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

Spot fixes for Arc 8, Extermination; additional general fixes

Browse source 
The additional general fixes include:

* Fixes to period and ellipsis usage (e.g. removing spaces before periods)
* Decapitalization of "loft"
* Hyphenation of "golden-skinned"
* Decapitalization after colons
This commit is contained in:
Domenic Denicola 2017-11-27 03:06:35 +01:00
commit 9c9fe2a936
2 changed files with 341 additions and 35 deletions
Download patch file Download diff file Expand all files Collapse all files

38
lib/convert.js
View file

@ -203,8 +203,12 @@ function getBodyXML(chapter, contentEl) {
xml = xml.replace(/\?”\s\s?She asked/g, "?” she asked");
xml = xml.replace(/\?”\s\s?He asked/g, "?” he asked");
// Fix extra periods at the end of paragraphs
// Fix bad periods and spacing/markup surrounding them
xml = xml.replace(/\.\.<\/p>/g, ".</p>");
xml = xml.replace(/\.\.”<\/p>/g, ".”</p>");
xml = xml.replace(/ \. /g, ". ");
xml = xml.replace(/ \.<\/p>/g, ".</p>");
xml = xml.replace(/\.<em>\.\./g, "<em>…");
// The author often fails to terminate a sentence, instead using a comma after a dialogue tag. For example,
// > “I didn’t get much done,” Greg said, “I got distracted by...
@ -232,8 +236,9 @@ function getBodyXML(chapter, contentEl) {
// This is sometimes missing its capitalization
xml = xml.replace(/the birdcage/g, "the Birdcage");
// There's no reason why this should be capitalized. (Note that it never appears at the beginning of any sentences.)
// There's no reason why these should be capitalized. (Note that they never appear at the beginning of any sentences.)
xml = xml.replace(/Halberd/g, "halberd");
xml = xml.replace(/Loft/g, "loft");
// Especially early in the story, PRT designations are capitalized; they should not be. This fixes the cases where we
// can be reasonably sure they don't start a sentence, although more specific instances are done in
@ -277,22 +282,29 @@ function getBodyXML(chapter, contentEl) {
// Signal(l)ing/signal(l)ed are spelled both ways. Both are acceptable in English. Let's standardize on single-L.
xml = xml.replace(/(S|s)ignall/g, "$1ignal");
// This is consistently missing hyphens.
// These are consistently missing hyphens.
xml = xml.replace(/shoulder mounted/g, "shoulder-mounted");
xml = xml.replace(/golden skinned/, "golden-skinned");
// One-off fixes
(substitutions[chapter.url] || []).forEach(substitution => {
const indexOf = xml.indexOf(substitution.before);
if (indexOf === -1) {
console.warn(`Could not find text "${substitution.before}" in ${chapter.url}. The chapter may have been ` +
`updated at the source, in which case, you should edit substitutions.json.`);
}
if (indexOf !== xml.lastIndexOf(substitution.before)) {
console.warn(`The text "${substitution.before}" occurred twice, and so the substitution was ambiguous. ` +
`Update substitutions.json for a more precise substitution.`);
}
if (substitution.before) {
const indexOf = xml.indexOf(substitution.before);
if (indexOf === -1) {
console.warn(`Could not find text "${substitution.before}" in ${chapter.url}. The chapter may have been ` +
`updated at the source, in which case, you should edit substitutions.json.`);
}
if (indexOf !== xml.lastIndexOf(substitution.before)) {
console.warn(`The text "${substitution.before}" occurred twice, and so the substitution was ambiguous. ` +
`Update substitutions.json for a more precise substitution.`);
}
xml = xml.replace(new RegExp(escapeRegExp(substitution.before)), substitution.after);
xml = xml.replace(new RegExp(escapeRegExp(substitution.before)), substitution.after);
} else if (substitution.regExp) {
xml = xml.replace(new RegExp(substitution.regExp, "g"), substitution.replacement);
} else {
console.warn(`Invalid substitution specified for ${chapter.url}`);
}
});
// Serializer inserts extra xmlns for us since it doesn't know we're going to put this into a <html>

338
lib/substitutions.json
View file

@ -78,6 +78,10 @@
{
"before": "simply, “Bug”",
"after": "simply, ‘Bug’."
},
{
"before": "powers: He could",
"after": "powers: he could"
}
],
"https://parahumans.wordpress.com/2011/07/12/insinuation-2-3/": [
@ -276,6 +280,10 @@
{
"before": "Not so unusual,",
"after": "Not so unusual;"
},
{
"before": "offer is this: Let me",
"after": "offer is this: let me"
}
],
"https://parahumans.wordpress.com/2011/10/15/shell-4-7/": [
@ -450,6 +458,10 @@
{
"before": "South",
"after": "south"
},
{
"before": "news: A tally",
"after": "news: a tally"
}
],
"https://parahumans.wordpress.com/2011/12/20/tangle-6-3/": [
@ -726,6 +738,10 @@
{
"before": "East",
"after": "east"
},
{
"before": "<em>listened to,</em>",
"after": "<em>listened to</em>."
}
],
"https://parahumans.wordpress.com/2012/03/06/extermination-8-2/": [
@ -736,9 +752,25 @@
{
"before": "West",
"after": "west"
},
{
"before": "And villains, too",
"after": "and villains, too"
},
{
"before": "featureless plain",
"after": "featureless plane"
}
],
"https://parahumans.wordpress.com/2012/03/08/interlude-7%C2%BD-bonus/": [
{
"before": "‘boyfriend’; Not",
"after": "‘boyfriend’; not"
},
{
"before": "thought &amp; the",
"after": "thought and the"
},
{
"before": "legs: Suit",
"after": "legs: suit"
@ -758,6 +790,104 @@
{
"before": "arms: Irritation",
"after": "arms: irritation"
},
{
"before": "goodwill bin",
"after": "Goodwill bin"
},
{
"before": "chedcked",
"after": "checked"
},
{
"before": "An envelope should arrive",
"after": "an envelope should arrive"
},
{
"before": "<em>Which</em>",
"after": "Which"
},
{
"before": "spoke, “My supervisor",
"after": "spoke. “My supervisor"
},
{
"before": "serious, “Turn",
"after": "serious. “Turn"
},
{
"before": "A third person gravely injured",
"after": "A third person was gravely injured"
},
{
"regExp": "\n<em>\\s*([^<]+)</em>(<br/>|</p>)",
"replacement": "\n<p style=\"padding-left:30px;\"><i>$1</i></p>"
},
{
"regExp": "\n<p style=\"padding-left:30px;\"><em>([^<]+)</em>(<br/>|</p>)",
"replacement": "\n<p style=\"padding-left:30px;\"><i>$1</i></p>"
},
{
"regExp": "\n<p style=\"padding-left:30px;\"><em>([^<\n]+)<br/>\n([^>\n]+)</em><br/>\n",
"replacement": "\n<p style=\"padding-left:30px;\"><i>$1</i></p>\n<p style=\"padding-left:30px;\"><i>$2</i></p>\n"
}
],
"https://parahumans.wordpress.com/2012/03/10/extermination-8-3/": [
{
"before": "Icouldn’t",
"after": "I couldn’t"
},
{
"before": "CD-5</em>…",
"after": "CD-5…</em>"
},
{
"before": "CD-5</em></p>",
"after": "CD-5.</em></p>"
},
{
"before": "<p>Brigandine deceased, CD-5.</p>",
"after": "<p><em>Brigandine deceased, CD-5.</em></p>"
},
{
"before": "deceased,</em> CD-6.",
"after": "deceased, CD-6.</em>"
},
{
"before": "CD-6,</em> The armbands",
"after": "CD-6</em>, the armbands"
},
{
"before": "CD-6</em></p>",
"after": "CD-6.</em></p>"
},
{
"before": "CD-6</em>.</p>",
"after": "CD-6.</em></p>"
},
{
"before": "<em>pushing.</em>",
"after": "<em>pushing</em>."
},
{
"before": "<em> Fuck</em>",
"after": " <em>Fuck</em>"
},
{
"before": "<em> Losses",
"after": " <em>Losses"
},
{
"before": "<em>Narwhal,</em>",
"after": "Narwhal,"
},
{
"before": "Tattletale</em>…",
"after": "Tattletale…</em>"
},
{
"before": "reeled—He",
"after": "reeled—he"
}
],
"https://parahumans.wordpress.com/2012/03/13/extermination-8-4/": [
@ -792,6 +922,34 @@
{
"before": "half South",
"after": "half south"
},
{
"before": "impression It had",
"after": "impression it had"
},
{
"before": "Aegis deceased</em>, CD-6",
"after": "Aegis deceased, CD-6</em>"
},
{
"before": "down,</em> CC-6.",
"after": "down, CC-6.</em>"
},
{
"before": "Kid win",
"after": "Kid Win"
},
{
"before": "of ‘a few blocks’</em>.",
"after": "of ‘a few blocks’.</em>"
},
{
"before": "<em>Negative</em>.",
"after": "<em>Negative.</em>"
},
{
"before": "<em>up.</em>",
"after": "<em>up</em>."
}
],
"https://parahumans.wordpress.com/2012/03/17/extermination-8-5/": [
@ -806,6 +964,60 @@
{
"before": "Master 5",
"after": "master five"
},
{
"before": "down, BW-8</em>",
"after": "down, BW-8.</em>"
},
{
"before": "deceased, BW-8</em>",
"after": "deceased, BW-8.</em>"
},
{
"before": "deceased,</em> CB-10.",
"after": "deceased, CB-10.</em>"
},
{
"before": "<em>Defensive perimeter, report</em>.",
"after": "<em>Defensive perimeter, report.</em>"
},
{
"before": "<em>Enemy location unknown,</em>",
"after": "<em>Enemy location unknown</em>,"
},
{
"before": "<em>Dad</em>.",
"after": "<em>Dad.</em>"
},
{
"before": "<em>The girl</em>.",
"after": "<em>The girl.</em>"
},
{
"before": "goggles, it",
"after": "goggles. It"
},
{
"before": "old, dirt and blood",
"after": "old dir and blood"
}
],
"https://parahumans.wordpress.com/2012/03/20/extermination-8-6/": [
{
"before": "and It—it’s",
"after": "and it—it’s"
},
{
"before": "glaring at me, “There’s",
"after": "glaring at me. “There’s"
},
{
"before": "<em>Ah</em>.",
"after": "<em>Ah.</em>"
},
{
"before": "<em>“All of us?”</em>",
"after": "“<em>All of us?</em>”"
}
],
"https://parahumans.wordpress.com/2012/03/24/extermination-8-7/": [
@ -824,22 +1036,94 @@
{
"before": "Master 5",
"after": "Master five"
}
],
"https://parahumans.wordpress.com/2012/03/08/interlude-7%c2%bd-bonus/": [
{
"before": "‘boyfriend’; Not",
"after": "‘boyfriend’; not"
},
{
"before": "thought &amp; the",
"after": "thought and the"
"before": "<em>This working? This on? Good.</em>",
"after": "<em>“This working? This on? Good.”</em>"
},
{
"before": "<em>Acknowledged</em>. Her",
"after": "<em>Acknowledged</em>, her"
},
{
"before": "struck me—If",
"after": "struck me—if"
},
{
"before": "went on, “Just",
"after": "went on, “just"
},
{
"before": "her, “That",
"after": "her, “that"
},
{
"before": "towards her, “Heroes died",
"after": "towards her. “Heroes died"
},
{
"before": "we get get in",
"after": "we get in"
},
{
"before": "at the heroine, “Let us",
"after": "at the heroine. “Let us"
},
{
"before": "ignoring Tattletale, “Frustrated",
"after": "ignoring Tattletale. “Frustrated"
},
{
"before": "humility, “If it",
"after": "humility. “If it"
},
{
"before": "hand at me, “She’s",
"after": "hand at me, “she’s"
}
],
"https://parahumans.wordpress.com/2012/03/27/extermination-8-8/": [
{
"before": "stick—hawk or some other",
"after": "stick ‘hawk’ or some other"
},
{
"before": "this: Knowing we had",
"after": "this: knowing we had"
},
{
"before": "of her head, “And my",
"after": "of her head. “And my"
}
],
"https://parahumans.wordpress.com/2012/03/31/interlude-8/": [
{
"before": "other: Another",
"after": "other: another"
},
{
"before": "reality: He",
"after": "reality: he"
},
{
"before": "prius",
"after": "Prius"
},
{
"before": "room,” it was",
"after": "room.” It was"
},
{
"before": "handicap accessible",
"after": "handicap-accessible"
},
{
"before": "“Hey,” her voice",
"after": "“Hey.” Her voice"
},
{
"before": "Mr. Pitter, “The Travelers’",
"after": "Mr. Pitter. “The Travelers’"
}
],
"https://parahumans.wordpress.com/2012/04/03/cell-9-1/": [
@ -990,6 +1274,12 @@
"after": "was… how"
}
],
"https://parahumans.wordpress.com/2012/05/26/infestation-11-3/": [
{
"before": "intimidating: A sea",
"after": "intimidating: a sea"
}
],
"https://parahumans.wordpress.com/2012/05/29/infestation-11-4/": [
{
"before": "of facility</em>‘",
@ -1420,6 +1710,12 @@
"after": "debate,” Kurt said, “people"
}
],
"https://parahumans.wordpress.com/2012/12/18/monarch-16-8/": [
{
"before": "decisions: Holding back",
"after": "decisions: holding back"
}
],
"https://parahumans.wordpress.com/2012/12/25/monarch-16-10/": [
{
"before": "it,” Trickster said. “You",
@ -1541,10 +1837,6 @@
"before": "MWBB <em>",
"after": "<em>MWBB "
},
{
"before": "kill us..",
"after": "kill us."
},
{
"before": "happens,” Krouse said. “Blame",
"after": "happens,” Krouse said, “blame"
@ -1562,6 +1854,10 @@
{
"before": "Well,” Krouse said. “Do",
"after": "Well,” Krouse said, “do"
},
{
"before": "the guy: A brown",
"after": "the guy: a brown"
}
],
"https://parahumans.wordpress.com/2013/01/15/migration-17-8/": [
@ -1663,8 +1959,8 @@
],
"https://parahumans.wordpress.com/2013/02/07/interlude-18-donation-bonus-3/": [
{
"before": "But..",
"after": "But…"
"before": "fighting. But.",
"after": "fighting. But…"
},
{
"before": "r,s",
@ -1895,10 +2191,6 @@
"before": "’20th’",
"after": "‘20th’"
},
{
"before": ".<em>..The reality",
"after": "<em>…The reality"
},
{
"before": "—A known murderer",
"after": "—a known murderer"
@ -2155,6 +2447,12 @@
"after": "it,” Imp said, “she’s"
}
],
"https://parahumans.wordpress.com/2013/06/06/interlude-23/": [
{
"before": "her mouth: A click",
"after": "her mouth: a click"
}
],
"https://parahumans.wordpress.com/2013/06/08/crushed-24-1/": [
{
"before": "both of you” I said",
@ -2848,10 +3146,6 @@
{
"before": "moved Case fifty-threes",
"after": "moved case fifty-threes"
},
{
"before": "future sight of his .",
"after": "future sight of his."
}
],
"https://parahumans.wordpress.com/2013/10/26/speck-30-6/": [