使用JQ在shell腳本中迭代JSON對象

Question

我試圖在使用jq的shell腳本程序中解析json文件。

的JSON結構看起來是這樣的：

{ 
    "items":[ 
     { 
     "playlists":[ 
      { 
       "title":"Level One Title", 
       "courses":[ 
        { 
        "title":"Level One Course One Title", 
        "lessons":[ 
         { 
          "title":"lesson 1 title", 
          "url":"lesson 1 url" 
         }, 
         { 
          "title":"lesson 2 title", 
          "url":"lesson 2 url" 
         } 
        ] 
        }, 
        { 
        "title":"Level One Course Two Title", 
        "lessons":[ 
         { 
          "title":"lesson 1 title", 
          "url":"lesson 1 url" 
         }, 
         { 
          "title":"lesson 2 title", 
          "url":"lesson 2 url" 
         } 
        ] 
        } 
       ] 
      }, 
      { 
       "title":"Level Two Title", 
       "courses":[ 
        { 
        "title":"Level Two Course One Title", 
        "lessons":[ 
         { 
          "title":"lesson 1 title", 
          "url":"lesson 1 url" 
         }, 
         { 
          "title":"lesson 2 title", 
          "url":"lesson 2 url" 
         } 
        ] 
        }, 
        { 
        "title":"Level Two Course Two Title", 
        "lessons":[ 
         { 
          "title":"lesson 1 title", 
          "url":"lesson 1 url" 
         }, 
         { 
          "title":"lesson 2 title", 
          "url":"lesson 2 url" 
         } 
        ] 
        } 
       ] 
      } 
     ] 
     } 
    ] 
} 

因爲我是一名iOS開發者，我寫了一個模擬件的雨燕代碼會達到我想要的。它看起來像這樣：

for level in levels { 
    let title = level["title"] 
    //TODO: create a folder with title and set as pwd 
    for course in level["courses"] { 
     let courseTitle = course["title"] 
     //TODO: create a folder with title and set as pwd 
     for lesson in course["lessons"] { 
      let title = lesson["title"] 
      let videoURL = lesson["url"] 
      //TODO: download video in current folder with title 
     } 
    } 
} 

levels是Dictionaries的Array，level是那些Dictionaries之一。每個級別包含Array的courses，其中每個是Dictionary，包含Array的lesson字典。

來自Swift，我發現jq的語法非常混亂，尤其是將函數返回值賦值給變量。這裏是我得到多遠：

levels=$(jq ".items[0].playlists" data.json) 
for level in $levels 
do 
    title=$($level | jq ".title") 
    echo $title 
done 

這似乎並沒有工作。我想我的邏輯使用for循環在這裏是完全錯誤的。任何想法如何完成？

Answer 1

代碼提取標題和URL到不同的shell變量可能看起來像：

jq_program=' 
.items[].playlists[]   # we only care about playlist contents 
| .title as $level_title  # store the level title before recursing 
| .courses[]     # ...into elements of the array in .courses... 
| .title as $course_title  # repeat to store the course title 
| .lessons[]     # ...before recursing into the lesson... 
# ...and emitting the stored level and course titles, plus the lesson title and url 
| [$level_title, $course_title, .title, .url] | @tsv        
' 

while IFS=$'\t' read -r level_title course_title lesson_title lesson_url <&3; do 
    mkdir -p -- "$level_title/$course_title" 
    ## For test purposes, just store the URL in the file 
    printf '%s\n' "$lesson_url" >"$level_title/$course_title/$lesson_title" 
    ## If we had real data, we might instead be running something like: 
    # curl -o "$level_title/$course_title/$lesson_title" "$lesson_url" 
done 3< <(jq -r "$jq_program" <input.json) 

在這裏，我們有jq做遞歸的工作，並散發出平（製表符分隔）列表，閱讀該列表在bash中。出於顯而易見的原因，這僅適用於內容不能包含製表符的情況;如果需要使用任意字符串（在bash可以存儲的範圍內），可以從jq生成NUL分隔的輸出，但需要使用stedolan/jq#1271中討論的其中一種解決方法。

請參閱BashFAQ #1討論我們在此處使用的技術，以從jq的輸出中讀取製表符分隔的數據。