| 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/; $_ } $self->selection); |
| 8 | |
8 | |
| 9 | () |
9 | () |
| 10 | } |
10 | } |
| 11 | |
11 | |
|
|
12 | sub on_init { |
|
|
13 | my ($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 | $self->{enabled} = 1; |
|
|
27 | |
|
|
28 | push @{ $self->{term}{option_popup_hook} }, sub { |
|
|
29 | ("new selection" => $self->{enabled}, sub { $self->{enabled} = shift }) |
|
|
30 | }; |
|
|
31 | |
|
|
32 | () |
|
|
33 | } |
|
|
34 | |
|
|
35 | # "find interesting things"-patterns |
| 12 | my @patterns = ( |
36 | my @mark_patterns = ( |
|
|
37 | # common types of "parentheses" |
|
|
38 | qr{ (?<![^[:space:]]) ‘ ([^‘’]+) ’ (?![^[:space]]) }x, |
|
|
39 | qr{ (?<![^[:space:]]) ` ([^`']+) ' (?![^[:space]]) }x, |
|
|
40 | qr{ (?<![^[:space:]]) (" [^[:space:]] [^"]* ") }x, |
|
|
41 | qr{ (" [^"]* [^[:space:]] ") (?![^[:space]]) }x, |
|
|
42 | qr{ \< ([^<>[:space:]]+) \> }x, |
|
|
43 | qr{ \{ ([^{}[:space:]]+) \} }x, |
|
|
44 | qr{ \[ ([^{}[:space:]]+) \] }x, |
|
|
45 | qr{ \( ([^()[:space:]]+) \) }x, |
|
|
46 | |
| 13 | # urls, just a heuristic |
47 | # urls, just a heuristic |
| 14 | qr{( |
48 | qr{( |
| 15 | (?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
49 | (?:https?://|ftp://|news://|mailto:|file://)[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
| 16 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27(),~] # do not include a trailing dot, its wrong too often |
50 | [ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~] # exclude some trailing characters (heuristic) |
| 17 | )}x, |
51 | )}x, |
| 18 | |
52 | |
| 19 | # common forms of quoting |
53 | # shell-like argument quoting, basically always matches |
| 20 | qr{(?:^|\s) [‘`] ([^‘`’']+) [’'] (?:\s|$)}x, |
|
|
| 21 | |
|
|
| 22 | # shell-like argument quoting |
|
|
| 23 | qr{\G [\ \t|&;<>()] *( |
54 | qr{\G [\ \t|&;<>()]* ( |
| 24 | (?: |
55 | (?: |
| 25 | [^\\"'\ \t|&;<>()]+ |
56 | [^\\"'\ \t|&;<>()]+ |
| 26 | | \\. |
57 | | \\. |
| 27 | | " (?: [^\\"]+ | \\. )* " |
58 | | " (?: [^\\"]+ | \\. )* " |
| 28 | | ' [^']* ' |
59 | | ' [^']* ' |
| 29 | )+ |
60 | )+ |
| 30 | )}x, |
61 | )}x, |
| 31 | ); |
62 | ); |
| 32 | |
63 | |
| 33 | my $x = "‘hallole’"; |
64 | # "correct obvious? crap"-patterns |
|
|
65 | my @simplify_patterns = ( |
|
|
66 | qr{^"([^\\"'\ \t|&;<>()*?]+)"$}, # "simple" => simple |
|
|
67 | qr{^(.*)[,\-]$}, # strip off trailing , and - |
|
|
68 | ); |
| 34 | |
69 | |
| 35 | sub on_sel_extend { |
70 | sub on_sel_extend { |
| 36 | my ($self) = @_; |
71 | my ($self, $time) = @_; |
|
|
72 | |
|
|
73 | $self->{enabled} |
|
|
74 | or return; |
| 37 | |
75 | |
| 38 | my ($row, $col) = $self->selection_mark; |
76 | my ($row, $col) = $self->selection_mark; |
| 39 | my $line = $self->line ($row); |
77 | my $line = $self->line ($row); |
| 40 | my $offset = $line->offset_of ($row, $col); |
|
|
| 41 | my $text = $line->t; |
78 | my $text = $line->t; |
|
|
79 | my $markofs = $line->offset_of ($row, $col); |
|
|
80 | my $curlen = $line->offset_of ($self->selection_end) |
|
|
81 | - $line->offset_of ($self->selection_beg); |
| 42 | |
82 | |
| 43 | for my $regex (@patterns) { |
83 | my @matches; |
|
|
84 | |
|
|
85 | if ($markofs < $line->l) { |
|
|
86 | # convert markofs form character to UTF-8 offset space |
|
|
87 | { |
|
|
88 | my $prefix = substr $text, 0, $markofs; |
|
|
89 | utf8::encode $prefix; |
|
|
90 | $markofs = length $prefix; |
|
|
91 | } |
|
|
92 | |
|
|
93 | # not doing matches in unicode mode helps speed |
|
|
94 | # enourmously here. working in utf-8 should be |
|
|
95 | # equivalent due to the magic of utf-8 encoding. |
|
|
96 | utf8::encode $text; |
|
|
97 | study $text; # _really_ helps, too :) |
|
|
98 | |
|
|
99 | for my $regex (@mark_patterns, @{ $self->{patterns} }) { |
| 44 | while ($text =~ /$regex/g) { |
100 | while ($text =~ /$regex/g) { |
| 45 | if ($-[1] <= $offset and $offset <= $+[1]) { |
101 | if ($-[1] <= $markofs and $markofs <= $+[1]) { |
| 46 | $self->selection_beg ($line->coord_of ($-[1])); |
102 | my $ofs = $-[1]; |
| 47 | $self->selection_end ($line->coord_of ($+[1])); |
103 | my $match = $1; |
| 48 | return 1; |
104 | |
|
|
105 | for my $regex (@simplify_patterns) { |
|
|
106 | if ($match =~ $regex) { |
|
|
107 | $match = $1; |
|
|
108 | $ofs += $-[1]; |
|
|
109 | } |
|
|
110 | } |
|
|
111 | |
|
|
112 | push @matches, [$ofs, length $match]; |
|
|
113 | } |
| 49 | } |
114 | } |
| 50 | } |
115 | } |
| 51 | } |
116 | } |
| 52 | |
117 | |
|
|
118 | # whole line |
|
|
119 | push @matches, [0, ($line->end - $line->beg + 1) * $self->ncol]; |
|
|
120 | |
|
|
121 | for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
|
|
122 | my ($ofs, $len) = @$_; |
|
|
123 | |
|
|
124 | next if $len <= $curlen; |
|
|
125 | |
|
|
126 | # convert back from UTF-8 offset space to character space |
|
|
127 | { |
|
|
128 | my $length = substr "$text ", $ofs, $len; |
|
|
129 | utf8::decode $length; |
|
|
130 | $len = length $length; |
|
|
131 | } |
|
|
132 | { |
|
|
133 | my $prefix = substr $text, 0, $ofs; |
|
|
134 | utf8::decode $prefix; |
|
|
135 | $ofs = length $prefix; |
|
|
136 | } |
|
|
137 | |
|
|
138 | $self->selection_beg ($line->coord_of ($ofs)); |
|
|
139 | $self->selection_end ($line->coord_of ($ofs + $len)); |
|
|
140 | return 1; |
|
|
141 | } |
|
|
142 | |
| 53 | () |
143 | () |
| 54 | } |
144 | } |
