting, basically always matches |
47 | # shell-like argument quoting, basically always matches |
29 | qr{\G [\ \t|&;<>()] *( |
48 | qr{\G [\ \t|&;<>()]* ( |
30 | (?: |
49 | (?: |
31 | [^\\"'\ \t|&;<>()]+ |
50 | [^\\"'\ \t|&;<>()]+ |
32 | | \\. |
51 | | \\. |
33 | | " (?: [^\\"]+ | \\. )* " |
52 | | " (?: [^\\"]+ | \\. )* " |
34 | | ' [^']* ' |
53 | | ' [^']* ' |
… | |
… | |
41 | qr{^"([^\\"'\ \t|&;<>()*?]+)"$}, # "simple" => simple |
60 | qr{^"([^\\"'\ \t|&;<>()*?]+)"$}, # "simple" => simple |
42 | qr{^(.*)[,\-]$}, # strip off trailing , and - |
61 | qr{^(.*)[,\-]$}, # strip off trailing , and - |
43 | ); |
62 | ); |
44 | |
63 | |
45 | sub on_sel_extend { |
64 | sub on_sel_extend { |
46 | my ($self) = @_; |
65 | my ($self, $time) = @_; |
47 | |
66 | |
48 | my ($row, $col) = $self->selection_mark; |
67 | my ($row, $col) = $self->selection_mark; |
49 | my $line = $self->line ($row); |
68 | my $line = $self->line ($row); |
50 | my $text = $line->t; |
69 | my $text = $line->t; |
51 | my $markofs = $line->offset_of ($row, $col); |
70 | my $markofs = $line->offset_of ($row, $col); |
52 | my $curlen = $line->offset_of ($self->selection_end) |
71 | my $curlen = $line->offset_of ($self->selection_end) |
53 | - $line->offset_of ($self->selection_beg); |
72 | - $line->offset_of ($self->selection_beg); |
54 | |
73 | |
55 | my @matches; |
74 | my @matches; |
56 | |
75 | |
57 | for my $regex (@mark_patterns) { |
76 | if ($markofs < $line->l) { |
58 | while ($text =~ /$regex/g) { |
77 | # convert markofs form character to UTF-8 offset space |
59 | if ($-[1] <= $markofs and $markofs <= $+[1]) { |
78 | { |
60 | my $ofs = $-[1]; |
79 | my $prefix = substr $text, 0, $markofs; |
61 | my $match = $1; |
80 | utf8::encode $prefix; |
|
|
81 | $markofs = length $prefix; |
|
|
82 | } |
62 | |
83 | |
63 | push @matches, [$ofs, length $match]; |
84 | # not doing matches in unicode mode helps speed |
|
|
85 | # enourmously here. working in utf-8 should be |
|
|
86 | # equivalent due to the magic of utf-8 encoding. |
|
|
87 | utf8::encode $text; |
|
|
88 | study $text; # _really_ helps, too :) |
64 | |
89 | |
|
|
90 | for my $regex (@mark_patterns, @{ $self->{patterns} }) { |
|
|
91 | while ($text =~ /$regex/g) { |
|
|
92 | if ($-[1] <= $markofs and $markofs <= $+[1]) { |
|
|
93 | my $ofs = $-[1]; |
|
|
94 | my $match = $1; |
|
|
95 | |
65 | for my $regex (@simplify_patterns) { |
96 | for my $regex (@simplify_patterns) { |
66 | if ($match =~ $regex) { |
97 | if ($match =~ $regex) { |
67 | $match = $1; |
98 | $match = $1; |
68 | $ofs += $-[1]; |
99 | $ofs += $-[1]; |
|
|
100 | } |
69 | } |
101 | } |
|
|
102 | |
|
|
103 | push @matches, [$ofs, length $match]; |
70 | } |
104 | } |
71 | |
|
|
72 | push @matches, [$ofs, length $match]; |
|
|
73 | } |
105 | } |
74 | } |
106 | } |
75 | } |
107 | } |
|
|
108 | |
|
|
109 | # whole line |
|
|
110 | push @matches, [0, ($line->end - $line->beg + 1) * $self->ncol]; |
76 | |
111 | |
77 | for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
112 | for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
78 | my ($ofs, $len) = @$_; |
113 | my ($ofs, $len) = @$_; |
79 | |
114 | |
80 | next if $len <= $curlen; |
115 | next if $len <= $curlen; |
81 | |
116 | |
|
|
117 | # convert back from UTF-8 offset space to character space |
|
|
118 | { |
|
|
119 | my $length = substr "$text ", $ofs, $len; |
|
|
120 | utf8::decode $length; |
|
|
121 | $len = length $length; |
|
|
122 | } |
|
|
123 | { |
|
|
124 | my $prefix = substr $text, 0, $ofs; |
|
|
125 | utf8::decode $prefix; |
|
|
126 | $ofs = length $prefix; |
|
|
127 | } |
|
|
128 | |
82 | $self->selection_beg ($line->coord_of ($ofs)); |
129 | $self->selection_beg ($line->coord_of ($ofs)); |
83 | $self->selection_end ($line->coord_of ($ofs + $len)); |
130 | $self->selection_end ($line->coord_of ($ofs + $len)); |
84 | return 1; |
131 | return 1; |
85 | } |
132 | } |
86 | |
133 | |
87 | # whole line |
134 | () |
88 | $self->selection_beg ($line->beg, 0); |
|
|
89 | $self->selection_end ($line->end, $self->ncol); |
|
|
90 | |
|
|
91 | return 1; |
|
|
92 | } |
135 | } |