[Repo Assist] Fix HTML parser dropping whitespace between inline elements (issue #1330)

[github-actions]

🤖 This is a draft PR from Repo Assist, an automated AI assistant.

Fixes #1330.

Root Cause

In HtmlParser.fs, the Emit() method normalises inter-element whitespace (e.g. the newline/spaces between two tags) to a single space, then checks whether the entire content was whitespace-only. Before this fix it silently dropped that Text "" token, which also discarded meaningful whitespace between inline elements such as the space in <span>Hello,</span> <span>World</span>.

Fix

A new InlineWhitespace case is added to the private HtmlToken discriminated union. When Emit() produces a whitespace-only DefaultMode text node, it now emits InlineWhitespace instead of Text "".

In the tree builder (parse'), an InlineWhitespace token is turned into a real HtmlText " " node only when both its nearest previous accumulated node and the next token in the stream represent inline content (i.e. a non-block-level element or non-whitespace text). Otherwise it is discarded, preserving the original behaviour for inter-block whitespace.

Character references that happen to produce whitespace (e.g. ,  , 	) go through the CharRefMode path and emit Text t — never InlineWhitespace — so they are never filtered by this logic.

The blockLevelElements set (introduced in an earlier iteration) remains to classify element names for the inline/block decision.

Examples fixed

// Before (both returned the wrong value):
HtmlDocument.Parse("<p>&lt; &gt;</p>").Body().InnerText()
// Before: "<>"   After: "< >"

HtmlDocument.Parse("<p><span>Hello,</span> <span>World</span></p>").Body().InnerText()
// Before: "Hello,World"   After: "Hello, World"

Test Status

Build: ✅ 0 errors, 0 warnings

Tests: ✅ 2846 passed, 0 failed (1 pre-existing infrastructure failure: Two custom header with different names don't throw an exception — 403 Forbidden, unrelated to this change)

Three regression tests added:

• Preserves space between entity references in inline content (issue 1330)
• Preserves space between adjacent inline elements (issue 1330)
• Drops inter-block whitespace but keeps inline whitespace

Generated by Repo Assist

To install this workflow, run gh aw add githubnext/agentics/workflows/repo-assist.md@828ac109efb43990f59475cbfce90ede5546586c. View source at https://github.com/githubnext/agentics/tree/828ac109efb43990f59475cbfce90ede5546586c/workflows/repo-assist.md.

Warning

⚠️ Firewall blocked 1 domain

The following domain was blocked by the firewall during workflow execution:

• www.google.com

[github-actions]

✅ Pull request created: #1630