cleanup

removed the "raw" subfolder for holding html

html is now expected at project root

Author: avindra

.gitignore

@@ -1,4 +1,4 @@
-raw/
+*.html
 node_modules
 
 *.pdf

README.md

@@ -16,7 +16,7 @@ Hence, this project exists and can be used create visualizations or progamatical
 
 1. yarn or npm install the dependencies
 2. view `config.js`              # Categories are limited to a default set. 
-3. `node save_data.js`           # Download pages containing raw HTML (seed files)
+3. `node save_data.js`           # Download pages containing input HTML data (seed files)
 4. `node run.js > schema.json`   # Generate structured representation (JSON artifact) from the seed files
 
 ## Constraints

save_data.js

@@ -3,14 +3,15 @@ const http = require('http');
 const {createWriteStream, existsSync} = require('fs');
 
 const {categories, doc_base} = require('./config');
-const {normalize} = require('./util');
+const {normalize} = require('./src/util');
 
 for (const _category of categories) {
 	const category = normalize(_category);
-	const targetFile = `./raw/${category}.html`;
-	const page = `${doc_base}/${category}.html`;
+	const fileName = `${category}.html`;
+	const targetFile = `./${fileName}`;
+	const page = `${doc_base}/${fileName}`;
 	if (existsSync(targetFile)) {
-		console.log(`skipped download cached entity [${category}] ${page}`);
+		console.log(` [${category}]: page in cache ${fileName} (skipped)`);
 		continue;
 	}
 

src/env.js

@@ -8,7 +8,7 @@ const {JSDOM} = require("jsdom");
  * in a browser-free environment.
  */
 const createDOM = (category) => {
-    const fData = readFileSync("./raw/" + category + ".html")
+    const fData = readFileSync("./" + category + ".html")
     const dom = new JSDOM(fData);
     return dom;
 }