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

Spot fixes for Arc 9, Sentinel; additional general fixes

Browse source 
The additional general fixes include:

* Fix "tv" and "T.V." to become "TV" like most other uses in the book
* Fix "I-I" to "I—I" and "I-uh" to "I—uh" conversions which weren't working properly.
* Add periods to the end of some paragraphs that were missing them.
* More of the never-ending italics and spacing work.
This commit is contained in:
Domenic Denicola 2017-12-22 13:39:44 -05:00
commit b9f79429b1
2 changed files with 351 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

19
lib/convert.js
View file

@ -157,6 +157,11 @@ function getBodyXML(chapter, contentEl) {
xml = xml.replace(/<em>‘([^<]+)’<\/em>/g, "‘<em>$1</em>’");
xml = xml.replace(/<em>‘([^<]+)<\/em>’/g, "‘<em>$1</em>’");
xml = xml.replace(/‘<em>([^<]+)’<\/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>");
xml = xml.replace(/<\/em> <\/p>/g, "</em></p>");
xml = xml.replace(/<em>([a-z]+),<\/em>/g, "<em>$1</em>,");
// Fix recurring poor quotes and apostrophes
xml = xml.replace(/<p>”/g, "<p>“");
@ -192,8 +197,8 @@ function getBodyXML(chapter, contentEl) {
xml = xml.replace(/-\s\s?/g, "—");
xml = xml.replace(/\s?\s-/g, "—");
xml = xml.replace(/\s+—”/g, "—”");
xml = xml.replace(/I-I”/g, "I—I");
xml = xml.replace(/I-uh”/g, "I—uh");
xml = xml.replace(/I-I/g, "I—I");
xml = xml.replace(/I-uh/g, "I—uh");
// Use <hr> for separators
xml = xml.replace(/<p>■<\/p>/g, "<hr/>");
@ -210,6 +215,10 @@ function getBodyXML(chapter, contentEl) {
xml = xml.replace(/ \.<\/p>/g, ".</p>");
xml = xml.replace(/\.<em>\.\./g, "<em>…");
// Fix extra spaces
xml = xml.replace(/ ? <\/p>/g, "</p>");
xml = xml.replace(/([a-z]) ,/g, "$1,");
// 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...
// This should instead be
@ -233,9 +242,13 @@ function getBodyXML(chapter, contentEl) {
// it's incorrect to capitalize in the one-off fixes.
xml = xml.replace(/the clairvoyant/g, "the Clairvoyant");
// This is sometimes missing its capitalization
// This is sometimes missing its capitalization.
xml = xml.replace(/the birdcage/g, "the Birdcage");
// This is usually spelled "TV" but sometimes the other ways. Normalize.
xml = xml.replace(/tv/g, "TV");
xml = xml.replace(/T\.V\./g, "TV");
// 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");

354
lib/substitutions.json
View file

@ -62,10 +62,6 @@
{
"before": "sue you, he gasped",
"after": "sue you,” he gasped"
},
{
"before": "<em>principles,</em> Glory",
"after": "<em>principles</em>, Glory"
}
],
"https://parahumans.wordpress.com/2011/07/05/insinuation-2-1/": [
@ -172,6 +168,12 @@
"after": "Lisa continued, “is"
}
],
"https://parahumans.wordpress.com/2011/09/03/agitation-3-8/": [
{
"before": "direction</p>",
"after": "direction.</p>"
}
],
"https://parahumans.wordpress.com/2011/09/10/agitation-3-10/": [
{
"before": "meaning his aim was",
@ -524,7 +526,7 @@
],
"https://parahumans.wordpress.com/2012/01/10/tangle-6-9/": [
{
"before": "I’m a<em> prisoner?</em>",
"before": "I’m a <em>prisoner?</em>",
"after": "I’m a <em>prisoner</em>?"
},
{
@ -550,6 +552,10 @@
{
"before": "It-it’s",
"after": "It—it’s"
},
{
"before": "envelope</p>",
"after": "envelope.</p>"
}
],
"https://parahumans.wordpress.com/2012/01/14/interlude-6/": [
@ -686,6 +692,10 @@
{
"before": "chimed in, “She",
"after": "chimed in. “She"
},
{
"before": "wounds</p>",
"after": "wounds.</p>"
}
],
"https://parahumans.wordpress.com/2012/02/18/buzz-7-10/": [
@ -820,7 +830,7 @@
"after": "A third person was gravely injured"
},
{
"regExp": "\n<em>\\s*([^<]+)</em>(<br/>|</p>)",
"regExp": "\n ?<em>\\s*([^<]+)</em>(<br/>|</p>)",
"replacement": "\n<p style=\"padding-left:30px;\"><i>$1</i></p>"
},
{
@ -869,14 +879,6 @@
"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,"
@ -1000,6 +1002,10 @@
{
"before": "old, dirt and blood",
"after": "old dir and blood"
},
{
"before": "collapse</p>",
"after": "collapse.</p>"
}
],
"https://parahumans.wordpress.com/2012/03/20/extermination-8-6/": [
@ -1210,6 +1216,18 @@
{
"before": "three Master classifications",
"after": "three master classifications"
},
{
"before": "files, “Here’s",
"after": "files. “Here’s"
},
{
"before": "youtube",
"after": "YouTube"
},
{
"before": "ground, “It",
"after": "ground, “it"
}
],
"https://parahumans.wordpress.com/2012/04/07/sentinel-9-2/": [
@ -1220,6 +1238,86 @@
{
"before": "Master 6",
"after": "master six"
},
{
"before": "build for a professional",
"after": "build up a professional"
},
{
"before": "need me, k?",
"after": "need me, ’kay?"
},
{
"before": "gentle walk</p>",
"after": "gentle walk.</p>"
},
{
"before": "real hair</p>",
"after": "real hair.</p>"
}
],
"https://parahumans.wordpress.com/2012/04/10/sentinel-9-3/": [
{
"before": "with my family</em>.",
"after": "with my family.</em>"
},
{
"before": "industrialized ones—Some of you",
"after": "industrialized ones—some of you"
},
{
"before": "the room, “You’ve",
"after": "the room, “you’ve"
},
{
"before": "dislike him</em>?",
"after": "dislike him?</em>"
},
{
"before": "answered, annoyed, “ He gets",
"after": "answered, annoyed. “He gets"
},
{
"before": "spoke, “Broad",
"after": "spoke. “Broad"
},
{
"before": "tipped his hat, “Then",
"after": "tipped his hat. “Then"
},
{
"before": "of this</em> guy.",
"after": "of this guy.</em>"
}
],
"https://parahumans.wordpress.com/2012/04/14/sentinel-9-4/": [
{
"before": "more focused one</em>.</p>",
"after": "more focused one.</em></p>"
},
{
"before": "groaning, “Fuck,",
"after": "groaning. “Fuck,"
},
{
"before": "smiled a little,",
"after": "smiled a little."
},
{
"before": "greeted him, “You’re",
"after": "greeted him. “You’re"
},
{
"before": "going to</em> get less interested. <em>Shit,",
"after": "going to get <em>less</em> interested. Shit,"
},
{
"before": "commented, “My tax",
"after": "commented. “My tax"
},
{
"before": "scratched-up hands, “No",
"after": "scratched-up hands. “No"
}
],
"https://parahumans.wordpress.com/2012/04/17/sentinel-9-5/": [
@ -1230,6 +1328,44 @@
{
"before": "That,” Piggot spoke. “And",
"after": "That,” Piggot spoke, “and"
},
{
"before": "hands together, “Thank",
"after": "hands together. “Thank"
},
{
"before": "Chariot replied, I’m sorry",
"after": "Chariot replied; I’m sorry"
},
{
"before": "said something</em>.",
"after": "said something.</em>"
},
{
"before": "across the room, “It was",
"after": "across the room. “It was"
}
],
"https://parahumans.wordpress.com/2012/04/21/sentinel-9-6/": [
{
"before": "Good—The woman’s",
"after": "Good—the woman’s"
},
{
"before": "face.</em> Shadow Stalker thought, <em>Records",
"after": "face,</em> Shadow Stalker thought. <em>Records"
},
{
"before": "Charlie horse",
"after": "charlie horse"
},
{
"before": "sarcastic, “How wonderfully",
"after": "sarcastic, “how wonderfully"
},
{
"before": "“…Things are",
"after": "“…things are"
}
],
"https://parahumans.wordpress.com/2012/04/28/parasite-10-2/": [
@ -1332,6 +1468,40 @@
"after": "my ‘candy’</em>"
}
],
"https://parahumans.wordpress.com/2012/06/22/interlude-11g/": [
{
"before": "<p><em>J’adore—</em></p>",
"after": "<p><i>J’adore—</i></p>"
},
{
"before": "<p><em>Sweet Honey—</em></p>",
"after": "<p><i>Sweet Honey—</i></p>"
},
{
"before": "<p><em>Love me, love me, you know you wanna love me…</em><br/>\n<em>Love me, love me, you know you wanna love me…</em></p>",
"after": "<p><i>Love me, love me, you know you wanna love me…<br/>\nLove me, love me, you know you wanna love me…</i></p>"
},
{
"before": "<p><em>Love me, you?<br/>\nLove me, true?</em></p>",
"after": "<p><i>Love me, you?<br/>\nLove me, true?</i></p>"
},
{
"before": "<p><em>Crazed, kooky, cracked, crazy,</em> <br/>\n<em>Nutty, barmy, mad for me…</em></p>",
"after": "<p><i>Crazed, kooky, cracked, crazy,<br/>\nNutty, barmy, mad for me…</i></p>"
},
{
"before": "<p><em><em>Crazed, kooky, cracked, crazy,</em><br/>\nMental, dotty, whacked, loopy…</em></p>",
"after": "<p><i>Crazed, kooky, cracked, crazy,<br/>\nMental, dotty, whacked, loopy…</i></p>"
},
{
"before": "<p><em>Crazed, kooky, cracked, crazy,<br/>\n<em>Nutty, screwy, mentally diseased…</em><br/>\n</em> She ",
"after": "<p><i>Crazed, kooky, cracked, crazy,<br/>\nNutty, screwy, mentally diseased…</i></p>\n<p>She "
},
{
"before": "<em>Ça va?</em>",
"after": "<i>Ça va?</i>"
}
],
"https://parahumans.wordpress.com/2012/06/26/plague-12-1/": [
{
"before": "need: Maybe",
@ -1392,6 +1562,10 @@
{
"before": "doesn’t,” Madcap said. “But",
"after": "doesn’t,” Madcap said, “but"
},
{
"before": "Second task.—c</p>",
"after": "Second task. —c</p>"
}
],
"https://parahumans.wordpress.com/2012/07/28/snare-13-1/": [
@ -1442,6 +1616,10 @@
{
"before": "Coil,” I said. “You",
"after": "Coil,” I said, “you"
},
{
"before": "<em>“Speak for",
"after": "“<em>Speak for"
}
],
"https://parahumans.wordpress.com/2012/08/21/snare-13-8/": [
@ -1512,6 +1690,10 @@
{
"before": "countermeasures,” I said. “We",
"after": "countermeasures,” I said, “we"
},
{
"before": "through the suit</em>?",
"after": "through the suit?</em>"
}
],
"https://parahumans.wordpress.com/2012/10/09/prey-14-11/": [
@ -1564,6 +1746,10 @@
{
"before": "rich,” Regent said. “Coming",
"after": "rich,” Regent said, “coming"
},
{
"before": "the street</p>",
"after": "the street.</p>"
}
],
"https://parahumans.wordpress.com/2012/10/27/colony-15-4/": [
@ -1584,6 +1770,10 @@
{
"before": "this,” Regent said. “I",
"after": "this,” Regent said, “I"
},
{
"before": "wonder <em>why?</em>",
"after": "wonder <em>why</em>?"
}
],
"https://parahumans.wordpress.com/2012/11/03/colony-15-6/": [
@ -1604,6 +1794,10 @@
{
"before": "Brooks,” I said. “Treat",
"after": "Brooks,” I said, “treat"
},
{
"before": "being kept</p>",
"after": "being kept.</p>"
}
],
"https://parahumans.wordpress.com/2012/11/08/interlude-15-donation-bonus-3/": [
@ -1716,6 +1910,12 @@
"after": "decisions: holding back"
}
],
"https://parahumans.wordpress.com/2012/12/22/monarch-16-9/": [
{
"before": "aisles</p>",
"after": "aisles.</p>"
}
],
"https://parahumans.wordpress.com/2012/12/25/monarch-16-10/": [
{
"before": "it,” Trickster said. “You",
@ -1812,7 +2012,7 @@
"after": "been replaying"
},
{
"before": "sort of stuff that , and",
"before": "sort of stuff that, and",
"after": "sort of stuff that ???, and"
},
{
@ -1840,6 +2040,14 @@
{
"before": "happens,” Krouse said. “Blame",
"after": "happens,” Krouse said, “blame"
},
{
"before": "one hallway</p>",
"after": "one hallway.</p>"
},
{
"before": "the glass vial</p>",
"after": "the glass vial.</p>"
}
],
"https://parahumans.wordpress.com/2013/01/14/migration-17-7/": [
@ -1858,6 +2066,10 @@
{
"before": "the guy: A brown",
"after": "the guy: a brown"
},
{
"before": "exaggerated shrug</p>",
"after": "exaggerated shrug.</p>"
}
],
"https://parahumans.wordpress.com/2013/01/15/migration-17-8/": [
@ -1886,6 +2098,10 @@
{
"before": "speaking,” I said. “Just",
"after": "speaking,” I said, “just"
},
{
"before": "or another</p>",
"after": "or another.</p>"
}
],
"https://parahumans.wordpress.com/2013/01/22/queen-18-2/": [
@ -2033,6 +2249,10 @@
{
"before": "better,” I said. “Eidolon",
"after": "better,” I said, “Eidolon"
},
{
"before": "giving her <em>us.</em>",
"after": "giving her <em>us</em>."
}
],
"https://parahumans.wordpress.com/2013/02/14/interlude-18-donation-bonus-4/": [
@ -2067,6 +2287,10 @@
{
"before": "anything,” Shamrock said. “I",
"after": "anything,” Shamrock said, “I"
},
{
"before": "the phone</p>",
"after": "the phone.</p>"
}
],
"https://parahumans.wordpress.com/2013/02/19/scourge-19-1/": [
@ -2127,7 +2351,7 @@
],
"https://parahumans.wordpress.com/2013/03/12/scourge-19-7/": [
{
"before": "they<em> did, but how we</em> dealt",
"before": "they <em>did, but how we</em> dealt",
"after": "they <em>did</em>, but <em>how</em> we dealt"
},
{
@ -2162,6 +2386,16 @@
{
"before": "you,” I said. “But",
"after": "you,” I said, “but"
},
{
"before": "completely recovered</em>.",
"after": "completely recovered.</em>"
}
],
"https://parahumans.wordpress.com/2013/03/23/chrysalis-20-2/": [
{
"before": "of anxiety</em>.",
"after": "of anxiety.</em>"
}
],
"https://parahumans.wordpress.com/2013/03/26/chrysalis-20-3/": [
@ -2347,6 +2581,10 @@
{
"before": "in,” I said. “Could",
"after": "in,” I said, “could"
},
{
"before": " “ <em>We’re",
"after": "“<em>We’re"
}
],
"https://parahumans.wordpress.com/2013/05/16/cell-22-6/": [
@ -2361,6 +2599,10 @@
{
"before": "capes,” Chevalier said. “Burdened",
"after": "capes,” Chevalier said, “burdened"
},
{
"before": "of it.</em>",
"after": "of it</em>."
}
],
"https://parahumans.wordpress.com/2013/05/18/interlude-22/": [
@ -2372,10 +2614,14 @@
"before": "it,” he said. “She",
"after": "it,” he said, “she"
},
{
"before": "“ I don’t",
"after": "“I don’t"
},
{
"before": "confirmed okay",
"after": "confirmed ok",
"_comment": "This is a forum post title and thus is probably informal and not event capitalized"
"_comment": "This is a forum post title and thus is probably informal and not even capitalized"
}
],
"https://parahumans.wordpress.com/2013/05/21/interlude-22-donation-bonus-1/": [
@ -2386,11 +2632,15 @@
{
"before": "frank,” Teacher said. “I",
"after": "frank,” Teacher said, “I"
},
{
"before": "their ears</p>",
"after": "their ears.</p>"
}
],
"https://parahumans.wordpress.com/2013/05/25/drone-23-1/": [
{
"before": "Defiant spoke , “Let’s",
"before": "Defiant spoke, “Let’s",
"after": "Defiant spoke. “Let’s"
},
{
@ -2532,6 +2782,10 @@
{
"before": "sheets pillows",
"after": "sheets, pillows"
},
{
"before": "Golem responded</p>",
"after": "Golem responded.</p>"
}
],
"https://parahumans.wordpress.com/2013/07/02/scarab-25-2/": [
@ -2554,6 +2808,10 @@
{
"before": "<em>bugs I don’t like</em>",
"after": "<em>bugs</em> I don’t like"
},
{
"before": "than this</em>.",
"after": "than this.</em>"
}
],
"https://parahumans.wordpress.com/2013/07/11/scarab-25-5/": [
@ -2592,6 +2850,12 @@
"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",
@ -2606,6 +2870,10 @@
{
"before": "And gauntlet</em>, to reply to the last one",
"after": "And gauntlet, to reply to the last one</em>"
},
{
"before": "a burst of alarm</em>.",
"after": "a burst of alarm.</em>"
}
],
"https://parahumans.wordpress.com/2013/07/23/sting-26-3/": [
@ -2676,6 +2944,10 @@
{
"before": "you,” he said. “I",
"after": "you,” he said, “I"
},
{
"before": "froze him</em>.",
"after": "froze him.</em>"
}
],
"https://parahumans.wordpress.com/2013/08/10/interlude-26/": [
@ -2686,6 +2958,18 @@
{
"before": "rain: The",
"after": "rain: the"
},
{
"before": "from one another</p>",
"after": "from one another.</p>"
},
{
"before": "fragmenting in greater numbers</p>",
"after": "fragmenting in greater numbers.</p>"
},
{
"before": "these past few years</p>",
"after": "these past few years.</p>"
}
],
"https://parahumans.wordpress.com/2013/08/13/extinction-27-1/": [
@ -2710,12 +2994,24 @@
{
"before": "Besides,” Tattletale said. “The",
"after": "Besides,” Tattletale said, “the"
},
{
"before": "shift position</p>",
"after": "shift position.</p>"
},
{
"before": "“ Convenient.",
"after": "“Convenient."
}
],
"https://parahumans.wordpress.com/2013/08/20/extinction-27-3/": [
{
"before": "guess,” Sophia said. “You",
"after": "guess,” Sophia said, “you"
},
{
"before": "said “ But",
"after": "said. “But"
}
],
"https://parahumans.wordpress.com/2013/08/22/extinction-27-4/": [
@ -2750,6 +3046,10 @@
{
"before": "Weavers",
"after": "Weaver’s"
},
{
"before": "Contessa’s power</em>.",
"after": "Contessa’s power.</em>"
}
],
"https://parahumans.wordpress.com/2013/08/31/cockroaches-28-1/": [
@ -2808,7 +3108,7 @@
"after": "No reaction from the Simurgh"
},
{
"before": "One hundred and eighty<em> integers",
"before": "One hundred and eighty <em>integers",
"after": "<em>One hundred and eighty integers"
},
{
@ -2876,6 +3176,10 @@
{
"before": "<em>willingly</em>,” Weld said. “Before",
"after": "<em>willingly</em>,” Weld said, “before"
},
{
"before": "<em>“Listen, we’re",
"after": "“<em>Listen, we’re"
}
],
"https://parahumans.wordpress.com/2013/09/19/venom-29-1/": [
@ -2986,6 +3290,10 @@
{
"before": "thing,” I said. “Was",
"after": "thing,” I said, “was"
},
{
"before": "by Golem’s bindings</p>",
"after": "by Golem’s bindings.</p>"
}
],
"https://parahumans.wordpress.com/2013/10/12/interlude-29/": [
@ -3030,6 +3338,10 @@
{
"before": "anyone alleged with",
"after": "anyone allied with"
},
{
"before": "coordinating two teams</em>.",
"after": "coordinating two teams.</em>"
}
],
"https://parahumans.wordpress.com/2013/10/22/speck-30-4/": [
@ -3308,6 +3620,10 @@
{
"before": "camera follows”",
"after": "camera follows.”"
},
{
"before": "background of his mind</p>",
"after": "background of his mind.</p>"
}
],
"https://parahumans.wordpress.com/2013/11/19/interlude-end/": [