Jq语言
jq是领域特定高级的纯函数式编程语言,它采用词法作用域,其中所有JSON值都是常量。jq支持回溯并可管理JSON数据的无限长字串流。jq支持基于名字空间的模块系统,并对闭包有一定支持,尤其是它的函数和泛函表达式可以用作其他函数的参数。 jq与Icon和Haskell编程语言有关。它最初采用Haskell实现[3],随即移植至C语言。gojq是“纯”Go实现,fq是针对二进制文件的jq[4]。jq还有用Rust实现的叫做jaq的方言[5],它规定了指称语义[6]。 历史jq由Stephen Dolan创建并在2012年10月发行[7][8] 。它被设计为“针对JSON数据的sed类似者”[9]。在jq版本1.5中增加支持了正则表达式。 针对jq的叫做yq的“包装器”[10],增加支持了YAML、XML和TOML。它首次发行于2017年[11]。 用Go实现的gojq最初发行于2019年[12],gojq显著的扩展jq包括了支持YAML。 用Rust实现的jaq,其项目目标是更快速和更准确的jq实现,仍保持与jq在大多数情况下的兼容性。在2024年3月于其目标中排除了jq的特定高级特征,比如模块、SQL风格算子和给非常大JSON文档的串流解析器[5]。 用jq实现的jqjq,最初发行于2022年。jqjq显著的可以运行自身,拥有REPL并支持 用法命令行用法jq典型的用于命令行,并可以协作于其他命令行实用工具,比如curl。下面的例子展示如何将 $ URL=$(echo 'jq语言' | jq -rR '@uri "https://zh.wikipedia.org/w/api.php?action=parse&page=\(.)&format=json"')
$ curl -s ${URL} | jq '.parse.categories[]."*"'
这里的流水线产生的输出,由JSON字符串的串流组成,它们是: "小寫標題"
"CS1英语来源_(en)"
"动态类型编程语言"
"函数式编程语言"
"面向文本编程语言"
"2012年建立的程式語言"
"数据查询语言"
"2012年软件"
上述 这里展示的jq过滤器的方法链是如下流水线(pipeline)的简写: .["parse"] | .["categories"] | .[] | .["*"]
这对应于 嵌入式用法C语言和Go实现二者都提供函数库,使得jq功能可以嵌入到其他应用和编程环境之中。 例如,gojq已经集成于SQLite,故而 运算的模态jq缺省的充当针对JSON输入的“串流编辑器”,非常像被当作多行文本的“串流编辑器”的sed实用工具。但是jq有一些其他运算模态:
“串流解析器”(streaming parser),在一个或多个JSON输入太大无法载入内存之时特别有用,因为它需求的内存典型的相当小。例如,对于任意大的JSON对象的阵列,峰值内存需求不比处理最大顶层对象所需要的多出很多。 这些运算模态可以在特定限制下组合起来。 语法和语义类型所有JSON值自身是jq中的值,它们从而有在下列表格中展示的类型[17]。gojq和jaq实现将数区分为整数和浮点数。gojq实现支持无界精度整数算术,同于jq采用Haskell的最初实现。
形式jq有特殊语法形式,比如:
jq中有两种类型的符号,可称为“变量”的值绑定和函数。二者都是词法作用域的,表达式只能提及其左侧即前面最近的定义的符号,但是函数可以提及自身来创建递归函数。 过滤器jq是面向JSON的编程语言,使用 $ echo '[1,2]' | jq 'add'
3
$ jq -n '[1,2] | add'
3
这里jq内的JSON阵列 尽管类似于Unix流水线,jq流水线允许将到来数据,如同并行的发送到在 $ jq -n '[1,2] | add/length'
1.5
$ jq -nc '[1,2] | [length, add, add/length]'
[2,3,1.5]
单独的点号 $ jq -nc '1 | [., .]'
[1,1]
$ jq -n '2 | pow(.; .)'
4
在jq中采用隐式编程风格,阶乘可以通过共递归运算 $ echo '0 1 2 6' | jq -c '[limit(.+1; [0,1] | recurse([first+1, last*(first+1)]) | last)]'
[1]
[1,1]
[1,1,2]
[1,1,2,6,24,120,720]
$ jq -n 'first([0,1] | recurse([first+1, last*(first+1)]) | last)'
1
$ jq -n '[0,1,2,6][] | nth(.; [0,1] | recurse([first+1, last*(first+1)]) | last)'
1
1
2
720
$ jq -nc '[0,1,2,6] | map(nth(.; [0,1] | recurse([first+1, last*(first+1)]) | last))'
[1,1,2,720]
斐波那契数列可以通过共递归运算 $ echo '0 1 2 6' | jq -c '[limit(.+1; [0,1] | recurse([last, add]) | first)]'
[0]
[0,1]
[0,1,1]
[0,1,1,2,3,5,8]
$ jq -n 'first([0,1] | recurse([last, add]) | first)'
0
$ jq -n '[0,1,2,6][] | nth(.; [0,1] | recurse([last, add]) | first)'
0
1
1
8
$ jq -nc '[0,1,2,6] | map(nth(.; [0,1] | recurse([last, add]) | first))'
[0,1,1,8]
这种无限列表不可以不加限定的使用,也不可以对其进行 def fac: nth(.; [0,1] | recurse([first+1, last*(first+1)]) | last);
def fib: nth(.; [0,1] | recurse([last, add]) | first);
共递归运算 $ jq -nc '[0 | recurse(.+1; . < 0)]'
[0]
$ jq -nc '[0 | recurse(select(. < 0) | .+1)]'
[0]
$ jq -nc '[0 | recurse(.+1; . <= 6)]'
[0,1,2,3,4,5,6]
$ jq -nc '[0 | recurse(select(. < 6) | .+1)]'
[0,1,2,3,4,5,6]
下面的例子展示如何定义参数化的命名过滤器,它格式化从2到36含二者的任何底数的整数: def tobase($b):
def digit: "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"[.:.+1];
def mod: .%$b;
def div: (.-mod)/$b;
def place_values: recurse(select(. >= $b) | div) | mod;
select(2 <= $b and $b <= 36)
| [place_values | digit] | reverse | add;
这里的 $ echo '15 16' | jq 'include "./tobase"; tobase(16)'
"F"
"10"
从 def send_more_money:
def choose(m; n; k): . as $used
| ([range(m; n+1)]-[$used[]])[] | $used+{(k):.};
def num: reduce .[] as $i (0; 10*. + $i);
{"m":1,"o":0}
| choose(8; 9; "s") | choose(2; 9; "e")
| choose(2; 9; "n") | choose(2; 9; "d")
| choose(2; 9; "r") | choose(2; 9; "y")
| select(([.s,.e,.n,.d] | num)
+ ([.m,.o,.r,.e] | num)
==([.m,.o,.n,.e,.y] | num))
| [.s,.e,.n,.d,"+",.m,.o,.r,.e,"=",.m,.o,.n,.e,.y];
send_more_money
将上述代码保存入 $ jq -nc -f ./send_more_money.jq
[9,5,6,7,"+",1,0,8,5,"=",1,0,6,5,2]
解析表达式文法在jq和解析表达式文法(PEG)形式化之间有密切关联[19]。这种关联源于下列表格中展示的PEG七个基本运算与jq构造之间的等价性。
八皇后问题例子回溯法求解八皇后问题,采用函数式编程风格的条件表达式和递归函数可以写为: def queens:
def q(r): . as $pl
| if r < 8 then
$pl[3] as $cl | $cl[] | . as $c
| (r+$c | tostring) as $k0
| (r-$c | tostring) as $k1
| def place:
($k0 | in($pl[1]) | not)
and ($k1 | in($pl[2]) | not);
select(place)
| [$pl[0]+[$c], $pl[1]+{$k0:null},
$pl[2]+{$k1:null}, $cl-[$c]]
| q(r+1)
else
.[0] end;
[[], {}, {}, [range(0; 8)]] | q(0)
| map("abcdefgh"[.:.+1]) | to_entries
| map(.value+(.key+1 | tostring)) | sort;
queens
将这段代码保存入 $ jq -nc -f ./queens.jq | wc -l
92
$ jq -nc -f ./queens.jq | sed -n '1,2p'
["a1","b7","c5","d8","e2","f4","g6","h3"]
["a1","b7","c4","d6","e8","f2","g5","h3"]
在能采用可逆(reversible)赋值和静态变量的语言比如Icon的实现中,其3个位置列表共有 进一步采用隐式编程风格,上述代码可以写为: def queens:
def q(r): . as $pl
| $pl[3] as $cl | $cl[] | . as $c
| (r+$c | tostring) as $k0
| (r-$c | tostring) as $k1
| def place:
($k0 | in($pl[1]) | not)
and ($k1 | in($pl[2]) | not);
select(place)
| [$pl[0]+[$c], $pl[1]+{$k0:null},
$pl[2]+{$k1:null}, $cl-[$c]];
def pipeline(i; n):
q(i) | if i+1 < n then pipeline(i+1; n) end;
[[], {}, {}, [range(0; 8)]] | pipeline(0; 8) | .[0]
| map("abcdefgh"[.:.+1]) | to_entries
| map(.value+(.key+1 | tostring)) | sort;
queens
它还可以采用共递归运算 def queens:
def q: . as $pl
| $pl[4] as $r
| $pl[3] as $cl | $cl[] | . as $c
| ($r+$c | tostring) as $k0
| ($r-$c | tostring) as $k1
| def place:
($k0 | in($pl[1]) | not)
and ($k1 | in($pl[2]) | not);
select(place)
| [$pl[0]+[$c], $pl[1]+{$k0:null},
$pl[2]+{$k1:null}, $cl-[$c], $r+1];
[[], {}, {}, [range(0; 8)], 0] | recurse(q)
| select(.[4] == 8) | .[0]
| map("abcdefgh"[.:.+1]) | to_entries
| map(.value+(.key+1 | tostring)) | sort;
queens
def queens(n):
def q: . as $pl
| $pl[4] as $r
| $pl[3] as $cl | $cl[] | . as $c
| ($r+$c | tostring) as $k0
| ($r-$c | tostring) as $k1
| def place:
($k0 | in($pl[1]) | not)
and ($k1 | in($pl[2]) | not);
select(place)
| [$pl[0]+[$c], $pl[1]+{$k0:null},
$pl[2]+{$k1:null}, $cl-[$c], $r+1];
def pipeline(n):
q | if n > 1 then pipeline(n-1) end;
def toletter:
"abcdefghijklmnopqrstuvwxyz"[.:.+1];
def fund_solut(f):
def inverse: . as $xl
| reduce range(0; n) as $i
([]; .+[$xl | index($i)]);
def variants:
[., inverse] | map(., reverse)
| map(., map(n-1-.))
| map(map(toletter) | add);
foreach f as $i
([null, {}]; .[1] as $ml
| ($i | variants) as $nl
| if all($nl[]; in($ml) | not) then
[$i, ($ml | .[$nl[]]=null)]
else
[null, $ml] end;
.[0])
| select (. != null);
fund_solut([[], {}, {}, [range(0; n)], 0]
| pipeline(n) | .[0])
| map(toletter) | to_entries
| map(.value+(.key+1 | tostring)) | sort;
queens(8)
下面演示其执行结果: $ jq -nc -f ./queens.jq | wc -l
12
$ jq -nc -f ./queens.jq
["a1","b7","c5","d8","e2","f4","g6","h3"]
["a1","b7","c4","d6","e8","f2","g5","h3"]
["a6","b1","c5","d2","e8","f3","g7","h4"]
["a4","b1","c5","d8","e2","f7","g3","h6"]
["a5","b1","c8","d4","e2","f7","g3","h6"]
["a3","b1","c7","d5","e8","f2","g4","h6"]
["a5","b1","c4","d6","e8","f2","g7","h3"]
["a7","b1","c3","d8","e6","f4","g2","h5"]
["a5","b1","c8","d6","e3","f7","g2","h4"]
["a5","b3","c1","d7","e2","f8","g6","h4"]
["a5","b7","c1","d4","e2","f8","g6","h3"]
["a6","b3","c1","d8","e4","f2","g7","h5"]
注释
参考书目
引用
外部链接
|
Portal di Ensiklopedia Dunia