Add FridaScript.post(message, data) for sending messages to scripts

[jpovixwm]

Not sure about the purpose of batchId, it seems to work OK with a hardcoded zero.
The zero length buffer is needed, because if you pass null, it fails somewhere further down due to the marshaller expecting an array:

C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:58
      for (var i = 0; i < data.length; ++i)
                               ^

TypeError: Cannot read properties of null (reading 'length')
    at write (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:58:32)
    at writeStruct (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:32:5)
    at write (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:42:7)
    at write (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:59:9)
    at writeStruct (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:32:5)
    at marshall (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\marshall.js:18:13)
    at Object.marshallMessage [as marshall] (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\message.js:86:16)
    at self.message (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\index.js:68:28)
    at bus.<anonymous> (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\bus.js:43:21)
    at bus.invoke (C:\github\httptoolkit\frida-js\node_modules\@httptoolkit\dbus-native\lib\promisify.js:12:16)

Lastly, the test suite is going to fail for this PR because it will fetch Frida v17, but frida-js only works with v16.

[CLAassistant]

All committers have signed the CLA.

[pimterry]

Thanks, this looks great! Nicely done.

Lastly, the test suite is going to fail for this PR because it will fetch Frida v17, but frida-js only works with v16.

Indeed they have failed. Have you already looked at this yourself already? If you have a fix for it, I'd love to merge that first rather than merging this with the tests failing. Or if not or you don't have time, don't worry, I can take a look myself and then merge this on top.

[jpovixwm]

@pimterry Locally, I made these changes to get the tests to work:

diff --git a/test/download-frida.spec.ts b/test/download-frida.spec.ts
index 9a56ede..ef4c0b1 100644
--- a/test/download-frida.spec.ts
+++ b/test/download-frida.spec.ts
@@ -26,8 +26,8 @@ if (isNode) {
             );
 
             await new Promise((resolve, reject) => {
-                writeStream.on('finish', resolve);
-                writeStream.on('error', reject);
+                writeStream.on('finish', () => {resolve(undefined)});
+                writeStream.on('error', () => {reject(undefined)});
             });
 
             const fileStat = await fs.promises.stat(tmpFile.path);
@@ -88,8 +88,8 @@ if (isNode) {
 
             const streamResult: any = await new Promise((resolve, reject) => {
                 fridaServerStream.on('error', reject);
-                writeStream.on('error', reject);
-                writeStream.on('finish', resolve);
+                writeStream.on('error', () => {reject(undefined)});
+                writeStream.on('finish', () => {resolve(undefined)});
             });
 
             // Stream should complete without errors:
diff --git a/test/download-test-frida.ts b/test/download-test-frida.ts
index b959392..fed2f01 100644
--- a/test/download-test-frida.ts
+++ b/test/download-test-frida.ts
@@ -26,7 +26,7 @@ async function setUpLocalEnv() {
         : null;
 
     try {
-        const latestServerDetails = await getFridaReleaseDetails('latest', process.env.GITHUB_TOKEN);
+        const latestServerDetails = await getFridaReleaseDetails('16.7.19', process.env.GITHUB_TOKEN);
         const latestServerVersion: string = latestServerDetails.tag_name;
 
         if (!serverExists || semver.gt(latestServerVersion, currentServerVersion)) {

If you think that's acceptable, I can make another PR. (Or you could just commit that on your own, as it's not all that much changes).
Though ideally, I think the tests should grab the most recent 16.x release rather than hardcoding a specific release like that. On the other hand, I took a brief look at Frida's past releases, and it looks like they don't really make new releases for previous major versions.