From 9c3f4614436d0566951d9062cf83f0cddaa0f18e Mon Sep 17 00:00:00 2001
From: Felix Yan <felixonmars@archlinux.org>
Date: Sat, 23 May 2020 15:38:21 +0800
Subject: [PATCH] Initial version

---
 Makefile   | 22 ++++++++++++++++++++++
 README     | 14 ++++++++++++++
 convert.py | 29 +++++++++++++++++++++++++++++
 3 files changed, 65 insertions(+)
 create mode 100644 Makefile
 create mode 100644 README
 create mode 100755 convert.py

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..4a59191
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,22 @@
+FILENAME=zhwiki-20200501-all-titles-in-ns0
+
+all: build
+
+build: zhwiki.dict
+
+download: $(FILENAME).gz
+
+$(FILENAME).gz:
+	wget https://dumps.wikimedia.org/zhwiki/20200501/$(FILENAME).gz
+
+$(FILENAME): $(FILENAME).gz
+	gzip -k -d $(FILENAME).gz
+
+zhwiki.raw: $(FILENAME)
+	./convert.py $(FILENAME) > zhwiki.raw
+
+zhwiki.dict: zhwiki.raw
+	libime_pinyindict zhwiki.raw zhwiki.dict
+
+install: zhwiki.dict
+	install -Dm644 zhwiki.dict -t $(DESTDIR)/usr/share/fcitx5/pinyin/dictionaries/
diff --git a/README b/README
new file mode 100644
index 0000000..87a536e
--- /dev/null
+++ b/README
@@ -0,0 +1,14 @@
+zhwiki dictionary for fcitx5-pinyin
+
+
+Build time requirements:
+
+Python modules:
+opencc
+pypinyin
+
+
+Installation:
+
+make
+sudo make install
diff --git a/convert.py b/convert.py
new file mode 100755
index 0000000..e968d19
--- /dev/null
+++ b/convert.py
@@ -0,0 +1,29 @@
+#!/bin/python
+import sys
+import re
+import opencc
+from pypinyin import lazy_pinyin
+converter = opencc.OpenCC('t2s.json')
+
+FILE = sys.argv[1]
+
+HANZI_RE = re.compile('^[\u4e00-\u9fa5]+$')
+count = 0
+with open(FILE) as f:
+    for line in f:
+        line = line.rstrip("\n")
+        if not HANZI_RE.match(line):
+            continue
+
+        pinyin = "'".join(lazy_pinyin(line))
+        if pinyin == line:
+            print("Failed to convert, ignoring:", pinyin, file=sys.stderr)
+            continue
+
+        print("\t".join((converter.convert(line), pinyin, "0")))
+        count += 1
+        if count % 1000 == 0:
+            print(str(count) + " converted", file=sys.stderr)
+
+if count % 1000 != 0:
+    print(str(count) + " converted", file=sys.stderr)