การแยกวิเคราะห์ JSON ด้วย jsonlite

JSON คืออะไร

library(jsonlite)
# JSON looks like:
# { "name": "Alice", "age": 30, "active": true }

# fromJSON parses JSON string to R object
json_str <- '{"name":"Alice","age":30,"active":true}'
result <- fromJSON(json_str)
class(result)   # 'list'
result$name     # 'Alice'
result$age      # 30 (numeric)
result$active   # TRUE (logical)

fromJSON(): การแยกวิเคราะห์พื้นฐาน

library(jsonlite)
# Parsing different JSON types
fromJSON('"hello"')         # character: 'hello'
fromJSON('42')               # numeric: 42
fromJSON('true')             # logical: TRUE
fromJSON('null')             # NA
fromJSON('[1, 2, 3]')        # numeric vector: c(1,2,3)
fromJSON('["a", "b", "c"]') # character vector

# Nested object
json <- '{"x": 1, "y": [2, 3], "z": null}'
obj <- fromJSON(json)
obj$x  # 1
obj$y  # c(2, 3)
obj$z  # NA

toJSON(): การทำให้ R เป็น JSON

library(jsonlite)
# Basic serialization
toJSON(c(1, 2, 3))          # '[1,2,3]'
toJSON(list(a=1, b='two'))  # '{"a":[1],"b":["two"]}'

# auto_unbox: length-1 vectors become scalars
toJSON(list(name='Bob', age=25), auto_unbox = TRUE)
# '{"name":"Bob","age":25}'

# pretty printing
cat(toJSON(list(x=1, y=c(2,3)), pretty = TRUE))
# {
#   "x": [1],
#   "y": [2, 3]
# }

อาร์เรย์ JSON เป็นกรอบข้อมูล

library(jsonlite)
# JSON array of objects -> data frame
json_array <- '[
  {"name": "Alice", "age": 30, "score": 95},
  {"name": "Bob",   "age": 25, "score": 87},
  {"name": "Carol", "age": 35, "score": 92}
]'

df <- fromJSON(json_array)
class(df)   # 'data.frame'
nrow(df)    # 3
print(df)
# name age score
# Alice  30    95
# Bob    25    87
# Carol  35    92

แปลงกรอบข้อมูลเป็น JSON

library(jsonlite)
df <- data.frame(
  product = c('Apple', 'Banana', 'Cherry'),
  price   = c(1.20, 0.50, 3.00),
  in_stock = c(TRUE, TRUE, FALSE),
  stringsAsFactors = FALSE
)

# Default: each value is an array (even scalars)
cat(toJSON(df, pretty = FALSE))

# auto_unbox: single values become JSON scalars
cat('\n---\n')
cat(toJSON(df, auto_unbox = TRUE, pretty = TRUE))

fromJSON() จาก URL

library(jsonlite)
# Read JSON directly from a URL (requires internet)
# Example: JSONPlaceholder (free test API)
# result <- fromJSON('https://jsonplaceholder.typicode.com/todos/1')
# result$title   # todo title
# result$completed  # TRUE/FALSE

# Example: GitHub API
# repos <- fromJSON('https://api.github.com/users/hadley/repos')
# repos$name   # vector of repo names
# nrow(repos)  # number of repos returned

# Simulate with local JSON string:
result <- fromJSON('{"id":1,"title":"Learn R","completed":false}')
cat('Task:', result$title, '\nDone:', result$completed)

prettify() และ minify()

library(jsonlite)
# Compact (minified) JSON
compact_json <- '{"name":"Alice","scores":[95,87,92],"active":true}'
cat('Minified:\n', compact_json, '\n\n')

# Prettify: add whitespace
cat('Prettified:\n')
cat(prettify(compact_json))
# {
#   "name": "Alice",
#   "scores": [95, 87, 92],
#   "active": true
# }

# Minify: remove whitespace
pretty_json <- prettify(compact_json)
cat('\nRe-minified:\n')
cat(minify(pretty_json))

flatten=TRUE สำหรับออบเจ็กต์ซ้อนกัน

library(jsonlite)
json <- '[
  {"id": 1, "user": {"name": "Alice", "city": "NY"}},
  {"id": 2, "user": {"name": "Bob",   "city": "LA"}}
]'

# Without flatten: nested list column
df_nested <- fromJSON(json)
class(df_nested$user)   # data.frame (nested!)

# With flatten: flat column names
df_flat <- fromJSON(json, flatten = TRUE)
names(df_flat)     # c('id', 'user.name', 'user.city')
print(df_flat)
# id user.name user.city
#  1     Alice        NY
#  2       Bob        LA

จัดการ JSON ที่มี NA และคีย์ที่หายไป

library(jsonlite)
# JSON with null values
json_nulls <- '{"a": 1, "b": null, "c": "hello"}'
obj <- fromJSON(json_nulls)
obj$b  # NA
is.na(obj$b)  # TRUE

# Array with missing keys
json_missing <- '[
  {"name": "Alice", "age": 30, "email": "a@b.com"},
  {"name": "Bob",   "age": 25}
]'
# 'email' missing from second record
df <- fromJSON(json_missing)
print(df)
# name age    email
# Alice  30  a@b.com
# Bob    25     <NA>

เขียน JSON ลงในไฟล์

library(jsonlite)
results <- list(
  model = 'linear_regression',
  coefficients = list(intercept = 2.5, slope = 1.3),
  r_squared = 0.87,
  n_obs = 100
)

# Convert to JSON string
json_output <- toJSON(results, auto_unbox = TRUE, pretty = TRUE)
cat(json_output)

# Save to file (uncomment):
# write_json(results, 'model_results.json',
#            auto_unbox = TRUE, pretty = TRUE)
# Or equivalently:
# writeLines(json_output, 'model_results.json')

ตรวจสอบความถูกต้องของ JSON

library(jsonlite)
# Valid JSON
validate('{"name": "Alice", "age": 30}')  # TRUE
validate('[1, 2, 3]')                       # TRUE
validate('"just a string"')                # TRUE

# Invalid JSON
validate('{name: Alice}')     # FALSE (unquoted keys)
validate('[1, 2, 3,]')        # FALSE (trailing comma)
validate('')                   # FALSE

# Safe parsing pattern
safe_parse <- function(json) {
  if (!validate(json)) {
    warning('Invalid JSON')
    return(NULL)
  }
  fromJSON(json)
}
safe_parse('{"x": 1}')
safe_parse('{bad json}')

ตรวจสอบความเข้าใจอย่างรวดเร็ว

ทบทวน: JSON ด้วย jsonlite

library(jsonlite)
# Complete jsonlite workflow:

# Parse
df <- fromJSON('[{"a":1,"b":2},{"a":3,"b":4}]')
print(df)

# Serialize
json <- toJSON(df, auto_unbox = TRUE, pretty = FALSE)
cat(json, '\n')

# Flatten nested
nested <- fromJSON(
  '[{"id":1,"x":{"val":10}},{"id":2,"x":{"val":20}}]',
  flatten = TRUE
)
print(nested)

# Validate before parse
validate('{"key": "value"}')  # TRUE