Merge pull request #7613 from quarto-dev/bugfix/7604

make longtable environments work with new crossref system

src/resources/filters/customnodes/floatreftarget.lua

@@ -266,10 +266,12 @@ end, function(float)
     arg = pandoc.Str(float.identifier)
   })
   latex_caption:insert(1, label_cmd)
+  local latex_caption_content = latex_caption
+
   latex_caption = quarto.LatexInlineCommand({
     name = caption_cmd_name,
     opt_arg = fig_scap,
-    arg = pandoc.Span(quarto.utils.as_inlines(latex_caption or {}) or {}) -- unnecessary to do the "or {}" bit but the Lua analyzer doesn't know that
+    arg = pandoc.Span(quarto.utils.as_inlines(latex_caption_content or {}) or {}) -- unnecessary to do the "or {}" bit but the Lua analyzer doesn't know that
   })
 
   if float.parent_id then
@@ -287,6 +289,56 @@ end, function(float)
     end
   end
 
+  -- we need Pandoc to render its table ahead of time in order to
+  -- do the longtable fixups below
+  float.content = _quarto.ast.walk(quarto.utils.as_blocks(float.content), {
+    Table = function(tbl)
+      return pandoc.RawBlock("latex", pandoc.write(pandoc.Pandoc({tbl}), "latex"))
+    end
+  })
+
+  if float_type == "tbl" then
+    local raw
+    -- have to redo this as_blocks() call here because assigning to float.content
+    -- goes through our AST metaclasses which coalesce a singleton list to a single AST element
+    _quarto.ast.walk(quarto.utils.as_blocks(float.content), {
+      RawBlock = function(el)
+        if _quarto.format.isRawLatex(el) and el.text:match(_quarto.patterns.latexLongtablePattern) then
+          raw = el
+        end
+      end
+    })
+    -- special case for singleton longtable floats
+    if raw then
+      local longtable_content = raw.text:gsub(_quarto.patterns.latexLongtablePattern, "%2", 1)
+      -- split the content into params and actual content
+      -- params are everything in the first line of longtable_content
+      -- actual content is everything else
+      local params, content = longtable_content:match("^(.-)\n(.*)$")
+      local cap_loc = cap_location(float)
+      if float.parent_id then
+        -- need to fixup subtables because longtables don't support subcaptions,
+        -- and longtable captions increment the wrong counter
+        -- we try our best here
+
+        fatal("longtables are not supported in subtables.\n" ..
+          "This is not a Quarto bug - the LaTeX longtable environment doesn't support subcaptions.\n")
+        return {}
+      else
+        local result = pandoc.Blocks({latex_caption, pandoc.RawInline("latex", "\\tabularnewline")})
+        -- if cap_loc is top, insert content on bottom
+        if cap_loc == "top" then
+          result:insert(pandoc.RawBlock("latex", content))        
+        else
+          result:insert(1, pandoc.RawBlock("latex", content))
+        end
+        result:insert(1, pandoc.RawBlock("latex", "\\begin{longtable}" .. params))
+        result:insert(pandoc.RawBlock("latex", "\\end{longtable}"))
+        return result
+      end
+    end 
+  end
+
   local figure_content
   local pt = pandoc.utils.type(float.content)
   if pt == "Block" then

src/resources/filters/main.lua

@@ -253,11 +253,7 @@ local quarto_pre_filters = {
   { name = "pre-table-captions", 
     filter = table_captions(),
     flags = { "has_table_captions" } },
-
-  { name = "pre-longtable-no-caption-fixup", 
-    filter = longtable_no_caption_fixup(),
-    flags = { "has_longtable_no_caption_fixup" } },
-
+
   { name = "pre-code-annotations", 
     filter = code_annotations(),
     flags = { "has_code_annotations" } },
@@ -346,6 +342,7 @@ local quarto_post_filters = {
   { name = "layout-meta-inject-latex-packages", filter = layout_meta_inject_latex_packages() },
 
   -- format fixups post rendering
+  { name = "post-render-latex-fixups", filter = render_latex_fixups() },
   { name = "post-render-html-fixups", filter = render_html_fixups() },
   { name = "post-render-ipynb-fixups", filter = render_ipynb_fixups() },
   { name = "post-render-typst-fixups", filter = render_typst_fixups() },

src/resources/filters/quarto-post/latex.lua

@@ -378,4 +378,24 @@ function render_latex()
       return pandoc.Div(calloutContents)
     end    
   }
-end
+end
+
+
+function render_latex_fixups()
+  if not _quarto.format.isLatexOutput() then
+    return {}
+  end
+
+  return {
+    RawBlock = function(raw)
+      if _quarto.format.isRawLatex(raw) then
+        if (raw.text:match(_quarto.patterns.latexLongtablePattern) and
+            not raw.text:match(_quarto.patterns.latexCaptionPattern)) then
+          raw.text = raw.text:gsub(
+            _quarto.patterns.latexLongtablePattern, "\\begin{longtable*}%2\\end{longtable*}", 1)
+          return raw
+        end
+      end
+    end
+  }
+end

src/resources/filters/quarto-pre/parsefiguredivs.lua

@@ -273,6 +273,9 @@ function parse_reftargets()
 
     -- respect single table in latex longtable fixups above
     if skip_outer_reftarget then
+      -- we also need to strip the div identifier here
+      -- or we end up with duplicate identifiers which latex doesn't like
+      div.identifier = ""
       div.content = content
       return div
     end

src/resources/filters/quarto-pre/table-captions.lua

@@ -6,21 +6,6 @@ local patterns = require("modules/patterns")
 kTblCap = "tbl-cap"
 kTblSubCap = "tbl-subcap"
 
-function longtable_no_caption_fixup()
-  return {
-    RawBlock = function(raw)
-      if _quarto.format.isRawLatex(raw) then
-        if (raw.text:match(_quarto.patterns.latexLongtablePattern) and
-            not raw.text:match(_quarto.patterns.latexCaptionPattern)) then
-          raw.text = raw.text:gsub(
-            _quarto.patterns.latexLongtablePattern, "\\begin{longtable*}%2\\end{longtable*}", 1)
-          return raw
-        end
-      end
-    end
-  }
-end
-
 function table_captions() 
   return {   
     Div = function(el)

tests/docs/smoke-all/2023/11/16/7604.qmd

@@ -0,0 +1,29 @@
+---
+title: "7604"
+format: pdf
+---
+
+```{r}
+library(kableExtra)
+```
+
+```{r}
+#| label: tbl-sad-broken
+#| tbl-cap: "This table is cross-referenceable but no longer breaks across pages because it's inside `\\centering{}`" 
+rbind(mtcars, mtcars) |>
+  kbl(longtable = TRUE, booktabs = TRUE) |>
+  kable_styling()
+```
+
+::: {#tbl-also-sad}
+```{r}
+rbind(mtcars, mtcars) |>
+  kbl(longtable = TRUE, booktabs = TRUE) |>
+  kable_styling()
+```
+
+This table is also cross-referenceable but also doesn't break across pages because of the same `\centering{}` problem
+:::
+
+
+see @tbl-sad-broken and @tbl-also-sad.

tests/docs/smoke-all/crossrefs/float/latex/latex-longtable-fixup-test.qmd

@@ -7,9 +7,6 @@ _quarto:
       ensureFileRegexMatches:
         - 
           - "\\label{tbl-test}"
-          - "\\label{tbl-test-two-tables}"
-          - "\\label{tbl-test-two-tables-1}"
-          - "\\label{tbl-test-two-tables-2}"
         - []
 ---
 
@@ -37,23 +34,3 @@ kable(df,
 ```
 
 See @tbl-test.
-
-```{r}
-#| echo: false
-#| label: tbl-test-two-tables
-#| tbl-cap: Overall caption
-
-kable(df, 
-  format = "latex", 
-  longtable = TRUE, 
-  booktabs = TRUE, 
-  caption = "Table one")
-
-kable(df, 
-  format = "latex", 
-  longtable = TRUE, 
-  booktabs = TRUE, 
-  caption = "Table two because screw you, that's why")
-```
-
-See @tbl-test-two-tables, @tbl-test-two-tables-1, and @tbl-test-two-tables-2.

tests/smoke/crossref/tables.test.ts

@@ -58,8 +58,9 @@ testRender(knitrTablesQmd.input, "html", false, [
 
 /* caption is inserted in the right place in table environment*/
 renderVerifyLatexOutput(docs("crossrefs/knitr-tables-latex.qmd"), [
-  /\\begin{table}.*\\caption{\\label{tbl-1}.*}.*\\begin{longtable}\[.*\]{.*}.*\\end{longtable}/s,
-  /\\begin{table}.*\\caption{\\label{tbl-2}.*}.*\\centering.*\\begin{tabular}{.*}/s,
-  /\\begin{table}.*\\caption{\\label{tbl-3}.*}.*\\centering.*\\begin{longtable\*}{.*}/s,
-  /\\begin{table}.*\\caption{\\label{tbl-4}.*}.*\\centering\n\\begin{tabular}\[c\]{.*}/s,
+  /\\begin{longtable}\[.*\]{.*}.*\n+\\caption{\\label{tbl-1}.*}\n+.*\\tabularnewline/,
+  /\\begin{table}\n+\\caption{\\label{tbl-2}.*}.*\n+\\centering{?\n+\\begin{tabular}{.*}/,
+  /\\begin{longtable}{.*}.*\n+\\caption{\\label{tbl-3}.*}\n+.*\\tabularnewline/,
+  // two centering calls here is ugly, but we don't control the input we get
+  /\\begin{table}\n+\\caption{\\label{tbl-4}.*}.*\n+\\centering{?\n+\\centering\n+\\begin{tabular}\[c\]{.*}/,
 ]);