svn merge -r113780:113785

svn+ssh://svn.lindenlab.com/svn/linden/branches/moss/gst3-t113732

QAR-1333 linux gstreamer compatibility improvements and ADD_BUILD_TEST
improvements - combo merge

3 files changed, 283 insertions, 0 deletions

diff --git a/indra/llmessage/tests/commtest.h b/indra/llmessage/tests/commtest.h
new file mode 100644
index 0000000000..7360230451
--- /dev/null
+++ b/indra/llmessage/tests/commtest.h
@@ -0,0 +1,60 @@
+/**
+ * @file   commtest.h
+ * @author Nat Goodspeed
+ * @date   2009-01-09
+ * @brief  
+ * 
+ * $LicenseInfo:firstyear=2009&license=viewergpl$
+ * Copyright (c) 2009, Linden Research, Inc.
+ * $/LicenseInfo$
+ */
+
+#if ! defined(LL_COMMTEST_H)
+#define LL_COMMTEST_H
+
+#include "networkio.h"
+#include "llevents.h"
+#include "llsd.h"
+#include "llhost.h"
+#include "stringize.h"
+#include <string>
+
+/**
+ * This struct is shared by a couple of standalone comm tests (ADD_COMM_BUILD_TEST).
+ */
+struct commtest_data
+{
+    NetworkIO& netio;
+    LLEventPumps& pumps;
+    LLEventStream replyPump, errorPump;
+    LLSD result;
+    bool success;
+    LLHost host;
+    std::string server;
+
+    commtest_data():
+        netio(NetworkIO::instance()),
+        pumps(LLEventPumps::instance()),
+        replyPump("reply"),
+        errorPump("error"),
+        success(false),
+        host("127.0.0.1", 8000),
+        server(STRINGIZE("http://" << host.getString() << "/"))
+    {
+        replyPump.listen("self", boost::bind(&commtest_data::outcome, this, _1, true));
+        errorPump.listen("self", boost::bind(&commtest_data::outcome, this, _1, false));
+    }
+
+    bool outcome(const LLSD& _result, bool _success)
+    {
+//      std::cout << "commtest_data::outcome(" << _result << ", " << _success << ")\n";
+        result = _result;
+        success = _success;
+        // Break the wait loop in NetworkIO::pump(), otherwise devs get
+        // irritated at making the big monolithic test executable take longer
+        pumps.obtain("done").post(success);
+        return false;
+    }
+};
+
+#endif /* ! defined(LL_COMMTEST_H) */
diff --git a/indra/llmessage/tests/networkio.h b/indra/llmessage/tests/networkio.h
new file mode 100644
index 0000000000..11c5cc07fc
--- /dev/null
+++ b/indra/llmessage/tests/networkio.h
@@ -0,0 +1,93 @@
+/**
+ * @file   networkio.h
+ * @author Nat Goodspeed
+ * @date   2009-01-09
+ * @brief  
+ * 
+ * $LicenseInfo:firstyear=2009&license=viewergpl$
+ * Copyright (c) 2009, Linden Research, Inc.
+ * $/LicenseInfo$
+ */
+
+#if ! defined(LL_NETWORKIO_H)
+#define LL_NETWORKIO_H
+
+#include "llmemory.h"               // LLSingleton
+#include "llapr.h"
+#include "llares.h"
+#include "llpumpio.h"
+#include "llhttpclient.h"
+
+/*****************************************************************************
+*   NetworkIO
+*****************************************************************************/
+// Doing this initialization in a class constructor makes sense. But we don't
+// want to redo it for each different test. Nor do we want to do it at static-
+// init time. Use the lazy, on-demand initialization we get from LLSingleton.
+class NetworkIO: public LLSingleton<NetworkIO>
+{
+public:
+    NetworkIO():
+        mServicePump(NULL),
+        mDone(false)
+    {
+        ll_init_apr();
+        if (! gAPRPoolp)
+        {
+            throw std::runtime_error("Can't initialize APR");
+        }
+
+        // Create IO Pump to use for HTTP Requests.
+        mServicePump = new LLPumpIO(gAPRPoolp);
+        LLHTTPClient::setPump(*mServicePump);
+        if (ll_init_ares() == NULL || !gAres->isInitialized())
+        {
+            throw std::runtime_error("Can't start DNS resolver");
+        }
+
+        // You can interrupt pump() without waiting the full timeout duration
+        // by posting an event to the LLEventPump named "done".
+        LLEventPumps::instance().obtain("done").listen("self",
+                                                       boost::bind(&NetworkIO::done, this, _1));
+    }
+
+    bool pump(F32 timeout=10)
+    {
+        // Reset the done flag so we don't pop out prematurely
+        mDone = false;
+        // Evidently the IO structures underlying LLHTTPClient need to be
+        // "pumped". Do some stuff normally performed in the viewer's main
+        // loop.
+        LLTimer timer;
+        while (timer.getElapsedTimeF32() < timeout)
+        {
+            if (mDone)
+            {
+//              std::cout << "NetworkIO::pump(" << timeout << "): breaking loop after "
+//                        << timer.getElapsedTimeF32() << " seconds\n";
+                return true;
+            }
+            pumpOnce();
+        }
+        return false;
+    }
+
+    void pumpOnce()
+    {
+        gAres->process();
+        mServicePump->pump();
+        mServicePump->callback();
+    }
+
+    bool done(const LLSD&)
+    {
+        mDone = true;
+        return false;
+    }
+
+private:
+    LLPumpIO* mServicePump;
+    bool mDone;
+};
+
+#endif /* ! defined(LL_NETWORKIO_H) */
diff --git a/indra/llmessage/tests/test_llsdmessage_peer.py b/indra/llmessage/tests/test_llsdmessage_peer.py
new file mode 100644
index 0000000000..e62f20912b
--- /dev/null
+++ b/indra/llmessage/tests/test_llsdmessage_peer.py
@@ -0,0 +1,130 @@
+#!/usr/bin/python
+"""\
+@file   test_llsdmessage_peer.py
+@author Nat Goodspeed
+@date   2008-10-09
+@brief  This script asynchronously runs the executable (with args) specified on
+        the command line, returning its result code. While that executable is
+        running, we provide dummy local services for use by C++ tests.
+
+$LicenseInfo:firstyear=2008&license=viewergpl$
+Copyright (c) 2008, Linden Research, Inc.
+$/LicenseInfo$
+"""
+
+import os
+import sys
+from threading import Thread
+from BaseHTTPServer import HTTPServer, BaseHTTPRequestHandler
+mydir = os.path.dirname(__file__)       # expected to be .../indra/llmessage/tests/
+sys.path.insert(0, os.path.join(mydir, os.pardir, os.pardir, "lib", "python"))
+from indra.util.fastest_elementtree import parse as xml_parse
+from indra.base import llsd
+
+def debug(*args):
+    sys.stdout.writelines(args)
+    sys.stdout.flush()
+# comment out the line below to enable debug output
+debug = lambda *args: None
+
+class TestHTTPRequestHandler(BaseHTTPRequestHandler):
+    """This subclass of BaseHTTPRequestHandler is to receive and echo
+    LLSD-flavored messages sent by the C++ LLHTTPClient.
+    """
+    def read(self):
+        # The following logic is adapted from the library module
+        # SimpleXMLRPCServer.py.
+        # Get arguments by reading body of request.
+        # We read this in chunks to avoid straining
+        # socket.read(); around the 10 or 15Mb mark, some platforms
+        # begin to have problems (bug #792570).
+        try:
+            size_remaining = int(self.headers["content-length"])
+        except (KeyError, ValueError):
+            return ""
+        max_chunk_size = 10*1024*1024
+        L = []
+        while size_remaining:
+            chunk_size = min(size_remaining, max_chunk_size)
+            chunk = self.rfile.read(chunk_size)
+            L.append(chunk)
+            size_remaining -= len(chunk)
+        return ''.join(L)
+        # end of swiped read() logic
+
+    def read_xml(self):
+        # This approach reads the entire POST data into memory first
+        return llsd.parse(self.read())
+##         # This approach attempts to stream in the LLSD XML from self.rfile,
+##         # assuming that the underlying XML parser reads its input file
+##         # incrementally. Unfortunately I haven't been able to make it work.
+##         tree = xml_parse(self.rfile)
+##         debug("Finished raw parse\n")
+##         debug("parsed XML tree %s\n" % tree)
+##         debug("parsed root node %s\n" % tree.getroot())
+##         debug("root node tag %s\n" % tree.getroot().tag)
+##         return llsd.to_python(tree.getroot())
+
+    def do_GET(self):
+        # Of course, don't attempt to read data.
+        self.answer(dict(reply="success", status=500,
+                         reason="Your GET operation requested failure"))
+
+    def do_POST(self):
+        # Read the provided POST data.
+        self.answer(self.read_xml())
+
+    def answer(self, data):
+        if "fail" not in self.path:
+            response = llsd.format_xml(data.get("reply", llsd.LLSD("success")))
+            self.send_response(200)
+            self.send_header("Content-type", "application/llsd+xml")
+            self.send_header("Content-Length", str(len(response)))
+            self.end_headers()
+            self.wfile.write(response)
+        else:                           # fail requested
+            status = data.get("status", 500)
+            reason = data.get("reason",
+                               self.responses.get(status,
+                                                  ("fail requested",
+                                                   "Your request specified failure status %s "
+                                                   "without providing a reason" % status))[1])
+            self.send_error(status, reason)
+
+    def log_request(self, code, size=None):
+        # For present purposes, we don't want the request splattered onto
+        # stderr, as it would upset devs watching the test run
+        pass
+
+    def log_error(self, format, *args):
+        # Suppress error output as well
+        pass
+
+class TestHTTPServer(Thread):
+    def run(self):
+        httpd = HTTPServer(('127.0.0.1', 8000), TestHTTPRequestHandler)
+        debug("Starting HTTP server...\n")
+        httpd.serve_forever()
+
+def main(*args):
+    # Start HTTP server thread. Note that this and all other comm server
+    # threads should be daemon threads: we'll let them run "forever,"
+    # confident that the whole process will terminate when the main thread
+    # terminates, which will be when the test executable child process
+    # terminates.
+    httpThread = TestHTTPServer(name="httpd")
+    httpThread.setDaemon(True)
+    httpThread.start()
+    # choice of os.spawnv():
+    # - [v vs. l] pass a list of args vs. individual arguments,
+    # - [no p] don't use the PATH because we specifically want to invoke the
+    #   executable passed as our first arg,
+    # - [no e] child should inherit this process's environment.
+    debug("Running %s...\n" % (" ".join(args)))
+    sys.stdout.flush()
+    rc = os.spawnv(os.P_WAIT, args[0], args)
+    debug("%s returned %s\n" % (args[0], rc))
+    return rc
+
+if __name__ == "__main__":
+    sys.exit(main(*sys.argv[1:]))