[Repo Assist] perf: compile regex instances to module-level singletons (#1162)

* perf: compile regex instances to module-level singletons

Five locations were creating a new uncompiled Regex on every call:

- PageContentList.mkPageContentMenu: new Regex per page rendered
- Formatting.fs (search index): Regex.Replace (static, uncompiled) per HTML page
- HtmlFormatting.formatAnchor: Regex.Matches per heading processed
- Menu.snakeCase: Regex.Replace (static, uncompiled) per menu item
- LlmsTxt (collapseBlankLines, normaliseTitle): two Regex.Replace per page entry

Each is now a module-level let binding with RegexOptions.Compiled, so the
pattern is compiled once at startup and reused for the lifetime of the
process.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* ci: trigger checks

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Don Syme <dsyme@users.noreply.github.com>

Author: 3 people

RELEASE_NOTES.md

@@ -3,6 +3,7 @@
 ## [Unreleased]
 
 ### Changed
+* Compile `Regex` instances to module-level singletons (with `RegexOptions.Compiled`) in `PageContentList`, `HtmlFormatting`, `Formatting`, `Menu`, and `LlmsTxt`. Previously a new, uncompiled `Regex` was constructed on every call (once per page heading, once per HTML page, once per menu item, once per llms.txt entry), incurring repeated JIT overhead. The patterns are now compiled once at module load and reused across all calls.
 * Replace deprecated `System.Net.WebClient` with `System.Net.Http.HttpClient` in the image downloader used by `--saveimages`. Removes the `#nowarn "44"` suppression.
 * Bump `Newtonsoft.Json` transitive-dependency pin from 13.0.3 to 13.0.4.
 * Bump `System.Memory` transitive-dependency pin from 4.5.5 to 4.6.3.0

src/FSharp.Formatting.Common/Menu.fs

@@ -12,8 +12,11 @@ type MenuItem =
       IsActive: bool }
 
 /// Converts a display string to a snake_case HTML id attribute value
+let private snakeCaseRegex =
+    System.Text.RegularExpressions.Regex("[A-Z]", System.Text.RegularExpressions.RegexOptions.Compiled)
+
 let private snakeCase (v: string) =
-    System.Text.RegularExpressions.Regex.Replace(v, "[A-Z]", "$0").Replace(" ", "_").ToLower()
+    snakeCaseRegex.Replace(v, "$0").Replace(" ", "_").ToLower()
 
 /// Renders an HTML navigation menu for the given header and items using template files in `input`
 let createMenu (input: string) (isCategoryActive: bool) (header: string) (items: MenuItem list) : string =

src/FSharp.Formatting.Common/PageContentList.fs

@@ -13,10 +13,11 @@ let EmptyContent = "<div class=\"empty\"></div>"
 /// We process the html to collect the table of content.
 /// We can't use the doc.MarkdownDocument because we cannot easily get the generated id values.
 /// It is safer to parse the html.
-let mkPageContentMenu (html: string) =
-    let headingLinkPattern = "<h(\\d)><a [^>]*href=\"([^\"]+)\">([^<]+)</a></h\\d>"
 
-    let regex = Regex(headingLinkPattern)
+// Compiled once at module load; reused across all pages.
+let private headingLinkRegex = Regex("<h(\\d)><a [^>]*href=\"([^\"]+)\">([^<]+)</a></h\\d>", RegexOptions.Compiled)
+
+let mkPageContentMenu (html: string) =
 
     let extractHeadingLinks (matchItem: Match) =
         let level = int matchItem.Groups.[1].Value
@@ -26,7 +27,7 @@ let mkPageContentMenu (html: string) =
         linkText, li [ Class $"level-%i{level}" ] [ a [ Href href ] [ !!linkText ] ]
 
     let headingTexts, listItems =
-        regex.Matches(html)
+        headingLinkRegex.Matches(html)
         |> Seq.cast<Match>
         |> Seq.map extractHeadingLinks
         |> Seq.toList

src/FSharp.Formatting.Literate/Formatting.fs

@@ -13,6 +13,9 @@ open FSharp.Formatting.Templating
 /// substitution key–value pairs used by the templating engine to populate page templates.
 module internal Formatting =
 
+    // Compiled once at module load; reused for every HTML page's search-index text extraction.
+    let private htmlTagRegex = Regex("<.*?>", RegexOptions.Compiled ||| RegexOptions.Singleline)
+
     /// Format document with the specified output kind
     let format (doc: MarkdownDocument) generateAnchors outputKind substitutions crefResolver mdlinkResolver =
         match outputKind with
@@ -290,7 +293,7 @@ module internal Formatting =
             (match ctx.OutputKind with
              | OutputKind.Html ->
                  // Strip the html tags
-                 let fullText = Regex.Replace(formattedDocument, "<.*?>", "")
+                 let fullText = htmlTagRegex.Replace(formattedDocument, "")
                  Some(IndexText(fullText, headingTexts))
              | _ -> None)
 

src/FSharp.Formatting.Markdown/HtmlFormatting.fs

@@ -129,9 +129,12 @@ let rec internal formatSpan (ctx: FormattingContext) span =
 and internal formatSpans ctx = List.iter (formatSpan ctx)
 
 /// generate anchor name from Markdown text
+// Compiled once at module load; reused for every heading anchor generated.
+let private wordRegex = Regex(@"\w+", RegexOptions.Compiled)
+
 let internal formatAnchor (ctx: FormattingContext) (spans: MarkdownSpans) =
     let extractWords (text: string) =
-        Regex.Matches(text, @"\w+") |> Seq.cast<Match> |> Seq.map (fun m -> m.Value)
+        wordRegex.Matches(text) |> Seq.cast<Match> |> Seq.map (fun m -> m.Value)
 
     let rec gather (span: MarkdownSpan) : string seq =
         seq {

src/fsdocs-tool/BuildCommand.fs

@@ -1317,6 +1317,13 @@ module Serve =
 /// Helpers for generating llms.txt and llms-full.txt content.
 module internal LlmsTxt =
 
+    // Compiled once at module load; reused across all llms.txt page entries.
+    let private multipleNewlinesRegex =
+        System.Text.RegularExpressions.Regex(@"\n{3,}", System.Text.RegularExpressions.RegexOptions.Compiled)
+
+    let private whitespaceRunRegex =
+        System.Text.RegularExpressions.Regex(@"\s+", System.Text.RegularExpressions.RegexOptions.Compiled)
+
     /// Decode HTML entities (e.g. " → ", > → >) in a string.
     let private decodeHtml (s: string) = System.Net.WebUtility.HtmlDecode(s)
 
@@ -1336,11 +1343,11 @@ module internal LlmsTxt =
 
     /// Collapse three or more consecutive newlines into at most two.
     let private collapseBlankLines (s: string) =
-        System.Text.RegularExpressions.Regex.Replace(s, @"\n{3,}", "\n\n")
+        multipleNewlinesRegex.Replace(s, "\n\n")
 
     /// Normalise a title: trim and collapse internal whitespace/newlines to a single space.
     let private normaliseTitle (s: string) =
-        System.Text.RegularExpressions.Regex.Replace(s.Trim(), @"\s+", " ")
+        whitespaceRunRegex.Replace(s.Trim(), " ")
 
     /// Decode HTML entities and remove --eval noise from content.
     let private cleanContent (s: string) =