jyapayne/worm-scraper
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Normalize Ward separators to <hr>

Browse source
This commit is contained in:
Domenic Denicola 2020-07-01 17:21:49 -04:00
commit ff991028af
2 changed files with 14 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

6
lib/convert.js
View file

@ -233,8 +233,14 @@ function getBodyXML(chapter, contentEl) {
xml = xml.replace(/I-uh/g, "I—uh"); xml = xml.replace(/I-uh/g, "I—uh");
// Use <hr> for separators // Use <hr> for separators
// https://www.parahumans.net/2019/12/21/interlude-18-z-radiation/ has "super-separators" ("⊙ ⊙ ⊙ ⊙ ⊙") which we
// leave untouched for now.
xml = xml.replace(/<p>■<\/p>/g, "<hr/>"); xml = xml.replace(/<p>■<\/p>/g, "<hr/>");
xml = xml.replace(/<p style="text-align:center;">■<\/p>/g, "<hr/>"); xml = xml.replace(/<p style="text-align:center;">■<\/p>/g, "<hr/>");
xml = xml.replace(/<p style="text-align: center;">⊙<\/p>/g, "<hr/>");
xml = xml.replace(/<p style="text-align: center;"><strong>⊙<\/strong><\/p>/g, "<hr/>");
xml = xml.replace(/<p style="text-align: center;"><em><strong>⊙<\/strong><\/em><\/p>/g, "<hr/>");
xml = xml.replace(/<p style="text-align: center;"><strong>⊙<\/strong><strong>⊙<\/strong><\/p>/g, "<hr/>");
// Fix recurring miscapitalization with questions // Fix recurring miscapitalization with questions
xml = xml.replace(/\?”\s\s?She asked/g, "?” she asked"); xml = xml.replace(/\?”\s\s?She asked/g, "?” she asked");

9
lib/substitutions.json
View file

@ -4118,7 +4118,7 @@
], ],
"https://www.parahumans.net/2017/09/11/daybreak-1-1/": [ "https://www.parahumans.net/2017/09/11/daybreak-1-1/": [
{ {
"before": "<p>Ward is the second work in the Parahumans series, and reading <strong><a href=\"https://parahumans.wordpress.com/\">Worm</a></strong> first is strongly recommended. A lot of this won’t make sense otherwise and if you do find yourself a fan of the universe, the spoilers in Ward will affect the reading of the other work.</p>\n<p>Ward is not recommended for young or sensitive readers.</p>\n<p style=\"text-align: center;\"><strong>⊙</strong></p>", "before": "<p>Ward is the second work in the Parahumans series, and reading <strong><a href=\"https://parahumans.wordpress.com/\">Worm</a></strong> first is strongly recommended. A lot of this won’t make sense otherwise and if you do find yourself a fan of the universe, the spoilers in Ward will affect the reading of the other work.</p>\n<p>Ward is not recommended for young or sensitive readers.</p>\n<hr/>",
"after": "", "after": "",
"_comment": "This is out of place in an eBook." "_comment": "This is out of place in an eBook."
}, },
@ -4138,5 +4138,12 @@
"before": "blunt instrument-I saw the", "before": "blunt instrument-I saw the",
"after": "blunt instrument—I saw the" "after": "blunt instrument—I saw the"
} }
],
"https://www.parahumans.net/2018/05/26/torch-7-4/": [
{
"before": "<p style=\"text-align: center;\"><a href=\"https://www.parahumans.net/2018/05/26/eclipse-x-1/\"><strong>⊙</strong></a></p>",
"after": "",
"_comment": "This appears to be intentional, but it just duplicates the Next Chapter link (i.e. turning the page in the eBook), so we remove it"
}
] ]
} }