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

Spot fixes for Ward through Sundown 17.z

Browse source
This commit is contained in:
Domenic Denicola 2020-12-19 13:42:36 -05:00
commit 3e06358fa2
2 changed files with 283 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

18
lib/convert-worker.js
View file

@ -302,6 +302,9 @@ function fixTruncatedWords(xml) {
// Short for "Birdcage"
xml = xml.replace(/[‘’][Cc]age(?![a-z])/g, "’Cage");
// We can't do "’Clear" (short for Crystalclear) here because it appears too much as a normal word preceded by an
// open quote, so we do that in substitutions.json.
return xml;
}
@ -396,6 +399,7 @@ function enDashJointNames(xml) {
// Joint names should use en dashes
xml = xml.replace(/Dallon-Pelham/g, "Dallon–Pelham");
xml = xml.replace(/Bet-Gimel/g, "Bet–Gimel");
xml = xml.replace(/Cheit-Gimel/g, "Bet–Gimel");
xml = xml.replace(/Tristan-Capricorn/g, "Tristan–Capricorn");
xml = xml.replace(/Capricorn-Byron/g, "Capricorn–Byron");
xml = xml.replace(/Tristan-Byron/g, "Tristan–Byron");
@ -417,6 +421,8 @@ function enDashJointNames(xml) {
xml = xml.replace(/Crawler-Breed/g, "Crawler–Breed");
xml = xml.replace(/Simurgh-Myrddin-plant/g, "Simurgh–Myrddin–plant");
xml = xml.replace(/Armsmaster-Defiant/g, "Armsmaster–Defiant");
xml = xml.replace(/([Aa])gent-parahuman/g, "$1gent–parahuman");
xml = xml.replace(/([Pp])arahuman-agent/g, "$1arahuman–agent");
return xml;
}
@ -425,7 +431,7 @@ function fixPossessives(xml) {
// Fix possessive of names ending in "s".
xml = xml.replace(
// eslint-disable-next-line max-len
/(?<!‘)(Judas|Brutus|Jess|Aegis|Dauntless|Circus|Sirius|Brooks|Genesis|Atlas|Lucas|Gwerrus|Chris|Eligos|Animos|Mags|Huntress|Hephaestus|Lord of Loss|John Combs|Mama Mathers|Monokeros|Goddess|Boundless|Paris|Tress|Harris|Antares|Nieves|Backwoods|Midas|Mrs. Sims|Ms. Stillons|Chuckles|Amias|Semiramis)’(?!s)/g,
/(?<!‘)(Judas|Brutus|Jess|Aegis|Dauntless|Circus|Sirius|Brooks|Genesis|Atlas|Lucas|Gwerrus|Chris|Eligos|Animos|Mags|Huntress|Hephaestus|Lord of Loss|John Combs|Mama Mathers|Monokeros|Goddess|Boundless|Paris|Tress|Harris|Antares|Nieves|Backwoods|Midas|Mrs. Sims|Ms. Stillons|Chuckles|Amias|Semiramis|Mother of Mothers)’(?!s)/g,
"$1’s"
);
@ -468,11 +474,6 @@ function fixCapitalization(xml, book) {
// The Speedrunners team name is missing its capitalization a couple times.
xml = xml.replace(/speedrunners/g, "Speedrunners");
// Capitalization is inconsistent, but shard names seems to usually be capitalized.
xml = xml.replace(/Grasping self/g, "Grasping Self");
xml = xml.replace(/Cloven stranger/g, "Cloven Stranger");
xml = xml.replace(/Princess shaper/g, "Princess Shaper");
// "patrol block" is capitalized three different ways: "patrol block", "Patrol block", and "Patrol Block". "patrol
// group" is always lowercased. It seems like "Patrol" is a proper name, and is used as a capitalized modifier in
// other contexts (e.g. Patrol leader). So let's standardize on "Patrol <lowercase>".
@ -525,6 +526,11 @@ function fixCapitalization(xml, book) {
"$1–$2"
);
// Capitalization is inconsistent, but shard names seems to usually be capitalized.
xml = xml.replace(/Grasping self/g, "Grasping Self");
xml = xml.replace(/Cloven stranger/g, "Cloven Stranger");
xml = xml.replace(/Princess shaper/g, "Princess Shaper");
// Place names need to always be capitalized
xml = xml.replace(/North end/g, "North End");
xml = xml.replace(/(Stonemast|Shale) avenue/g, "$1 Avenue");

276
lib/substitutions.json
View file

@ -4162,6 +4162,10 @@
{
"before": "on the other earth",
"after": "on the other Earth"
},
{
"before": "name? did it start",
"after": "name? Did it start"
}
],
"https://parahumans.wordpress.com/2013/10/29/30-7/": [
@ -6227,6 +6231,10 @@
{
"before": "<em>apparently I</em>",
"after": "<em>apparently</em> I"
},
{
"before": "Hey Little V",
"after": "Hey little V"
}
],
"https://www.parahumans.net/2019/01/29/blinding-11-10/": [
@ -6272,7 +6280,7 @@
},
{
"before": "<em>Go you, Little</em> V.",
"after": "<em>Go you, Little V.</em>"
"after": "<em>Go you, little V.</em>"
},
{
"before": "the King on the chess board",
@ -6657,10 +6665,6 @@
{
"before": "For some of the heartbroken",
"after": "For some of the Heartbroken"
},
{
"before": "agent-parahuman",
"after": "agent–parahuman"
}
],
"https://www.parahumans.net/2019/04/20/black-13-6/": [
@ -7556,36 +7560,172 @@
{
"before": "Some—<em>many-</em> were unwillingly",
"after": "Some—<em>many</em>—were unwillingly"
},
{
"before": "looking down at my hand at her arm and my hand",
"after": "looking down at her arm and my hand"
},
{
"before": "more more thing hanging over our heads",
"after": "one more thing hanging over our heads"
},
{
"before": "my mothers cheeks",
"after": "my mother’s cheeks"
},
{
"before": "—Understand, please",
"after": "—understand, please"
},
{
"before": "“Victoria,” Crystal’s voice was gentle",
"after": "“Victoria.” Crystal’s voice was gentle"
}
],
"https://www.parahumans.net/2019/10/05/sundown-17-2/": [
{
"before": "egg for Crystal and mom",
"after": "egg for Crystal and Mom"
},
{
"before": "<p><strong>Me:</strong><br />",
"after": "<p style=\"padding-left: 30px;\"><strong>Me:</strong><br />"
},
{
"before": "<p><strong>Vista (Little V):</strong><br />",
"after": "<p style=\"padding-left: 30px;\"><strong>Vista (Little V):</strong><br />"
},
{
"before": "few minutes of peace, damn it.—",
"after": "few minutes of peace, damn it—"
},
{
"regExp": "‘Clear",
"replacement": "’Clear"
}
],
"https://www.parahumans.net/2019/10/08/sundown-17-3/": [
{
"before": "say, Math class",
"after": "say, math class"
},
{
"before": "hang outs",
"after": "hangouts"
},
{
"before": "<em>‘</em>",
"after": "’"
},
{
"before": "“—Which I’m not.”",
"after": "“—which I’m not.”"
},
{
"before": "Vista for the kid’s sake",
"after": "Vista for the kids’ sake"
},
{
"before": "Day to day activities",
"after": "Day-to-day activities"
},
{
"before": "tower crescent avenue",
"after": "Tower Crescent Avenue"
},
{
"before": "all had to deal with ‘heavy’.",
"after": "all had to deal with ‘heavy’.”"
},
{
"before": "stuff?” I spoke up, “it get into that control",
"after": "stuff?” I spoke up. “It gets into that control"
},
{
"before": "talked to me a like",
"after": "talked to me like"
},
{
"before": "call me Big V",
"after": "call me big V"
}
],
"https://www.parahumans.net/2019/10/12/sundown-17-4/": [
{
"before": "on resisting Master influence",
"after": "on resisting master influence"
},
{
"before": "CDs and Vinyls",
"after": "CDs and vinyls"
},
{
"before": "“P.R.T.”",
"after": "“PRT.”"
},
{
"before": "second or third string Wardens",
"after": "second- or third-string Wardens"
},
{
"before": "intern level clerks",
"after": "intern-level clerks"
},
{
"before": "half way",
"after": "halfway"
},
{
"before": "her tone didn’t soft",
"after": "her tone didn’t soften"
},
{
"before": "Jessica, Mr…",
"after": "Jessica, Mr.…"
},
{
"before": "—Captive",
"after": "—captive"
},
{
"before": "come from? what language",
"after": "come from? What language"
}
],
"https://www.parahumans.net/2019/10/15/sundown-17-5/": [
{
"before": "Is mom walking without difficulty",
"after": "Is Mom walking without difficulty"
},
{
"before": "Through valkyrie",
"after": "Through Valkyrie"
},
{
"before": "“‘Lo.”",
"after": "“’Lo.”"
}
],
"https://www.parahumans.net/2019/10/19/sundown-17-6/": [
{
"before": "Probably dad",
"after": "Probably Dad"
},
{
"before": "and no and no rejiggered rat",
"after": "and no rejiggered rat"
},
{
"before": "Harder headed",
"after": "Harder-headed"
},
{
"before": "<em>you proved my sister</em> right? <em>Tired,",
"after": "<em>you proved my sister </em>right</em>? Tired,"
},
{
"before": "boundary,” He whispered",
"after": "boundary,” he whispered"
}
],
"https://www.parahumans.net/2019/10/22/sundown-17-7/": [
@ -7596,6 +7736,10 @@
{
"before": "Golem explained</p>",
"after": "Golem explained.</p>"
},
{
"before": "see the Lab",
"after": "see the lab"
}
],
"https://www.parahumans.net/2019/10/26/sundown-17-8/": [
@ -7606,6 +7750,40 @@
{
"before": "She’s aggressive If anyone makes",
"after": "She’s aggressive. If anyone makes"
},
{
"before": "—They mobbed Lookout",
"after": "—they mobbed Lookout"
},
{
"before": "Of Course Kenzie",
"after": "Of course Kenzie"
},
{
"before": "Rain about Girls",
"after": "Rain about girls"
},
{
"before": "I typed: The <em>one",
"after": "I typed: <em>The one"
},
{
"before": "realize,” Seir called out. “If a fight",
"after": "realize,” Seir called out, “if a fight"
},
{
"before": "was emotion resistant",
"after": "was emotion-resistant"
}
],
"https://www.parahumans.net/2019/10/29/sundown-17-9/": [
{
"before": "—Are you going to be wishing",
"after": "—are you going to be wishing"
},
{
"before": "Asking question?",
"after": "Asking a question?"
}
],
"https://www.parahumans.net/2019/11/02/sundown-17-10/": [
@ -7616,6 +7794,40 @@
{
"before": "that didn’t betray too much</p>",
"after": "that didn’t betray too much.</p>"
},
{
"before": "say Molasses",
"after": "say molasses"
},
{
"before": "‘nother",
"after": "’nother"
},
{
"before": "‘she peed on me!’",
"after": "‘she peed on me!’."
},
{
"before": "madam Mayor",
"after": "Madam Mayor"
}
],
"https://www.parahumans.net/2019/11/05/interlude-17-x-sundown/": [
{
"before": "five-o’clock shadow",
"after": "five o’clock shadow"
},
{
"before": "nervous looking guy",
"after": "nervous-looking guy"
},
{
"before": "you have <em>drugs? Alcohol? Cigarettes</em>? I have",
"after": "you have <em>drugs</em>? <em>Alcohol?</em> <em>Cigarettes?</em> I have"
},
{
"before": "Not by mom or dad",
"after": "Not by Mom or Dad"
}
],
"https://www.parahumans.net/2019/11/09/interlude-17-y-sundown/": [
@ -7626,6 +7838,14 @@
{
"before": "new titan",
"after": "new Titan"
},
{
"before": "anti parahumans",
"after": "anti-parahumans"
},
{
"before": "hit the Anti-parahumans",
"after": "hit the anti-parahumans"
}
],
"https://www.parahumans.net/2019/11/12/interlude-17-z-sundown/": [
@ -7636,6 +7856,42 @@
{
"regExp": "titan(?![a-z])",
"replacement": "Titan"
},
{
"before": "What if you <em>weren’t?</em>",
"after": "What if you <em>weren’t</em>?"
},
{
"before": "Victor, Can you go join Tribute",
"after": "Victor, can you go join Tribute"
},
{
"before": "The forcefield above him elaborate and getting",
"after": "The forcefield above him was elaborate and getting"
},
{
"before": "Their faculties were diminished, but</p>",
"after": "Their faculties were diminished, but—</p>"
},
{
"before": "Flashbang, he saw was one of the",
"after": "Flashbang, he saw, was one of the"
},
{
"before": "the palm, Out each foot, and out of the mouth.lifting one",
"after": "the palm. Out each foot, and out of the mouth. Lifting one"
},
{
"before": "no eyes; only a blindfold",
"after": "no eyes, only a blindfold"
},
{
"before": "lightness of a feather landing, came to perch",
"after": "lightness of a feather landing, it came to perch"
},
{
"before": "…The only things that are",
"after": "…the only things that are"
}
],
"https://www.parahumans.net/2019/11/16/radiation-18-1/": [
@ -7816,6 +8072,16 @@
{
"before": "the horrifying titan",
"after": "the horrifying Titan"
},
{
"before": "that stranger",
"after": "that Stranger"
}
],
"https://www.parahumans.net/2020/01/07/infrared-19-4/": [
{
"before": "talked to them? or done",
"after": "talked to them? Or done"
}
],
"https://www.parahumans.net/2020/01/11/infrared-19-b/": [