Trim down Emscripten SDK asset size

These cuts reduce the number of files by 2/3 and the size by 1/3

There's not too much more that could be trimmed - now the size
is dominated by the specially compiled LLVM/Clang binaries

Bug: skia:7186
Change-Id: Ie88fb6f2277eafbefac0f676daaca809dcb53f62
Reviewed-on: https://skia-review.googlesource.com/107061
Reviewed-by: Ben Wagner <benjaminwagner@google.com>
Commit-Queue: Kevin Lubick <kjlubick@google.com>

3 files changed, 32 insertions, 13 deletions

diff --git a/infra/bots/assets/emscripten_sdk/VERSION b/infra/bots/assets/emscripten_sdk/VERSION
index 56a6051ca2..d8263ee986 100644
--- a/infra/bots/assets/emscripten_sdk/VERSION
+++ b/infra/bots/assets/emscripten_sdk/VERSION
@@ -1 +1 @@
-1
\ No newline at end of file
+2
\ No newline at end of file
diff --git a/infra/bots/assets/emscripten_sdk/create_and_upload.py b/infra/bots/assets/emscripten_sdk/create_and_upload.py
index 56e41a232b..1d17deb86d 100755
--- a/infra/bots/assets/emscripten_sdk/create_and_upload.py
+++ b/infra/bots/assets/emscripten_sdk/create_and_upload.py
@@ -12,6 +12,7 @@
 import argparse
 import common
 import os
+import shutil
 import subprocess
 import sys
 import utils
@@ -26,17 +27,35 @@ def main():
   parser.add_argument('--sdk_path', '-s', required=True)
   args = parser.parse_args()
 
-  with utils.tmp_dir():
-    upload_script = os.path.join(common.FILE_DIR, 'upload.py')
-
-    try:
-      cmd = ['python', upload_script, '-t', args.sdk_path]
-      if args.gsutil:
-        cmd.extend(['--gsutil', args.gsutil])
-      subprocess.check_call(cmd)
-    except subprocess.CalledProcessError:
-      # Trap exceptions to avoid printing two stacktraces.
-      sys.exit(1)
+  upload_script = os.path.join(common.FILE_DIR, 'upload.py')
+
+  try:
+    # Remove large directories that we don't need to use the sdk
+    # (e.g. intermediate build products)
+    rmpath = os.path.join(args.sdk_path, 'clang', 'fastcomp',
+                          'build_incoming_64')
+    # We can ignore errors, for example, if the folders were already deleted.
+    shutil.rmtree(os.path.join(rmpath, 'lib'), ignore_errors=True)
+    shutil.rmtree(os.path.join(rmpath, 'tools') , ignore_errors=True)
+
+    # Remove the source code, which has lots of small files which slows
+    # extraction.  We don't need the source code - we mostly care about the
+    # binaries in $SDK/clang/fastcomp/build_incoming_64/bin.
+    src = os.path.join(args.sdk_path, 'clang', 'fastcomp', 'src')
+    for name in os.listdir(src):
+      p = os.path.join(src, name)
+      # Purposely don't delete src/emscripten-version.txt, which can cause
+      # compilation warnings about "can't verify version".
+      if os.path.isdir(p):
+        shutil.rmtree(p)
+
+    cmd = ['python', upload_script, '-t', args.sdk_path]
+    if args.gsutil:
+      cmd.extend(['--gsutil', args.gsutil])
+    subprocess.check_call(cmd)
+  except subprocess.CalledProcessError:
+    # Trap exceptions to avoid printing two stacktraces.
+    sys.exit(1)
 
 
 if __name__ == '__main__':
diff --git a/infra/bots/tasks.json b/infra/bots/tasks.json
index 9d48b34e6b..a2b70eb575 100644
--- a/infra/bots/tasks.json
+++ b/infra/bots/tasks.json
@@ -4587,7 +4587,7 @@
         {
           "name": "skia/bots/emscripten_sdk",
           "path": "emscripten_sdk",
-          "version": "version:1"
+          "version": "version:2"
         }
       ],
       "dimensions": [