1 |
#! perl |
2 |
|
3 |
sub on_keyboard_command { |
4 |
my ($self, $cmd) = @_; |
5 |
|
6 |
$cmd eq "selection:rot13" |
7 |
and $self->selection (map { y/A-Za-z/N-ZA-Mn-za-m/; $_ } $self->selection); |
8 |
|
9 |
() |
10 |
} |
11 |
|
12 |
sub on_init { |
13 |
my ($self) = @_; |
14 |
|
15 |
for (my $idx = 0; defined (my $res = $self->x_resource ("selection.pattern-$idx")); $idx++) { |
16 |
no re 'eval'; # just to be sure |
17 |
$res = utf8::encode $self->locale_decode ($res); |
18 |
push @{ $self->{patterns} }, qr/$res/; |
19 |
} |
20 |
|
21 |
() |
22 |
} |
23 |
|
24 |
# "find interetsing things"-patterns |
25 |
my @mark_patterns = ( |
26 |
# common types of "parentheses" |
27 |
qr{ (?<![^[:space:]]) ‘ ([^‘’]+) ’ (?![^[:space]]) }x, |
28 |
qr{ (?<![^[:space:]]) ` ([^`']+) ' (?![^[:space]]) }x, |
29 |
qr{ (?<![^[:space:]]) (" [^[:space:]] [^"]* ") }x, |
30 |
qr{ (" [^"]* [^[:space:]] ") (?![^[:space]]) }x, |
31 |
qr{ \< ([^<>[:space:]]+) \> }x, |
32 |
qr{ \{ ([^{}[:space:]]+) \} }x, |
33 |
qr{ \[ ([^{}[:space:]]+) \] }x, |
34 |
qr{ \( ([^()[:space:]]+) \) }x, |
35 |
|
36 |
# urls, just a heuristic |
37 |
qr{( |
38 |
(?:https?|ftp|news|mailto|file)://[ab-zA-Z0-9\-\@;\/?:&=%\$_.+!*\x27(),~]+ |
39 |
[ab-zA-Z0-9\-\@;\/?:&=%\$_+!*\x27()~] # exclude some trailing characters (heuristic) |
40 |
)}x, |
41 |
|
42 |
# shell-like argument quoting, basically always matches |
43 |
qr{\G [\ \t|&;<>()] *( |
44 |
(?: |
45 |
[^\\"'\ \t|&;<>()]+ |
46 |
| \\. |
47 |
| " (?: [^\\"]+ | \\. )* " |
48 |
| ' [^']* ' |
49 |
)+ |
50 |
)}x, |
51 |
); |
52 |
|
53 |
# "correct obvious? crap"-patterns |
54 |
my @simplify_patterns = ( |
55 |
qr{^"([^\\"'\ \t|&;<>()*?]+)"$}, # "simple" => simple |
56 |
qr{^(.*)[,\-]$}, # strip off trailing , and - |
57 |
); |
58 |
|
59 |
sub on_sel_extend { |
60 |
my ($self, $time) = @_; |
61 |
|
62 |
my ($row, $col) = $self->selection_mark; |
63 |
my $line = $self->line ($row); |
64 |
my $text = $line->t; |
65 |
my $markofs = $line->offset_of ($row, $col); |
66 |
my $curlen = $line->offset_of ($self->selection_end) |
67 |
- $line->offset_of ($self->selection_beg); |
68 |
|
69 |
my @matches; |
70 |
|
71 |
# not doing matches in unicode mode helps speed |
72 |
# enourmously here. working in utf-8 should be |
73 |
# equivalent due to the magic of utf-8 encoding. |
74 |
utf8::encode $text; |
75 |
study $text; # _really_ helps, too :) |
76 |
|
77 |
for my $regex (@mark_patterns, @{ $self->{patterns} }) { |
78 |
while ($text =~ /$regex/g) { |
79 |
if ($-[1] <= $markofs and $markofs <= $+[1]) { |
80 |
my $ofs = $-[1]; |
81 |
my $match = $1; |
82 |
|
83 |
for my $regex (@simplify_patterns) { |
84 |
if ($match =~ $regex) { |
85 |
$match = $1; |
86 |
$ofs += $-[1]; |
87 |
} |
88 |
} |
89 |
|
90 |
push @matches, [$ofs, length $match]; |
91 |
} |
92 |
} |
93 |
} |
94 |
|
95 |
# whole line |
96 |
push @matches, [0, ($line->end - $line->beg + 1) * $self->ncol]; |
97 |
|
98 |
for (sort { $a->[1] <=> $b->[1] or $b->[0] <=> $a->[0] } @matches) { |
99 |
my ($ofs, $len) = @$_; |
100 |
|
101 |
next if $len <= $curlen; |
102 |
|
103 |
$self->selection_beg ($line->coord_of ($ofs)); |
104 |
$self->selection_end ($line->coord_of ($ofs + $len)); |
105 |
return 1; |
106 |
} |
107 |
|
108 |
() |
109 |
} |