From ba387d3555c7a9ec3ace14bacf7e4667462b1268 Mon Sep 17 00:00:00 2001
From: Domenic Denicola <d@domenic.me>
Date: Fri, 1 Jan 2021 16:24:33 -0500
Subject: [PATCH] Improve deletion of empty-ish elements

The previous heuristic of replacing them with a space character caused spaces to be inserted in the middle of words. Also, various cases were missed. This should help.
---
 lib/convert-worker.js  | 26 +++++++++++++++++++-------
 lib/substitutions.json | 41 +++++++++--------------------------------
 2 files changed, 28 insertions(+), 39 deletions(-)

diff --git a/lib/convert-worker.js b/lib/convert-worker.js
index f46312a..19a0cd4 100644
--- a/lib/convert-worker.js
+++ b/lib/convert-worker.js
@@ -77,13 +77,21 @@ function getBodyXML(chapter, book, contentEl) {
     }
   }
 
-  // Remove empty <em>s and <i>s
-  // Remove style attributes from them, as they're always messed up.
-  for (const em of contentEl.querySelectorAll("em, i")) {
-    if (em.textContent.trim() === "") {
-      em.replaceWith(contentEl.ownerDocument.createTextNode(" "));
+  // Remove empty inline elements.
+  // Remove style attributes from inline elements, as they're always messed up.
+  for (const el of contentEl.querySelectorAll("em, i, strong, b")) {
+    const { textContent } = el;
+
+    if (textContent === "") {
+      el.remove();
+    } else if (textContent.trim() === "") {
+      if (el.childElementCount === 0) {
+        el.replaceWith(" ");
+      } else if (el.childElementCount === 1 && el.children[0].localName === "br") {
+        el.outerHTML = "<br />\n";
+      }
     } else {
-      em.removeAttribute("style");
+      el.removeAttribute("style");
     }
   }
 
@@ -221,9 +229,13 @@ function getBodyXML(chapter, book, contentEl) {
   xml = xml.replace(/(\s*)<\/strong>/g, "</strong>$1");
   xml = xml.replace(/><strong>(.*)<\/strong>:</g, "><strong>$1:</strong><");
 
-  // No need for line breaks before paragraph ends
+  // No need for line breaks before paragraph ends or after paragraph starts
   // These often occur with the <br>s inside <b>/<strong>/<em>/<i> fixed above.
   xml = xml.replace(/<br \/>\s*<\/p>/g, "</p>");
+  xml = xml.replace(/<p><br \/>\s*/g, "<p>");
+
+  // This is another quote fix but it needs to happen after the line break deletion... so entangled, ugh.
+  xml = xml.replace(/<\/em>\s*“\s*<\/p>/g, "</em>”</p>");
 
   // Fix missing spaces after commas
   xml = xml.replace(/([a-zA-Z]+),([a-zA-Z]+)/g, "$1, $2");
diff --git a/lib/substitutions.json b/lib/substitutions.json
index 38aa1c0..30f027e 100644
--- a/lib/substitutions.json
+++ b/lib/substitutions.json
@@ -1713,12 +1713,6 @@
       "after": "Dinah being kidnapped, and leaving"
     }
   ],
-  "https://parahumans.wordpress.com/2012/05/22/infestation-11-2/": [
-    {
-      "before": "attentio n",
-      "after": "attention"
-    }
-  ],
   "https://parahumans.wordpress.com/2012/05/26/infestation-11-3/": [
     {
       "before": "intimidating:  A sea",
@@ -1943,7 +1937,7 @@
       "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 ",
+      "before": "<p><em>Crazed, kooky, cracked, crazy,<br />\n<em>Nutty, screwy, mentally diseased…</em><br />\n<br />\n</em>She ",
       "after": "<p><i>Crazed, kooky, cracked, crazy,<br />\nNutty, screwy, mentally diseased…</i></p>\n<p>She "
     },
     {
@@ -2765,11 +2759,6 @@
     {
       "before": "stranger class",
       "after": "stranger-class"
-    },
-    {
-      "before": "real ly",
-      "after": "really",
-      "comment": "There's an empty em element in the middle of this word that gets converted to a space"
     }
   ],
   "https://parahumans.wordpress.com/2013/02/05/monarch-18-6/": [
@@ -3284,10 +3273,6 @@
     {
       "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/": [
@@ -3317,10 +3302,6 @@
       "before": "it,” he said. “She",
       "after": "it,” he said, “she"
     },
-    {
-      "before": "“ I don’t",
-      "after": "“I don’t"
-    },
     {
       "before": "confirmed okay",
       "after": "confirmed ok",
@@ -3579,6 +3560,10 @@
     {
       "before": "avoided-",
       "after": "avoided"
+    },
+    {
+      "before": "<span style=\"text-decoration:underline;\"><strong>Lord</strong></span><span style=\"text-decoration:underline;\"> <strong>Walston</strong></span>",
+      "after": "<span style=\"text-decoration:underline;\"><strong>Lord Walston</strong></span>"
     }
   ],
   "https://parahumans.wordpress.com/2013/07/18/sting-26-1/": [
@@ -3649,8 +3634,8 @@
   ],
   "https://parahumans.wordpress.com/2013/07/30/sting-26-5/": [
     {
-      "before": "</em>  Wait…",
-      "after": "  Wait…</em>"
+      "before": "</em>  Wait…“</p>",
+      "after": "  Wait…</em>”</p>"
     },
     {
       "before": "This,” Imp said. “Is",
@@ -3758,10 +3743,6 @@
     {
       "before": "shift position</p>",
       "after": "shift position.</p>"
-    },
-    {
-      "before": "“ Convenient.",
-      "after": "“Convenient."
     }
   ],
   "https://parahumans.wordpress.com/2013/08/20/extinction-27-3/": [
@@ -3770,8 +3751,8 @@
       "after": "guess,” Sophia said, “you"
     },
     {
-      "before": "said  “ But",
-      "after": "said. “But"
+      "before": "“Maybe,” I said  “But",
+      "after": "“Maybe,” I said.  “But"
     }
   ],
   "https://parahumans.wordpress.com/2013/08/24/extinction-27-5/": [
@@ -3954,10 +3935,6 @@
     {
       "before": "<em>Once the bead was in place, every</em> bullet hit.",
       "after": "Once the bead was in place, every bullet hit."
-    },
-    {
-      "before": "I ‘m",
-      "after": "I’m"
     }
   ],
   "https://parahumans.wordpress.com/2013/09/24/venom-29-3/": [