misc-mirrors/article_scraper
1
0
Fork 0
mirror of https://gitlab.com/news-flash/article_scraper.git synced 2025-07-07 16:15:32 +02:00
Code Releases Activity

fix alternative top candidate calcs

Browse source
This commit is contained in:
Jan Lukas Gernert 2023-02-28 18:28:01 +01:00
parent f4ccd22837
commit a1c07d436f
1 changed files with 18 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

22
src/full_text_parser/readability/mod.rs
View file

@ -279,9 +279,10 @@ impl Readability {
let mut alternative_candidate_ancestors = Vec::new();
// Find a better top candidate node if it contains (at least three) nodes which belong to `topCandidates` array
// and whose scores are quite closed with current `topCandidate` node.
for top_candidate in &top_candidates {
if let Some(score) = Self::get_content_score(top_candidate) {
if score >= 0.75 {
if let Some(top_score) = Self::get_content_score(&top_candidate) {
for candidate in top_candidates.iter().skip(1) {
let score = Self::get_content_score(candidate).unwrap_or(0.0);
if score / top_score >= 0.75 {
if let Some(ancestor) = top_candidate.get_parent() {
alternative_candidate_ancestors.push(ancestor);
}
@ -289,6 +290,7 @@ impl Readability {
}
}
if alternative_candidate_ancestors.len() >= constants::MINIMUM_TOPCANDIDATES {
let mut parent_of_top_candidate = top_candidate.get_parent();
@ -325,10 +327,22 @@ impl Readability {
let mut parent_of_top_candidate = top_candidate.get_parent();
let mut last_score = Self::get_content_score(&top_candidate).unwrap_or(0.0);
// let html = document.to_string_with_options(libxml::tree::SaveOptions {
// format: true,
// no_declaration: false,
// no_empty_tags: true,
// no_xhtml: false,
// xhtml: false,
// as_xml: false,
// as_html: true,
// non_significant_whitespace: false,
// });
// std::fs::write("doc.html", &html).unwrap();
// The scores shouldn't get too low.
let score_threshold = last_score / 3.0;
while Util::has_tag_name(parent_of_top_candidate.as_ref(), "BODY") {
while !Util::has_tag_name(parent_of_top_candidate.as_ref(), "BODY") {
if parent_of_top_candidate
.as_ref()
.map(|n| Self::get_content_score(n).is_none())