Commit | Line | Data |
920d6222 |
1 | package Web::Simple::DispatchParser; |
2 | |
3 | use strict; |
4 | use warnings FATAL => 'all'; |
5 | |
a5917caa |
6 | sub DEBUG () { 0 } |
7 | |
8 | BEGIN { |
9 | if ($ENV{WEB_SIMPLE_DISPATCHPARSER_DEBUG}) { |
10 | no warnings 'redefine'; |
11 | *DEBUG = sub () { 1 } |
12 | } |
13 | } |
14 | |
15 | sub diag { if (DEBUG) { warn $_[0] } } |
16 | |
920d6222 |
17 | sub new { bless({}, ref($_[0])||$_[0]) } |
18 | |
19 | sub _blam { |
20 | my ($self, $error) = @_; |
a4ec359d |
21 | my $hat = (' ' x (pos||0)).'^'; |
920d6222 |
22 | die "Error parsing dispatch specification: ${error}\n |
23 | ${_} |
24 | ${hat} here\n"; |
25 | } |
26 | |
27 | sub parse_dispatch_specification { |
28 | my ($self, $spec) = @_; |
c6ea9542 |
29 | return $self->_parse_spec($spec); |
30 | } |
31 | |
32 | sub _parse_spec { |
b0420ad6 |
33 | my ($self, $spec, $nested) = @_; |
c6ea9542 |
34 | for ($_[1]) { |
920d6222 |
35 | my @match; |
920d6222 |
36 | /^\G\s*/; # eat leading whitespace |
37 | PARSE: { do { |
c6ea9542 |
38 | push @match, $self->_parse_spec_section($_) |
920d6222 |
39 | or $self->_blam("Unable to work out what the next section is"); |
b0420ad6 |
40 | if (/\G\)/gc) { |
41 | $self->_blam("Found closing ) with no opening (") unless $nested; |
42 | last PARSE; |
43 | } |
920d6222 |
44 | last PARSE if (pos == length); |
c6ea9542 |
45 | $match[-1] = $self->_parse_spec_combinator($_, $match[-1]) |
46 | or $self->_blam('No valid combinator - expected + or |'); |
920d6222 |
47 | } until (pos == length) }; # accept trailing whitespace |
b0420ad6 |
48 | if ($nested and pos == length) { |
a4ec359d |
49 | pos = $nested - 1; |
b0420ad6 |
50 | $self->_blam("No closing ) found for opening ("); |
51 | } |
920d6222 |
52 | return $match[0] if (@match == 1); |
53 | return sub { |
54 | my $env = { %{$_[0]} }; |
55 | my $new_env; |
56 | my @got; |
57 | foreach my $match (@match) { |
58 | if (my @this_got = $match->($env)) { |
59 | my %change_env = %{shift(@this_got)}; |
60 | @{$env}{keys %change_env} = values %change_env; |
61 | @{$new_env}{keys %change_env} = values %change_env; |
62 | push @got, @this_got; |
63 | } else { |
64 | return; |
65 | } |
66 | } |
67 | return ($new_env, @got); |
68 | }; |
69 | } |
70 | } |
71 | |
9b9866ae |
72 | sub _parse_spec_combinator { |
73 | my ($self, $spec, $match) = @_; |
74 | for ($_[1]) { |
75 | |
76 | /\G\+/gc and |
77 | return $match; |
78 | |
79 | /\G\|/gc and |
80 | return do { |
81 | my @match = $match; |
82 | PARSE: { do { |
83 | push @match, $self->_parse_spec_section($_) |
84 | or $self->_blam("Unable to work out what the next section is"); |
85 | last PARSE if (pos == length); |
86 | last PARSE unless /\G\|/gc; # give up when next thing isn't | |
87 | } until (pos == length) }; # accept trailing whitespace |
88 | return sub { |
89 | foreach my $try (@match) { |
90 | if (my @ret = $try->(@_)) { |
91 | return @ret; |
92 | } |
93 | } |
94 | return; |
95 | }; |
96 | }; |
97 | } |
98 | return; |
99 | } |
100 | |
920d6222 |
101 | sub _parse_spec_section { |
102 | my ($self) = @_; |
103 | for ($_[1]) { |
104 | |
105 | # GET POST PUT HEAD ... |
106 | |
107 | /\G([A-Z]+)/gc and |
108 | return $self->_http_method_match($_, $1); |
109 | |
110 | # /... |
111 | |
112 | /\G(?=\/)/gc and |
113 | return $self->_url_path_match($_); |
114 | |
c6ea9542 |
115 | # .* and .html |
116 | |
117 | /\G\.(\*|\w+)/gc and |
920d6222 |
118 | return $self->_url_extension_match($_, $1); |
b0420ad6 |
119 | |
2ee4ab06 |
120 | # (...) |
b0420ad6 |
121 | |
122 | /\G\(/gc and |
123 | return $self->_parse_spec($_, pos); |
2ee4ab06 |
124 | |
125 | # !something |
126 | |
127 | /\G!/gc and |
128 | return do { |
129 | my $match = $self->_parse_spec_section($_); |
130 | return sub { |
131 | return {} unless $match->(@_); |
132 | return; |
133 | }; |
134 | }; |
920d6222 |
135 | |
9b9866ae |
136 | # ?<param spec> |
137 | /\G\?/gc and |
138 | return $self->_parse_param_handler($_, 'query'); |
c6ea9542 |
139 | } |
9b9866ae |
140 | return; # () will trigger the blam in our caller |
c6ea9542 |
141 | } |
142 | |
920d6222 |
143 | sub _http_method_match { |
144 | my ($self, $str, $method) = @_; |
920d6222 |
145 | sub { shift->{REQUEST_METHOD} eq $method ? {} : () }; |
146 | } |
147 | |
148 | sub _url_path_match { |
149 | my ($self) = @_; |
920d6222 |
150 | for ($_[1]) { |
151 | my @path; |
da8429c9 |
152 | my $full_path = '$'; |
153 | PATH: while (/\G\//gc) { |
154 | /\G\.\.\./gc |
155 | and do { |
156 | $full_path = ''; |
157 | last PATH; |
158 | }; |
920d6222 |
159 | push @path, $self->_url_path_segment_match($_) |
160 | or $self->_blam("Couldn't parse path match segment"); |
161 | } |
da8429c9 |
162 | my $re = '^()'.join('/','',@path).($full_path ? '$' : '(/.*)$'); |
163 | $re = qr/$re/; |
164 | if ($full_path) { |
165 | return sub { |
166 | if (my @cap = (shift->{PATH_INFO} =~ /$re/)) { |
167 | $cap[0] = {}; return @cap; |
168 | } |
169 | return (); |
170 | }; |
171 | } |
920d6222 |
172 | return sub { |
173 | if (my @cap = (shift->{PATH_INFO} =~ /$re/)) { |
da8429c9 |
174 | $cap[0] = { PATH_INFO => pop(@cap) }; return @cap; |
920d6222 |
175 | } |
176 | return (); |
177 | }; |
178 | } |
179 | return; |
180 | } |
181 | |
182 | sub _url_path_segment_match { |
183 | my ($self) = @_; |
184 | for ($_[1]) { |
185 | # trailing / -> require / on end of URL |
186 | /\G(?:(?=\s)|$)/gc and |
187 | return '$'; |
188 | # word chars only -> exact path part match |
189 | /\G(\w+)/gc and |
190 | return "\Q$1"; |
28f3dfd5 |
191 | # ** -> capture unlimited path parts |
192 | /\G\*\*/gc and |
193 | return '(.*?[^/])'; |
920d6222 |
194 | # * -> capture path part |
195 | /\G\*/gc and |
196 | return '([^/]+)'; |
197 | } |
198 | return (); |
199 | } |
200 | |
201 | sub _url_extension_match { |
202 | my ($self, $str, $extension) = @_; |
c6ea9542 |
203 | if ($extension eq '*') { |
204 | sub { |
205 | if ((my $tmp = shift->{PATH_INFO}) =~ s/\.(\w+)$//) { |
206 | ({ PATH_INFO => $tmp }, $1); |
207 | } else { |
208 | (); |
209 | } |
210 | }; |
211 | } else { |
212 | sub { |
213 | if ((my $tmp = shift->{PATH_INFO}) =~ s/\.\Q${extension}\E$//) { |
214 | ({ PATH_INFO => $tmp }); |
215 | } else { |
216 | (); |
217 | } |
218 | }; |
219 | } |
920d6222 |
220 | } |
221 | |
9b9866ae |
222 | sub _parse_param_handler { |
223 | my ($self, $spec, $type) = @_; |
224 | |
225 | require Web::Simple::ParamParser; |
226 | my $unpacker = Web::Simple::ParamParser->can("get_unpacked_${type}_from"); |
227 | |
228 | for ($_[1]) { |
a5917caa |
229 | my (@required, @single, %multi, $star, $multistar); |
9b9866ae |
230 | PARAM: { do { |
231 | |
232 | # per param flag |
233 | |
234 | my $multi = 0; |
235 | |
236 | # ?@foo or ?@* |
237 | |
238 | /\G\@/gc and $multi = 1; |
239 | |
240 | # @* or * |
241 | |
a5917caa |
242 | if (/\G\*/gc) { |
9b9866ae |
243 | |
244 | $multi ? ($multistar = 1) : ($star = 1); |
a5917caa |
245 | |
246 | if ($star && $multistar) { |
247 | $self->_blam("Can't use * and \@* in the same parameter match"); |
248 | } |
9b9866ae |
249 | } else { |
250 | |
251 | # @foo= or foo= or @foo~ or foo~ |
252 | |
253 | /\G(\w+)/gc or $self->_blam('Expected parameter name'); |
254 | |
255 | my $name = $1; |
256 | |
257 | # check for = or ~ on the end |
258 | |
259 | /\G\=/gc |
260 | ? push(@required, $name) |
261 | : (/\G\~/gc or $self->_blam('Expected = or ~ after parameter name')); |
262 | |
263 | # record the key in the right category depending on the multi (@) flag |
264 | |
a5917caa |
265 | $multi ? ($multi{$name} = 1) : (push @single, $name); |
9b9866ae |
266 | } |
267 | } while (/\G\&/gc) } |
268 | |
269 | return sub { |
270 | my $raw = $unpacker->($_[0]); |
271 | foreach my $name (@required) { |
272 | return unless exists $raw->{$name}; |
273 | } |
274 | my %p; |
275 | foreach my $name ( |
276 | @single, |
277 | ($star |
278 | ? (grep { !exists $multi{$_} } keys %$raw) |
279 | : () |
280 | ) |
281 | ) { |
282 | $p{$name} = $raw->{$name}->[-1] if exists $raw->{$name}; |
283 | } |
284 | foreach my $name ( |
285 | keys %multi, |
286 | ($multistar |
287 | ? (grep { !exists $p{$_} } keys %$raw) |
288 | : () |
289 | ) |
290 | ) { |
291 | $p{$name} = $raw->{$name}||[]; |
292 | } |
293 | return ({}, \%p); |
294 | }; |
295 | } |
296 | } |
297 | |
920d6222 |
298 | 1; |