From cc9abdff120ccc5292431d82bdfcc5693a56c143 Mon Sep 17 00:00:00 2001
From: "Reece H. Dunn" <msclrhd@gmail.com>
Date: Mon, 24 Dec 2012 14:47:06 +0000
Subject: [PATCH] Fetch UnicodeData.txt from unicode.org if not present to make
 the build fully automated.

---
 .gitignore          |  1 +
 Makefile.am         | 15 +++++++++++----
 src/case.cpp        |  2 +-
 src/categories.cpp  |  2 +-
 tools/case.py       |  2 +-
 tools/categories.py |  2 +-
 6 files changed, 16 insertions(+), 8 deletions(-)

diff --git a/.gitignore b/.gitignore
index 3d6f555d..2a39e037 100644
--- a/.gitignore
+++ b/.gitignore
@@ -12,6 +12,7 @@
 
 # build output:
 
+data/
 src/libucd.la
 tests/printucddata
 
diff --git a/Makefile.am b/Makefile.am
index 07930c27..bcd233ed 100644
--- a/Makefile.am
+++ b/Makefile.am
@@ -54,14 +54,21 @@ EXTRA_DIST += ChangeLog
 
 ############################# Unicode Character Database ######################
 
-src/case.cpp: tools/case.py tools/ucd.py
-	tools/case.py ${UCD_ROOTDIR} > $@
+UCD_VERSION=6.2.0
+UCD_ROOTDIR=data/ucd
 
-src/categories.cpp: tools/categories.py tools/ucd.py
-	tools/categories.py ${UCD_ROOTDIR} > $@
+data/ucd/UnicodeData.txt:
+	mkdir -pv data/ucd
+	wget -O $@ http://www.unicode.org/Public/${UCD_VERSION}/ucd/UnicodeData.txt
 
 ############################# libucd ##########################################
 
+src/case.cpp: tools/case.py tools/ucd.py data/ucd/UnicodeData.txt
+	tools/case.py ${UCD_ROOTDIR} ${UCD_VERSION} > $@
+
+src/categories.cpp: tools/categories.py tools/ucd.py data/ucd/UnicodeData.txt
+	tools/categories.py ${UCD_ROOTDIR} ${UCD_VERSION} > $@
+
 libucd_includedir = $(includedir)/ucd
 libucd_include_HEADERS = \
 	src/include/ucd/ucd.h
diff --git a/src/case.cpp b/src/case.cpp
index 73d11297..b7935655 100644
--- a/src/case.cpp
+++ b/src/case.cpp
@@ -27,7 +27,7 @@
 
 using namespace ucd;
 
-// Unicode Character Data 6.2
+// Unicode Character Data 6.2.0
 
 struct case_conversion_entry
 {
diff --git a/src/categories.cpp b/src/categories.cpp
index 1d7e1183..16647897 100644
--- a/src/categories.cpp
+++ b/src/categories.cpp
@@ -27,7 +27,7 @@
 
 using namespace ucd;
 
-// Unicode Character Data 6.2
+// Unicode Character Data 6.2.0
 
 static const uint8_t categories_Cn[256] =
 {
diff --git a/tools/case.py b/tools/case.py
index 6c03aad5..16dbccbf 100755
--- a/tools/case.py
+++ b/tools/case.py
@@ -22,7 +22,7 @@ import sys
 import ucd
 
 ucd_rootdir = sys.argv[1]
-ucd_version = ucd_rootdir.split('-')[-1]
+ucd_version = sys.argv[2]
 
 unicode_chars = {}
 null = ucd.CodePoint('0000')
diff --git a/tools/categories.py b/tools/categories.py
index 6b53ec88..dac66853 100755
--- a/tools/categories.py
+++ b/tools/categories.py
@@ -22,7 +22,7 @@ import sys
 import ucd
 
 ucd_rootdir = sys.argv[1]
-ucd_version = ucd_rootdir.split('-')[-1]
+ucd_version = sys.argv[2]
 
 unicode_chars = {}
 for data in ucd.parse_ucd_data(ucd_rootdir, 'UnicodeData'):