Fix autocaps around brackets and other delimiters

ajyoon · May 12, 2024 · 6c32031 · 6c32031
1 parent dec25b8
commit 6c32031
Show file tree

Hide file tree

Showing 3 changed files with 239 additions and 217 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,8 +3,9 @@
 ### 0.1.10 (unreleased)
 * Improve interactive view error/warning capture
 * Update capitalization cleanup postprocessing to support sentences
-  ending before quotation marks. For instance, `"Test." test` will now
-  correct to `"Test." Test` where previously it would not.
+  ending before quotation marks and other matching delimiters like brackets.
+  For instance, `"Test." test` will now correct to `"Test." Test` where
+  previously it would not.
 * Fix visual line breaks when backslash is followed by whitespace
 * Update punctuation cleanup postprocessing to shift punctuation to the
   right of quotes (single, double, or underscores). For example `"test" .`

diff --git a/src/postprocessing.ts b/src/postprocessing.ts
@@ -12,63 +12,63 @@ const TRAILING_WHITESPACE_RE = /\s+$/;
  * 5. Ensuring the text ends with a single line break
  */
 export function whitespaceCleanup(text: string): string {
-    let out = '';
-    let atDocStart = true;
-    let lastLineWasBlank = false;
-    for (let line of text.split('\n')) {
-        let isBlank = BLANK_LINE_RE.test(line);
-
-        if (atDocStart) {
-            if (isBlank) {
-                // Skip blank lines at start of document
-                continue;
-            } else {
-                atDocStart = false;
-            }
-        }
+  let out = '';
+  let atDocStart = true;
+  let lastLineWasBlank = false;
+  for (let line of text.split('\n')) {
+    let isBlank = BLANK_LINE_RE.test(line);
+
+    if (atDocStart) {
+      if (isBlank) {
+        // Skip blank lines at start of document
+        continue;
+      } else {
+        atDocStart = false;
+      }
+    }
 
-        if (isBlank) {
-            if (lastLineWasBlank) {
-                // Skip runs of blank lines
-                continue;
-            }
-            // Lines consisting of only whitespace should
-            // become simply blank lines
-            line = '';
+    if (isBlank) {
+      if (lastLineWasBlank) {
+        // Skip runs of blank lines
+        continue;
+      }
+      // Lines consisting of only whitespace should
+      // become simply blank lines
+      line = '';
+    } else {
+      // intra-line cleanups
+      line = line.replace(TRAILING_WHITESPACE_RE, '');
+      let rewrittenLine = '';
+      let atLineStart = true;
+      let lastCharWasSpace = false;
+      for (let char of line) {
+        let charIsSpace = char === ' ';
+        if (!atLineStart && lastCharWasSpace && charIsSpace) {
+          continue;
         } else {
-            // intra-line cleanups
-            line = line.replace(TRAILING_WHITESPACE_RE, '');
-            let rewrittenLine = '';
-            let atLineStart = true;
-            let lastCharWasSpace = false;
-            for (let char of line) {
-                let charIsSpace = char === ' ';
-                if (!atLineStart && lastCharWasSpace && charIsSpace) {
-                    continue;
-                } else {
-                    if (!charIsSpace) {
-                        atLineStart = false;
-                    }
-                    rewrittenLine += char;
-                    lastCharWasSpace = charIsSpace;
-                }
-            }
-            line = rewrittenLine;
+          if (!charIsSpace) {
+            atLineStart = false;
+          }
+          rewrittenLine += char;
+          lastCharWasSpace = charIsSpace;
         }
+      }
+      line = rewrittenLine;
+    }
 
-        lastLineWasBlank = isBlank;
+    lastLineWasBlank = isBlank;
 
-        out += line + '\n';
-    }
+    out += line + '\n';
+  }
 
-    // Edge case: if input ended with a line break already, above code
-    // will result in \n\n ending the output. Correct this so output
-    // always terminates with a single \n
-    if (out.endsWith('\n\n')) {
-        out = out.substring(0, out.length - 1);
-    }
+  // Edge case: if input ended with a line break already, above code
+  // will result in \n\n ending the output. Correct this so output
+  // always terminates with a single \n
+  if (out.endsWith('\n\n')) {
+    out = out.substring(0, out.length - 1);
+  }
 
-    return out;
+  return out;
 }
 
 // Note the 3 dashes here are the different kinds, not the same character
@@ -86,12 +86,12 @@ const MISPLACED_WORD_ENDING_PUNC_RE = /([a-zA-Z0-9\xA0-\uFFFF"'_])(\s+)([.,:;!?\
  * dependent on style and context.
  */
 export function punctuationCleanup(text: string): string {
-    return text.replace(MISPLACED_WORD_ENDING_PUNC_RE, '$1$3$2');
+  return text.replace(MISPLACED_WORD_ENDING_PUNC_RE, '$1$3$2');
 }
 
 
 // \p{Ll} matches unicode lowercase letters which have uppercase variants.
-const INCORRECT_CAPS_RE = /([.!?]["'_]?\s+|^\s*)(\p{Ll})/gu;
+const INCORRECT_CAPS_RE = /([.!?]["'_\]\)\}\s]*?\s+|^\s*)(\p{Ll})/gu;
 
 
 /**
@@ -101,18 +101,18 @@ const INCORRECT_CAPS_RE = /([.!?]["'_]?\s+|^\s*)(\p{Ll})/gu;
  * following a sentence-ending punctuation mark.
  */
 export function capitalizationCleanup(text: string): string {
-    // Conforms to `text.replace` replacer function interface
-    function correctCaps(_match: string, p1: string, p2: string) {
-        return p1 + p2.toUpperCase();
-    }
+  // Conforms to `text.replace` replacer function interface
+  function correctCaps(_match: string, p1: string, p2: string) {
+    return p1 + p2.toUpperCase();
+  }
 
-    return text.replace(INCORRECT_CAPS_RE, correctCaps);
+  return text.replace(INCORRECT_CAPS_RE, correctCaps);
 }
 
 const VISUAL_LINE_BREAK_RE = /\\ *(\r?\n|\r)[ \t]*/g
 
 export function replaceVisualLineBreaks(text: string): string {
-    return text.replace(VISUAL_LINE_BREAK_RE, ' ');
+  return text.replace(VISUAL_LINE_BREAK_RE, ' ');
 }
 
 const INDEFINITE_ARTICLE_RE = /\b(a|an) ([\p{L}0-9]+)\b/igu
@@ -121,27 +121,27 @@ const INDEFINITE_ARTICLE_RE = /\b(a|an) ([\p{L}0-9]+)\b/igu
  * Attempt to correct English indefinite articles (a / an)
  */
 export function correctIndefiniteArticles(text: string) {
-    function upcaseFirstLetter(s: string): string {
-        if (s.length === 0) {
-            return s;
-        } else if (s.length === 1) {
-            return s.toUpperCase();
-        } else {
-            return s[0].toUpperCase() + s.slice(1);
-        }
+  function upcaseFirstLetter(s: string): string {
+    if (s.length === 0) {
+      return s;
+    } else if (s.length === 1) {
+      return s.toUpperCase();
+    } else {
+      return s[0].toUpperCase() + s.slice(1);
     }
-    // Conforms to `text.replace` replacer function interface
-    function correctArticle(_match: string, originalArticle: string, word: string) {
-        let article = indefinite(word, { articleOnly: true });
-        if (originalArticle === 'a' || originalArticle === 'an') {
-            return article + ' ' + word;
-        } else if (originalArticle === 'A' || originalArticle === 'An') {
-            return upcaseFirstLetter(article) + ' ' + word;
-        } else {
-            // All caps
-            return article.toUpperCase() + ' ' + word;
-        }
+  }
+  // Conforms to `text.replace` replacer function interface
+  function correctArticle(_match: string, originalArticle: string, word: string) {
+    let article = indefinite(word, { articleOnly: true });
+    if (originalArticle === 'a' || originalArticle === 'an') {
+      return article + ' ' + word;
+    } else if (originalArticle === 'A' || originalArticle === 'An') {
+      return upcaseFirstLetter(article) + ' ' + word;
+    } else {
+      // All caps
+      return article.toUpperCase() + ' ' + word;
     }
+  }
 
-    return text.replace(INDEFINITE_ARTICLE_RE, correctArticle);
+  return text.replace(INDEFINITE_ARTICLE_RE, correctArticle);
 }