ner-node - npm Package Compare versions

Comparing version 0.0.4 to 0.0.5

package.json

 {
   "name": "ner-node",
-  "version": "0.0.4",
+  "version": "0.0.5",
   "description": "Node library for spawning NER server and pinging it to get entities out of raw text.",
@@ -13,5 +13,5 @@ "main": "SocketNER.js",
     "type": "git",
-    "url" : "https://github.com/Vikasg7/NER-Node.git"
+    "url": "https://github.com/Vikasg7/NER-Node.git"
   },
-  "keywords":[
+  "keywords": [
     "Stanford NER",
@@ -25,4 +25,4 @@ "Named Entity Recognition",
   "dependencies": {
-    "deasync": "^0.1.4"
+    "synchronize": "^0.9.15"
   }
 }

README.md

 <h1>NER-Node</h2>
 <h5>Library to connect to Stanford NER local Server, send in the Raw Text and get back Entity JSON</h5>
 
+<h4><i>4X performance boost with same syntax in versions 0.0.5 and further. Please do upgrade.</i></h4>
+
 <h4>Installation</h4>
@@ -12,3 +14,3 @@ <ol>
 
-```
+````javascript
 
@@ -19,9 +21,9 @@ var socketNER = require("ner-node")
 	obj.parser = function (taggedText) {..... return entities}
-	// A Sync function to get the Entities JSON
+	// Synchronous function to get the Entities JSON
 	var entitiesJSON = obj.getEntities(rawText, requiredEntity)
-	// closes the server and socket when done
+	// closes the server and client when done
 	obj.close()
 })
 
-```
+````
 <p>
@@ -32,2 +34,5 @@ <strong>Note:-</strong>
 
+<h4>Updates</h4>
+<p>Its 20-Feb-2016. I have pushed an update to make the library run 4x faster than previous versions. so please <strong>use version 0.0.5</strong> or lastest for production purposes. There has been no change in the Syntax.</p>
+
 <h4>Issues & Suggestions</h4>
@@ -39,2 +44,5 @@ <p>If you find an issues using the Library OR if you have any suggestions to make it perform better, then you can write to us in the Issues Section.</p>
 	<a href="https://www.paypal.com/cgi-bin/webscr?cmd=_s-xclick&hosted_button_id=F3QQCWFPWHBYE" target="_blank">Buy Me Coffee</a>
+OR you can use 
+	<a href="https://gratipay.com/~xcelancer/" target="_blank">Gratipay</a>
+to show your appreciation and gratitude.
 </p>

SocketNER.js

 var spawn = require("child_process").spawn
-var socket = require("net").Socket()
-var deasync = require("deasync")
-// The purpose of deasync is to make the getEntities function synchronous 
-// without guessing the time it would take to finish. Previously, I was using
-// socket.on(data, function () {...}) with a callback. But the downside was when I 
-// used the getEntities function in a loop, NodeJs warns me of Memory leak, as I was 
-// registering so many on("data") events. So I register on("data") event only one time
-// by taking the code out of getEntities function and by using a global variable jsonEntities.
+var sync = require("synchronize")
 
-
 function SocketNER(port, classifierFileName, pathToNER, callback) {
@@ -17,8 +9,3 @@ //defining defaults if arguments is a false value
    pathToNER = pathToNER || "/"
-   // using jsonEntities as a list is a clever technique of mine
-   // to empty using .pop() while returning the value in getEntities function**
-   var jsonEntities = []
-   var taggedText
-   var requiredEntityG
-   var rawTextG
+   var client
 
@@ -32,40 +19,52 @@ // starting server as a seperate process
          "-loadClassifier", pathToNER + classifierFileName, 
-            "-port", port, "-outputFormat", "inlineXML"
-         ]
-      )
+         "-port", port, "-outputFormat", "inlineXML"
+      ]
+   )
 
-   // Setup a Socket Connection after Server loads the Classifier
-   // I don't know why server's stderr stream gets all output and why stdout don't
+   // I don't know why server's stderr stream gets 
+   // all output and why stdout don't
    server.stderr.on("data", function (data) {
-      // Server would finish loading, when it flushes out 'done [x secs]'
-      if (data.toString().search("done") !== -1) { callback(socketNER) }
+      // Server would finish loading, 
+      // when it flushes out 'done [x secs]'
+      if (data.toString().search("done") > -1) { 
+         startNERClient()
+      }
    })
    
+   function startNERClient() {
+      client = spawn(
+         "java",[
+            "-cp",
+            pathToNER + "stanford-ner.jar", 
+            "edu.stanford.nlp.ie.NERServer", 
+            "-port", port, "-client"
+         ]
+      )
 
-   // Adding event handlers to the socket
-   socket.on("connect", function () { socket.write(rawTextG) })
-   socket.on("error", function (err) { console.log(err.toString()) })
-   socket.on("data", function (data) { taggedText = data.toString() })
-   // The NER server (ie. the other end) sends FIN signal after each write request, thereby closes 
-   // the socket at its end. So I am using the 'end' event to process the returned taggedText
-   socket.on("end", function (err) {
-      jsonEntities.push(socketNER.parser(taggedText, requiredEntityG))
-   })
+      client.stdout.once("data", function (data) {
+         if (data.toString().search("Input some text") > -1) {
+            // Running Callback in fiber to make it sync aware
+            sync.fiber(function () {
+               callback(socketNER)
+            })
+         }
+      })
+   }
 
+   function tagIt(rawText, reqEntity, cb) {
+      client.stdin.write(rawText)
+      client.stdout.once("data", function (data) {
+         taggedText = data.toString()
+         // Synchronize module follows (err, data) format for cb.
+         // Trim() is necessary to avoid leading and follwing 
+         // line breaks.
+         cb(null, socketNER.parser(taggedText.trim(), reqEntity))
+      })
+   }
 
    var socketNER = {}
-      
-   socketNER.getEntities = function (rawText, requiredEntity) {
-      // taking a copy of requiredEntity preference
-      requiredEntityG = requiredEntity
-      // replacing line breaks with spaces and adding two line breaks at the end
-      // for an unknown reason. May be, it relates to how a request should be sent through socket.
-      rawTextG = rawText.replace(/[\r\n\f\t\v]/g, " ") + "\n\n"
-      // Reconnecting on each getEntities function call, that's neccessary as the NER sever closes
-      // the socket after each such call.
-      socket.connect(port)       
-      // deasync would won't halt the Event Loop ie. it would also next events to be met
-      // but at the same time, it won't exit this function untill jsonEntities comes back.
-      deasync.loopWhile(function () { return (jsonEntities.length !== 1) })
-      return jsonEntities.pop()  // ** or jsonEntities.shift()
+
+   socketNER.getEntities = function (rawText, reqEntity) {
+      rawText = rawText.replace(/[\r\n\f\t\v]/g, " ") + "\n"      
+      return sync.await(tagIt(rawText, reqEntity, sync.defer()))      
    }
@@ -75,3 +74,3 @@
    socketNER.close = function () {
-      socket.end()
+      client.kill()
       server.kill()
@@ -83,5 +82,4 @@ }
    socketNER.parser = function (taggedText, requiredEntity) {
-      var matches, entities = {} //return value of parser function
-      // Change the regex scope according to user's Entitry requirements
-      // Please always pass the requiredEntity in Upper case as NER uses upper cased Tags
+      var matches, entities = {} // return value of parser function
+      requiredEntity = requiredEntity.toUpperCase()
       var re = requiredEntity ? new RegExp(["<(",requiredEntity,"?)>(.*?)<\/",requiredEntity,"?>"].join(""), "g") 
@@ -101,5 +99,4 @@ : /<([A-Z]+?)>(.*?)<\/[A-Z]+?>/g
    }
-
 }
 
 module.exports = SocketNER

test.js

@@ -29,3 +29,3 @@ // Terminal command to run the test file :- node test.js
 
-var nerServer = require("ner-node")
+var nerServer = require("./SocketNER.js")
 nerServer(8080, null , "./StanfordNER/", function (ner) {
@@ -32,0 +32,0 @@ var jsonEntities = ner.getEntities(rawText, "")

test2.js

 // Terminal command to run the test file :- node test2.js
 
-var rawText = 
-[
+var rawText = [
 	'Steven Paul Jobs (/\ˈdʒɒbz/; February 24, 1955 – October 5,',
@@ -33,3 +32,3 @@ '2011) was an American technology entrepreneur, visionary and',
 
-var nerServer = require("ner-node")
+var nerServer = require("./SocketNER.js")
 
@@ -36,0 +35,0 @@ nerServer(8080, null , "./StanfordNER/", function (ner) {

Fixed alerts

Network access

Supply chain risk

This module accesses the network.

Found 1 instance in 1 package

Improved metrics

Number of package files

7

increased by40%

Number of lines in readme file

44

increased by22.22%

Number of medium supply chain risk alerts

1

decreased by-50%

Worsened metrics

Total package byte prevSize

14469

decreased by-4.12%

Lines of code

260

decreased by-0.76%

Dependency changes

• + Addedsynchronize@^0.9.15

• + Addedfibers@1.0.15(transitive)

• + Addedsynchronize@0.9.15(transitive)

• - Removeddeasync@^0.1.4

• - Removedbindings@1.5.0(transitive)

• - Removeddeasync@0.1.30(transitive)

• - Removedfile-uri-to-path@1.0.0(transitive)

• - Removednode-addon-api@1.7.2(transitive)