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

Standardize on one style for "Case Fifty-Three" and friends

Browse source
This commit is contained in:
Domenic Denicola 2020-11-14 17:23:22 -05:00
commit 1a2b6de78e
2 changed files with 26 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

26
lib/convert-worker.js
View file

@ -236,6 +236,7 @@ function getBodyXML(chapter, book, contentEl) {
xml = fixMispellings(xml); xml = fixMispellings(xml);
xml = fixHyphens(xml); xml = fixHyphens(xml);
xml = standardizeSpellings(xml); xml = standardizeSpellings(xml);
xml = fixCaseNumbers(xml);
// One-off fixes // One-off fixes
for (const substitution of substitutions[chapter.url] || []) { for (const substitution of substitutions[chapter.url] || []) {
@ -563,6 +564,31 @@ function standardizeSpellings(xml) {
return xml; return xml;
} }
function fixCaseNumbers(xml) {
// Case numbers are very inconsistent. For "Case Fifty-Three", the breakdown is:
// * 9 Case-53
// * 6 Case 53
// * 2 case-53
// * 1 Case-Fifty-Three
// * 41 Case Fifty-Three
// * 1 Case Fifty Three
// * 13 Case fifty-three
// * 119 case fifty-three
// * 4 case-fifty-three
// * 1 case fifty three
// We standardize on "Case Fifty-Three"; although it isn't the most common, it seems best to treat these as proper
// nouns.
xml = xml.replace(/case[ -](?:fifty[ -]three|53)(?!’)/ig, "Case Fifty-Three");
xml = xml.replace(/case[ -](?:thirty[ -]two|53)(?!’)/ig, "Case Thirty-Two");
xml = xml.replace(/case[ -](?:sixty[ -]nine|53)(?!’)/ig, "Case Sixty-Nine");
xml = xml.replace(/(?<!in )case[ -](zero|one|two|three|four|twelve|fifteen|seventy|ninety)(?!-)/ig,
(_, caseNumber) => "Case " + caseNumber[0].toUpperCase() + caseNumber.substring(1));
return xml;
}
function isEmptyOrGarbage(el) { function isEmptyOrGarbage(el) {
const text = el.textContent.trim(); const text = el.textContent.trim();
return text === "" || return text === "" ||

16
lib/substitutions.json
View file

@ -3723,10 +3723,6 @@
"before": "I suspect It’s a", "before": "I suspect It’s a",
"after": "I suspect it’s a" "after": "I suspect it’s a"
}, },
{
"before": "Case Fifty-three",
"after": "case fifty-three"
},
{ {
"before": "reality,” Tattletale said. “What’s", "before": "reality,” Tattletale said. “What’s",
"after": "reality,” Tattletale said, “what’s" "after": "reality,” Tattletale said, “what’s"
@ -3751,10 +3747,6 @@
} }
], ],
"https://parahumans.wordpress.com/2013/09/17/interlude-28/": [ "https://parahumans.wordpress.com/2013/09/17/interlude-28/": [
{
"before": "Case fifty-threes",
"after": "case fifty-threes"
},
{ {
"before": "than,” Revel paused. “Six", "before": "than,” Revel paused. “Six",
"after": "than,” Revel paused, “six" "after": "than,” Revel paused, “six"
@ -4045,14 +4037,6 @@
"before": "gun build", "before": "gun build",
"after": "gun built" "after": "gun built"
}, },
{
"before": "hunchbacked Case fifty-three",
"after": "hunchbacked case fifty-three"
},
{
"before": "moved Case fifty-threes",
"after": "moved case fifty-threes"
},
{ {
"before": "I revoked my control over her, leaving in in the middle", "before": "I revoked my control over her, leaving in in the middle",
"after": "I revoked my control over her, leaving her in the middle" "after": "I revoked my control over her, leaving her in the middle"