@warren-bank/hls-proxy 3.2.3 → 3.2.5

package/README.md

@@ -600,7 +600,8 @@ curl --silent --insecure "$URL"
     * internal `proxy` module exports an Object containing event listeners to process requests that can be either:
       - added to an instance of [`http.Server`](https://nodejs.org/api/http.html#class-httpserver)
       - added to an [`Express.js`](https://github.com/expressjs/express) application as middleware to handle a custom route
-        * important limitation: since `/` is a valid character in a base64 encoded URL, the path for a custom route needs to end with a character that is not allowed in base64 encoding (ex: `'/proxy_/*'`)
+        * important requirement: the path for a custom route needs to include exactly one unnamed [parameter](https://expressjs.com/en/guide/routing.html#route-parameters) that matches the base64 encoded URL and (optionally) a file extension (ex: `'/proxy/*'`)
+        * the use of nested routers is supported
   - system requirements:
     * Node.js v16.0.0 and higher
       - required features: [`Proxy` constructor](https://node.green/#ES2015-built-ins-Proxy-constructor-requires-new), [`Proxy` 'apply' handler](https://node.green/#ES2015-built-ins-Proxy--apply--handler), [`Reflect.apply`](https://node.green/#ES2015-built-ins-Reflect-Reflect-apply), [`RegExp` 'd' flag](https://node.green/#ES2022-features-RegExp-Match-Indices---hasIndices-----d--flag-)

package/hls-proxy/expressjs_utils.js

@@ -0,0 +1,32 @@
+const get_full_req_url = function(req) {
+  return req.originalUrl || req.url
+}
+
+const has_req_param = function(req, key) {
+  return (req.params && (typeof req.params === 'object') && req.params[key])
+}
+
+const get_proxy_req_url = function(req) {
+  const key = "0"
+  return has_req_param(req, key)
+    ? `/${req.params[key]}`
+    : req.url
+}
+
+const get_base_req_url = function(req) {
+  let base_url = ''
+  const key = "0"
+
+  if (req.path && has_req_param(req, key)) {
+    base_url  = req.baseUrl || ''
+    base_url += req.path.substring(0, (req.path.length - req.params[key].length - 1))
+  }
+
+  return base_url
+}
+
+module.exports = {
+  get_full_req_url,
+  get_proxy_req_url,
+  get_base_req_url
+}

package/hls-proxy/manifest_parser.js

@@ -16,9 +16,6 @@ const url_location_landmarks = {
       '#EXT-X-RENDITION-REPORT:',
       '#EXT-X-DATERANGE:',
       '#EXT-X-CONTENT-STEERING:'
-    ],
-    next_line: [
-      '#EXT-X-STREAM-INF:'
     ]
   },
   ts: {
@@ -26,26 +23,20 @@ const url_location_landmarks = {
       '#EXT-X-MAP:',
       '#EXT-X-PART:',
       '#EXT-X-PRELOAD-HINT:'
-    ],
-    next_line: [
-      '#EXTINF:'
     ]
   },
   json: {
     same_line: [
       '#EXT-X-SESSION-DATA:'
-    ],
-    next_line: []
+    ]
   },
   key: {
     same_line: [
       '#EXT-X-KEY:'
-    ],
-    next_line: []
+    ]
   },
   other: {
-    same_line: [],
-    next_line: []
+    same_line: []
   }
 }
 
@@ -148,63 +139,82 @@ const parse_manifest = function(m3u8_content, m3u8_url, referer_url, hooks, cach
 const extract_embedded_urls = function(m3u8_lines, m3u8_url, referer_url, meta_data) {
   const embedded_urls = []
 
-  let m3u8_line, has_next_m3u8_line, next_m3u8_line, matches, matching_landmark, matching_url
+  // one of: ['master','media']
+  // determined by the detection of either: ['#EXT-X-STREAM-INF','#EXTINF'], respectively
+  // until the type is determined, URI lines are ignored (as per the HLS spec)
+  let manifest_type = null
+
+  let m3u8_line, matches, matching_landmark, matching_url
 
   for (let i=0; i < m3u8_lines.length; i++) {
-    m3u8_line          = m3u8_lines[i]
-    has_next_m3u8_line = ((i+1) < m3u8_lines.length)
-
-    matches = regexs.m3u8_line_landmark.exec(m3u8_line)
-    if (!matches) continue
-    matching_landmark = matches[1]
-
-    matches = regexs.m3u8_line_url.exec(m3u8_line)
-    matching_url = matches
-      ? matches[1]
-      : null
-
-    if (meta_data !== null)
-      extract_meta_data(meta_data, m3u8_line, matching_landmark)
-
-    for (let url_type in url_location_landmarks) {
-      if (matching_url && (url_location_landmarks[url_type]['same_line'].indexOf(matching_landmark) >= 0)) {
-        embedded_urls.push({
-          line_index:         i,
-          url_indices:        matches.indices[1],
-          url_type:           url_type,
-          original_match_url: matching_url,
-          resolved_match_url: (new URL(matching_url, m3u8_url)).href,
-          redirected_url:     null,
-          referer_url:        referer_url,
-          encoded_url:        null
-        })
-        break
+    m3u8_line = m3u8_lines[i]
+
+    if (is_m3u8_line_a_blank(m3u8_line) || is_m3u8_line_a_comment(m3u8_line))
+      continue
+
+    if (is_m3u8_line_a_tag(m3u8_line)) {
+      matches = regexs.m3u8_line_landmark.exec(m3u8_line)
+      if (!matches) continue
+      matching_landmark = matches[1]
+
+      if (manifest_type === null) {
+        if (matching_landmark === '#EXT-X-STREAM-INF:')
+          manifest_type = 'master'
+        else if (matching_landmark === '#EXTINF:')
+          manifest_type = 'media'
       }
-      if (has_next_m3u8_line && (url_location_landmarks[url_type]['next_line'].indexOf(matching_landmark) >= 0)) {
-        next_m3u8_line = m3u8_lines[i+1].trim()
 
-        if (next_m3u8_line && (next_m3u8_line[0] !== '#')) {
-          i++
+      if (meta_data !== null)
+        extract_meta_data(meta_data, m3u8_line, matching_landmark)
 
+      matches = regexs.m3u8_line_url.exec(m3u8_line)
+      if (!matches) continue
+      matching_url = matches[1]
+
+      for (let url_type in url_location_landmarks) {
+        if (url_location_landmarks[url_type]['same_line'].indexOf(matching_landmark) >= 0) {
           embedded_urls.push({
             line_index:         i,
-            url_indices:        null,
+            url_indices:        matches.indices[1],
             url_type:           url_type,
-            original_match_url: next_m3u8_line,
-            resolved_match_url: (new URL(next_m3u8_line, m3u8_url)).href,
+            original_match_url: matching_url,
+            resolved_match_url: (new URL(matching_url, m3u8_url)).href,
             redirected_url:     null,
             referer_url:        referer_url,
             encoded_url:        null
           })
+          break
         }
-        break
       }
     }
+    else {
+      // line is a URI
+      if (manifest_type === null) continue
+
+      const url_type = (manifest_type === 'master') ? 'm3u8' : 'ts'
+
+      embedded_urls.push({
+        line_index:         i,
+        url_indices:        null,
+        url_type:           url_type,
+        original_match_url: m3u8_line,
+        resolved_match_url: (new URL(m3u8_line, m3u8_url)).href,
+        redirected_url:     null,
+        referer_url:        referer_url,
+        encoded_url:        null
+      })
+    }
   }
 
   return embedded_urls
 }
 
+const is_m3u8_line_a_blank = (line) => (!line || !line.trim())
+
+const is_m3u8_line_a_comment = (line) => ((line.indexOf('#') === 0) && !is_m3u8_line_a_tag(line))
+
+const is_m3u8_line_a_tag = (line) => (line.indexOf('#EXT') === 0)
+
 const extract_meta_data = function(meta_data, m3u8_line, matching_landmark) {
   for (let meta_data_key in meta_data_location_landmarks) {
     if (meta_data_location_landmarks[meta_data_key]['same_line'].indexOf(matching_landmark) >= 0) {

package/hls-proxy/utils.js

@@ -1,4 +1,5 @@
 const parse_url = require('@warren-bank/url').parse
+const expressjs = require('./expressjs_utils')
 
 const regexs = {
   req_url: new RegExp('^(.*?)/([a-zA-Z0-9\\+/=%]+)(?:[\\._]([^/\\?#]*))?(?:[\\?#].*)?$'),
@@ -20,10 +21,10 @@ const parse_req_url = function(params, req) {
 
   const result = {redirected_base_url: '', url_type: '', url: '', referer_url: ''}
 
-  const matches = regexs.req_url.exec(req.url)
+  const matches = regexs.req_url.exec( expressjs.get_proxy_req_url(req) )
 
   if (matches) {
-    result.redirected_base_url = `${ is_secure ? 'https' : 'http' }://${host || req.headers.host}${matches[1] || ''}`
+    result.redirected_base_url = `${ is_secure ? 'https' : 'http' }://${host || req.headers.host}${expressjs.get_base_req_url(req) || matches[1] || ''}`
 
     if (matches[3])
       result.url_type = matches[3].toLowerCase().trim()

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@warren-bank/hls-proxy",
   "description": "Node.js server to proxy HLS video streams",
-  "version":     "3.2.3",
+  "version":     "3.2.5",
   "scripts": {
     "start":     "node hls-proxy/bin/hlsd.js",
     "sudo": "sudo node hls-proxy/bin/hlsd.js"