Merge pull request #12090 from quarto-dev/crossref/raw_latex

Author: cderv

news/changelog-1.7.md

@@ -80,3 +80,4 @@ All changes included in 1.7:
 - ([fb38eb5](https://github.com/quarto-dev/quarto-cli/commit/fb38eb56c11e09f44cef58fd3b697ff24bb5a3f3)) Use the `latest` parser for Acorn when analyzing JS code imported from OJS blocks.
 - ([#10532](https://github.com/quarto-dev/quarto-cli/issues/10532)): Quarto changed default of `--headless=old` to `--headless` as [Chrome 132 has removed old headless mode](https://developer.chrome.com/blog/removing-headless-old-from-chrome) and only support new mode. For using old mode, `QUARTO_CHROMIUM` could be set to a [new `chrome-headless-shell` binary](https://developer.chrome.com/blog/chrome-headless-shell) or too an older chrome version (between 128 and 132) and `QUARTO_CHROMIUM_HEADLESS_MODE` set to `old` for using old headless mode with that compatabitle version.
 - ([#10961](https://github.com/quarto-dev/quarto-cli/issues/10961)): Add more information on which Chrome Headless will be used in `quarto check install`. This is helpful to help debug mermaid issues.
+- ([#11951](https://github.com/quarto-dev/quarto-cli/issues/11951)): Raw LaTeX table without `tbl-` prefix label for using Quarto crossref are now correctly passed through unmodified.

src/command/render/pandoc.ts

@@ -398,6 +398,16 @@ export async function runPandoc(
     ...options.flags?.metadata,
   } as Metadata;
 
+  const cleanQuartoTestsMetadata = (metadata: Metadata) => {
+    // remove any metadata that is only used for testing
+    if (metadata["_quarto"] && typeof metadata["_quarto"] === "object") {
+      delete (metadata._quarto as { [key: string]: unknown })?.tests;
+      if (Object.keys(metadata._quarto).length === 0) {
+        delete metadata._quarto;
+      }
+    }
+  };
+
   // remove some metadata that are used as parameters to our lua filters
   const cleanMetadataForPrinting = (metadata: Metadata) => {
     delete metadata.params;
@@ -409,6 +419,7 @@ export async function runPandoc(
     delete metadata[kRevealJsScripts];
     deleteProjectMetadata(metadata);
     deleteCrossrefMetadata(metadata);
+    removeFilterParams(metadata);
 
     // Don't print empty reveal-js plugins
     if (
@@ -417,7 +428,12 @@ export async function runPandoc(
     ) {
       delete metadata[kRevealJSPlugins];
     }
+
+    // Don't print _quarto.tests
+    // This can cause issue on regex test for printed output
+    cleanQuartoTestsMetadata(metadata);
   };
+
   cleanMetadataForPrinting(printMetadata);
 
   // Forward flags metadata into the format
@@ -1270,11 +1286,9 @@ export async function runPandoc(
   delete pandocPassedMetadata.project;
   delete pandocPassedMetadata.website;
   delete pandocPassedMetadata.about;
-  if (pandocPassedMetadata._quarto) {
-    // these shouldn't be visible because they are emitted on markdown output
-    // and it breaks ensureFileRegexMatches
-    delete pandocPassedMetadata._quarto.tests;
-  }
+  // these shouldn't be visible because they are emitted on markdown output
+  // and it breaks ensureFileRegexMatches
+  cleanQuartoTestsMetadata(pandocPassedMetadata);
 
   Deno.writeTextFileSync(
     metadataTemp,
@@ -1377,7 +1391,7 @@ export async function runPandoc(
 
 // this mutates metadata[kClassOption]
 function cleanupPandocMetadata(metadata: Metadata) {
-  // pdf classoption can end up with duplicaed options
+  // pdf classoption can end up with duplicated options
   const classoption = metadata[kClassOption];
   if (Array.isArray(classoption)) {
     metadata[kClassOption] = ld.uniqBy(
@@ -1668,8 +1682,6 @@ function runPandocMessage(
     const printMetadata = ld.cloneDeep(metadata) as Metadata;
     delete printMetadata.format;
 
-    // remove filter params
-    removeFilterParams(printMetadata);
     // print message
     if (Object.keys(printMetadata).length > 0) {
       info("metadata", { bold: true });

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

@@ -702,36 +702,39 @@ function parse_floatreftargets()
         return nil
       end
 
+      -- prevent raw mutation
+      local rawText = raw.text
+
       -- first we check if all of the expected bits are present
 
       -- check for {#...} or \label{...}
-      if raw.text:find(patterns.latex_label) == nil and 
-         raw.text:find(patterns.attr_identifier) == nil then
+      if rawText:find(patterns.latex_label) == nil and 
+         rawText:find(patterns.attr_identifier) == nil then
         return nil
       end
 
       -- check for \caption{...}
-      if raw.text:find(patterns.latex_caption) == nil then
+      if rawText:find(patterns.latex_caption) == nil then
         return nil
       end
 
       -- check for tabular or longtable
-      if raw.text:find(patterns.latex_long_table) == nil and
-         raw.text:find(patterns.latex_tabular) == nil then
+      if rawText:find(patterns.latex_long_table) == nil and
+         rawText:find(patterns.latex_tabular) == nil then
         return nil
       end
 
       -- if we're here, then we're going to parse this as a FloatRefTarget
       -- and we need to remove the label and caption from the raw block
       local identifier = ""
-      local b, e, match1, label_identifier = raw.text:find(patterns.latex_label)
+      local b, e, _ , label_identifier = rawText:find(patterns.latex_label)
       if b ~= nil then
-        raw.text = raw.text:sub(1, b - 1) .. raw.text:sub(e + 1)
+        rawText = rawText:sub(1, b - 1) .. rawText:sub(e + 1)
         identifier = label_identifier
       else
-        local b, e, match2, attr_identifier = raw.text:find(patterns.attr_identifier)
+        local b, e, _ , attr_identifier = rawText:find(patterns.attr_identifier)
         if b ~= nil then
-          raw.text = raw.text:sub(1, b - 1) .. raw.text:sub(e + 1)
+          rawText = rawText:sub(1, b - 1) .. rawText:sub(e + 1)
           identifier = attr_identifier
         else
           internal_error()
@@ -749,9 +752,9 @@ function parse_floatreftargets()
       end
 
       local caption
-      local b, e, match3, caption_content = raw.text:find(patterns.latex_caption)
+      local b, e, _, caption_content = rawText:find(patterns.latex_caption)
       if b ~= nil then
-        raw.text = raw.text:sub(1, b - 1) .. raw.text:sub(e + 1)
+        rawText = rawText:sub(1, b - 1) .. rawText:sub(e + 1)
         caption = pandoc.RawBlock("latex", caption_content)
       else
         internal_error()
@@ -760,16 +763,16 @@ function parse_floatreftargets()
 
       -- finally, if the user passed a \\begin{table} float environment
       -- we just remove it because we'll re-emit later ourselves  
-      local matched, _ = _quarto.modules.patterns.match_in_list_of_patterns(raw.text, _quarto.patterns.latexTableEnvPatterns)
+      local matched, _ = _quarto.modules.patterns.match_in_list_of_patterns(rawText, _quarto.patterns.latexTableEnvPatterns)
       if matched then
         -- table_body is second matched element.
-        raw.text = matched[2]
+        rawText = matched[2]
       end
 
       return construct({
         attr = pandoc.Attr(identifier, {}, {}),
         type = "Table",
-        content = pandoc.Blocks({ raw }),
+        content = pandoc.Blocks({ pandoc.RawBlock(raw.format, rawText) }),
         caption_long = quarto.utils.as_blocks(caption)
       }), false
     end

tests/docs/smoke-all/2024/01/22/8389.md.snapshot

@@ -1,5 +1,4 @@
 ---
-_quarto: {}
 foo: bar
 toc-title: Table of contents
 ---

tests/docs/smoke-all/crossrefs/float/latex/latex-raw-table-env-no-tbl-label.qmd

@@ -0,0 +1,44 @@
+---
+title: Raw LaTeX table with not `tbl` in label id
+format: pdf
+keep-tex: true
+_quarto:
+  tests:
+    pdf:
+      ensureLatexFileRegexMatches:
+        - ['[\s\S]+\\begin\{table\}\[htbp\][\s\S]+', '[\s\S]+\\caption\{\\label\{mod\}']
+        - []
+      printsMessage:
+        - INFO
+        - 'WARNING(.*)Raw LaTeX table found with non-tbl label:' 
+---
+
+This document has a raw LaTeX table with no intent to use quarto crossref. It uses a label with `tbl-` id.
+
+Table will be identified by floatreftarget processing but raw TeX should be left untouched in the output, while a warning still is emitted. 
+
+```{=latex}
+\begin{table}[htbp]
+   \caption{\label{mod} no title}
+   \centering
+   \begin{tabular}{lc}
+      \tabularnewline \midrule \midrule
+      Dependent Variable: & disp\\  
+      Model:              & (1)\\  
+      \midrule
+      \emph{Variables}\\
+      Constant            & 580.9$^{***}$\\   
+                          & (41.74)\\   
+      mpg                 & -17.43$^{***}$\\   
+                          & (1.993)\\   
+      \midrule
+      \emph{Fit statistics}\\
+      Observations        & 32\\  
+      R$^2$               & 0.71834\\  
+      Adjusted R$^2$      & 0.70895\\  
+      \midrule \midrule
+      \multicolumn{2}{l}{\emph{IID standard-errors in parentheses}}\\
+      \multicolumn{2}{l}{\emph{Signif. Codes: ***: 0.01, **: 0.05, *: 0.1}}\\
+   \end{tabular}
+\end{table}
+```