Add big endian support

scripts/make_data_assembly.py

Index: scripts/make_data_assembly.py
diff --git a/scripts/make_data_assembly.py b/scripts/make_data_assembly.py
new file mode 100755
index 0000000000000000000000000000000000000000..65ca058d9a995e2fc61b1bd4f4b23db446bfae98
--- /dev/null
+++ b/scripts/make_data_assembly.py
@@ -0,0 +1,57 @@
+#!/usr/bin/env python

[jungshik at Google, 2016/05/17 18:09:52]

Add these 3 lines after the 1st line. 

# Copyright 2016 The Chromium Authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.

style guide is here: (search for Python) 
https://sites.google.com/a/chromium.org/dev/developers/coding-style

+
+import binascii
+import sys
+
+if len(sys.argv)<2:

[jungshik at Google, 2016/05/17 18:09:52]

nit: Please, use spaces around a binary operator (here and elsewhere). 

if len(sys.argv) < 2: 

( see
https://www.python.org/dev/peps/pep-0008/#whitespace-in-expressions-and-state...
)

+	sys.exit("Provide input file name as argument to use the script!");

[jungshik at Google, 2016/05/17 17:51:59]

nit:  Chromium python style is to use 2-space indentation instead of 8.
Oh. you're using <tab>.  Please, replace all the tabs with 2 spaces. 

( http://dev.chromium.org/chromium-os/python-style-guidelines )

And, perhaps, this is clearer? 

sys.exit("Usage: %s icu_data_file" % sys.argv[0]); 

+
+input_file = sys.argv[1]
+n = input_file.find(".dat")
+if n==-1:
+	sys.exit("Not a .dat file!");

[jungshik at Google, 2016/05/17 17:51:58]

sys.exit("%s is not a ICU .dat file." % input_file); 

+
+output_file = input_file[0:n] + "_dat.S"

[jungshik at Google, 2016/05/17 20:33:12]

For use with gyp (automatic generation during a build), this script needs to
accept an explicit output file path. For a local test run, you can keep this
automatic derivation of the output file, but if sys.argv[2] is given, that has
to be used for the output. 

+
+input = open(input_file, 'rb').read()
+n = input.find("icudt")
+if n==-1:
+	sys.exit("Cannot find version number in the input file!");

[jungshik at Google, 2016/05/17 17:51:58]

sys.exit("Cannot find a version number in %s." % input_file); 

+
+version_number = input[n+5:n+7]
+output = open(output_file, 'w')
+
+output.write(".globl icudt" + version_number + "_dat\n"
+			 "\t.section .note.GNU-stack,\"\",%progbits\n"

[jungshik at Google, 2016/05/17 17:51:58]

You can align this and other lines in the string with ".globl. icudt", can't
you? 

+			 "\t.section .rodata\n"
+			 "\t.balign 16\n"
+			 "#ifdef U_HIDE_DATA_SYMBOL\n"
+			 "\t.hidden icudt" + version_number + "_dat\n"
+			 "#endif\n"
+			 "\t.type icudt" + version_number + "_dat,%object\n"
+			 "icudt" + version_number + "_dat:\n")
+
+split = [binascii.hexlify(input[i:i+4]).upper().lstrip('0') for i in range(0, len(input), 4)]

[jungshik at Google, 2016/05/17 17:51:58]

nit: each line has to be 80 chars or fewer. 

split = [binascii.hexlify(input[i:i+4]).upper().lstrip('0')
         for i in range(0, len(input), 4)]

And, to make this work for both little and big endians, you need to 
determine  earlier the endianness (based on the input file name suffix, 'l.dat'
vs 'b.dat', maybe) Then, depending on the endianness, 

input[i:i+4] has to be either input[i:i+4][::1] (big endian) or
input[i:i+4][::-1] (little endian)

+
+for i in range(len(split)):
+	if (len(split[i])==0):
+		value='0'
+	elif (len(split[i])==1):
+		if not any((c in '123456789') for c in split[i]):
+			value='0x0'+split[i]
+		else:
+			value=split[i]
+	elif (len(split[i])%2==1):
+		value='0x0'+split[i]
+	else:
+		value='0x'+split[i]
+
+	if (i%32==0):
+		output.write("\n.long ")
+	else:
+		output.write(",")
+	output.write(value)
+
+output.write("\n")
+output.close()
+print "Generated " + output_file