Merge pull request #1 from nealwp/updates

Updates

README.md

@@ -10,7 +10,17 @@ This CLI tool reads a JSON file and produces BigQuery compatible SQL views from
 ## Usage
 
 ```bash
-npx @nealwp/blobview <filepath> 
+npx @nealwp/blobview@latest [options] <filepath> 
+```
+
+```text
+Arguments:
+  filepath      path to valid JSON file
+
+Options:
+  -t TABLE, --table=TABLE           specify a table name to use in FROM clause. default: "<table>"
+  -d DATASET, --dataset=DATASET     specify a dataset to use in FROM clause. default: "<dataset>" 
+  -h, --help                        show help
 ```
 
 ## Examples:
@@ -19,9 +29,16 @@ Default output to STDOUT:
 npx @nealwp/blobview ./path/to/file.json
 ```
 
+Dataset and table as input options:
+```bash
+npx @nealwp/blobview --dataset=myDataset --table=myTable ./path/to/file.json
+# shorthand options
+npx @nealwp/blobview -d myDataset -t myTable ./path/to/file.json
+```
+
 Redirect output to file:
 ```bash
-npx @nealwp/blobview ./path/to/file.json > my-view-file.sql
+npx @nealwp/blobview@latest ./path/to/file.json > my-view-file.sql
 ```
 
 ## Features
@@ -32,6 +49,7 @@ npx @nealwp/blobview ./path/to/file.json > my-view-file.sql
 * Auto-formats column names to snake_case from camelCase and PascalCase
 * Detects deeply-nested objects and formats to JSON string
 * Pre-populates FROM clause with BigQuery-style placeholders
+* BigQuery dataset and table name can be supplied as input options
 
 ## Limitations
 * Does not detect DATE or TIMESTAMP types, or other types like BOOLEAN
@@ -40,7 +58,7 @@ npx @nealwp/blobview ./path/to/file.json > my-view-file.sql
 * Does not create SQL views in any syntax other than BigQuery
 * Requires a local JSON file to read
 * Does not include option to write queries to separate files instead of STDOUT
-* BigQuery project, dataset, and datastream names cannot be supplied as input
+* BigQuery project name cannot be supplied as input
 
 ## Example Output
 
@@ -100,20 +118,52 @@ SELECT
     , CAST(JSON_VALUE(json_blob.integerField) as INTEGER) as integer_field
     , CAST(JSON_VALUE(json_blob.decimalField) as DECIMAL) as decimal_field
     , TO_JSON_STRING(json_blob.exampleGeoJson) as example_geo_json 
-FROM <project>.<datastream>.<dataset>
---------
+FROM <project>.<dataset>.<table>
+/**/
+SELECT
+    CAST(JSON_VALUE(json_blob.childField1.gender) as STRING) as gender 
+    , CAST(JSON_VALUE(json_blob.childField1.latitude) as DECIMAL) as latitude 
+FROM <project>.<dataset>.<table>
+/**/
+SELECT
+    CAST(JSON_VALUE(json_blob.childField2.favoriteFruit) as STRING) as favorite_fruit 
+    , CAST(JSON_VALUE(json_blob.childField2.longitude) as DECIMAL) as longitude 
+FROM <project>.<dataset>.<table>
+/**/
+SELECT
+    CAST(JSON_VALUE(json_blob.childWithNestedObject.isNormal) as STRING) as is_normal 
+    , TO_JSON_STRING(json_blob.childWithNestedObject.nestedObject) as nested_object 
+FROM <project>.<dataset>.<table>
+```
+
+```bash
+# terminal command with input options
+npx @nealwp/blobview --dataset=myDataset --table=myTable sample-data.json
+```
+
+Will produce the following output:
+
+```sql
+/* stdout */
+SELECT
+    CAST(JSON_VALUE(json_blob.stringField) as STRING) as string_field
+    , CAST(JSON_VALUE(json_blob.integerField) as INTEGER) as integer_field
+    , CAST(JSON_VALUE(json_blob.decimalField) as DECIMAL) as decimal_field
+    , TO_JSON_STRING(json_blob.exampleGeoJson) as example_geo_json 
+FROM <project>.myDataset.myTable
+/**/
 SELECT
     CAST(JSON_VALUE(json_blob.childField1.gender) as STRING) as gender 
     , CAST(JSON_VALUE(json_blob.childField1.latitude) as DECIMAL) as latitude 
-FROM <project>.<datastream>.<dataset>
---------
+FROM <project>.myDataset.myTable
+/**/
 SELECT
     CAST(JSON_VALUE(json_blob.childField2.favoriteFruit) as STRING) as favorite_fruit 
     , CAST(JSON_VALUE(json_blob.childField2.longitude) as DECIMAL) as longitude 
-FROM <project>.<datastream>.<dataset>
---------
+FROM <project>.myDataset.myTable
+/**/
 SELECT
     CAST(JSON_VALUE(json_blob.childWithNestedObject.isNormal) as STRING) as is_normal 
     , TO_JSON_STRING(json_blob.childWithNestedObject.nestedObject) as nested_object 
-FROM <project>.<datastream>.<dataset>
+FROM <project>.myDataset.myTable
 ```

package.json

@@ -1,6 +1,6 @@
 {
     "name": "@nealwp/blobview",
-    "version": "0.1.0",
+    "version": "0.2.0",
     "description": "Generate BigQuery SQL views from JSON",
     "bin": {
         "@nealwp/blobview": "./index.js"
@@ -23,6 +23,7 @@
     "license": "ISC",
     "devDependencies": {
         "@types/jest": "^29.5.0",
+        "@types/minimist": "^1.2.2",
         "@typescript-eslint/eslint-plugin": "^5.57.1",
         "@typescript-eslint/parser": "^5.57.1",
         "copyfiles": "^2.4.1",
@@ -36,5 +37,8 @@
         "ts-jest": "^29.1.0",
         "ts-node": "^10.9.1",
         "typescript": "^5.0.3"
+    },
+    "dependencies": {
+        "minimist": "^1.2.8"
     }
 }

src/index.spec.ts

@@ -5,7 +5,8 @@ describe('main', () => {
     it('should log error if no filepath is passed', () => {
         process.argv = ['foo', 'bar']
         console.log = jest.fn()
-        main()
+        const args = {_: []}
+        main(args)
         expect(console.log).toHaveBeenCalled()
     })
 })

src/index.ts

@@ -1,30 +1,72 @@
 #!/usr/bin/env node
 
+import minimist from 'minimist'
 import { readJsonFileContent, writeToStdOut } from "./io"
 import { jsonToSqlView } from "./parser"
 
-export const main = () => {
-    const args = process.argv.slice(2)
-    const inputFilePath = args[0]
+export const main = (args: minimist.ParsedArgs) => {
+    const inputFilePath = args._[0]
+
+    let table = '<table>'
+    let dataset = '<dataset>'
 
+    if(args.help || args.h) {
+        printHelp()
+        return
+    }
+
     if(!inputFilePath) {
-        printHelp();
+        printError("No file path provided!");
         return
     }
 
+    if (args.table) {
+        table = args.table
+    } else if (args.t) {
+        table = args.t
+    }
+
+    if (args.dataset) {
+        dataset = args.dataset
+    } else if (args.d) {
+        dataset = args.d
+    }
+
     try {
         const fileContent = readJsonFileContent(inputFilePath)
-        const sqlOutput = jsonToSqlView(fileContent)
+        const sqlOutput = jsonToSqlView(fileContent, {table, dataset})
         writeToStdOut(sqlOutput)
     } catch(err) {
         console.log(err)
     }
 }
 
 function printHelp() {
-    const msg = `Error: no file path provided.\nUsage:\n\nnpx @nealwp/blobview <filepath>\n`
-    console.log(msg)
+   const helpText = `Generate BigQuery SQL views from JSON.
+
+Usage:    
+  @nealwp/blobview [options] <filepath> 
+
+Arguments:
+  filepath      path to valid JSON file
+
+Options:
+  -t TABLE, --table=TABLE           specify a table name to use in FROM clause. default: "<table>"
+  -d DATASET, --dataset=DATASET     specify a dataset to use in FROM clause. default: "<dataset>" 
+  -h, --help                        show help` 
+
+    console.log(helpText)
 }
 
-main(); 
+function printError(message: string) {
+    console.log(`Error: ${message}`)
+    console.log('Run "@nealwp/blobview --help" to display help.')
+}
+
+const args = minimist(process.argv.slice(2), {
+    stopEarly: true,
+    boolean: false
+})
+
+main(args); 
 

src/io.ts

@@ -8,7 +8,7 @@ export function readJsonFileContent(path: string) {
 export function writeToStdOut(data: {parentSql: string, childQueries: string[]}) {
     console.log(data.parentSql)
     data.childQueries.forEach(q => { 
-        console.log("--------")
+        console.log("/**/")
         console.log(q)
     })
 }