robin/ugc-scaper
1
0
Fork 0
mirror of https://codeberg.org/icewind/ugc-scaper.git synced 2026-06-03 10:14:11 +02:00
Code Issues Projects Releases Packages Wiki Activity

handle empty name changes

Browse source
This commit is contained in:
Robin Appelman 2025-04-15 14:48:19 +02:00
commit bee06f18f9
6 changed files with 4071 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

11
src/parser/mod.rs
View file

@ -48,6 +48,17 @@ fn select_text<'a>(el: ElementRef<'a>, selector: &Selector) -> Option<&'a str> {
.map(str::trim)
}
fn select_text_empty<'a>(el: ElementRef<'a>, selector: &Selector) -> Option<&'a str> {
el.select(selector)
.next()
.map(|item| {
item.text()
.find(|s| !s.trim().is_empty())
.unwrap_or_default()
})
.map(str::trim)
}
fn select_last_text<'a>(el: ElementRef<'a>, selector: &Selector) -> Option<&'a str> {
el.select(selector)
.next()

8
src/parser/team.rs
View file

@ -1,4 +1,4 @@
use super::{ElementExt, Parser};
use super::{select_text_empty, ElementExt, Parser};
use crate::data::{Membership, NameChange, Record, Team};
use crate::parser::{
select_text, steam_id_from_link, DATE_FORMAT, MEMBER_DATE_ALT_FORMAT, MEMBER_DATE_FORMAT,
@ -320,17 +320,17 @@ impl Parser for TeamParser {
.map(|row| {
let from_tag =
select_text(row, &self.selector_team_name_from_tag).unwrap_or_default();
let from_name = select_text(row, &self.selector_team_name_from_name).ok_or(
let from_name = select_text_empty(row, &self.selector_team_name_from_name).ok_or(
ParseError::ElementNotFound {
selector: SELECTOR_TEAM_NAME_FROM_NAME,
role: "team name change from name",
},
)?;
let to_tag = select_text(row, &self.selector_team_name_to_tag).unwrap_or_default();
let to_name = select_text(row, &self.selector_team_name_to_name).ok_or(
let to_name = select_text_empty(row, &self.selector_team_name_to_name).ok_or(
ParseError::ElementNotFound {
selector: SELECTOR_TEAM_NAME_TO_NAME,
role: "team name change from name",
role: "team name change to name",
},
)?;
let date = select_text(row, &self.selector_team_name_date).ok_or(