organicmaps/wikiparser
13
Fork 0
Code Issues 13 Pull requests 2 Projects Releases Wiki Activity Actions

Remove header ids

Browse source 
Signed-off-by: Evan Lloyd New-Schmidt <evan@new-schmidt.com>
This commit is contained in:
Evan Lloyd New-Schmidt 2023-08-15 18:06:56 -04:00 • committed by Evan Lloyd New-Schmidt
parent b96c2cf4db
commit c7fe34f3ad
3 changed files with 12 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/html.rs
View file

@ -1,7 +1,7 @@
use std::collections::{BTreeMap, BTreeSet};
use ego_tree::NodeId;
use markup5ever::{local_name, LocalName, Namespace, QualName};
use markup5ever::{LocalName, Namespace, QualName};
use once_cell::sync::Lazy;
use scraper::{ElementRef, Html, Node, Selector};
use serde::Deserialize;
@ -263,12 +263,12 @@ fn remove_attrs(document: &mut Html) {
}
}
for (k, v) in el.attrs.iter() {
if (k.local == local_name!("id") && v.starts_with("mw"))
|| k.local.starts_with("data-mw")
|| ["prefix", "typeof", "about", "rel"]
for (k, _v) in el.attrs.iter() {
if k.local.starts_with("data-mw")
// TODO: To keep ids for linking to headers, only remove ones that start with "mw".
|| ["id", "prefix", "typeof", "about", "rel"]
.iter()
.any(|id| *id == &k.local)
.any(|attr| *attr == &k.local)
{
to_remove.push(k.to_owned());
}

6
tests/data/Q4185820-en/output.html
View file

@ -5,7 +5,7 @@
<p>
It has been described as ‘the most important public building in Ireland’ by late Nobel laureate Seamus Heaney.
</p>
<h2 id="History">
<h2>
History
</h2>
<p>
@ -32,7 +32,7 @@
<p>
Mary Hanley (1914-1979) was the founder of the Kiltartan Society. A native of Carron, County Clare, Hanley founded the society in 1961 to foster interest in the literary history of the district, especially that of Lady Gregory, Edward Martyn and W.B. Yeats. She was responsible for the restoration of Thoor Ballylee (with the aid of Bord Fáilte and the Yeats family). At the time, the Office of Public Works was owner of the property. Hanley persuaded the poet Padraic Colum to open the castle on Sunday 20 June 1965, the centenary of Yeats’s birth, as <i>Yeats Tower</i> to appear as it was when he lived there and refitted as a Yeats museum containing a collection of first editions and items of furniture. The adjoining miller's cottage became a tea room and shop. This was later expanded by a newly constructed building in the back.
</p>
<h2 id="Today">
<h2>
Today
</h2>
<p>
@ -50,7 +50,7 @@
<p>
In early December 2015, Storm Desmond devastated parts of Ireland with flooding rain and damaging winds. Thoor Ballylee, and the adjacent cottage, were both damaged by several feet of flood water.
</p>
<h2 id="Architecture">
<h2>
Architecture
</h2>
<p>

6
tests/data/Q748282-en/output.html
View file

@ -36,7 +36,7 @@
Qarabiy yayla
</li>
</ul>
<h2 id="Highest_peaks">
<h2>
Highest peaks
</h2>
<p>
@ -65,7 +65,7 @@
Ai-Petri (Ukrainian: <span lang="uk">Ай-Петрі</span>, Russian: <span lang="ru">Ай-Петри</span>, Crimean Tatar: <span lang="crh">Ay Petri</span>) 1,234 m in the Ay Petri Yaylası.
</li>
</ul>
<h2 id="Passes_and_rivers">
<h2>
Passes and rivers
</h2>
<p>
@ -85,7 +85,7 @@
<p>
Rivers of the Crimean Mountains include the Alma River, Chernaya River, and Salhir River on the northern slope and Uchan-su River on the southern slope which forms the Uchan-su waterfall, and the highest waterfall in Crimea.
</p>
<h2 id="History">
<h2>
History
</h2>
<p>