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

Restore original scene breaks, instead of using <hr>

Browse source
This commit is contained in:
Domenic Denicola 2020-11-07 16:12:41 -05:00
commit 547b7e9518
2 changed files with 16 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

22
lib/convert-worker.js
View file

@ -288,15 +288,19 @@ function getBodyXML(chapter, book, contentEl) {
xml = xml.replace(/Simurgh-Myrddin-plant/g, "Simurgh–Myrddin–plant");
xml = xml.replace(/Armsmaster-Defiant/g, "Armsmaster–Defiant");
// 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 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><\/p>/g, "<hr />");
// Normalize scene breaks. <hr> would be more semantically appropriate, but loses the author's intent. This is
// especially the case in Ward, which uses a variety of different scene breaks.
xml = xml.replace(/<p(?:[^>]*)>■<\/p>/g, `<p style="text-align: center;">■</p>`);
xml = xml.replace(/<p style="text-align: center;">⊙<\/p>/g, `<p style="text-align: center;">⊙</p>`);
xml = xml.replace(/<p style="text-align: center;"><strong>⊙<\/strong><\/p>/g, `<p style="text-align: center;">⊙</p>`);
xml = xml.replace(/<p style="text-align: center;"><em><strong>⊙<\/strong><\/em><\/p>/g,
`<p style="text-align: center;">⊙</p>`);
xml = xml.replace(/<p style="text-align: center;"><strong>⊙⊙<\/strong><\/p>/g,
`<p style="text-align: center;">⊙</p>`);
xml = xml.replace(/<p style="text-align: center;"><strong>⊙ *⊙ *⊙ *⊙ *⊙<\/strong><\/p>/g,
`<p style="text-align: center;">⊙ ⊙ ⊙ ⊙ ⊙</p>`);
// Fix recurring miscapitalization with questions
xml = xml.replace(/\?”\s\s?She asked/g, "?” she asked");

6
lib/substitutions.json
View file

@ -5,7 +5,7 @@
"after": "bugs: flies, ants"
},
{
"before": "<p><em>Brief note from the author: This story isn’t intended for young or sensitive readers. Readers who are on the lookout for <a href=\"http://www.urbandictionary.com/define.php?term=Trigger%20warning&amp;defid=6195009\">trigger warnings</a> are advised to give Worm a pass.</em></p>\n<hr />\n",
"before": "<p><em>Brief note from the author: This story isn’t intended for young or sensitive readers. Readers who are on the lookout for <a href=\"http://www.urbandictionary.com/define.php?term=Trigger%20warning&amp;defid=6195009\">trigger warnings</a> are advised to give Worm a pass.</em></p>\n<p style=\"text-align: center;\">■</p>\n",
"after": "",
"_comment": "The pseudo-trigger warning is out of place in an eBook."
}
@ -4334,7 +4334,7 @@
],
"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<hr />",
"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;\">⊙</p>\n",
"after": "",
"_comment": "This is out of place in an eBook."
}
@ -5013,7 +5013,7 @@
],
"https://www.parahumans.net/2018/06/05/torch-7-5/": [
{
"before": "<hr />\n<p><em>I don’t",
"before": "<p style=\"text-align: center;\">⊙</p>\n<p><em>I don’t",
"after": "<p><em>I don’t"
},
{