WIP: plagiarism detection

src/collect-code-from-firebase-dump.js

@@ -0,0 +1,26 @@
+// Extracts students' code submissions from a single JSON firebase dump
+// and returns them as a JSON array of `file` entries,
+// for evaluate-plagiarism-post.js to consume.
+
+const fs = require('fs');
+
+const SUBMISSIONS_DUMP_FILE = './exam-data/eemi-own-exam-export.json';
+const EXERCISE_IDENTIFIER = 'code11';
+
+// returns [ { name, content, type } ]
+async function parse(file) {
+  const { submissions } = require(`./${file}`); // await fs.promises.readFile(`${path}/${dirent.name}`);
+  return Object.keys(submissions).map(name => {
+    const submission = submissions[name];
+    return {
+      name,
+      contents: submission[EXERCISE_IDENTIFIER], // code
+      type: 'js'
+    };
+  });
+}
+
+console.warn(`Parsing code submissions from ${SUBMISSIONS_DUMP_FILE}...`)
+parse(SUBMISSIONS_DUMP_FILE)
+  .then(files => console.log(JSON.stringify(files, null, 2)))
+  .catch(console.error);

src/collect-code-from-json-files.js

@@ -0,0 +1,28 @@
+// Extracts students' code submissions from ./students/*.json
+// and returns them as a JSON array of `file` entries,
+// for evaluate-plagiarism-post.js to consume.
+
+const fs = require('fs');
+
+const SUBMISSIONS_DIRECTORY = './students';
+const EXERCISE_IDENTIFIER = 'code11';
+
+// returns [ { name, content, type } ]
+async function ls(path) {
+  const dir = await fs.promises.opendir(path);
+  const files = []; // structure: [ { name, content, type } ]
+  for await (const dirent of dir) {
+    const submission = await fs.promises.readFile(`${path}/${dirent.name}`);
+    files.push({
+      name: `${dirent.name.split('@')[0]}.js`, // extract student name from email address, and add .js file extension
+      contents: JSON.parse(submission)[EXERCISE_IDENTIFIER], // code
+      type: 'js'
+    });
+  }
+  return files;
+}
+
+console.warn(`Parsing code submissions from ${SUBMISSIONS_DIRECTORY}...`)
+ls(SUBMISSIONS_DIRECTORY)
+  .then(files => console.log(JSON.stringify(files, null, 2)))
+  .catch(console.error);

src/evaluate-plagiarism-post.js

@@ -0,0 +1,45 @@
+// Reads students' code submissions from ./student-code-entries.json
+// (generated by collect-code-from-*.js from ./students/*.json)
+// and POSTs them to Zino's plagiarism detection API.
+
+const fs = require('fs');
+const https = require('https');
+
+const ENTRIES_FILE = './student-code-entries.json';
+const SUBMIT_URL = 'https://plagiat.krugazor.eu/api/submit';
+
+// test:
+// $ curl -v -X POST --header "Content-Type: application/json" --header "P-Token: ${PLAGIAT_TOKEN}" --data "{\"files\":[]}" https://plagiat.krugazor.eu/api/submit
+// $ curl -v -X POST --header "Content-Type: application/json" --header "P-Token: ${PLAGIAT_TOKEN}" --data '{"files":[{"name":"1.js","contents":"console.log(\"a\");"},{"name":"2.js","contents":"console.log(\"b\");"}]}' https://plagiat.krugazor.eu/api/submit
+
+const post = (body) => new Promise((resolve, reject) => {
+  const data = JSON.stringify(body);
+  const options = {
+    method: 'POST',
+    hostname: 'plagiat.krugazor.eu',
+    port: 443,
+    path: '/api/submit',
+    headers: {
+      'Content-Type': 'application/json',
+      'Content-Length': data.length,
+      'P-Token': process.env.PLAGIAT_TOKEN // secret
+    }
+  };
+  const req = https.request(options, res => {
+    console.log(`statusCode: ${res.statusCode}`);
+    res.on('data', responseData => resolve(responseData.toString('utf8')));
+  })
+  req.on('error', reject);
+  req.write(data);
+  req.end();
+});
+
+async function submitEntries(entriesFile) {
+  const files = await fs.promises.readFile(entriesFile, { encoding: 'utf8' });
+  return await post({ files });
+}
+
+console.warn(`Sending code submissions from ${ENTRIES_FILE} to ${SUBMIT_URL}...`)
+submitEntries(ENTRIES_FILE)
+  .then(result => console.log(result))
+  .catch(console.error);