linkeysearch: 2 step implementation; prepare scripts to run as Google Cloud Funcctions

1 files changed, 175 insertions, 103 deletions

diff --git a/javascript/keygen.js b/javascript/keygen.js
index 16e2490..39174b1 100644
--- a/javascript/keygen.js
+++ b/javascript/keygen.js
@@ -1,4 +1,19 @@
-// requirements
+/** keygen
+ * 
+ * this script constucts a linked-wordkeys structure
+ * -----------------------------------------------------------------------------
+ * 
+ * Contents:
+ * #1 Requirements
+ * #2.1 Personalized constants and parametres
+ * #2.2 Preloaded Data
+ * #3 Supporting functions
+ * #4 Output functions
+ * #5 Entry-point function main()
+ */
+
+// #1
+// REQUIREMENTS
 ////////////////////////////////////////////////////////////////////////////////
 
 const fs = require('fs');
@@ -7,17 +22,45 @@ const fetch = require('node-fetch');
 const { ECDH } = require('crypto');
 const {Storage} = require('@google-cloud/storage');     // Google Cloud Storage
 
-// (setup/personalize) constants
+
+
+// #2.1
+// SETUP PERSONALIZED CONSTANTS AND PARAMETRES
 ////////////////////////////////////////////////////////////////////////////////
 
+
+// get products FROM api/endpoint parametres
+const frequency_endpoint = 'https://storage.googleapis.com/pythia-files/uploads/json/freq.json';
+const products_endpoint = "https://emarket-laravel-dlqjpfxz5q-oa.a.run.app/api/v1/productsSearch";
+const request_settings = { method: "Get" };
+
+
 // google-storage parametres
 // --- -- -- - - -
 const projectId = 'pythia-251711';
 const keyFilename = '../auth/pythia-251711-047e3d5e6608.json';
 
 
+// temp files
+// need to be created prior to save-to-google-storage proccess
+// because upload_file() uploats an existing local file to G-Storage
+// --- -- -- - - -
+const temp_kw_file = '../results/keywords.json';
+const temp_prod_file = '../results/products.json';
+
+
+// Global Variables
+// -----------------------------------------------------------------------------
+
+// keyword links (word-links dictionary; array of objects)
+// -----------------------------------------------------------------------------
+var kwlinks_ = [];      ////////////////////////////// MAIN OUTPUT OF THE SCRIPT
+var prods_ = [];
+var fr_ =[];
+
 
-// preloaded data
+// #2.2
+// PRELOADED DATA
 ////////////////////////////////////////////////////////////////////////////////
 
 // any-character to keyboard-latin mapping
@@ -228,10 +271,19 @@ removeOriginals = 'κατά παρά υπό από μετά προς μας με
 removeOriginals.forEach( w => { removeList.push( kb_trans(w)); });
 
 
-// read frequency data
-let rawdata = fs.readFileSync('../results/freq.json');
-let fr_ = JSON.parse(rawdata);      // frequencies
-console.log('frequencies loaded');
+// depricated
+/// // read frequency (local) data
+/// let rawdata = fs.readFileSync('../results/freq.json');
+/// let fr_ = JSON.parse(rawdata);      // frequencies
+/// console.log('frequencies loaded');
+
+
+
+
+// #3
+// SUPPORTING FUNCTIONS
+////////////////////////////////////////////////////////////////////////////////
+
 
 /** freq
  * 
@@ -247,57 +299,15 @@ function freq_of(id) {
 }
 
 
-// get products from api/endpoint (in Json format)
-// --- -- -- - - -
-let url = "https://emarket-laravel-dlqjpfxz5q-oa.a.run.app/api/v1/productsSearch";
-
-let settings = { method: "Get" };
-
-fetch(url, settings)
-    .then(res => res.json())
-    .then((json) => {
-
-        let temp_file = '../results/keywords.json';
-        let temp_prod_file = '../results/products.json';
-
-        console.log('products data loaded from emarket api/endpoint');
-
-        // do something with JSON
-        var linkeys = do_proccess(json.data);
-
-        save_local(kwlinks_, temp_kw_file);
-        save_local(kwlinks_, temp_prod_file);
-
-        // upload_file('pythia-files', os.tmpdir()+'/keywords.json', 'uploads/orders/keywords.json');
-        upload_file('pythia-files', temp_kw_file, 'uploads/orders/emarket-keywords.json');
-        upload_file('pythia-files', temp_prod_file, 'uploads/orders/emarket-keywords.json');
-
-        // check
-        // echo 100 most frequent
-        let i = 0;
-        kwlinks_.forEach(rec => {
-            if (i<100) {
-                console.log(i, JSON.stringify(rec.w))
-            }
-            i++;
-        })
-});
-
-
-
-// keyword links (word-links dictionary; array of objects)
-// -----------------------------------------------------------------------------
-var kwlinks_ = [];      ////////////////////////////// MAIN OUTPUT OF THE SCRIPT
-var prods_ = [];
 
+/* ---
 // connect;
 // get records to proccess;
 // call main proccess function;
 // save dictionary;
 // end script;
 // -----------------------------------------------------------------------------
-
-exports.main = () => {
+var main = () => {
 
     con.connect(function(err) {
         // connect;
@@ -340,13 +350,13 @@ exports.main = () => {
         });
     });
 }
+--- */
 
 
-var unmatched = 0;
-
-// main proccess
+// proccess data
+// The MAIN proccess 
 // -----------------------------------------------------------------------------
-function do_proccess(obj) {   
+function extract_linked_keywords(obj) {   
     // console.log(JSON.stringify(obj, null, 2));
 
     obj.forEach( rec => {
@@ -362,8 +372,6 @@ function do_proccess(obj) {
                 w: description
             })
 
-            // if (fq == 0) console.log(pid, description);
-
             var keys = [];
             var words = description.split(' ');
 
@@ -418,52 +426,6 @@ function do_proccess(obj) {
 }
 
 
-
-// save proccesses
-////////////////////////////////////////////////////////////////////////////////
-
-// Save Local
-// -----------------------------------------------------------------------------
-function save_local(jsonArray, filePath) {
-    let fileStr = JSON.stringify(jsonArray);    // convert json to string
-    
-    // write string to file
-    fs.writeFileSync(filePath, fileStr, 'utf8', (err) => {
-        if (err) {
-            console.log("An error occured while writing keywords.json");
-            return console.log(err);
-        }
-        console.log("JSON file has been saved.");
-    });
-
-    return true;
-}
-
-
-
-// Save to Cloud Storage
-// -----------------------------------------------------------------------------
-
-
-async function upload_file( bucketName, srcFilePath, trgFilePath ) {
-    // Creates a client
-    const storage = new Storage({projectId, keyFilename});
-
-    try {
-        await storage.bucket(bucketName).upload(srcFilePath, {
-            destination: trgFilePath, 
-            gzip: true,     // serve compressed
-            metadata: {     // cache for 8 hours
-                cacheControl: 'public, max-age=60'      // production set: 28800
-            }
-        });
-        console.log(`${srcFilePath} uploaded to ${bucketName}`);
-    }
-    catch(err) {
-        console.error('ERROR:', err);
-    }
-}
-
 // functions for linking words in keywords dictionary
 ////////////////////////////////////////////////////////////////////////////////
 
@@ -620,3 +582,113 @@ function mark_link(lws, source) {
     else return source;
 }
 
+
+
+// #4
+// OUTPUT FUNCTIONS
+////////////////////////////////////////////////////////////////////////////////
+
+
+// Save Local
+// -----------------------------------------------------------------------------
+function save_local(jsonArray, filePath) {
+    let fileStr = JSON.stringify(jsonArray);    // convert json to string
+    
+    // write string to file
+    fs.writeFileSync(filePath, fileStr, 'utf8', (err) => {
+        if (err) {
+            console.log("An error occured while writing keywords.json");
+            return console.log(err);
+        }
+        // console.log("JSON file has been saved.");
+    });
+
+    return true;
+}
+
+
+
+// Save to Google Cloud Storage
+// -----------------------------------------------------------------------------
+async function upload_file( bucketName, srcFilePath, trgFilePath ) {
+    // Creates a client
+    const storage = new Storage({projectId, keyFilename});
+
+    try {
+        await storage.bucket(bucketName).upload(srcFilePath, {
+            destination: trgFilePath, 
+            gzip: true,     // serve compressed
+            metadata: {     // cache for 8 hours
+                cacheControl: 'public, max-age=60'      // production set: 28800
+            }
+        });
+        console.log(`${srcFilePath} uploaded to ${bucketName}`);
+    }
+
+    catch(err) {
+        console.error('ERROR:', err);
+    }
+
+}
+
+
+
+
+// #5
+// MAIN function (exposed function to run the whole proccess)
+////////////////////////////////////////////////////////////////////////////////
+
+
+/** main()
+ * 
+ * nodejs's exported function
+ * used as entry-point function (in case of Google Cloud function)
+ */
+
+// keep only 1 of next 2 lines
+// exports.main = () => {       // entry point functioon when google cloud function
+var main = () => {              // concise function when direct script
+
+    fetch(frequency_endpoint, request_settings)
+    .then(res => res.json())
+    .then((json) => {
+        fr_ = json;
+        console.log('frequencies loaded from CDN');
+
+        fetch(products_endpoint, request_settings)
+        .then(res => res.json())
+        .then((json) => {
+
+            console.log('products data loaded from emarket api/endpoint');
+
+            // proccess data
+            extract_linked_keywords(json.data);
+
+            save_local(kwlinks_, temp_kw_file);
+            save_local(kwlinks_, temp_prod_file);
+
+            // upload temp files to cloud
+            upload_file('pythia-files', temp_kw_file, 'uploads/json/emarket-keywords.json');
+            upload_file('pythia-files', temp_prod_file, 'uploads/json/emarket-products.json');
+
+            /// // check
+            /// // echo 100 most frequent
+            /// let i = 0;
+            /// kwlinks_.forEach(rec => {
+            ///     if (i<100) {
+            ///         console.log(i, JSON.stringify(rec.w))
+            ///     }
+            ///     i++;
+            /// })
+        });
+
+    });
+
+}
+
+
+// NOTE:
+// if running the script directry you need to call the main function
+// if running through google cloud-functions you do NOT need to call main()
+// (you need define main() as the entry-point function instead)
+main();
\ No newline at end of file