1 files changed, 80 insertions, 0 deletions
diff --git a/html2json.js b/html2json.js
new file mode 100644
index 0000000..841149a
--- /dev/null
+++ b/html2json.js
@@ -0,0 +1,80 @@
+var cheerio = require('cheerio')
+var path = require('path')
+
+// Read the HTML report
+fs = require('fs')
+var htmlContent;
+try {
+    htmlContent = fs.readFileSync(process.argv[2], 'utf8');
+} catch(e) {
+    console.log('Error:', e.stack);
+}
+
+// Get the directory containing the results to make pathes relative to it later.
+report_directory = path.dirname(process.argv[2])
+
+const $ = cheerio.load(htmlContent)
+
+// Extract licenses and the number of occurences.
+var licenses = [];
+$('div.summary div.row').children().first().find('ul li').each(function(i, doc) {
+    tmp = $(this).text();
+    matches = tmp.match(/^([0-9]+) ((\s|\S)*)/m)
+    licenses.push({
+        count: parseInt(matches[1], 10),
+        name: matches[2]
+    })
+})
+
+// Extract dependencies info.
+var dependencies = []
+$('div.dependencies div').each(function(i, doc) {
+    // Get license name.
+    license = $(this).find('blockquote p').text().trim();
+    license = license.split("\n")[0];
+
+    // Get URL.
+    license_url = $(this).find('blockquote p a[href]').attr('href');
+
+    // Get dependency name.
+    dependency_name = $(this).find('h2').text().trim();
+    dependency_name = dependency_name.split("\n")[0];
+
+    // Get dependency URL.
+    dependency_url = $(this).find('h2 a[href]').attr('href');
+
+    // Get dependency description.
+    dependency_description = $(this).find('dl').first().next().text().trim();
+
+    // Get dependency location relative to the project root path
+    dependency_pathes = []
+    $(this).find('dl').first().find('dd').each(function(i, doc) {
+        dependency_path = path.relative(report_directory, $(this).text().trim());
+
+        // Whitespace path means current directory
+        if (!dependency_path) {
+            dependency_path = ".";
+        }
+
+        dependency_pathes.push(dependency_path);
+    })
+    dependencies.push({
+        license: {
+            name: license,
+            url: license_url
+        },
+        dependency: {
+            name: dependency_name,
+            url: dependency_url,
+            description: dependency_description,
+            pathes: dependency_pathes
+        }
+    })
+})
+
+console.log(JSON.stringify({
+    licenses: licenses,
+    dependencies: dependencies}, null, 4))
+
+
+