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

Fix many erroneously-italicized closing quotes

Browse source
This commit is contained in:
Domenic Denicola 2020-11-14 18:13:12 -05:00
commit d2402b1bb2
2 changed files with 31 additions and 17 deletions
Download patch file Download diff file Expand all files Collapse all files

8
lib/convert-worker.js
View file

@ -161,9 +161,9 @@ function getBodyXML(chapter, book, contentEl) {
xml = xml.replace(/<em>‘\s*([^<]+)\s*’<\/em>/g, "‘<em>$1</em>’");
xml = xml.replace(/<em>‘\s*([^<]+)\s*<\/em>\s*’/g, "‘<em>$1</em>’");
xml = xml.replace(/‘\s*<em>\s*([^<]+)\s*’<\/em>/g, "‘<em>$1</em>’");
xml = xml.replace(/<em>“\s*([^<]+)\s*”<\/em>/g, "“<em>$1</em>”");
xml = xml.replace(/<em>“\s*([^<]+)\s*<\/em>\s*”/g, "“<em>$1</em>”");
xml = xml.replace(/“\s*<em>\s*([^<]+)\s*”<\/em>/g, "“<em>$1</em>”");
xml = xml.replace(/<em>“\s*([^<”]+)\s*”<\/em>/g, "“<em>$1</em>”");
xml = xml.replace(/<em>“\s*([^<”]+)\s*<\/em>\s*”/g, "“<em>$1</em>”");
xml = xml.replace(/“\s*<em>\s*([^<”]+)\s*”<\/em>/g, "“<em>$1</em>”");
xml = xml.replace(/([^\n>])<em> ?/g, "$1 <em>");
xml = xml.replace(/ ?<\/em>/g, "</em> ");
xml = xml.replace(/<p([^>]+)> <em>/g, "<p$1><em>");
@ -197,6 +197,8 @@ function getBodyXML(chapter, book, contentEl) {
xml = xml.replace(/‘<em>([^<]+)<\/em>‘/g, "‘<em>$1</em>’");
xml = xml.replace(/I”m/g, "I’m");
xml = xml.replace(/<em>([a-z]+)!<\/em>/g, "<em>$1</em>!");
xml = xml.replace(/(?<! {2})<em>([\w ’]+)([!.?])”<\/em>/g, "<em>$1</em>$2”");
xml = xml.replace(/<em>([\w ’]+[!.?])”<\/em>/g, "<em>$1</em>”");
// Similar problems occur in Ward with <b> and <strong> as do in Worm with <em>s
xml = xml.replace(/<b \/>/g, "");

40
lib/substitutions.json
View file

@ -2544,6 +2544,10 @@
{
"before": "city,” Jess said. “And",
"after": "city,” Jess said, “and"
},
{
"before": "“No<em>!”</em>",
"after": "“No!”"
}
],
"https://parahumans.wordpress.com/2013/01/10/migration-17-3/": [
@ -3438,6 +3442,10 @@
{
"before": "White capped",
"after": "White-capped"
},
{
"before": "<em>. “Bosses are worried.”</em>",
"after": ". “<em>Bosses are worried</em>.”"
}
],
"https://parahumans.wordpress.com/2013/07/09/scarab-25-4/": [
@ -3486,12 +3494,6 @@
"after": "avoided"
}
],
"https://parahumans.wordpress.com/2013/07/16/interlude-25/": [
{
"before": "your <em>manners?”</em>",
"after": "your <em>manners</em>?”"
}
],
"https://parahumans.wordpress.com/2013/07/18/sting-26-1/": [
{
"before": "up,” I said. “The",
@ -5798,10 +5800,6 @@
{
"before": "Be safe—Nat.",
"after": "Be safe —Nat."
},
{
"before": "<em>kill each other.”</em>",
"after": "<em>kill each other</em>.”"
}
],
"https://www.parahumans.net/2018/11/17/polarize-10-8/": [
@ -5904,10 +5902,6 @@
{
"before": "how it were more active",
"after": "how it was more active"
},
{
"before": "running <em>to the same places.”</em>",
"after": "running <em>to the same places</em>.”"
}
],
"https://www.parahumans.net/2018/12/04/polarize-10-12/": [
@ -6287,6 +6281,10 @@
{
"before": "you have have surmised",
"after": "you have surmised"
},
{
"before": "pound of flesh, <em>at least—!”</em>",
"after": "pound of flesh, <em>at least</em>—!”"
}
],
"https://www.parahumans.net/2019/04/02/black-13-1/": [
@ -6649,6 +6647,10 @@
{
"before": "looking for mom in the crowd",
"after": "looking for Mom in the crowd"
},
{
"before": "“E<em>ntrapment</em>.”",
"after": "“<em>Entrapment.</em>”"
}
],
"https://www.parahumans.net/2019/11/23/radiation-18-3/": [
@ -6803,6 +6805,10 @@
{
"before": "you know about” my Aunt Sarah said",
"after": "you know about,” my Aunt Sarah said"
},
{
"before": "so they grow <em>in…”</em>",
"after": "so they grow <em>in</em>…”"
}
],
"https://www.parahumans.net/2020/02/11/infrared-19-f/": [
@ -6841,6 +6847,12 @@
"after": "across this clearing. Eyes, cameras"
}
],
"https://www.parahumans.net/2020/03/07/last-20-4/": [
{
"before": "peace<em>!”</em> Cryptid growled",
"after": "peace!” Cryptid growled"
}
],
"https://www.parahumans.net/2020/03/31/last-20-10/": [
{
"before": "shaping it as it rolled out. it became a circular",