From 2b84a6d52771818a824aa404a396f569d86bb3bf Mon Sep 17 00:00:00 2001
From: Baudouin Raoult <baudouin.raoult@ecmwf.int>
Date: Mon, 30 Aug 2021 08:34:39 +0100
Subject: [PATCH] Better handling of memfs

---
 memfs.py             | 72 ++++++++++++++++++++++++++++++++++----------
 memfs/CMakeLists.txt | 17 +++++++----
 2 files changed, 67 insertions(+), 22 deletions(-)

diff --git a/memfs.py b/memfs.py
index ea4800abb..df7f42d8d 100755
--- a/memfs.py
+++ b/memfs.py
@@ -1,12 +1,54 @@
 #!/usr/bin/env python
 from __future__ import print_function
+
+import argparse
+import binascii
 import os
 import re
 import sys
-import binascii
 import time
 
-assert len(sys.argv) > 2
+parser = argparse.ArgumentParser()
+
+parser.add_argument(
+    "-n",
+    "--count",
+    type=int,
+    default=10,
+    help="Number of files to generate",
+)
+
+parser.add_argument(
+    "-C",
+    "--chunk",
+    type=int,
+    default=16,
+    help="Chunk size (MB)",
+)
+
+parser.add_argument(
+    "-o",
+    "--output",
+    type=str,
+    default="memfs_gen",
+    help="Name of C file to generate",
+)
+
+parser.add_argument(
+    "-e",
+    "--exclude",
+    help="Exclude packages",
+)
+
+parser.add_argument(
+    "dirs",
+    type=str,
+    nargs="+",
+    help="The list of directories to process",
+)
+
+args = parser.parse_args()
+
 
 start = time.time()
 print("MEMFS: starting")
@@ -15,25 +57,23 @@ print("MEMFS: starting")
 # The BUFR codetables is not used in the engine
 EXCLUDED = ["grib3", "codetables", "taf", "stations"]
 
-pos = 1
-if sys.argv[1] == "-exclude":
-    product = sys.argv[2]
-    if product == "bufr":
-        EXCLUDED.append(product)
-    elif product == "grib":
-        EXCLUDED.extend(["grib1", "grib2"])
-    else:
-        assert False, "Invalid product %s" % product
-    pos = 3
+EXCLUDE = {
+    None: [],
+    "bufr": ["bufr"],
+    "grib": ["grib1", "grib2"],
+}
 
-dirs = [os.path.realpath(x) for x in sys.argv[pos:-1]]
+EXCLUDED.extend(EXCLUDE[args.exclude])
+
+
+dirs = [os.path.realpath(x) for x in args.dirs]
 print("Directories: ", dirs)
 print("Excluding: ", EXCLUDED)
 
 FILES = {}
 SIZES = {}
 NAMES = []
-CHUNK = 14 * 1024 * 1024  # chunk size in bytes
+CHUNK = args.chunk * 1024 * 1024  # chunk size in bytes
 
 # Binary to ASCII function. Different in Python 2 and 3
 try:
@@ -48,11 +88,11 @@ def get_outfile_name(base, count):
 
 
 # The last argument is the base name of the generated C file(s)
-output_file_base = sys.argv[-1]
+output_file_base = args.output
 
 buffer = None
 fcount = 0
-MAX_FCOUNT = 10
+MAX_FCOUNT = args.count
 
 for directory in dirs:
 
diff --git a/memfs/CMakeLists.txt b/memfs/CMakeLists.txt
index e3fe7dd41..04aba8494 100644
--- a/memfs/CMakeLists.txt
+++ b/memfs/CMakeLists.txt
@@ -1,3 +1,8 @@
+list( APPEND memfs_args
+      --count 10
+      --output ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen
+      )
+
 set( generated_c_files
     ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen_000.c
     ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen_001.c
@@ -11,21 +16,21 @@ set( generated_c_files
     ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen_009.c
     ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen_final.c)
 
-set( exclude "" )
 if( NOT HAVE_PRODUCT_BUFR )
-  set( exclude -exclude bufr )
-elseif( NOT HAVE_PRODUCT_GRIB )
-  set( exclude -exclude grib )
+  list( APPEND memfs_args --exclude bufr)
+endif()
+
+if( NOT HAVE_PRODUCT_GRIB )
+  list( APPEND memfs_args --exclude grib)
 endif()
 
 add_custom_command(
   OUTPUT ${generated_c_files}
   COMMAND ${PYTHON_EXECUTABLE} ${PROJECT_SOURCE_DIR}/memfs.py
-    ${exclude}
+    ${memfs_args}
     ${PROJECT_SOURCE_DIR}/definitions
     ${PROJECT_SOURCE_DIR}/samples
     ${PROJECT_SOURCE_DIR}/ifs_samples
-    ${CMAKE_CURRENT_BINARY_DIR}/memfs_gen
   DEPENDS ${PROJECT_SOURCE_DIR}/memfs.py)
 
 set_source_files_properties(