git: initial commit

.gitignore

@@ -0,0 +1,4 @@
+/.venv
+/.pdm.toml
+/onomatopoeia.json
+/public/onomatopoeia.json

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "data/nihongoresources.com"]
+	path = data/nihongoresources.com
+	url = https://github.com/Pomax/nihongoresources.com.git

README.md

@@ -0,0 +1,8 @@
+# jp-onomatopoeia
+
+Compiled dataset of Japanese onomatopoeia and sound effects.
+
+## Data sources
+
+- [nihongoresources.com](https://github.com/Pomax/nihongoresources.com)
+- [Japanese Manga SFX spreadsheet](https://docs.google.com/spreadsheets/d/1iXFIVmSzukN1PYD1JakdqMmrNNlc7TtkutDKevuKM0c/edit)

compile.py

@@ -0,0 +1,122 @@
+import csv
+import dataclasses
+import glob
+import json
+import re
+from collections import defaultdict
+from itertools import islice
+from pathlib import Path
+from typing import Optional
+
+
+@dataclasses.dataclass()
+class RawEntry:
+    katakana: str
+    english: str
+    details: Optional[str]
+
+
+def compile_nihongoresources_com(entries: list[RawEntry]):
+    with open(Path("data") / "nihongoresources.com" / "giongo.txt") as f:
+        for line in f.readlines()[1:]:
+            [_katakana, katakana, translation, *details] = line.split("\t")
+            entries.append(RawEntry(
+                katakana=katakana,
+                english=translation,
+                details=", ".join(details).strip(),
+            ))
+
+
+def compile_japanese_manga_sfx_sheet(entries: list[RawEntry]):
+    pattern = Path("data") / "japanese-manga-sfx" / "*.csv"
+    for file in glob.glob(str(pattern)):
+        with open(file) as f:
+            # Structured with two header rows.
+            rows = list(csv.reader(f))
+
+            # Find all columns of entries.
+            katakana_columns = []
+            for index, cell in enumerate(rows[1]):
+                if cell == "Japanese":
+                    katakana_columns.append(index)
+
+            # Compile each column.
+            new_entries = []
+            for index in katakana_columns:
+                katakana, english, details = None, "", ""
+                for row_index, row in islice(enumerate(rows), 2, None):
+                    if row[index + 1].strip():
+                        # Assume each romaji row starts a new entry.
+                        if katakana is not None:
+                            new_entries.append(RawEntry(katakana, english, details))
+                            english, details = "", ""
+                        katakana = row[index].removesuffix(",")
+
+                    # Append text to existing entry.
+                    english += row[index + 2] + " "
+                    details += row[index + 3] + " "
+
+                # Some columns have no entries.
+                if katakana is not None:
+                    new_entries.append(RawEntry(katakana, english, details))
+
+            # Refine new entries.
+            for entry in new_entries:
+                # Normalize english.
+                english = entry.english
+                english = re.sub(r"\s+", " ", english.strip())
+
+                # Normalize details.
+                details = entry.details
+                details = details.replace("More »", "")
+                details = re.sub(r"\s+", " ", details.strip())
+
+                # Consider numbering.
+                english_split = re.split(r"\((\d+)\)", english)
+                if len(english_split) == 1:
+                    # No numbering.
+                    entries.append(RawEntry(entry.katakana, english, details or None))
+                    continue
+
+                details_items = {}
+                details_split = re.split(r"\((\d+)\)", details)
+                for index, string in enumerate(details_split):
+                    if string.isdigit():
+                        value = details_split[index + 1]
+                        value = value.strip().removesuffix(";")
+                        details_items[int(string)] = value
+
+                for index, string in enumerate(english_split):
+                    if string.isdigit():
+                        english = english_split[index + 1]
+                        english = english.strip().removesuffix(";")
+                        details = details_items.get(int(string), None)
+                        entries.append(RawEntry(entry.katakana, english, details))
+
+
+def compile_raw_entries() -> list[RawEntry]:
+    entries = []
+    compile_nihongoresources_com(entries)
+    compile_japanese_manga_sfx_sheet(entries)
+    return entries
+
+
+def compile_entries(raw_entries: list[RawEntry]):
+    entries = defaultdict(list)
+    for entry in raw_entries:
+        entries[entry.katakana].append({
+            "english": entry.english,
+            "details": entry.details,
+        })
+    return entries
+
+
+def main():
+    raw_entries = compile_raw_entries()
+    entries = compile_entries(raw_entries)
+    with open("onomatopoeia.json", "w") as f:
+        json.dump(entries, f, indent=2)
+
+
+if __name__ == '__main__':
+    main()

data/japanese-manga-sfx/Japanese Manga SFX - Sheet1 (3).csv

@@ -0,0 +1,95 @@
+,,,,ラ,,,,,レ,,,,,ワ,,,,,ン,,,,,リ,,,,,ル,,,,,ロ,
+,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation,,Japanese,Romaji*,English,Explanation
+,,"ラララ,",ra ra ra,*la di da*,SFX for skipping and humming a happy tune. ,,"レロレロ,",rero rero,*bleh bleh*,(1) Rolling something with/on the tongue ,,"ワ,",wa,*wow!* *ack!* *whoa!*,More »,,"ンッ,",n,*grunt*,"As of surprise, effort, sleepiness, pain, or passion.",,,,,,,,,,,,,,,
+,,ららら,,*la la la*,,,れろれろ,,*roll roll*,,,わ,,,,,んっ,,,Usually in context of exerting physical effort. ,,,,,,,,,,,,,,,
+,,,,,More »,,,,,More »,,"ワ,",wa,(1) *boo!*,More »,,,,,,,,,,,,,,,,,,,,
+,,"ラブラブ,",rabu rabu,*romantic*,"SFX that literally derives from the Japanese pronunciation of ""love-love"" as ""rah-bu rah-bu."" ",,,,,,,わ,,(2) *ack!*,,,,,,More »,,,,,,,,,,,,,,,
+,,らぶらぶ,,*love love*,,,,,,,,"ワッ,",wa,*wah!* *yay!* *yeeahh!* *cheer* *woot!*,"SFX for a cheer, or a happy atmosphere. ",,"ン？,",n,*hey?*,"Grunt, usually coming out in the form of a question. ",,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,わっ,,,,,ん？,,*what?*,,,,,,,,,,,,,,,,
+,,"ラクラク,",raku raku,*too easy*,"SFX derived from the adjective ""easy"" or 「楽な」. ",,,,,,,,,,More »,,,,*hm?*,More »,,,,,,,,,,,,,,,
+,,らくらく,,,,,,,,,,"ワー,",wa-,(1) *roar*;,(1) (A crowd's) excited roar. See also *wai wai* and *wara wara*;,,,,*hmm?*,,,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,わー,,(2) *wow*;,(4) SFX for cheering or SFX for cry of dismay or despair. ,,"ン,",n,(1) *uhn*; ,"Crying out, usually when exerting physical effort. ",,,,,,,,,,,,,,,
+,,"ランッ,",ran,(1) *beam* *intensely*,"(1) SFX for sudden attention in one's gaze, like snapping to attention dramatically (can be used for comedic effect) ",,,,,,,,,(3) *yeaahh* *cheer* *applause*;,,,ん,,(2) *ughn*,,,,,,,,,,,,,,,,
+,,らんっ,,,,,,,,,,,,(4) *waaah*,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,"ワーッ,",wa-,*wow* *yay* *delight*,More »,,"ンボーッ,",nbo-,*spaced out* *daydreaming* *dazed*,"Kind of a slow reaction state, when one is very tired or groggy from lack of sleep or exhaustion. (Even the SFX is exhausted with the ""n"" sound before the *bo-* SFX.) See also *bo-*. ",,,,,,,,,,,,,,,
+,,"ランラン,",ran ran,*la la*,SFX for skipping and humming a happy tune. ,,,,,,,わーっ,,,,,んぼーっ,,,,,,,,,,,,,,,,,,
+,,らんらん,,,,,,,,,,"ワーイ,",wa-i,*yay* *woohoo* *yes* *yeah*,"Positive cheering, expression of cheerfulness or happiness. ",,,,,More »,,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,わーい,,,,,"ンチュ,",nchu,*kiss*,"See also *buchu*, *chu*, and *uchu*. ",,,,,,,,,,,,,,,
+,,"ランラン,",ranran,*intense* *fiery*,"SFX for an eye expression, for intense, fiery eyes. ",,,,,,,,,,More »,,んちゅ,,,,,,,,,,,,,,,,,,
+,,らんらん,,,,,,,,,,"ワーンッ,",wa-n,(1) *bawling* *waahh*;,"(1) SFX for a long, wailing cry of despair;",,,,,More »,,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,わーんっ,,(2) *waahh* *grrr*,(2) SFX for a loud cry of outrage or unhappiness with something ,,"ンゴーンガー,",ngo- nga-,*hrrm mrmm*,"SFX for unintelligible muttering or murmuring sounds, such as the sounds made when one is tossing and turning in one's sleep. ",,,,,,,,,,,,,,,
+,,"ララリ,",rarari,*ching ching*,"SFX for a light tinkling, ringing, bell-like, chime-like sound. Can be mechanical or digital, like from a video game. ",,,,,,,,,,,,んごーんがー,,*mph mmm*,,,,,,,,,,,,,,,,
+,,ららり,,*cha-ching ching*,,,,,,,,,,,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,More »,,,,,,,"ワーンワーン,",wa-n wa-n,*waah waah*,SFX for crying loudly and hysterically. ,,"ンク,",nku,*sniff* ,"See also *funka*, *hunka*, and *kunka*. ",,,,,,,,,,,,,,,
+,,,,,,,,,,,,わーんわーん,,*booh hoo*,,,んく,,*inhale*,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*bawling*,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワーワー,",wa-wa-,(1) *roar*;,(1) (A crowd's) excited roar. See also *wai wai* and *wara wara*;,,"ンマッ,",nma,*geh* *gah*,SFX for shock. ,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わーわー,,(2) *chatter* *rustle*,(2) SFX for background noise of pedestrians on the street. ,,んまっ,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,"ンモ,",nmo,*munch* ,SFX for eating with mouth closed. ,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワア,",waa,(1) *wow*;,"(2) Like the war cries of soldiers in battle, or cry of protest. ",,んも,,*mmm munch*,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わあ,,(2) *roar* *yaahh!*;,,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,(3) *cheer* *applause*,More »,,"ンン,",nn,(1) *grunt* *urgh* *ugh* *argh* *arg* *umph* *mmph* *nn*;,"(1) SFX for a grunt of effort, stress, or pain;",,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワッフン,",waffun,*grunt*,Light SFX...strange grunt. ,,んん,,(2) *hm?*,(2) Can be used as a question 「んん？」 ,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わっふん,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワフ,",wafu,(1) *grunt*;,(1) Light SFX...strange grunt;,,"ンヌー,",nnu-,*menacing*,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わふ,,(2) *woof*,"(2) SFX for dog ""woof-ing"", see also *Wafu wafu*. ",,んぬー,,*scary*,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,,,,,*rawr*,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,"ンショ,",nsho,(1) *grunt*;,Sound made when exerting physical effort or strain. ,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワフワフ,",wafu wafu,(1) *grunt grunt*,(1) Light SFX...strange grunt;,,んしょ,,(2) oof! umph!,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わふわふ,,*mumble mumble*;,"(2) SFX for dog ""woof-ing"", see also *Wafu*. ",,,,,Same as *Yoisho*; see also *nisho* and *unsho*. ,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,(2) *woof woof*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワフーン,",wafu-n,*whine*,"High-pitched, sad whine of a dog. ",,"ンッショ,",nsho,*grunt*,Sound made when exerting physical effort.,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わふーん,,,,,んっしょ,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワハハ,",wahaha,*hahaha*,"Full-on, loud laughter.",,"ンショット,",nshotto,(1) *effort*; ,"(1) See also *nisho*, *unsho*, and *yoisho*;",,,,,,,,,,,,,,,
+,,,,,,,,,,,,わはは,,*bwahaha*,,,んしょっと,,(2) oof! umph!,"(2) Effort, strain. ",,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*ahaha*,See also *hahaha* (ハハハ). ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワイ,",wai,*yay!* *delight*,"Feminine exclamation of delight, spoken by a character. ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わい,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワイワイ,",wai wai,*chatter chatter*,"To explain lively action/noise, like *Gaya gaya* ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わいわい,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワク,",waku,*excitement* *happy*,Cute anticipation. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わく,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワクワク,",waku waku,*excited* ,"Happy, cute anticipation.",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わくわく,,*nervous* ,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*trembling*,See also *baku baku*. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワン！,",wan,*woof!*,SFX for the call of a dog or wolf. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わん！,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワンワン,",wan wan,*bow-wow*,SFX for the sound of a dog calling. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わんわん,,*woof woof*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*wuff wuff*,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*ruff ruff*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワナワナ,",wana wana,*tremble tremble* ,Trembling from fear or anger; trembling all over. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わなわな,,*shake shake*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワラワラ,",wara wara,*chatter chatter*,"Crowded people, usually small kids. See also *waa* and *wai wai*. ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わらわら,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワサワサ,",wasa wasa,*rustle rustle*,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わさわさ,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワッシ,",wasshi,*one-two*,SFX for the sound one makes while motivating oneself to use physical effort or to exert oneself physically.,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わっし,,*one-and-a-two*,"Ex: Counting out ""one-two-three, and a one-two-three"" while doing jumping jacks. ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*one-two-three*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*hee haa*,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,*huff huff*,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワタワタ,",wata wata,(1) *panic*;,(1) Moving busily;,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わたわた,,(2) *flap flap*,"(2) SFX for arms flailing around, trying to cover a mistake. ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワワ,",wawa,*whoa*,"In a ""what the heck"" way. ",,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わわ,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,"ワザワザ,",waza waza,*expressly* *intentionally*,To go through the trouble of doing something. ,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,わざわざ,,,,,,,,,,,,,,,,,,,,,,,
+,,,,,,,,,,,,,,,More »,,,,,,,,,,,,,,,,,,,,