📜 Improvements to LaTeX parsing (#672)

* 📜 Improvements to LaTeX parsing * 🏷 SI units in JATS
jupyter-book · Oct 13, 2023 · 6fa758f · 6fa758f
1 parent 0b090ca
commit 6fa758f
Show file tree

Hide file tree

Showing 8 changed files with 41 additions and 2 deletions.
diff --git a/.changeset/gentle-beans-sip.md b/.changeset/gentle-beans-sip.md
@@ -0,0 +1,5 @@
+---
+'myst-to-jats': patch
+---
+
+Improve SI units for JATS export
diff --git a/.changeset/silly-jars-deliver.md b/.changeset/silly-jars-deliver.md
@@ -0,0 +1,5 @@
+---
+'tex-to-myst': patch
+---
+
+Updates to various latex parsing functions
diff --git a/packages/myst-to-jats/src/index.ts b/packages/myst-to-jats/src/index.ts
@@ -438,7 +438,7 @@ const handlers: Record<string, Handler> = {
   si(node, state) {
     // <named-content content-type="quantity">5 <abbrev content-type="unit" alt="milli meter">mm</abbrev></named-content>
     state.openNode('named-content', { 'content-type': 'quantity' });
-    state.text(`${node.number} `);
+    if (node.number != null) state.text(`${node.number} `);
     state.openNode('abbrev', { 'content-type': 'unit', alt: node.alt });
     state.text(node.unit);
     state.closeNode();

diff --git a/packages/tex-to-myst/src/misc.ts b/packages/tex-to-myst/src/misc.ts
@@ -61,6 +61,7 @@ export const MISC_HANDLERS: Record<string, Handler> = {
   // Some size options, not respecting at the moment
   macro_vspace: pass,
   macro_hfill: pass,
+  macro_tiny: pass,
   macro_small: pass,
   macro_footnotesize: pass,
   macro_normalsize: pass,
@@ -75,4 +76,7 @@ export const MISC_HANDLERS: Record<string, Handler> = {
   macro_textwidth: pass,
   macro_onecolumn: pass,
   macro_linewidth: pass,
+  // line numbers
+  macro_linenumbers: pass,
+  macro_nolinenumbers: pass,
 };
diff --git a/packages/tex-to-myst/src/siunitx.ts b/packages/tex-to-myst/src/siunitx.ts
@@ -17,7 +17,7 @@ const UNITS: Record<string, string> = {
   mole: 'mol',
   second: 's',
   becquerel: 'Bq',
-  degreeCelsius: '°C',
+  degreeCelsius: '℃',
   coulomb: 'C',
   farad: 'F',
   gray: 'Gy',
@@ -54,6 +54,7 @@ const UNITS: Record<string, string> = {
   arcsecond: '″', // second (plane angle) U+2033
   neper: 'Np',
   tonne: 't',
+  celsius: '℃',
 };
 
 // SI prefixes
@@ -327,6 +328,18 @@ const SIUNITX_HANDLERS: Record<string, Handler> = {
     });
     return;
   },
+  macro_si(node, state) {
+    state.openParagraph();
+    const { units, alt, organized } = createSiUnitNode(state.file, node, 0);
+    const translated = organized.map(unitToString).join(NARROW_NO_BREAK_SPACE);
+    state.addLeaf<SiUnit>('si', {
+      unit: translated,
+      alt,
+      units: units.map((n: GenericNode) => n.content),
+      value: translated,
+    });
+    return;
+  },
   macro_qty(node, state) {
     state.openParagraph();
     const { units, alt, organized } = createSiUnitNode(state.file, node, 1);

diff --git a/packages/tex-to-myst/src/tables.ts b/packages/tex-to-myst/src/tables.ts
@@ -59,6 +59,13 @@ export const TABLE_HANDLERS: Record<string, Handler> = {
     state.closeParagraph();
     state.closeNode();
   },
+  ['env_table*'](node, state) {
+    state.closeParagraph();
+    state.openNode('container', { kind: 'table' });
+    state.renderChildren(node);
+    state.closeParagraph();
+    state.closeNode();
+  },
   env_tabular: createTable,
   env_tabularx: createTable,
   env_supertabular: createTable,

diff --git a/packages/tex-to-myst/src/tex.ts b/packages/tex-to-myst/src/tex.ts
@@ -72,6 +72,7 @@ const macros: Record<string, number> = {
   affiliation: 1,
   framebox: 1,
   tnote: 1,
+  ['table*']: 1,
   arraystretch: 1,
   multirow: 5,
   multicolumn: 3,
@@ -83,6 +84,7 @@ const macros: Record<string, number> = {
   captionof: 2,
   // SI Units: https://texdoc.org/serve/siunitx/0
   SI: 2,
+  si: 1,
   qty: 2,
   tothe: 1,
   raiseto: 1,

diff --git a/packages/tex-to-myst/tests/siunitx.yml b/packages/tex-to-myst/tests/siunitx.yml
@@ -87,3 +87,6 @@ cases:
   - title: kg⁻¹
     tex: \qty[per-mode = symbol]{1.99}{ \per \kilogram }
     text: 1.99 kg⁻¹
+  - title: 90 ms
+    tex: 90~\si{\milli \second}
+    text: 90 ms