calais-entity-extractor - npm Package Compare versions

Comparing version 1.1.1 to 1.2.0

example.js

@@ -5,3 +5,3 @@
 
-var Calais = require('calais-entity-extractor').Calais;
+var Calais = require('./lib/calais-entity-extractor.js').Calais;
 
@@ -18,2 +18,22 @@ //You can enter options as the second parameter.
 
+// Example entity lookup
+calais.lookup("1-4295861160", function(result, err) {
+    if (err)
+        return console.log("Error: " + err);
+
+    console.log("Result: " + util.inspect(result, false, null));
+});
+
+
+// Example entity searching
+calais.search("IBM", function(result, err) {
+    if (err)
+        return console.log("Error: " + err);
+    console.log("Result: " + util.inspect(result, false, null));
+});
+
+
+/*
+    //Example text entity tagging functionality.
+
 calais.extractFromText(function(result, err) {     //perform the request
@@ -52,2 +72,3 @@ if (err) {
     });
-});
+});
+*/

lib/calais-entity-extractor.js

@@ -21,2 +21,5 @@ var request = require('request');
 
+        this.entitySearchUrl = "https://api.thomsonreuters.com:443/permid/search";
+        this.entityLookupUrl = "https://permid.org:443/";
+
         this._setDefaultOptions(options);
@@ -44,2 +47,4 @@ },
         var tags = [ ];
+        var industries = [ ];
+
         for(var i in result) {
@@ -86,2 +91,11 @@ var p = result[i];
                         }
+                    } else if (p._typeGroup == 'industry') {
+                        var industry = {
+                            'name' : p['name'],
+                            'rcscode' : p['rcscode'],
+                            'trbccode' : p['trbccode'],
+                            'permid' : p['permid'],
+                            'relevance' : p['relevance']
+                        };
+                        industries.push(industry);
                     }
@@ -92,5 +106,167 @@ }
 
-        return {'entities' : entities, 'tags' : tags };
+        return {'entities' : entities, 'tags' : tags, 'industries' : industries };
     },
 
+    _parseCalaisSearchResult: function(data) {
+        var organizations = [];
+        var instruments = [];
+        var quotes = [];
+
+        if (!data.hasOwnProperty('result'))
+            return { 'organizations' : organizations, 'instruments' : instruments, 'quotes' : quotes };
+
+        if (data.result.hasOwnProperty('organizations') && data.result.organizations.hasOwnProperty('entities')) {
+            var orgs = data.result.organizations.entities;
+            for (var i = 0, len = orgs.length; i < len; i++) {
+                var entity = { };
+
+                if (orgs[i]["@id"]) {
+                    entity['id'] = orgs[i]["@id"];
+                    if(entity['id'].indexOf("/") != -1)
+                        entity['id'] = entity['id'].substring(entity['id'].lastIndexOf("/")+1);
+                }
+                if (orgs[i].hasOwnProperty("organizationName"))
+                    entity['name'] = orgs[i].organizationName;
+                if (orgs[i].hasOwnProperty("primaryTicker"))
+                    entity['ticker'] = orgs[i].primaryTicker;
+                if (orgs[i].hasOwnProperty("orgSubtype"))
+                    entity['type'] = orgs[i].orgSubtype;
+                if (orgs[i].hasOwnProperty("hasHoldingClassification"))
+                    entity['public'] = true;
+                else
+                    entity['public'] = false;
+                if (orgs[i].hasOwnProperty("hasURL"))
+                    entity['url'] = orgs[i].hasURL;
+
+                organizations.push(entity);
+            }
+        }
+
+        if (data.result.hasOwnProperty('instruments') && data.result.instruments.hasOwnProperty('entities')) {
+            var ins = data.result.instruments.entities;
+            for (var i = 0, len = ins.length; i < len; i++) {
+                var entity = { };
+                if (ins[i].hasOwnProperty("@id")) {
+                    entity['id'] = ins[i]["@id"];
+                    if(entity['id'].indexOf("/") != -1)
+                        entity['id'] = entity['id'].substring(entity['id'].lastIndexOf("/")+1);
+                }
+                if (ins[i].hasOwnProperty("hasName"))
+                    entity['name'] = ins[i].hasName;
+                if (ins[i].hasOwnProperty("assetClass"))
+                    entity['assetClass'] = ins[i].assetClass;
+                if (ins[i].hasOwnProperty("isIssuedByName"))
+                    entity['issuerName'] = ins[i].isIssuedByName;
+                if (ins[i].hasOwnProperty("isIssuedBy")) {
+                    entity['issuerId'] = ins[i].isIssuedBy;
+                    if(entity.issuerId.indexOf("/") != -1)
+                        entity.issuerId = entity.issuerId.substring(entity.issuerId.lastIndexOf("/")+1);
+                }
+                if (ins[i].hasPrimaryQuote) {
+                    entity['primaryQuoteId'] = ins[i].hasPrimaryQuote;
+                    if (entity.primaryQuoteId.indexOf("/") != -1)
+                        entity.primaryQuoteId = entity.primaryQuoteId.substring(entity.primaryQuoteId.lastIndexOf("/")+1);
+                }
+                if (ins[i].hasOwnProperty("primaryTicker"))
+                    entity['primaryTicker'] = ins[i].primaryTicker;
+
+                instruments.push(entity);
+            }
+        }
+
+        if (data.result.hasOwnProperty('quotes') && data.result.quotes.hasOwnProperty('entities')) {
+            var quo = data.result.quotes.entities;
+            for (var i = 0, len = quo.length; i < len; i++) {
+                var entity = { };
+
+                if (quo[i]["@id"]) {
+                    entity['id'] = ins[i]["@id"];
+                    if(entity['id'].indexOf("/") != -1)
+                        entity['id'] = entity['id'].substring(entity['id'].lastIndexOf("/")+1);
+                }
+                if (quo[i].hasName)
+                    entity['name'] = quo[i].hasName;
+                if (quo[i].assetClass)
+                    entity['class'] = quo[i].assetClass;
+                if (quo[i].isQuoteOfInstrumentName)
+                    entity['instrumentName'] = quo[i].isQuoteOfInstrumentName;
+                if (quo[i].isIssuedByName)
+                    entity['issuer'] = quo[i].isIssuedByName;
+                if (quo[i].hasRIC)
+                    entity["ric"] = quo[i].hasRIC;
+                if (quo[i].hasMic)
+                    entity["mic"] = quo[i].hasMic;
+                if (quo[i].hasExchangeTicker)
+                    entity["ticker"] = quo[i].hasExchangeTicker;
+                if (quo[i].isQuoteOf) {
+                    entity["quoteId"] = quo[i].isQuoteOf;
+                    if (entity.quoteId.indexOf("/") != -1)
+                        entity.quoteId = entity.quoteId.substring(entity.quoteId.lastIndexOf("/")+1);
+                }
+
+                quotes.push(entity);
+            }
+        }
+
+        return { 'organizations' : organizations, 'instruments' : instruments, 'quotes' : quotes };
+    },
+
+
+    _parseCalaisLookupResult: function(data) {
+
+        var result = { };
+        if (data.hasOwnProperty("@id")) {
+            result['id'] = data["@id"];
+            if(result.id.indexOf("/") != -1)
+                result.id = result.id.substring(result.id.lastIndexOf("/")+1);
+        }
+
+        if (data.hasOwnProperty("@type")) {
+            result['type'] = data["@type"];
+            if(result.type.indexOf(":") != -1)
+                result.type = result.type.substring(result.type.lastIndexOf(":")+1);
+        }
+
+        if (data.hasOwnProperty("mdaas:HeadquartersAddress"))
+            result.headquarters = data["mdaas:HeadquartersAddress"];
+        if (data.hasOwnProperty("mdaas:RegisteredAddress"))
+            result.address = data["mdaas:RegisteredAddress"];
+        if (data.hasOwnProperty("hasOrganizationPrimaryQuote")) {
+            result.primaryQuoteId = data["hasOrganizationPrimaryQuote"];
+            if(result.primaryQuoteId.indexOf("/") != -1)
+                result.primaryQuoteId = result.primaryQuoteId.substring(result.primaryQuoteId.lastIndexOf("/")+1);
+        }
+        if (data.hasOwnProperty("hasPrimaryInstrument")) {
+            result.primaryInstrumentId = data["hasPrimaryInstrument"];
+            if(result.primaryInstrumentId.indexOf("/") != -1)
+                result.primaryInstrumentId = result.primaryInstrumentId.substring(result.primaryInstrumentId.lastIndexOf("/")+1);
+        }
+        if (data.hasOwnProperty("hasActivityStatus")) {
+            result.status = data["hasActivityStatus"];
+            if(result.status.indexOf("status") != -1)
+                result.status = result.status.substring(result.status.lastIndexOf("status")+6).toLowerCase();
+        }
+        if (data.hasOwnProperty("tr-org:hasHeadquartersPhoneNumber"))
+            result.headquartersPhoneNumber = data["tr-org:hasHeadquartersPhoneNumber"];
+        if (data.hasOwnProperty("hasHoldingClassification"))
+            result.public = true;
+        if (data.hasOwnProperty("hasIPODate"))
+            result.ipoDate = data["hasIPODate"];
+        if (data.hasOwnProperty("hasLatestOrganizationFoundedDate"))
+            result.founded = data["hasLatestOrganizationFoundedDate"];
+        if (data.hasOwnProperty("tr-org:hasRegisteredPhoneNumber"))
+            result.registeredPhoneNumber = data["tr-org:hasRegisteredPhoneNumber"];
+        if (data.hasOwnProperty("isIncorporatedIn"))
+            result.incorporatedLocation = data["isIncorporatedIn"];
+        if (data.hasOwnProperty("isDomiciledIn"))
+            result.domiciledLocation = data["isDomiciledIn"];
+        if (data.hasOwnProperty("hasURL"))
+            result.url = data["hasURL"];
+        if (data.hasOwnProperty("vcard:organization-name"))
+            result.name = data["vcard:organization-name"];
+
+        return result;
+    },
+
+
     set: function (key, value) {
@@ -240,2 +416,95 @@ this.options[key] = value;
 
+    },
+
+    search: function(query, cb) {
+        var calais = this;
+
+        if (!calais.validateOptions())
+            return cb({}, 'Bad options');
+
+        //Make sure we were given a URL.
+        if (this._undefinedOrNull(query) || typeof query != 'string' || query.length == 0)
+            return cb({}, 'Invalid query given.');
+
+
+        var compositeUrl = calais.entitySearchUrl + "?q=" + query;
+
+        //We can upload the html directly to Calais if we set the contentType as text/html
+        var params = {
+            'Host'                   : calais.options.apiHost,
+            'x-ag-access-token'      : calais.apiKey,
+            'x-calais-language'      : calais.options.language,
+            'Accept'                 : 'application/json',
+            'OutputFormat'           : 'application/json'
+        };
+
+        var options = {
+            uri    : compositeUrl,
+            method : 'GET',
+            headers: params
+        };
+
+
+        request(options, function(error, response, calaisData) {
+
+            if (error)
+                return cb({}, error);
+
+            if (response === undefined) {
+                return cb({}, 'Undefined Calais response');
+            } else if (response.statusCode === 200) {
+
+                // parse to a Javascript object if requested
+                var result = JSON.parse(calaisData);
+                result = (typeof result === 'string') ? JSON.parse(result) : result;
+
+                var parsedResult = calais._parseCalaisSearchResult(result);
+
+                return cb(parsedResult, calais.errors);
+            } else
+                return cb({}, 'Request error: ' + (typeof response === 'string' ? response : JSON.stringify(response)));
+
+
+        });
+    },
+
+    lookup: function(identifier, cb) {
+        var calais = this;
+
+        if (!calais.validateOptions())
+            return cb({}, 'Bad options');
+
+        //Make sure we were given a URL.
+        if (this._undefinedOrNull(identifier) || typeof identifier != 'string' || identifier.length == 0)
+            return cb({}, 'Invalid identifier given.');
+
+
+        var compositeUrl = calais.entityLookupUrl + identifier + "?format=json-ld&access-token=" + calais.apiKey;
+
+        var options = {
+            uri    : compositeUrl,
+            method : 'GET'
+        };
+
+        request(options, function(error, response, calaisData) {
+            if (error)
+                return cb({}, error);
+
+            if (response === undefined) {
+                return cb({}, 'Undefined Calais response');
+            } else if (response.statusCode === 200) {
+
+                // parse to a Javascript object if requested
+                var result = JSON.parse(calaisData);
+                result = (typeof result === 'string') ? JSON.parse(result) : result;
+
+                var parsedResult = calais._parseCalaisLookupResult(result);
+                return cb(parsedResult, calais.errors);
+            } else
+                return cb({}, 'Request error: ' + (typeof response === 'string' ? response : JSON.stringify(response)));
+
+
+        });
+
+
     }
@@ -242,0 +511,0 @@ };

package.json

 {
   "name": "calais-entity-extractor",
-  "version": "1.1.1",
+  "version": "1.2.0",
   "description": "Extract entities from text using Open Calais.",
@@ -5,0 +5,0 @@ "scripts": {

No alert changes

Improved metrics

Total package byte prevSize

32392

increased by58.3%

Lines of code

473

increased by97.91%

No dependency changes