1use camino::Utf8Path;
7use chardetng::EncodingDetector;
8use std::{collections::HashSet, fs, io};
9
10pub fn read_text_auto(path: &Utf8Path) -> io::Result<String> {
12 let bytes = fs::read(path)?;
13 let mut det = EncodingDetector::new();
14 det.feed(&bytes, true);
15 Ok(det.guess(None, true).decode(&bytes).0.into_owned())
16}
17
18#[derive(Debug, Default, PartialEq, Eq, Clone)]
22enum Section {
23 Styles,
24 Events,
25 #[default]
26 Other,
27}
28
29impl Section {
30 fn from_header(s: &str) -> Self {
32 match s {
33 "[V4+ Styles]" | "[V4 Styles]" | "[V4++ Styles]" => Self::Styles,
34 "[Events]" => Self::Events,
35 _ => Self::Other,
36 }
37 }
38}
39
40#[derive(Debug, PartialEq, Eq, Clone)]
44struct ColFormat {
45 target_index: usize,
47 columns_count: usize,
49}
50
51impl ColFormat {
52 fn style_default() -> Self {
54 Self {
56 target_index: 1,
57 columns_count: usize::MAX,
58 }
59 }
60
61 fn event_default() -> Self {
63 Self {
65 target_index: 9,
66 columns_count: 10,
67 }
68 }
69
70 fn update_from_format_line(&mut self, format_str: &str, needle: &str) {
72 let cols: Vec<&str> = format_str.split(',').map(str::trim).collect();
73 self.columns_count = cols.len();
74 self.target_index = cols
75 .iter()
76 .position(|c| c.eq_ignore_ascii_case(needle))
77 .unwrap_or(self.columns_count.saturating_sub(1));
78 }
79}
80
81pub fn extract_fonts(ass: &str) -> Vec<String> {
85 let mut fonts = HashSet::<String>::new();
86 let mut section = Section::default();
87 let mut style_fmt = ColFormat::style_default();
88 let mut event_fmt = ColFormat::event_default();
89
90 for line in ass.lines().map(str::trim).filter(|l| !l.is_empty()) {
91 if line.starts_with('[') && line.ends_with(']') {
92 section = Section::from_header(line);
93 continue;
94 }
95
96 match section {
97 Section::Styles => handle_style(line, &mut style_fmt, &mut fonts),
98 Section::Events => handle_event(line, &mut event_fmt, &mut fonts),
99 Section::Other => {}
100 }
101 }
102
103 let mut out: Vec<_> = fonts.into_iter().collect();
104 out.sort();
105 out
106}
107
108fn handle_style(line: &str, fmt: &mut ColFormat, fonts: &mut HashSet<String>) {
112 if let Some(rest) = line.strip_prefix("Format:") {
113 fmt.update_from_format_line(rest, "fontname");
114 } else if let Some(rest) = line.strip_prefix("Style:")
115 && let Some(name) = rest.split(',').nth(fmt.target_index)
116 {
117 add_font(fonts, name);
118 }
119}
120
121fn handle_event(line: &str, fmt: &mut ColFormat, fonts: &mut HashSet<String>) {
123 if let Some(rest) = line.strip_prefix("Format:") {
124 fmt.update_from_format_line(rest, "text");
125 }
126 else if matches!(line.split_once(':'), Some(("Dialogue", _))) {
128 let content = line[line.find(':').unwrap() + 1..].trim_start();
129 if let Some(text) = content.splitn(fmt.columns_count, ',').nth(fmt.target_index) {
130 scan_inline_tags(text, fonts);
131 }
132 }
133}
134
135fn scan_inline_tags(text: &str, fonts: &mut HashSet<String>) {
139 let mut rest = text;
141 while let Some(open) = rest.find('{') {
142 rest = &rest[open + 1..];
143 let Some(close) = rest.find('}') else { break };
144
145 rest[..close]
146 .split('\\')
147 .filter_map(|tag| tag.strip_prefix("fn"))
148 .for_each(|name| add_font(fonts, name));
149
150 rest = &rest[close + 1..];
151 }
152}
153
154fn add_font(fonts: &mut HashSet<String>, raw: &str) {
158 let trimmed = raw.trim();
159 let name = trimmed.strip_prefix('@').unwrap_or(trimmed);
160 if !name.is_empty() {
161 fonts.insert(name.to_owned());
162 }
163}
164
165#[cfg(test)]
166mod tests {
167 use super::*;
168
169 #[test]
170 fn test_extract_fonts_lucky_star() {
171 let ass = r"
172 [Script Info]
173 Title:lucky star
174 Original Script:CASO&I.G
175 Synch Point:0
176 ScriptType:v4.00+
177 Collisions:Normal
178 PlayResX:704
179 PlayResY:396
180 Timer:100.0000
181
182 [V4+ Styles]
183 Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
184 Style: zhengwen,方正准圆_GBK,26,&H00FFFFFF,&H004080FF,&H20B71082,&H99B71082,-1,0,0,0,100,100,0,0.00,0,3,1,2,30,30,8,134
185 Style: zhushi,方正准圆_GBK,19,&H00FFFFFF,&H004080FF,&H20E22E1B,&H99E22E1B,-1,0,0,0,100,100,0,0.00,0,3,1,8,30,30,8,134
186 Style: jinggao,方正准圆_GBK,17,&H00FFFFFF,&H00FFFFFF,&H00000000,&HFF000000,-1,0,0,0,100,100,0,0.00,1,3,1,8,15,15,5,134
187 Style: staff,@DFGKanTeiRyu-XB,18,&H00FFFFFF,&H00000000,&H407A0748,&HA0FFFFFF,0,0,0,0,105,105,1,0.00,1,3,0,7,30,30,10,128
188 Style: OPJ,DFGKanTeiRyu-XB,18,&H00FFFFFF,&H00000000,&H208B0C66,&H66666666,0,0,0,0,100,105,2,0.00,1,3,1,8,30,30,10,128
189 Style: OPC,方正少儿_GBK,24,&H00FFFFFF,&H00000000,&H208B0C66,&H66666666,0,0,0,0,100,100,0,0.00,1,3,1,2,30,30,7,134
190 Style: EDJ,MS Gothic,21,&H00EEEEEE,&H90FFFFFF,&H12333333,&H20000000,-1,0,0,0,100,100,1,0.00,0,3,0,8,30,30,8,128
191 Style: EDC,方正黑体_GBK,19,&H00EEEEEE,&HFF000000,&H12333333,&H20000000,0,0,0,0,100,100,2,0.00,0,3,0,8,30,30,12,134
192
193 [Events]
194 Format: Layer, Start, End, Style, Actor, MarginL, MarginR, MarginV, Effect, Text
195 Dialogue: 0,0:00:06.08,0:00:08.78,OPJ,NTP,0000,0000,0000,,曖昧3センチ そりゃぷにってコトかい? ちょっ!
196 Dialogue: 0,0:00:08.95,0:00:11.97,OPJ,NTP,0000,0000,0000,,らっぴんぐが制服…だぁぁ不利ってことない ぷ。
197 Dialogue: 0,0:00:12.20,0:00:13.55,OPJ,NTP,0000,0000,0000,,がんばっちゃ やっちゃっちゃ
198 ";
199 let fonts_extracted = extract_fonts(ass);
200 let mut fonts_expected = vec![
201 "DFGKanTeiRyu-XB",
202 "MS Gothic",
203 "方正少儿_GBK",
204 "方正准圆_GBK",
205 "方正黑体_GBK",
206 ];
207 fonts_expected.sort();
208 assert_eq!(fonts_extracted, fonts_expected);
209 }
210
211 #[test]
212 fn test_extract_fonts_make_heroine_ga_oosugiru() {
213 let ass = r"
214 [Script Info]
215 ; Script generated by Aegisub 9706-cibuilds-20caaabc0
216 ; http://www.aegisub.org/
217 Title: [KitaujiSub] Make Heroine ga Oosugiru! - 12
218 ScriptType: v4.00+
219 WrapStyle: 2
220 ScaledBorderAndShadow: yes
221 YCbCr Matrix: TV.709
222 PlayResX: 1920
223 PlayResY: 1080
224 LayoutResX: 1920
225 LayoutResY: 1080
226
227 [V4+ Styles]
228 Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
229 Style: Text - CN,Source Han Sans TC Medium,80,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,1,0,1,2.5,0,2,10,10,52,1
230 Style: Text - JP,Source Han Sans Medium,52,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,1,0,1,2.2,0,2,10,10,10,1
231 Style: Text - CN - UP,Source Han Sans TC Medium,80,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,1,0,1,2.5,0,8,10,10,10,1
232 Style: Text - JP - UP,Source Han Sans Medium,52,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,1,0,1,2.2,0,8,10,10,80,1
233 Style: Screen,Source Han Sans TC Medium,80,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,1,0,1,2.5,0,2,10,10,53,1
234 Style: Title,华康翩翩体W5-A,72,&H00B9B9B9,&H000000FF,&H00B99A57,&H00B99A57,0,0,0,0,100,100,5,0,1,0,0,2,10,10,285,1
235 Style: Ruby,Source Han Sans TC Medium,45,&H00F5F5F5,&H000000FF,&H00252525,&H00000000,0,0,0,0,100,100,2,0,1,2.2,0,2,10,10,115,1
236 Style: Staff,华康翩翩体W5-A,60,&H00B9B9B9,&H000000FF,&H00B99A57,&H00B99A57,0,0,0,0,100,100,0,0,1,0,0,8,10,10,30,1
237
238 [Events]
239 Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
240 Comment: 0,0:00:00.00,0:00:00.00,Note,,0,0,0,,--------------- Subtitle Staff&Title ---------------
241 Comment: 0,0:05:41.45,0:05:47.37,Screen,,0,0,0,,# 參見這裡 https://www.nonhoi.jp/amusement/
242 Dialogue: 0,0:05:41.44,0:05:41.48,Screen,,0,0,0,,{=0}{\alpha&H00&\an4\fs70\bord0\c&H312C33&\pos(455,330)}兒童列車
243 Dialogue: 0,0:05:41.48,0:05:41.52,Screen,,0,0,0,,{=0}{\alpha&H00&\an4\fs70\bord0\c&H312C33&\pos(457.06,310)}兒童列車
244 Dialogue: 0,0:05:41.52,0:05:41.57,Screen,,0,0,0,,{=0}{\alpha&H00&\an4\fs70\bord0\c&H312C33&\pos(459.13,290.16)}兒童列車
245 Dialogue: 0,0:05:42.78,0:05:42.82,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(175.71,1125.35)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
246 Dialogue: 0,0:05:42.82,0:05:42.86,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(177.77,1105.37)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
247 Dialogue: 0,0:05:42.86,0:05:42.90,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(179.85,1085.56)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
248 Dialogue: 0,0:05:42.90,0:05:42.94,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(181.98,1065.73)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
249 Dialogue: 0,0:05:42.94,0:05:42.98,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(184.07,1045.73)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
250 Dialogue: 0,0:05:42.98,0:05:43.03,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(186.17,1025.88)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
251 Dialogue: 0,0:05:43.03,0:05:43.07,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(188.31,1006.03)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
252 Dialogue: 0,0:05:43.07,0:05:43.11,Screen,,0,0,0,,{=2}{\alpha&H00&\an4\fnSource Han Sans TC\fs40\bord0\fsp0\c&H38434A&\pos(190.43,986.13)}隨著「嘟嘟—」的汽笛聲前進!乘上列車沿著遊樂場內鋪設的軌道周遊,好好探險吧。
253";
254 let fonts_extracted = extract_fonts(ass);
255 let mut fonts_expected = vec![
256 "Source Han Sans TC",
257 "Source Han Sans Medium",
258 "华康翩翩体W5-A",
259 "Source Han Sans TC Medium",
260 ];
261 fonts_expected.sort();
262 assert_eq!(fonts_extracted, fonts_expected);
263 }
264
265 #[test]
266 fn test_scan_inline_tags() {
267 let mut fonts = HashSet::new();
268 scan_inline_tags("{\\", &mut fonts);
269 scan_inline_tags("{fn", &mut fonts);
270 scan_inline_tags("{\\fn}", &mut fonts);
271 scan_inline_tags("{\\fn\\}", &mut fonts);
272 assert!(fonts.is_empty());
273
274 let mut fonts = HashSet::new();
275 scan_inline_tags("{\\fnSource Han Sans TC\\b2e}", &mut fonts);
276 assert_eq!(fonts, HashSet::from(["Source Han Sans TC".to_string()]));
277
278 let mut fonts = HashSet::new();
279 scan_inline_tags("{{\\fnArial}text}", &mut fonts);
280 assert_eq!(fonts, HashSet::from(["Arial".to_string()]));
281 }
282}