From 1d094f2e9e18c5247002271b2eefed74e2eaece7 Mon Sep 17 00:00:00 2001
From: Paul O'Leary McCann <polm@dampfkraft.com>
Date: Tue, 1 Aug 2017 23:00:33 +0900
Subject: [PATCH] Fix handling of node-format when output-format-type is set

Current behavior is that if output-format-type is set node-format is
ignored. This includes the case where output-format-type is set in a
config file, so if your config file sets a default format it's
impossible to override it on the command line.

One solution is to avoid setting a default output format in config
files. While that works, it's worth noting unidic-mecab's default config
file includes a default output format. If someone tries to use
node-format without being aware of this they might just think it doesn't
work.

This change makes it so that if node-format is explicitly set
output-format-type is ignored. If output-format-type isn't used behavior
is unchanged; this only changes things when output-format-type and
node-format are both set. -POLM
---
 mecab/src/tagger.cpp |  4 ++--
 mecab/src/writer.cpp | 10 ++++++----
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/mecab/src/tagger.cpp b/mecab/src/tagger.cpp
index 2780dbb..8aa4fef 100644
--- a/mecab/src/tagger.cpp
+++ b/mecab/src/tagger.cpp
@@ -49,9 +49,9 @@ const MeCab::Option long_options[] = {
     "output marginal probability (default false)" },
   { "max-grouping-size",  'M',  "24",
     "INT",  "maximum grouping size for unknown words (default 24)" },
-  { "node-format",        'F',  "%m\\t%H\\n", "STR",
+  { "node-format",        'F',  "", "STR",
     "use STR as the user-defined node format" },
-  { "unk-format",        'U',  "%m\\t%H\\n", "STR",
+  { "unk-format",        'U',  "", "STR",
     "use STR as the user-defined unknown node format"   },
   { "bos-format",        'B',  "", "STR",
     "use STR as the user-defined beginning-of-sentence format"   },
diff --git a/mecab/src/writer.cpp b/mecab/src/writer.cpp
index ed68767..592294a 100644
--- a/mecab/src/writer.cpp
+++ b/mecab/src/writer.cpp
@@ -23,7 +23,8 @@ void Writer::close() {
 }
 
 bool Writer::open(const Param &param) {
-  const std::string ostyle = param.get<std::string>("output-format-type");
+  const std::string node_format_option = param.get<std::string>("node-format");
+  const std::string ostyle = node_format_option.empty() ? param.get<std::string>("output-format-type") : "";
   write_ = &Writer::writeLattice;
 
   if (ostyle == "wakati") {
@@ -77,7 +78,8 @@ bool Writer::open(const Param &param) {
     if (node_format != node_format2 || bos_format != bos_format2 ||
         eos_format != eos_format2 || unk_format != unk_format2) {
       write_ = &Writer::writeUser;
-      if (node_format != node_format2) {
+
+      if (!node_format2.empty() && node_format != node_format2) {
         node_format = node_format2;
       }
       if (bos_format != bos_format2) {
@@ -86,9 +88,9 @@ bool Writer::open(const Param &param) {
       if (eos_format != eos_format2) {
         eos_format = eos_format2;
       }
-      if (unk_format != unk_format2) {
+      if (!unk_format2.empty() && unk_format != unk_format2) {
         unk_format = unk_format2;
-      } else if (node_format != node_format2) {
+      } else if (!node_format2.empty() && node_format != node_format2) {
         unk_format = node_format2;
       } else {
         unk_format = node_format;