Fix \ref with underscore in citation affixes (#74)

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+
+- Fix `\ref` with underscore in citation affixes ([#74](https://github.com/zepinglee/citeproc-lua/issues/74)).
+
 ## [0.6.2] - 2024-08-21
 
 ### Added
@@ -15,14 +19,14 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
-- Fix note position in multiple chapters ([#72](https://github.com/zepinglee/citeproc-lua/discussions/72)).
+- Fix note position in multiple chapters ([#72](https://github.com/zepinglee/citeproc-lua/issues/72)).
 - Fix incorrect locale map of UKenglish.
 
 ## [0.6.1] - 2024-08-15
 
 ### Fixed
 
-- Fix parsing quotation marks ([#71](https://github.com/zepinglee/citeproc-lua/discussions/71)).
+- Fix parsing quotation marks ([#71](https://github.com/zepinglee/citeproc-lua/issues/71)).
 
 ## [0.6.0] - 2024-07-31
 

citeproc/citeproc-latex-parser.lua

@@ -133,10 +133,11 @@ function latex_parser.get_latex_grammar()
   local latex_grammar = P{
     "latex_text";
     latex_text = Ct((specials + control_sequence + math + ligatures + specials + V"group" + plain_text)^0),
-    group = P"{" * V"latex_text" * P"}" / function (group_contents)
+    group = C(P"{" * V"latex_text" * P"}") / function (raw, group_contents)
       return {
         type = "group",
         contents = group_contents,
+        raw = raw,
       }
     end,
   }
@@ -348,9 +349,18 @@ function latex_parser.convert_cs_to_inlines(tokens, i, strict, case_protection)
   else
     -- Unrecognized command
     if strict then
-      table.insert(inlines, markup.Code:new(token.raw))
+      local raw = token.raw
+      for j = i + 1, #tokens do
+        if type(tokens[j]) == "table" and tokens[j].type == "group" then
+          i = j
+          raw = raw .. latex_parser.convert_token_to_raw(tokens[j])
+        else
+          break
+        end
+      end
+      table.insert(inlines, markup.Code:new(raw))
     else
-      -- Gobble following groupsas pandoc does:
+      -- Gobble following groups as pandoc does:
       -- "Foo \unrecognized{bar}{baz} quz" -> "Foo  quz"
       for j = i + 1, #tokens do
         if type(tokens[j]) == "table" and tokens[j].type == "group" then
@@ -365,6 +375,31 @@ function latex_parser.convert_cs_to_inlines(tokens, i, strict, case_protection)
   return inlines, i
 end
 
+---@param token any
+---@return string
+function latex_parser.convert_token_to_raw(token)
+  if type(token) == "string" then
+    return token
+  elseif type(token) == "table" then
+    if token.raw then
+      return token.raw
+    elseif token.type == "group" then
+      local raw = "{"
+      for _, token_ in ipairs(token.contents) do
+        raw = raw .. latex_parser.convert_token_to_raw(token_)
+      end
+      raw = raw .. "}"
+      return raw
+    else
+      error(token.type)
+      return ""
+    end
+  else
+    error(type(token))
+    return ""
+  end
+end
+
 -- TODO: raise a warning for unrecognized LaTeX command like `\switchargs`
 function latex_parser.convert_group_to_inlines(token, strict, case_protection, force_add_braces)
   if case_protection then

tests/bibtex/xampl.json

@@ -8,7 +8,7 @@
 				"given": "L[eslie] A."
 			}
 		],
-		"container-title": "<code>\\mbox{</code><span class=\"nocase\">G-Animal’s</span><code>}</code> Journal",
+		"container-title": "<code>\\mbox{G-Animal's}</code> Journal",
 		"issued": {
 			"date-parts": [
 				[
@@ -27,7 +27,7 @@
 				"given": "L[eslie] A."
 			}
 		],
-		"container-title": "<code>\\mbox{</code><span class=\"nocase\">G-Animal’s</span><code>}</code> Journal",
+		"container-title": "<code>\\mbox{G-Animal's}</code> Journal",
 		"issue": "7",
 		"issued": {
 			"date-parts": [
@@ -58,7 +58,7 @@
 	{
 		"id": "whole-journal",
 		"type": "article-journal",
-		"container-title": "<code>\\mbox{</code><span class=\"nocase\">G-Animal’s</span><code>}</code> Journal",
+		"container-title": "<code>\\mbox{G-Animal's}</code> Journal",
 		"issue": "7",
 		"issued": {
 			"date-parts": [
@@ -786,6 +786,6 @@
 	{
 		"id": "random-note-crossref",
 		"type": "document",
-		"note": "Volume 2 is listed under Knuth <code>\\cite{</code><span class=\"nocase\">book-full</span><code>}</code>"
+		"note": "Volume 2 is listed under Knuth <code>\\cite{book-full}</code>"
 	}
 ]

tests/latex-parser-test.lua

@@ -91,7 +91,7 @@ describe("LaTeX parser", function ()
     end)
 
     it("command with arguments", function ()
-      assert.same("<code>\\mbox{</code><span class=\"nocase\">G-Animal’s</span><code>}</code> Journal",
+      assert.same("<code>\\mbox{G-Animal's}</code> Journal",
         latex_parser.latex_to_pseudo_html("\\mbox{G-Animal's} Journal", true, true)
       )
     end)