Handle auxiliary environment in bbot-worker

2 files changed, 189 insertions, 29 deletions

diff --git a/bbot/worker/worker.cxx b/bbot/worker/worker.cxx
index 8217c45..8388d03 100644
--- a/bbot/worker/worker.cxx
+++ b/bbot/worker/worker.cxx
@@ -1219,6 +1219,9 @@ upload_manifest (tracer& trace,
   }
 }
 
+static strings
+parse_auxiliary_environment (const string&, const char*); // See below.
+
 static const string worker_checksum ("5"); // Logic version.
 
 static int bbot::
@@ -2356,6 +2359,26 @@ build (size_t argc, const char* argv[])
       operation_result* pr (&add_result ("configure"));
       operation_result& r (*pr); // @@ TMP: Apple Clang 14.0.3 ICE
 
+      // If we have auxiliary environment, show it in the logs.
+      //
+      if (tm.auxiliary_environment)
+      {
+        strings es (parse_auxiliary_environment (*tm.auxiliary_environment,
+                                                 comment_begin));
+        if (!es.empty ())
+        {
+          for (const string& e: es)
+          {
+            r.log += e;
+            r.log += '\n';
+          }
+
+          // Add a trailing blank line to separate this from the rest.
+          //
+          r.log += '\n';
+        }
+      }
+
       // Noop, just for the log record.
       //
       change_wd (trace, &r.log, rwd);
@@ -5942,6 +5965,104 @@ build (size_t argc, const char* argv[])
   return 3;
 }
 
+// Parse the task_manifest::auxiliary_environment value into the list of
+// environment variable assignments as expected by the process API. Throw
+// invalid_argument if the auxiliary environment is invalid.
+//
+// If comment is not NULL, then add blank and comment lines prefixed with this
+// string (which is normally either '#' or 'rem'). This mode is used to print
+// the environment into the build log.
+//
+static strings
+parse_auxiliary_environment (const string& s, const char* comment = nullptr)
+{
+  strings r;
+
+  // Note: parse observing blanks.
+  //
+  for (size_t b (0), e (0), m (0), n (s.size ());
+       next_word (s, n, b, e, m, '\n', '\r'), b != n; )
+  {
+    string line (s, b, e - b);
+
+    if (trim (line).empty ()) // Blank.
+    {
+      if (comment != nullptr)
+        r.push_back (comment);
+
+      continue;
+    }
+
+    if (line.front () == '#') // Comment.
+    {
+      if (comment != nullptr)
+      {
+        line.erase (0, 1);
+        line.insert (0, comment);
+        r.push_back (move (line));
+      }
+
+      continue;
+    }
+
+    size_t p (line.find ('='));
+
+    if (p == string::npos)
+      throw invalid_argument ("missing '=' in '" + line + '\'');
+
+    string name (line, 0, p);
+
+    if (trim_right (name).empty () ||
+        name.find_first_not_of (
+          "ABCDEFGHIJKLMNOPQRSTUVWXYZ_0123456789") != string::npos)
+    {
+      throw invalid_argument ("invalid variable name '" + name + '\'');
+    }
+
+    // Disallow certain well-known environment variables.
+    //
+    if (name == "PATH"
+#if   defined(_WIN32)
+#elif defined(__APPLE__)
+        || name == "DYLD_LIBRARY_PATH"
+#else // Linux, FreeBSD, NetBSD, OpenBSD
+        || name == "LD_LIBRARY_PATH"
+#endif
+    )
+    {
+      throw invalid_argument ("disallowed variable name '" + name + '\'');
+    }
+
+    line.erase (0, p + 1); // Value.
+
+    // Note: we allow empty values.
+    //
+    if (!trim_left (line).empty ())
+    {
+      // Unquote.
+      //
+      char c (line.front ());
+      if (c == '"' || c == '\'')
+      {
+        if (line.size () == 1 || line.back () != c)
+          throw invalid_argument ("invalid quoted value '" + line + '\'');
+
+        line.pop_back ();
+        line.erase (0, 1);
+      }
+    }
+
+    // Reassemble.
+    //
+    line.insert (0, 1, '=');
+    line.insert (0, name);
+
+    r.push_back (move (line));
+  }
+
+  return r;
+}
+
 static int
 startup ()
 {
@@ -5951,11 +6072,13 @@ startup ()
   //
   // 1. Download the task manifest into the build directory (CWD).
   //
-  // 2. Parse it and get the target.
+  // 2. Parse it and get the target, environment name, and auxiliary
+  //    environment.
   //
-  // 3. Find the environment setup executable for this target.
+  // 3. Find the environment setup executable for this name.
   //
-  // 4. Execute the environment setup executable.
+  // 4. Execute the environment setup executable for this target in the
+  //    auxiliary environment.
   //
   // 5. If the environment setup executable fails, then upload the (failed)
   //    result ourselves.
@@ -5970,6 +6093,33 @@ startup ()
   //
   task_manifest tm;
 
+  auto upload_result = [&trace, &tm] (result_status rs,
+                                      operation_results&& ors)
+  {
+    const string url ("tftp://" + ops.tftp_host () + "/result.manifest.lz4");
+
+    // If we failed before being able to parse the task manifest, use the
+    // "unknown" values for the package name and version.
+    //
+    result_manifest rm {
+      tm.name.empty ()    ? bpkg::package_name ("unknown") : tm.name,
+      tm.version.empty () ? bpkg::version ("0")            : tm.version,
+      rs,
+      move (ors),
+      worker_checksum,
+      nullopt /* dependency_checksum */
+    };
+
+    try
+    {
+      upload_manifest (trace, url, rm, "result");
+    }
+    catch (const io_error& e)
+    {
+      fail << "unable to upload result manifest to " << url << ": " << e;
+    }
+  };
+
   try
   {
     // Download the task.
@@ -6051,6 +6201,31 @@ startup ()
         fail << "no default environment setup executable in " << env_dir;
     }
 
+    // Auxiliary environment.
+    //
+    strings aux_env;
+    if (tm.auxiliary_environment)
+    {
+      try
+      {
+        aux_env = parse_auxiliary_environment (*tm.auxiliary_environment);
+      }
+      catch (const invalid_argument& e)
+      {
+        // Note: include (unparsed) environment into the log so that we can
+        // see what we are dealing with.
+        //
+        operation_result r {
+          "configure",
+          result_status::abort,
+          *tm.auxiliary_environment + "\n" +
+          "error: invalid auxiliary environment: " + e.what () + '\n'};
+
+        upload_result (result_status::abort, {move (r)});
+        return 1;
+      }
+    }
+
     // Run it.
     //
     strings os;
@@ -6088,7 +6263,12 @@ startup ()
     // result manifest. There is no reason to retry (most likely there is
     // nobody listening on the other end anymore).
     //
-    switch (run_io_exit (trace, 0, 2, 2, pp, tg, argv0.effect_string (), os))
+    switch (run_io_exit (trace,
+                         0 /* stdin */, 2 /* stdout */, 2 /* stderr */,
+                         process_env (pp, aux_env),
+                         tg,
+                         argv0.effect_string (),
+                         os))
     {
     case 3:
     case 2:  return 1;
@@ -6098,29 +6278,7 @@ startup ()
   }
   catch (const failed&)
   {
-    const string url ("tftp://" + ops.tftp_host () + "/result.manifest.lz4");
-
-    // If we failed before being able to parse the task manifest, use the
-    // "unknown" values for the package name and version.
-    //
-    result_manifest rm {
-      tm.name.empty ()    ? bpkg::package_name ("unknown") : tm.name,
-      tm.version.empty () ? bpkg::version ("0")            : tm.version,
-      result_status::abnormal,
-      operation_results {},
-      worker_checksum,
-      nullopt /* dependency_checksum */
-    };
-
-    try
-    {
-      upload_manifest (trace, url, rm, "result");
-    }
-    catch (const io_error& e)
-    {
-      fail << "unable to upload result manifest to " << url << ": " << e;
-    }
-
+    upload_result (result_status::abnormal, operation_results {});
     return 1;
   }
 }
diff --git a/doc/manual.cli b/doc/manual.cli
index 0de6d0b..2fa3248 100644
--- a/doc/manual.cli
+++ b/doc/manual.cli
@@ -621,8 +621,10 @@ one per line, in the form:
 \
 
 Whitespaces before \c{<name>}, around \c{=}, and after \c{<value>} as well as
-blank lines and lines that start with \c{#} are ignored. The \c{<value>} part
-as a whole can be single ('\ ') or double (\"\ \") quoted. For example:
+blank lines and lines that start with \c{#} are ignored. The \c{<name>} part
+must only contain capital alphabetic, numeric, and \c{_} characters. The
+\c{<value>} part as a whole can be single ('\ ') or double (\"\ \")
+quoted. For example:
 
 \
 DATABASE_HOST=192.168.0.1