Merge pull request #232 from derduher/parse-existing-sitemap

Introduce parsing xml and outputing a compat conf

CHANGELOG.md

@@ -2,6 +2,12 @@
 Add a pretty print option to `toString(false)`
 pass true pretty print
 
+Add an xmlparser that will output a config that would generate that same file
+cli:
+  use --parser to output the complete config --line-separated to print out line
+  separated config compatible with the --json input option for cli
+lib: import parseSitemap and pass it a stream
+
 # 4.0.2
 Fix npx script error - needs the shebang
 

README.md

@@ -27,6 +27,8 @@ Table of Contents
     * [Sitemap](#sitemap)
     * [buildSitemapIndex](#buildsitemapindex)
     * [createSitemapIndex](#createsitemapindex)
+    * [xmlLint](#xmllint)
+    * [parseSitemap](#parsesitemap)
     * [Sitemap Item Options](#sitemap-item-options)
     * [ISitemapImage](#isitemapimage)
     * [IVideoItem](#ivideoitem)
@@ -327,6 +329,33 @@ createSitemapIndex({
 })
 ```
 
+### xmlLint
+Resolve or reject depending on whether the passed in xml is a valid sitemap.
+This is just a wrapper around the xmllint command line tool and thus requires
+xmllint.
+```
+const { createReadStream } = require('fs')
+const { xmlLint } = require('sitemap')
+xmlLint(createReadStream('./example.xml')).then(
+  () => console.log('xml is valid'),
+  ([err, stderr]) => console.error('xml is invalid', stderr)
+)
+```
+
+### parseSitemap
+Read xml and resolve with the configuration that would produce it or reject with
+an error
+```
+const { createReadStream } = require('fs')
+const { parseSitemap, createSitemap } = require('sitemap')
+parseSitemap(createReadStream('./example.xml')).then(
+  // produces the same xml
+  // you can, of course, more practically modify it or store it
+  (xmlConfig) => console.log(createSitemap(xmlConfig).toString()),
+  (err) => console.log(err)
+)
+```
+
 ### Sitemap Item Options
 
 |Option|Type|eg|Description|

cli.ts

@@ -5,7 +5,8 @@ import { Readable } from 'stream'
 import { createReadStream } from 'fs'
 import { xmlLint } from './lib/xmllint'
 import { XMLLintUnavailable } from './lib/errors'
-console.warn('CLI is in new and likely to change quite a bit. Please send feature/bug requests to https://github.com/ekalinin/sitemap.js/issues')
+import { parseSitemap } from './lib/sitemap-parser'
+console.warn('CLI is new and likely to change quite a bit. Please send feature/bug requests to https://github.com/ekalinin/sitemap.js/issues')
 /* eslint-disable-next-line @typescript-eslint/no-var-requires */
 const arg = require('arg')
 
@@ -39,27 +40,50 @@ const argSpec = {
   '--help':    Boolean,
   '--version': Boolean,
   '--json': Boolean,
-  '--validate': Boolean
+  '--validate': Boolean,
+  '--parse': Boolean,
+  '--line-separated': Boolean
 }
 const argv = arg(argSpec)
+
+function getStream (): Readable {
+  if (argv._ && argv._.length) {
+    return createReadStream(argv._[0])
+  } else {
+    console.warn('Reading from stdin. If you are not piping anything in, this command is not doing anything')
+    return process.stdin
+  }
+}
 if (argv['--version']){
   /* eslint-disable-next-line @typescript-eslint/no-var-requires */
   const packagejson = require('../package.json')
   console.log(packagejson.version)
 } else if (argv['--help']) {
+  // TODO stream a full JSON configuration in
+  // TODO allow user to append entry to existing xml
   console.log(`
 Turn a list of urls into a sitemap xml.
 Options:
-  --help Print this text
-  --version Print the version
-  --json Parse each line as json and feed to Sitemap
+  --help           Print this text
+  --version        Print the version
+  --json           Parse each line as json and feed to Sitemap
+  --parse          Parse fed xml and spit out config
+  --line-separated When used with parse, it spits out each entry as json rather
+                   than the whole json. This can be then consumed with --json by
+                   the cli
 `)
+} else if (argv['--parse']) {
+  parseSitemap(getStream()).then((items): void => {
+    if (argv['--line-separated'] && items.urls) {
+      items.urls.forEach((url): void => {
+        console.log(JSON.stringify(url))
+      })
+    } else {
+      console.log(JSON.stringify(items))
+    }
+  })
 } else if (argv['--validate']) {
-  let xml = process.stdin
-  if (argv._ && argv._.length) {
-    xml = argv._[0]
-  }
-  xmlLint(xml)
+  xmlLint(getStream())
     .then((): void => console.log('valid'))
     .catch(([error, stderr]: [Error|null, Buffer]): void => {
       if (error instanceof XMLLintUnavailable) {

index.ts

@@ -10,5 +10,6 @@ export * from './lib/sitemap-index'
 export * from './lib/errors'
 export * from './lib/types'
 export { xmlLint } from './lib/xmllint'
+export { parseSitemap } from './lib/sitemap-parser'
 
 export default createSitemap