Compare commits

...

2 commits

Author SHA1 Message Date
da8b2a27fc fix: Swahili duration text parsing
All checks were successful
ci/woodpecker/push/woodpecker Pipeline was successful
2023-05-22 17:44:14 +02:00
2c4d70cc0d fix tests 2023-05-22 15:17:05 +02:00
3 changed files with 27 additions and 6 deletions

View file

@ -531,8 +531,15 @@ impl<T> YouTubeListMapper<T> {
}); });
let length = video.accessibility.and_then(|acc| { let length = video.accessibility.and_then(|acc| {
let parts = ACCESSIBILITY_SEP_REGEX.split(&acc).collect::<Vec<_>>(); // The video title has to be stripped from the beginning because in Swahili
if parts.len() > 2 { // the duration follows the title with no separator (probably a bug).
// Example: `what I do with leftoversdakika 1 - cheza video`
let parts = ACCESSIBILITY_SEP_REGEX
.split(acc.trim_start_matches(&video.headline))
.collect::<Vec<_>>();
if parts.len() > 1 {
// In Russian, the duration is the last part
// Example: `Воспроизвести видео \"hangover food\". Его продолжительность 58 секунд.`
let i = match self.lang { let i = match self.lang {
Language::Ru => 1, Language::Ru => 1,
_ => 2, _ => 2,

View file

@ -344,7 +344,21 @@ struct DurationTxtSegment {
word: String, word: String,
} }
fn split_duration_txt(txt: &str, start_c: bool) -> Vec<DurationTxtSegment> { /// Split a video duration string into its segments.
///
/// Each segment consists of a word and a string of digits (one of them may be empty).
///
/// The `start_word` parameter determines whether the segments should start with a word
/// instead of a number. This is the case in Swahili and Singhalese.
///
/// Example (start_word=false):
/// - `1 minute, 13 seconds` -> `{1;minute} {13;seconds}`
/// - `foo 1 minute, 13 seconds bar` -> `{foo} {1;minute} {13;seconds bar}`
///
/// Example (start_word=true):
/// - `dakika 1 na sekunde 1` -> `{1;dakika} {1;na sekunde}`
/// - `foo dakika 1 na sekunde 1 bar` -> `{1;foo dakika} {1;na sekunde} {bar}`
fn split_duration_txt(txt: &str, start_word: bool) -> Vec<DurationTxtSegment> {
let mut segments = Vec::new(); let mut segments = Vec::new();
// 1: parse digits, 2: parse word // 1: parse digits, 2: parse word
@ -353,14 +367,14 @@ fn split_duration_txt(txt: &str, start_c: bool) -> Vec<DurationTxtSegment> {
for c in txt.trim().chars() { for c in txt.trim().chars() {
if c.is_ascii_digit() { if c.is_ascii_digit() {
if state == 2 && (!seg.digits.is_empty() || (!start_c && segments.is_empty())) { if state == 2 && (!seg.digits.is_empty() || (!start_word && segments.is_empty())) {
segments.push(seg); segments.push(seg);
seg = DurationTxtSegment::default(); seg = DurationTxtSegment::default();
} }
seg.digits.push(c); seg.digits.push(c);
state = 1; state = 1;
} else { } else {
if (state == 1) && (!seg.word.is_empty() || (start_c && segments.is_empty())) { if (state == 1) && (!seg.word.is_empty() || (start_word && segments.is_empty())) {
segments.push(seg); segments.push(seg);
seg = DurationTxtSegment::default(); seg = DurationTxtSegment::default();
} }

View file

@ -1001,7 +1001,7 @@ fn channel_order(
)) ))
.unwrap(); .unwrap();
// Upload dates should be in descending order // Upload dates should be in descending order
if tab != ChannelVideoTab::Shorts { if tab == ChannelVideoTab::Videos {
let mut latest_items = latest.items.iter().peekable(); let mut latest_items = latest.items.iter().peekable();
while let (Some(v), Some(next_v)) = (latest_items.next(), latest_items.peek()) { while let (Some(v), Some(next_v)) = (latest_items.next(), latest_items.peek()) {
if !v.is_upcoming && !v.is_live && !next_v.is_upcoming && !next_v.is_live { if !v.is_upcoming && !v.is_live && !next_v.is_upcoming && !next_v.is_live {