Add compound keys to dat import

Author: okdistribute

bin/import.js

@@ -33,6 +33,11 @@ module.exports = {
       boolean: false,
       abbr: 'k'
     },
+    {
+      name: 'keys',
+      boolean: false,
+      abbr: 'ks'
+    },
     {
       name: 'batch',
       boolean: false,
@@ -49,11 +54,9 @@ module.exports = {
 function handleImport (args) {
   debug('handleImport', args)
 
-  if (args.help || args._.length === 0) {
-    return usage()
-  }
-
+  if (args.help || args._.length === 0) return usage()
   if (!args.dataset) abort(new Error('Error: Must specify dataset (-d)'), args)
+  if (args.keys) args.keys = args.keys.split(',')
 
   openDat(args, function (err, db) {
     if (err) abort(err, args)

docs/cli-docs.md

@@ -480,7 +480,8 @@ dat import <filename> --dataset=<name>
 
 ### Options
 
-- `key`/`k` - specify which field to use as the primary key (false for no key)
+- `key`/`k` - specify which column to use as the primary key (defaults to auto-generated keys)
+- `keys`/`ks` - comma-separated list of column names to craft a compound key. sorted ascending by default
 - `message`/`m` - a short description of this import
 
 Examples:

lib/import.js

@@ -4,14 +4,26 @@ var through = require('through2')
 var debug = require('debug')('lib/import')
 var parseInputStream = require('../lib/util/parse-input-stream.js')
 
+var COMPOUND_KEY_SEPARATOR = '+'
+
 module.exports = function (db, opts) {
   if (!opts) opts = {}
   if (!opts.dataset) throw new Error('Error: Must specify dataset (-d)')
   if (opts.dataset === 'files') throw new Error('Cannot import into the \'files\' dataset')
 
   var transform = through.obj(function (obj, enc, next) {
     debug('heres my obj!', obj)
-    var key = obj[opts.key] || obj.key || uuid()
+
+    var key
+    if (opts.keys) {
+      key = opts.keys.sort().map(function (key) {
+        return obj[key] || ''
+      }).join(COMPOUND_KEY_SEPARATOR)
+    } else {
+      key = obj[opts.key] || obj.key
+    }
+
+    if (!key || key === COMPOUND_KEY_SEPARATOR) key = uuid()
     var doc = {type: 'put', key: key, value: obj}
     next(null, doc)
   })

tests/import.js

@@ -41,6 +41,25 @@ test('import: dat import json', function (t) {
   st.end()
 })
 
+test('import: dat import json with compound key', function (t) {
+  var json = path.resolve(__dirname + '/fixtures/all_hour.json')
+  var st = spawn(t, dat + ' import ' + json + ' --keys=latitude,longitude -d compound', {cwd: dat2})
+  st.stdout.empty()
+  st.stderr.match(/Done importing data/)
+  st.end()
+})
+
+test('import: dat keys get integer id', function (t) {
+  var st = spawn(t, dat + ' keys -d compound', {cwd: dat2})
+  st.stdout.match(function (output) {
+    var keys = output.split('\n')
+    t.same(keys[0], '33.9233322+-117.9376678')
+    return true
+  })
+  st.stderr.empty()
+  st.end()
+})
+
 test('import: dat import json with integer id', function (t) {
   var json = path.resolve(__dirname + '/fixtures/all_hour.json')
   var st = spawn(t, dat + ' import ' + json + ' --key=int --dataset=int-id', {cwd: dat2})

usage/export.txt

@@ -2,4 +2,5 @@ dat export
   -d <dataset-name> (required) # the name of the dataset
   --limit=<number> # the number of rows to output. default: infinity
   --format=[ndjson, csv, json] # how to parse the output. default: ndjson
+  --full # output full metadata with each row, including autogenerated keys
 

usage/import.txt

@@ -1,3 +1,5 @@
 dat import <filename or - for stdin> (required)
   -d <dataset-name> (required) # the name of the dataset to import
   -f [ndjson, csv, json] # how to parse the file. guessed if not supplied.
+  --key <column name># the primary key to use. defaults to auto-generated keys
+  --keys <column,names,separated> # a comma-separated list of column names for a compound key