| 1 | #! perl |
1 | #! perl |
| 2 | |
2 | |
| 3 | sub on_keyboard_command { |
3 | sub on_user_command { |
| 4 | my ($self, $cmd) = @_; |
4 | my ($self, $cmd) = @_; |
| 5 | |
5 | |
| 6 | $cmd eq "selection:rot13" |
6 | $cmd eq "selection:rot13" |
| 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/; $_ } $self->selection); |
| 8 | |
8 | |
| … | |
… | |
| 10 | } |
10 | } |
| 11 | |
11 | |
| 12 | sub on_init { |
12 | sub on_init { |
| 13 | my ($self) = @_; |
13 | my ($self) = @_; |
| 14 | |
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 | |
| 15 | for (my $idx = 0; defined (my $res = $self->x_resource ("selection.pattern-$idx")); $idx++) { |
20 | for (my $idx = 0; defined (my $res = $self->x_resource ("selection.pattern-$idx")); $idx++) { |
| 16 | no re 'eval'; # just to be sure |
21 | $res = $self->locale_decode ($res); |
|
|
22 | utf8::encode $res; |
| 17 | push @{ $self->{patterns} }, qr/$res/; |
23 | push @{ $self->{patterns} }, qr/$res/; |
| 18 | } |
24 | } |
|
|
25 | |
|
|
26 | $self->{enabled} = 1; |
|
|
27 | |
|
|
28 | push @{ $self->{term}{option_popup_hook} }, sub { |
|
|
29 | ("new selection" => $self->{enabled}, sub { $self->{enabled} = shift }) |
|
|
30 | }; |
| 19 | |
31 | |
| 20 | () |
32 | () |
| 21 | } |
33 | } |
| 22 | |
34 | |
| 23 | # "find interetsing things"-patterns |
35 | # "find interesting things"-patterns |
| 24 | my @mark_patterns = ( |
36 | my @mark_patterns = ( |
| 25 | qr{([[:word:]]+)}, |
37 | # qr{ ([[:word:]]+) }x, |
|
|
38 | qr{ ([^[:space:]]+) }x, |
| 26 | |
39 | |
| 27 | # common "parentheses" |
40 | # common types of "parentheses" |
| 28 | qr{(?:^|\s) ‘ ([^‘’]+?) ’ (?:\s|\)|$)}x, |
41 | qr{ (?<![^[:space:]]) [`'] ([^`']+) [`'] (?![^[:space:]]) }x, |
| 29 | qr{(?:^|\s) ` ([^`']+?) ' (?:\s|\)|$)}x, |
42 | qr{ (?<![^[:space:]]) ‘ ([^‘’]+) ’ (?![^[:space:]]) }x, |
|
|
43 | qr{ (?<![^[:space:]]) “ ([^“”]+) ” (?![^[:space:]]) }x, |
|
|
44 | |
|
|
45 | qr{ (?<![^[:space:]]) (' [^[:space:]] [^']* ') }x, |
|
|
46 | qr{ (' [^']* [^[:space:]] ') (?![^[:space:]]) }x, |
|
|
47 | qr{ (?<![^[:space:]]) (` [^[:space:]] [^']* ') }x, |
|
|
48 | qr{ (` [^']* [^[:space:]] ') (?![^[:space:]]) }x, |
|
|
49 | qr{ (?<![^[:space:]]) (" [^[:space:]] [^"]* ") }x, |
|
|
50 | qr{ (" [^"]* [^[:space:]] ") (?![^[:space:]]) }x, |
|
|
51 | |
| 30 | qr{ \{ ([^{}]+?) \} }x, |
52 | qr{ \{ ([^\{\}]+) \} }x, |
| 31 | qr{ \[ ([^{}]+?) \] }x, |
|
|
| 32 | qr{ \( ([^()]+?) \) }x, |
53 | qr{ \( ([^\(\)]+) \) }x, |
|
|
54 | qr{ \[ ([^\[\]]+) \] }x, |
|
|
55 | qr{ \< ([^\<\>]+) \> }x, |
| 33 | |
56 | |
| 34 | # urls, just a heuristic |
57 | # urls, just a heuristic |
| 35 | qr{( |
58 | qr{( |
| 36 | (?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
59 | (?:https?://|ftp://|news://|mailto:|file://|\bwww\.)[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~#]+ |
| 37 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~] # exclude some trailing characters (heuristic) |
60 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+*()~] # exclude some trailing characters (heuristic) |
| 38 | )}x, |
61 | )}x, |
| 39 | |
62 | |
| 40 | # shell-like argument quoting, basically always matches |
63 | # shell-like argument quoting, basically always matches |
| 41 | qr{\G [\ \t|&;<>()] *( |
64 | qr{\G [\ \t|&;<>()]* ( |
| 42 | (?: |
65 | (?: |
| 43 | [^\\"'\ \t|&;<>()]+ |
66 | [^\\"'\ \t|&;<>()]+ |
| 44 | | \\. |
67 | | \\. |
| 45 | | " (?: [^\\"]+ | \\. )* " |
68 | | " (?: [^\\"]+ | \\. )* " |
| 46 | | ' [^']* ' |
69 | | ' [^']* ' |
| … | |
… | |
| 55 | ); |
78 | ); |
| 56 | |
79 | |
| 57 | sub on_sel_extend { |
80 | sub on_sel_extend { |
| 58 | my ($self, $time) = @_; |
81 | my ($self, $time) = @_; |
| 59 | |
82 | |
|
|
83 | $self->{enabled} |
|
|
84 | or return; |
|
|
85 | |
| 60 | my ($row, $col) = $self->selection_mark; |
86 | my ($row, $col) = $self->selection_mark; |
| 61 | my $line = $self->line ($row); |
87 | my $line = $self->line ($row); |
| 62 | my $text = $line->t; |
88 | my $text = $line->t; |
| 63 | my $markofs = $line->offset_of ($row, $col); |
89 | my $markofs = $line->offset_of ($row, $col); |
| 64 | my $curlen = $line->offset_of ($self->selection_end) |
90 | my $curlen = $line->offset_of ($self->selection_end) |
| 65 | - $line->offset_of ($self->selection_beg); |
91 | - $line->offset_of ($self->selection_beg); |
| 66 | |
92 | |
| 67 | my @matches; |
93 | my @matches; |
| 68 | |
94 | |
| 69 | for my $regex (@mark_patterns, @{ $self->{patterns} }) { |
95 | if ($markofs < $line->l) { |
| 70 | while ($text =~ /$regex/g) { |
96 | # convert markofs from character to UTF-8 offset space |
| 71 | if ($-[1] <= $markofs and $markofs <= $+[1]) { |
97 | { |
| 72 | my $ofs = $-[1]; |
98 | my $prefix = substr $text, 0, $markofs; |
| 73 | my $match = $1; |
99 | utf8::encode $prefix; |
|
|
100 | $markofs = length $prefix; |
|
|
101 | } |
| 74 | |
102 | |
| 75 | push @matches, [$ofs, length $match]; |
103 | # not doing matches in unicode mode helps speed |
|
|
104 | # enourmously here. working in utf-8 should be |
|
|
105 | # equivalent due to the magic of utf-8 encoding. |
|
|
106 | utf8::encode $text; |
|
|
107 | study $text; # _really_ helps, too :) |
| 76 | |
108 | |
|
|
109 | for my $regex (@mark_patterns, @{ $self->{patterns} }) { |
|
|
110 | while ($text =~ /$regex/g) { |
|
|
111 | if ($-[1] <= $markofs and $markofs <= $+[1]) { |
|
|
112 | my $ofs = $-[1]; |
|
|
113 | my $match = $1; |
|
|
114 | |
| 77 | for my $regex (@simplify_patterns) { |
115 | for my $regex (@simplify_patterns) { |
| 78 | if ($match =~ $regex) { |
116 | if ($match =~ $regex) { |
| 79 | $match = $1; |
117 | $match = $1; |
| 80 | $ofs += $-[1]; |
118 | $ofs += $-[1]; |
|
|
119 | } |
| 81 | } |
120 | } |
|
|
121 | |
|
|
122 | push @matches, [$ofs, length $match]; |
| 82 | } |
123 | } |
| 83 | |
|
|
| 84 | push @matches, [$ofs, length $match]; |
|
|
| 85 | } |
124 | } |
| 86 | } |
125 | } |
| 87 | } |
126 | } |
| 88 | |
127 | |
| 89 | # whole line |
128 | # whole line |
| … | |
… | |
| 92 | for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
131 | for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
| 93 | my ($ofs, $len) = @$_; |
132 | my ($ofs, $len) = @$_; |
| 94 | |
133 | |
| 95 | next if $len <= $curlen; |
134 | next if $len <= $curlen; |
| 96 | |
135 | |
|
|
136 | # convert back from UTF-8 offset space to character space |
|
|
137 | { |
|
|
138 | my $length = substr "$text ", $ofs, $len; |
|
|
139 | utf8::decode $length; |
|
|
140 | $len = length $length; |
|
|
141 | } |
|
|
142 | { |
|
|
143 | my $prefix = substr $text, 0, $ofs; |
|
|
144 | utf8::decode $prefix; |
|
|
145 | $ofs = length $prefix; |
|
|
146 | } |
|
|
147 | |
| 97 | $self->selection_beg ($line->coord_of ($ofs)); |
148 | $self->selection_beg ($line->coord_of ($ofs)); |
| 98 | $self->selection_end ($line->coord_of ($ofs + $len)); |
149 | $self->selection_end ($line->coord_of ($ofs + $len)); |
| 99 | return 1; |
150 | return 1; |
| 100 | } |
151 | } |
| 101 | |
152 | |
