有一段结构化的文本, 写一个 Grammar 来解析它：
name = Animal Facts
author = jnthn
[cat]
desc = The smartest and cutest
cuteness = 100000
[dugong]
desc = The cow of the sea
cuteness = -10
[magpie]
desc = crow; raven; rook; jackdaw; chough; magpie; jay
cuteness = 99
每一段都是一个章节, 有的章节没有[cat]这样的标题, 要求 grammar 生成一个散列, 散列的键是方括号中的单词, 如果没有就默认为 _ , 散列的值是一个散列的数组, 数组里面的每个散列的键为等号左边的单词, 键值为等号右边的字符。Grammar 如下:
use v6;
#use Grammar::Debugger;
grammar INIFile::Grammar {
token TOP {
^
<entries> # 条目
<section>+ # 章节
$
}
token section {
'[' ~ ']' <key> \n
<entries> # 每个章节含有多个条目 entry
}
token entries {
[
| <entry> \n
| \n # entry 可以为空
]+
}
token entry { <key> \h* '=' \h* <value> }
token key { \w+ }
token value { \N+ }
}
class INIFileActions {
method entries($/) {
my %entries;
for $<entry> -> $e {
%entries{$e<key>} := ~$e<value>;
}
make %entries;
}
method TOP($/) {
my %result;
%result<_> := $<entries>.ast;
for $<section> -> $sec {
%result{$sec<key>} := $sec<entries>.ast;
}
make %result;
}
}
my $m := INIFile::Grammar.parse(Q{
name = Animal Facts
author = jnthn
[cat]
desc = The smartest and cutest
cuteness = 100000
[dugong]
desc = The cow of the sea
cuteness = -10
[magpie]
desc = crow; raven; rook; jackdaw; chough; magpie; jay
cuteness = 99
}, :actions(INIFileActions));
my %sections := $m.ast;
for %sections -> $sec {
say("章节: {$sec.key}");
for $sec.value -> $entry {
say(" {$entry.key}: {$entry.value}");
}
}
make 是一个函数, 接收单个参数, make 的作用是, 对于每一个 method 中对应的 $_ , 存储生成的抽象语法树(AST)(片段)到 $/ 中。 .ast 用于从已保存的 AST 抽象语法树中检索提取 AST (片段), » 相当于一个循环, 即检索每一个 $<entry> 之类的语法树。
use v6;
grammar INIFile::Grammar {
token TOP {
^
<section>+ # 章节
$
}
token section {
[ '[' ~ ']' <key> \n ]? # [key] 这一行是可选的
<entries> # 每个章节含有多个条目 entry
}
token entries {
[
| <entry> \n
| \n # entry 可以为空
]+
}
token entry { <key> \h* '=' \h* <value> }
token key { \w+ }
token value { \N+ }
}
class INIFileActions {
method key ($/) { $/.make: ~$/ }
method value ($/) { $/.make: ~$/ }
method entry ($/) { $/.make: $<key>.ast => $<value>.ast }
method entries($/) { $/.make: $<entry>».ast }
method section($/) { $/.make: $<key>.ast // '_' => $<entries>.ast } # 如果 key 不存在就默认为 `_`
method TOP($/) {
$/.make: $<section>».ast; # 等价于 $/.make($<section>».ast);
# '-' => $<entries>.ast # '_' 没有 ast 方法
}
}
my $m = INIFile::Grammar.parse(Q{
name = Animal Facts
author = jnthn
[cat]
desc = The smartest and cutest
cuteness = 100000
[dugong]
desc = The cow of the sea
cuteness = -10
[magpie]
desc = crow; raven; rook; jackdaw; chough; magpie; jay
cuteness = 99
}, :actions(INIFileActions)).ast;
for @$m -> $sec {
say("章节: {$sec.key}");
for $sec.value -> $entry {
say(" {$entry.key}: {$entry.value}");
}
}
结果输出为:
章节: dugong
desc: The cow of the sea
cuteness: -10
章节: cat
desc: The smartest and cutest
cuteness: 100000
章节: _
name: Animal Facts
author: jnthn
章节: magpie
desc: crow; raven; rook; jackdaw; chough; magpie; jay
cuteness: 99
Grammar 的解析是从上至下的, 从 top-level (TOP) 开始, 到分支(branch)。 Actions 中的方法是随着解析而执行的, 但是抽象语法树(AST) 的存储和检索是自下而上的, 只有底部的存储完($/.make:)了, 其上层部分才可以使用 .ast 或 .made 方法进行检索, 检索到之后各自进行处理后再次存储, 以供它的上层部分使用, 以此类推。
注意, 第一段代码中 $m 存储的是散列, 而第二段代码中, $m 存储的是数组! 这说明可以返回散列和数组两种形式。留待改天。
