fenrir/internal/punctuation/text_processing_test.go

// Copyright (c) 2026 Storm Dragon
// SPDX-License-Identifier: GPL-3.0-or-later

package punctuation

import (
	"os"
	"path/filepath"
	"strings"
	"testing"

	"git.stormux.org/storm/fenrir/internal/i18n"
)

func TestTextSubstitutionsAndBoundaries(t *testing.T) {
	processor, _, err := Load([]string{"../../config/punctuation/default.toml"})
	if err != nil {
		t.Fatal(err)
	}
	options := TextOptions{Emoticons: true, RepeatedSymbols: true, RepeatedSymbolsMin: 4}
	for _, trial := range []struct{ text, want string }{
		{":)", "smile"},
		{":) :) :)", "smile smile smile"},
		{":) :(", "smile frown"},
		{"hello :) there", "hello smile there"},
		{"你好 :) 世界", "你好 smile 世界"},
		{"#####", "5 number"},
		{"#### ####", "4 number 4 number"},
		{"##### title #####", "5 number title 5 number"},
		{"////", "4 slash"},
		{"+++++-----", "5 plus 5 dash"},
		{"+++++--", "5 plus  dash  dash "},
		{"--+++++", " dash  dash  5 plus"},
		{"+++++-----++++", "5 plus 5 dash 4 plus"},
		{"[####)", "  4 number  "},
		{"[#####]", " 5 number "},
		{"(#####)", " 5 number "},
		{"{#####}", " 5 number "},
		{"<#####>", " 5 number "},
		{"[#####-----]", " 5 number 5 dash "},
		{"[-----#####]", " 5 dash 5 number "},
		{"[#####----#####]", " 5 number 4 dash 5 number "},
		{"[#####--]", " 5 number  dash  dash  "},
		{"Oo", "WTF?"},
	} {
		if got := processor.Process_text(trial.text, "some", options); got != trial.want {
			t.Errorf("%q: %q, want %q", trial.text, got, trial.want)
		}
	}
	for _, text := range []string{"hello :)!", "path/to:)file", "https://example.invalid/:)", "https://example.invalid/:-/", "(o_O)", "###", "##--##", "a####b", "https://example.invalid/####", "10000", "aaaa", "🙂🙂🙂🙂", "", "[]", "[###]", "[##--##]", "[a####b]", "[https://example.invalid/####]"} {
		for _, level := range Levels() {
			if got, want := processor.Process_text(text, level, options), processor.Text(text, level); got != want {
				t.Errorf("%q, %s changed ordinary processing: %q, want %q", text, level, got, want)
			}
		}
	}
	if got := processor.Process_text("\t:)\u00a0:) \n", "none", options); got != "\tsmile\u00a0smile \n" {
		t.Fatal("whitespace changed", got)
	}
	if got := processor.Process_text("##### ####", "all", options); got != "5 number space 4 number" {
		t.Fatal("generated names were processed again", got)
	}
	if got := processor.Process_text("Oo", "all", options); got != "WTF?" {
		t.Fatal("punctuation in emoticon name processed again", got)
	}
}

func TestSubstitutionOptionsAndProfileNames(t *testing.T) {
	processor, _, err := Load([]string{"../../config/punctuation/default.toml"})
	if err != nil {
		t.Fatal(err)
	}
	text := ":) #####"
	for _, level := range Levels() {
		if got := processor.Process_text(text, level, TextOptions{}); got != processor.Text(text, level) {
			t.Fatal("disabled substitutions changed text", level, got)
		}
	}
	options := TextOptions{RepeatedSymbols: true, RepeatedSymbolsMin: 5}
	if got := processor.Process_text("####", "some", options); got != processor.Text("####", "some") {
		t.Fatal("minimum ignored", got)
	}
	if got := processor.Process_text("#####", "none", options); got != processor.Text("#####", "none") {
		t.Fatal("summarized unselected symbol", got)
	}
	for _, minimum := range []int{0, 1} {
		options.RepeatedSymbolsMin = minimum
		if got := processor.Process_text("#####", "some", options); got != processor.Text("#####", "some") {
			t.Fatal("invalid options summarized text", got)
		}
	}
	options.RepeatedSymbolsMin = 4
	data, err := os.ReadFile("../../config/punctuation/default.toml")
	if err != nil {
		t.Fatal(err)
	}
	data = []byte(strings.Replace(string(data), `"/" = "slash"`, `"/" = "path/separator"`, 1))
	custom, err := Decode(data)
	if err != nil {
		t.Fatal(err)
	}
	if got := custom.Process_text("////", "some", options); got != "4 path/separator" {
		t.Fatal("custom name changed or processed twice", got)
	}
	if got := custom.Process_text("[////####]", "some", options); got != " 4 path/separator 4 number " {
		t.Fatal("mixed summary reprocessed custom name", got)
	}
	if got := custom.Process_text("////####", "some", options); got != "4 path/separator 4 number" {
		t.Fatal("unbracketed summary reprocessed custom name", got)
	}
	french, _, err := Load([]string{"../../config/punctuation/fr.toml"})
	if err != nil {
		t.Fatal(err)
	}
	if got := french.Process_text("#####", "most", options); got != "5 dièse" {
		t.Fatal("localized profile name lost", got)
	}
	if got := processor.Character('#'); got != "hash" {
		t.Fatal("summary changed character naming", got)
	}
}

func TestRepeatedSymbolsRespectOptionsAndLevels(t *testing.T) {
	processor, _, err := Load([]string{"../../config/punctuation/default.toml"})
	if err != nil {
		t.Fatal(err)
	}
	for _, level := range Levels() {
		for _, options := range []TextOptions{
			{},
			{Emoticons: true},
			{RepeatedSymbols: true, RepeatedSymbolsMin: 1},
			{RepeatedSymbols: true, RepeatedSymbolsMin: 6},
			{RepeatedSymbols: true, RepeatedSymbolsMin: 4},
		} {
			for _, text := range []string{"[#####]", "[#####-----]", "+++++-----", "+++++..", "../...."} {
				want := processor.Text(text, level)
				if options.RepeatedSymbols && options.RepeatedSymbolsMin == 4 && level != "none" {
					switch text {
					case "[#####]":
						want = processor.Text("[", level) + "5 number" + processor.Text("]", level)
					case "[#####-----]":
						want = processor.Text("[", level) + "5 number 5 dash" + processor.Text("]", level)
					case "+++++-----":
						want = "5 plus 5 dash"
					case "+++++..":
						want = "5 plus " + processor.Text("..", level)
					case "../....":
						if level == "most" || level == "all" {
							want = processor.Text("..", level) + " " + processor.Text("/", level) + " 4 dot"
						}
					}
				}
				if got := processor.Process_text(text, level, options); got != want {
					t.Errorf("%q, level %s, options %+v: got %q, want %q", text, level, options, got, want)
				}
			}
		}
	}
}

func TestTextSubstitutionTranslationsAndFieldOrder(t *testing.T) {
	root := t.TempDir()
	directory := filepath.Join(root, "xx", "LC_MESSAGES")
	if err := os.MkdirAll(directory, 0700); err != nil {
		t.Fatal(err)
	}
	catalog := `msgid ""
msgstr ""
"Content-Type: text/plain; charset=UTF-8\n"

msgid "smile"
msgstr "translated-smile"

msgid "number"
msgstr "translated-number"

msgid "{0} {1}"
msgstr "{1}: {0}"
`
	if err := os.WriteFile(filepath.Join(directory, "fenrir.po"), []byte(catalog), 0600); err != nil {
		t.Fatal(err)
	}
	previous := i18n.Use(i18n.New(root, "xx"))
	t.Cleanup(func() { i18n.Use(previous) })
	processor, _, err := Load([]string{"../../config/punctuation/default.toml"})
	if err != nil {
		t.Fatal(err)
	}
	options := TextOptions{Emoticons: true, RepeatedSymbols: true, RepeatedSymbolsMin: 4}
	if got := processor.Process_text(":) #####", "some", options); got != "translated-smile translated-number: 5" {
		t.Fatal("translation or positional order lost", got)
	}
}