diff --git a/sudachi-cli/src/build.rs b/sudachi-cli/src/build.rs
index eea11ecf..dbb03444 100644
--- a/sudachi-cli/src/build.rs
+++ b/sudachi-cli/src/build.rs
@@ -27,6 +27,8 @@ use sudachi::dic::build::report::DictPartReport;
 use sudachi::dic::build::DictBuilder;
 use sudachi::dic::dictionary::JapaneseDictionary;
 use sudachi::dic::grammar::Grammar;
+use sudachi::dic::header::HeaderVersion;
+use sudachi::dic::lexicon::word_infos::WordInfo;
 use sudachi::dic::lexicon_set::LexiconSet;
 use sudachi::dic::word_id::WordId;
 use sudachi::dic::DictionaryLoader;
@@ -76,9 +78,17 @@ pub(crate) enum BuildCli {
 
     #[command(name = "dump")]
     Dump {
-        dict: PathBuf,
+        /// target dictionary to dump
+        dictionary: PathBuf,
+        /// dump target (matrix, pos, winfo)
         part: String,
+        /// output file
         output: PathBuf,
+
+        /// reference system dictionary.
+        /// required to dump winfo of an user dictionary
+        #[arg(short = 's', long = "system")]
+        system: Option<PathBuf>,
     },
 }
 
@@ -101,7 +111,12 @@ pub fn build_main(subcommand: BuildCli) {
     match subcommand {
         BuildCli::System { common, matrix } => build_system(common, matrix),
         BuildCli::User { common, dictionary } => build_user(common, dictionary),
-        BuildCli::Dump { dict, part, output } => dump_part(dict, part, output),
+        BuildCli::Dump {
+            dictionary,
+            part,
+            output,
+            system,
+        } => dump_part(dictionary, system, part, output),
     }
 }
 
@@ -172,31 +187,36 @@ fn output_file(p: &Path) -> File {
     OpenOptions::new()
         .write(true)
         .create_new(true)
-        .open(&p)
+        .open(p)
         .unwrap_or_else(|e| panic!("failed to open {:?} for writing:\n{:?}", p, e))
 }
 
-fn dump_part(dict: PathBuf, part: String, output: PathBuf) {
-    let file = File::open(&dict).expect("open failed");
-    let data = unsafe { Mmap::map(&file) }.expect("mmap failed");
+fn dump_part(dict: PathBuf, system: Option<PathBuf>, part: String, output: PathBuf) {
+    let file = File::open(dict).expect("open dict failed");
+    let data = unsafe { Mmap::map(&file) }.expect("mmap dict failed");
     let loader =
         unsafe { DictionaryLoader::read_any_dictionary(&data) }.expect("failed to load dictionary");
-    let dict = loader.to_loaded().expect("should contain grammar");
 
     let outf = output_file(&output);
     let mut writer = BufWriter::new(outf);
 
     match part.as_str() {
-        "pos" => dump_pos(dict.grammar(), &mut writer),
-        "matrix" => dump_matrix(dict.grammar(), &mut writer),
-        "winfo" => dump_word_info(dict.lexicon(), &mut writer).unwrap(),
+        "pos" => dump_pos(loader, &mut writer),
+        "matrix" => dump_matrix(loader, &mut writer),
+        "winfo" => dump_word_info(loader, system, &mut writer).unwrap(),
         _ => unimplemented!(),
     }
     writer.flush().unwrap();
 }
 
-fn dump_pos<W: Write>(grammar: &Grammar, w: &mut W) {
-    for p in grammar.pos_list.iter() {
+fn dump_pos<W: Write>(dict: DictionaryLoader, w: &mut W) {
+    let dict = dict
+        .to_loaded()
+        .expect("target dict should contain grammar");
+    let grammar = dict.grammar();
+
+    for (id, p) in grammar.pos_list.iter().enumerate() {
+        write!(w, "{},", id).unwrap();
         for (i, e) in p.iter().enumerate() {
             w.write_all(e.as_bytes()).unwrap();
             if (i + 1) == p.len() {
@@ -208,35 +228,86 @@ fn dump_pos<W: Write>(grammar: &Grammar, w: &mut W) {
     }
 }
 
-fn dump_matrix<W: Write>(grammar: &Grammar, w: &mut W) {
+fn dump_matrix<W: Write>(dict: DictionaryLoader, w: &mut W) {
+    if let HeaderVersion::UserDict(_) = dict.header.version {
+        panic!("user dictionary does not have connection matrix.")
+    }
+
+    let dict = dict
+        .to_loaded()
+        .expect("target dict should contain grammar");
+    let grammar = dict.grammar();
     let conn = grammar.conn_matrix();
-    write!(w, "{} {}", conn.num_left(), conn.num_right()).unwrap();
 
+    writeln!(w, "{} {}", conn.num_left(), conn.num_right()).unwrap();
     for left in 0..conn.num_left() {
         for right in 0..conn.num_right() {
             let cost = conn.cost(left as _, right as _);
-            write!(w, "{} {} {}\n", left, right, cost).unwrap();
+            writeln!(w, "{} {} {}", left, right, cost).unwrap();
         }
     }
 }
 
-fn dump_word_info<W: Write>(lex: &LexiconSet, w: &mut W) -> SudachiResult<()> {
-    let size = lex.size();
+fn dump_word_info<W: Write>(
+    dict: DictionaryLoader,
+    system: Option<PathBuf>,
+    w: &mut W,
+) -> SudachiResult<()> {
+    let is_user = match dict.header.version {
+        HeaderVersion::UserDict(_) => true,
+        HeaderVersion::SystemDict(_) => false,
+    };
+    let did = if is_user { 1 } else { 0 };
+    let size = dict.lexicon.size();
+
+    let data = system.map(|system_path| {
+        let file = File::open(system_path).expect("open system failed");
+        unsafe { Mmap::map(&file) }.expect("mmap system failed")
+    });
+    let system = data.as_ref().map(|data| {
+        let loader = DictionaryLoader::read_system_dictionary(data)
+            .expect("failed to load system dictionary");
+        loader
+            .to_loaded()
+            .expect("failed to load system dictionary")
+    });
+
+    let (base, user) = if is_user {
+        (
+            system.expect("system dictionary is required to dump user dictionary lexicon"),
+            Some(dict),
+        )
+    } else {
+        (dict.to_loaded().expect("failed to load dictionary"), None)
+    };
+
+    let mut lex = base.lexicon_set;
+    let mut grammar = base.grammar;
+    if let Some(udic) = user {
+        lex.append(udic.lexicon, grammar.pos_list.len())?;
+        if let Some(g) = udic.grammar {
+            grammar.merge(g)
+        }
+    }
+
     for i in 0..size {
-        let wid = WordId::checked(0, i)?;
+        let wid = WordId::checked(did, i)?;
         let (left, right, cost) = lex.get_word_param(wid);
         let winfo = lex.get_word_info(wid)?;
+        write!(w, "{},", unicode_escape(winfo.surface()))?;
         write!(w, "{},{},{},", left, right, cost)?;
-        write!(w, "{},", winfo.surface())?;
-        write!(w, "{},", winfo.head_word_length())?;
-        write!(w, "{},", winfo.normalized_form())?;
-        write!(w, "{},", winfo.dictionary_form_word_id())?;
-        write!(w, "{},", winfo.reading_form())?;
-        dump_wids(w, winfo.a_unit_split())?;
+        write!(w, "{},", unicode_escape(winfo.surface()))?; // writing
+        write!(w, "{},", pos_string(&grammar, winfo.pos_id()))?;
+        write!(w, "{},", unicode_escape(winfo.reading_form()))?;
+        write!(w, "{},", unicode_escape(winfo.normalized_form()))?;
+        let dict_form = dictionary_form_string(&grammar, &lex, winfo.dictionary_form_word_id());
+        write!(w, "{},", dict_form)?;
+        write!(w, "{},", split_mode(&winfo))?;
+        dump_wids(w, &grammar, &lex, winfo.a_unit_split())?;
         w.write_all(b",")?;
-        dump_wids(w, winfo.b_unit_split())?;
+        dump_wids(w, &grammar, &lex, winfo.b_unit_split())?;
         w.write_all(b",")?;
-        dump_wids(w, winfo.word_structure())?;
+        dump_wids(w, &grammar, &lex, winfo.word_structure())?;
         w.write_all(b",")?;
         dump_gids(w, winfo.synonym_group_ids())?;
         w.write_all(b"\n")?;
@@ -244,23 +315,76 @@ fn dump_word_info<W: Write>(lex: &LexiconSet, w: &mut W) -> SudachiResult<()> {
     Ok(())
 }
 
-fn dump_wids<W: Write>(w: &mut W, data: &[WordId]) -> SudachiResult<()> {
+fn unicode_escape(raw: &str) -> String {
+    // replace '"' and ','
+    raw.to_string()
+        .replace('"', "\\u0022")
+        .replace(',', "\\u002c")
+}
+
+fn split_mode(winfo: &WordInfo) -> &str {
+    let asplits = winfo.a_unit_split();
+    if asplits.is_empty() {
+        return "A";
+    }
+    let bsplits = winfo.b_unit_split();
+    if bsplits.is_empty() {
+        return "B";
+    }
+    "C"
+}
+
+fn pos_string(grammar: &Grammar, posid: u16) -> String {
+    let pos_parts = grammar.pos_components(posid);
+    pos_parts.join(",")
+}
+
+fn dictionary_form_string(grammar: &Grammar, lex: &LexiconSet, wid: i32) -> String {
+    if wid < 0 {
+        return "*".to_string();
+    }
+    let wid_with_dic = WordId::checked(0, wid as u32).expect("invalid wordid");
+    format!("\"{}\"", wordref_string(grammar, lex, &wid_with_dic))
+}
+
+fn wordref_string(grammar: &Grammar, lex: &LexiconSet, wid: &WordId) -> String {
+    let winfo = lex.get_word_info(*wid).expect("failed to get wordinfo");
+    format!(
+        "{},{},{}",
+        unicode_escape(winfo.surface()),
+        pos_string(grammar, winfo.pos_id()),
+        unicode_escape(winfo.reading_form()),
+    )
+}
+
+fn dump_wids<W: Write>(
+    w: &mut W,
+    grammar: &Grammar,
+    lex: &LexiconSet,
+    data: &[WordId],
+) -> SudachiResult<()> {
+    if data.is_empty() {
+        write!(w, "*")?;
+        return Ok(());
+    }
+    w.write_all(b"\"")?;
     for (i, e) in data.iter().enumerate() {
-        let prefix = match e.dic() {
-            0 => "",
-            _ => "U",
-        };
-        write!(w, "{}{}", prefix, e.word())?;
+        write!(w, "{}", wordref_string(grammar, lex, e))?;
         if i + 1 != data.len() {
             w.write_all(b"/")?;
         }
     }
+    w.write_all(b"\"")?;
     Ok(())
 }
 
 fn dump_gids<W: Write>(w: &mut W, data: &[u32]) -> SudachiResult<()> {
+    if data.is_empty() {
+        write!(w, "*")?;
+        return Ok(());
+    }
     for (i, e) in data.iter().enumerate() {
-        write!(w, "{}", e)?;
+        write!(w, "{:06}", e)?;
         if i + 1 != data.len() {
             w.write_all(b"/")?;
         }