fix(wikilinks): handle wikilinks inside tables seperately from other wikilinks (#1005)

* fix(wikilinks): handle wikilinks inside tables seperately from other wikilinks

* Prettier

* Cleaned up duplicate code

* Remove test logging

* Refactored and fixed for non-aliased wikilinks inside table

* Updated naming and comments

* Updated comment of wikilink regex

* Updated regex to match previous formatting

* Match table even if EOF is immediately after the table.

* Update quartz/plugins/transformers/ofm.ts

Co-authored-by: Jacky Zhao <j.zhao2k19@gmail.com>

* Change table escape replace to non-regex version

* Prettier

* Prettier

---------

Co-authored-by: Jacky Zhao <j.zhao2k19@gmail.com>

Author: saberzero1

quartz/plugins/transformers/ofm.ts

@@ -99,17 +99,27 @@ export const externalLinkRegex = /^https?:\/\//i
 
 export const arrowRegex = new RegExp(/(-{1,2}>|={1,2}>|<-{1,2}|<={1,2})/, "g")
 
-// (\|[^\|\[\n]*)?   -> optional check if wikilink is inside a table cell
-// !?                -> optional embedding
-// \[\[              -> open brace
-// ([^\[\]\|\#]+)    -> 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)
-// ([^\|\n]*\|)?     -> optional check if wikilink is inside a table cell
+// !?                 -> optional embedding
+// \[\[               -> open brace
+// ([^\[\]\|\#]+)     -> one or more non-special characters ([,],|, or #) (name)
+// (#[^\[\]\|\#]+)?   -> # then one or more non-special characters (heading link)
+// (\\?\|[^\[\]\#]+)? -> optional escape \ then | then one or more non-special characters (alias)
 export const wikilinkRegex = new RegExp(
-  /(\|[^\|\[\n]*)?!?\[\[([^\[\]\|\#\\]+)?(#+[^\[\]\|\#\\]+)?(\\?\|[^\[\]\#]+)?\]\]([^\|\n]*\|)?/,
+  /!?\[\[([^\[\]\|\#\\]+)?(#+[^\[\]\|\#\\]+)?(\\?\|[^\[\]\#]+)?\]\]/,
   "g",
 )
+
+// ^\|([^\n])+\|\n(\|) -> matches the header row
+// ( ?:?-{3,}:? ?\|)+  -> matches the header row separator
+// (\|([^\n])+\|\n)+   -> matches the body rows
+export const tableRegex = new RegExp(
+  /^\|([^\n])+\|\n(\|)( ?:?-{3,}:? ?\|)+\n(\|([^\n])+\|\n?)+/,
+  "gm",
+)
+
+// matches any wikilink, only used for escaping wikilinks inside tables
+export const tableWikilinkRegex = new RegExp(/(!?\[\[[^\]]*?\]\])/, "g")
+
 const highlightRegex = new RegExp(/==([^=]+)==/, "g")
 const commentRegex = new RegExp(/%%[\s\S]*?%%/, "g")
 // from https://github.com/escwxyz/remark-obsidian-callout/blob/main/src/index.ts
@@ -171,27 +181,34 @@ export const ObsidianFlavoredMarkdown: QuartzTransformerPlugin<Partial<Options>
           src = src.toString()
         }
 
+        // replace all wikilinks inside a table first
+        src = src.replace(tableRegex, (value) => {
+          // escape all aliases and headers in wikilinks inside a table
+          return value.replace(tableWikilinkRegex, (value, ...capture) => {
+            const [raw]: (string | undefined)[] = capture
+            let escaped = raw ?? ""
+            escaped = escaped.replace("#", "\\#")
+            escaped = escaped.replace("|", "\\|")
+
+            return escaped
+          })
+        })
+
+        // replace all other wikilinks
         src = src.replace(wikilinkRegex, (value, ...capture) => {
-          const [rawTablePre, rawFp, rawHeader, rawAlias, rawTablePost]: (string | undefined)[] =
-            capture
+          const [rawFp, rawHeader, rawAlias]: (string | undefined)[] = capture
 
           const fp = rawFp ?? ""
           const anchor = rawHeader?.trim().replace(/^#+/, "")
           const blockRef = Boolean(anchor?.startsWith("^")) ? "^" : ""
           const displayAnchor = anchor ? `#${blockRef}${slugAnchor(anchor)}` : ""
-          let displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
+          const displayAlias = rawAlias ?? rawHeader?.replace("#", "|") ?? ""
           const embedDisplay = value.startsWith("!") ? "!" : ""
 
           if (rawFp?.match(externalLinkRegex)) {
             return `${embedDisplay}[${displayAlias.replace(/^\|/, "")}](${rawFp})`
           }
 
-          // transform `[[note#^block_ref|^block_ref]]` to `[[note#^block_ref\|^block_ref]]`,
-          // when the wikilink with alias is inside a table.
-          if (displayAlias && displayAlias.startsWith("|") && rawTablePre && rawTablePost) {
-            displayAlias = `\\${displayAlias}`
-          }
-
           return `${embedDisplay}[[${fp}${displayAnchor}${displayAlias}]]`
         })
       }
@@ -211,7 +228,7 @@ export const ObsidianFlavoredMarkdown: QuartzTransformerPlugin<Partial<Options>
             replacements.push([
               wikilinkRegex,
               (value: string, ...capture: string[]) => {
-                let [_rawTablePre, rawFp, rawHeader, rawAlias, _rawTablePost] = capture
+                let [rawFp, rawHeader, rawAlias] = capture
                 const fp = rawFp?.trim() ?? ""
                 const anchor = rawHeader?.trim() ?? ""
                 const alias = rawAlias?.slice(1).trim()