fix: miscテスト修正（HR4+制限・backslash改行・末尾空白除去・bibliography除外）

Author: r74tech

packages/parser/src/lexer/lexer.ts

@@ -310,8 +310,8 @@ export class Lexer {
       return;
     }
 
-    // Horizontal rule --- or more (check before --)
-    if (isLineStart && this.match("---")) {
+    // Horizontal rule ---- or more (4+ hyphens, check before --)
+    if (isLineStart && this.match("----")) {
       let dashes = "";
       while (this.current() === "-") {
         dashes += this.advance();
@@ -582,6 +582,13 @@ export class Lexer {
       return;
     }
 
+    // Backslash line break marker (U+E000, inserted by preproc)
+    if (char.charCodeAt(0) === 0xe000) {
+      this.advance();
+      this.addToken("BACKSLASH_BREAK", char);
+      return;
+    }
+
     // Identifier: alphanumeric sequence
     if (this.isAlphanumeric(char)) {
       let ident = "";

packages/parser/src/lexer/tokens.ts

@@ -73,6 +73,7 @@ export type TokenType =
 
   // Other
   | "UNDERSCORE" // _ (single underscore, for line break)
+  | "BACKSLASH_BREAK" // U+E000 (preproc marker for \ at end of line)
 
   // Comment
   | "COMMENT_OPEN" // [!--

packages/parser/src/parser/preprocess/whitespace.ts

@@ -59,8 +59,9 @@ export function substitute(text: string): string {
   // Strip lines with only whitespace
   result = result.replace(WHITESPACE_ONLY_LINE, "");
 
-  // Join concatenated lines (ending with '\')
-  result = result.replace(CONCAT_LINES, "");
+  // Backslash at end of line → line break marker (U+E000)
+  // Wikidot treats \ at end of line as <br />
+  result = result.replace(CONCAT_LINES, String.fromCharCode(0xe000));
 
   // Tabs to spaces
   result = result.replace(TABS, "    ");

packages/parser/src/parser/rules/block/paragraph.ts

@@ -91,6 +91,16 @@ export const paragraphRule: BlockRule = {
       elements.pop();
     }
 
+    // Remove trailing whitespace-only text nodes
+    while (
+      elements.length > 0 &&
+      elements[elements.length - 1]?.element === "text" &&
+      typeof elements[elements.length - 1]?.data === "string" &&
+      (elements[elements.length - 1]?.data as string).trim() === ""
+    ) {
+      elements.pop();
+    }
+
     // Remove leading line-breaks
     while (elements.length > 0 && elements[0]?.element === "line-break") {
       elements.shift();

packages/parser/src/parser/rules/inline/index.ts

@@ -11,7 +11,7 @@ import { linkSingleRule } from "./link-single";
 import { linkAnchorRule } from "./link-anchor";
 import { linkStarRule } from "./link-star";
 import { colorRule } from "./color";
-import { newlineLineBreakRule, underscoreLineBreakRule } from "./line-break";
+import { backslashLineBreakRule, newlineLineBreakRule, underscoreLineBreakRule } from "./line-break";
 import { commentRule } from "./comment";
 import { rawRule } from "./raw";
 import { spanRule, closeSpanRule } from "./span";
@@ -39,7 +39,7 @@ export { linkSingleRule } from "./link-single";
 export { linkAnchorRule } from "./link-anchor";
 export { linkStarRule } from "./link-star";
 export { colorRule } from "./color";
-export { newlineLineBreakRule, underscoreLineBreakRule } from "./line-break";
+export { backslashLineBreakRule, newlineLineBreakRule, underscoreLineBreakRule } from "./line-break";
 export { commentRule } from "./comment";
 export { rawRule } from "./raw";
 export { spanRule, closeSpanRule } from "./span";
@@ -71,6 +71,7 @@ export const inlineRules: InlineRule[] = [
   linkAnchorRule,
   linkStarRule,
   colorRule,
+  backslashLineBreakRule,
   underscoreLineBreakRule,
   newlineLineBreakRule, // Must come after underscore rule
   commentRule,

packages/parser/src/parser/rules/inline/line-break.ts

@@ -6,8 +6,8 @@
  * - " _\n" pattern → line-break
  * - "^_\n" pattern → line-break (underscore at start of line)
  *
- * Note: Backslash line break (\ at end of line) is handled by preproc,
- * which removes \\\n and joins lines.
+ * Note: Backslash line break (\ at end of line) is preprocessed to U+E000
+ * by preproc, then handled by backslashLineBreakRule.
  */
 import type { Element } from "@wdpr/ast";
 import type { InlineRule, ParseContext, RuleResult } from "../types";
@@ -103,6 +103,29 @@ export const newlineLineBreakRule: InlineRule = {
   },
 };
 
+/**
+ * Backslash line break: \ at end of line (preprocessed to U+E000)
+ */
+export const backslashLineBreakRule: InlineRule = {
+  name: "backslashLineBreak",
+  startTokens: ["BACKSLASH_BREAK"],
+
+  parse(ctx: ParseContext): RuleResult<Element> {
+    const currentTok = ctx.tokens[ctx.pos];
+    if (!currentTok || currentTok.type !== "BACKSLASH_BREAK") {
+      return { success: false };
+    }
+
+    const lb: any = { element: "line-break" };
+    lb._preservedTrailingBreak = true;
+    return {
+      success: true,
+      elements: [lb],
+      consumed: 1,
+    };
+  },
+};
+
 /**
  * Underscore line break: _ at end of line
  * Syntax: " _\n" (space + underscore + newline)

tests/fixtures/misc/bibliography/expected.json

@@ -214,25 +214,19 @@
       }
     },
     {
-      "element": "container",
-      "data": {
-        "type": "paragraph",
-        "attributes": {},
-        "elements": [
-          {
-            "element": "text",
-            "data": "[["
-          },
-          {
-            "element": "text",
-            "data": "bibliography"
-          },
-          {
-            "element": "text",
-            "data": "]]"
-          }
-        ]
-      }
+      "element": "text",
+      "data": "[["
+    },
+    {
+      "element": "text",
+      "data": "bibliography"
+    },
+    {
+      "element": "text",
+      "data": "]]"
+    },
+    {
+      "element": "line-break"
     },
     {
       "element": "definition-list",

tests/fixtures/misc/comment/expected.json

@@ -20,14 +20,6 @@
           {
             "element": "text",
             "data": "Empty"
-          },
-          {
-            "element": "text",
-            "data": " "
-          },
-          {
-            "element": "text",
-            "data": " "
           }
         ]
       }

tests/fixtures/misc/hr/expected.json

@@ -127,7 +127,11 @@
           },
           {
             "element": "text",
-            "data": "—-"
+            "data": "—"
+          },
+          {
+            "element": "text",
+            "data": "-"
           }
         ]
       }

tests/fixtures/misc/symbols/output.html

@@ -1 +1 @@
-`~!@#$%^&amp;*()-=+[]{}|/<br />
+<p>`~!@#$%^&amp;*()-=+[]{}|/<br /></p>