From d979331dc781e92296166e50512c4aebd7aa66a5 Mon Sep 17 00:00:00 2001
From: Jacky Zhao <j.zhao2k19@gmail.com>
Date: Sun, 17 Dec 2023 20:54:52 +0000
Subject: [PATCH] fix: remove whitespace unicode from tag regex

---
 quartz/plugins/transformers/ofm.ts |   15 ++++++++++-----
 1 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/quartz/plugins/transformers/ofm.ts b/quartz/plugins/transformers/ofm.ts
index a8399d0..c1bec16 100644
--- a/quartz/plugins/transformers/ofm.ts
+++ b/quartz/plugins/transformers/ofm.ts
@@ -110,7 +110,10 @@
 // ([^\[\]\|\#]+)   -> one or more non-special characters ([,],|, or #) (name)
 // (#[^\[\]\|\#]+)? -> # then one or more non-special characters (heading link)
 // (|[^\[\]\|\#]+)? -> | then one or more non-special characters (alias)
-const wikilinkRegex = new RegExp(/!?\[\[([^\[\]\|\#]+)?(#[^\[\]\|\#]+)?(\|[^\[\]\|\#]+)?\]\]/, "g")
+export const wikilinkRegex = new RegExp(
+  /!?\[\[([^\[\]\|\#]+)?(#+[^\[\]\|\#]+)?(\|[^\[\]\|\#]+)?\]\]/,
+  "g",
+)
 const highlightRegex = new RegExp(/==([^=]+)==/, "g")
 const commentRegex = new RegExp(/%%(.+)%%/, "g")
 // from https://github.com/escwxyz/remark-obsidian-callout/blob/main/src/index.ts
@@ -118,8 +121,8 @@
 const calloutLineRegex = new RegExp(/^> *\[\!\w+\][+-]?.*$/, "gm")
 // (?:^| )              -> non-capturing group, tag should start be separated by a space or be the start of the line
 // #(...)               -> capturing group, tag itself must start with #
-// (?:[-_\p{L}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters, hyphens and/or underscores
-// (?:\/[-_\p{L}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
+// (?:[-_\p{L}\d\p{Z}])+       -> non-capturing group, non-empty string of (Unicode-aware) alpha-numeric characters and symbols, hyphens and/or underscores
+// (?:\/[-_\p{L}\d\p{Z}]+)*)   -> non-capturing group, matches an arbitrary number of tag strings separated by "/"
 const tagRegex = new RegExp(/(?:^| )#((?:[-_\p{L}\d])+(?:\/[-_\p{L}\d]+)*)/, "gu")
 const blockReferenceRegex = new RegExp(/\^([A-Za-z0-9]+)$/, "g")
 
@@ -178,8 +181,9 @@
         src = src.replaceAll(wikilinkRegex, (value, ...capture) => {
           const [rawFp, rawHeader, rawAlias] = capture
           const fp = rawFp ?? ""
-          const anchor = rawHeader?.trim().slice(1)
-          const displayAnchor = anchor ? `#${slugAnchor(anchor)}` : ""
+          const anchor = rawHeader?.trim().replace(/^#+/, "")
+          const blockRef = Boolean(anchor?.startsWith("^")) ? "^" : ""
+          const displayAnchor = anchor ? `#${blockRef}${slugAnchor(anchor)}` : ""
           const displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
           const embedDisplay = value.startsWith("!") ? "!" : ""
           return `${embedDisplay}[[${fp}${displayAnchor}${displayAlias}]]`
@@ -401,6 +405,7 @@
               if (/^\d+$/.test(tag)) {
                 return false
               }
+
               tag = slugTag(tag)
               if (file.data.frontmatter && !file.data.frontmatter.tags.includes(tag)) {
                 file.data.frontmatter.tags.push(tag)

--
Gitblit v1.10.0