robin/ugc-scaper
1
0
Fork 0
mirror of https://codeberg.org/icewind/ugc-scaper.git synced 2026-06-03 18:24:10 +02:00
Code Issues Projects Releases Packages Wiki Activity

handle nameless teams...

Browse source
This commit is contained in:
Robin Appelman 2023-11-18 21:56:59 +01:00
commit cba291d6b1
1 changed files with 18 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

37
src/parser/team.rs
View file

@ -110,17 +110,20 @@ impl Parser for TeamParser {
fn parse(&self, document: &str) -> Result<Self::Output> {
let document = Html::parse_document(document);
let root = document.root_element();
let name = select_text(root, &self.selector_name)
.ok_or(ScrapeError::NotFound)?
let mut name = select_text(root, &self.selector_name)
.unwrap_or_default()
.to_string();
let tag = select_text(root, &self.selector_tag)
.ok_or(ParseError::ElementNotFound {
selector: SELECTOR_TEAM_TAG,
role: "team tag",
})?
.unwrap_or_default()
.to_string();
match (tag.as_str(), name.as_str()) {
("", "") => return Err(ScrapeError::NotFound),
(_, "") => name = tag.clone(),
_ => {}
};
let image =
document
.select(&self.selector_image)
@ -157,24 +160,20 @@ impl Parser for TeamParser {
.to_string();
let description = select_text(root, &self.selector_team_description)
.ok_or(ParseError::ElementNotFound {
selector: SELECTOR_TEAM_DESCRIPTION,
role: "team description",
})?
.unwrap_or_default()
.replace('\n', " ");
let titles = document
.select(&self.selector_team_titles)
.next()
.ok_or(ParseError::ElementNotFound {
selector: SELECTOR_TEAM_TITLES,
role: "team titles",
})?
.text()
.map(str::trim)
.filter(|s| !s.is_empty())
.map(String::from)
.collect();
.map(|el| {
el.text()
.map(str::trim)
.filter(|s| !s.is_empty())
.map(String::from)
.collect()
})
.unwrap_or_default();
let results = document
.select(&self.selector_team_records)