temml 0.10.2 → 0.10.4

package/dist/temml.mjs

@@ -188,7 +188,7 @@ class Settings {
     this.leqno = utils.deflt(options.leqno, false);                // boolean
     this.errorColor = utils.deflt(options.errorColor, "#b22222");  // string
     this.macros = options.macros || {};
-    this.wrap = utils.deflt(options.wrap, "none");         // "none" | "tex" | "="
+    this.wrap = utils.deflt(options.wrap, "tex");                    // "tex" | "="
     this.xml = utils.deflt(options.xml, false);                     // boolean
     this.colorIsTextColor = utils.deflt(options.colorIsTextColor, false);  // booelean
     this.strict = utils.deflt(options.strict, false);    // boolean
@@ -1747,10 +1747,12 @@ for (let i = 0; i < 10; i++) {
  * Then the top level of a <math> element can be occupied by <mrow> elements, and the browser
  * will break after a <mrow> if the expression extends beyond the container limit.
  *
- * We want the expression to render with soft line breaks after each top-level binary or
+ * The default is for soft line breaks after each top-level binary or
  * relational operator, per TeXbook p. 173. So we gather the expression into <mrow>s so that
  * each <mrow> ends in a binary or relational operator.
  *
+ * An option is for soft line breaks before an "=" sign. That changes the <mrow>s.
+ *
  * Soft line breaks will not work in Chromium and Safari, only Firefox.
  *
  * Hopefully browsers will someday do their own linebreaking and we will be able to delete
@@ -1962,6 +1964,48 @@ const consolidateText = mrow => {
   return mtext
 };
 
+const numberRegEx$1 = /^[0-9]$/;
+const isCommaOrDot = node => {
+  return (node.type === "atom" && node.text === ",") ||
+         (node.type === "textord" && node.text === ".")
+};
+const consolidateNumbers = expression => {
+  // Consolidate adjacent numbers. We want to return <mn>1,506.3</mn>,
+  // not <mn>1</mn><mo>,</mo><mn>5</mn><mn>0</mn><mn>6</mn><mi>.</mi><mn>3</mn>
+  if (expression.length < 2) { return }
+  const nums = [];
+  let inNum = false;
+  // Find adjacent numerals
+  for (let i = 0; i < expression.length; i++) {
+    const node = expression[i];
+    if (node.type === "textord" && numberRegEx$1.test(node.text)) {
+      if (!inNum) { nums.push({ start: i }); }
+      inNum = true;
+    } else {
+      if (inNum) { nums[nums.length - 1].end = i - 1; }
+      inNum = false;
+    }
+  }
+  if (inNum) { nums[nums.length - 1].end = expression.length - 1; }
+
+  // Determine if numeral groups are separated by a comma or dot.
+  for (let i = nums.length - 1; i > 0; i--) {
+    if (nums[i - 1].end === nums[i].start - 2 && isCommaOrDot(expression[nums[i].start - 1])) {
+      // Merge the two groups.
+      nums[i - 1].end = nums[i].end;
+      nums.splice(i, 1);
+    }
+  }
+
+  // Consolidate the number nodes
+  for (let i = nums.length - 1; i >= 0; i--) {
+    for (let j = nums[i].start + 1; j <= nums[i].end; j++) {
+      expression[nums[i].start].text += expression[j].text;
+    }
+    expression.splice(nums[i].start + 1, nums[i].end - nums[i].start);
+  }
+};
+
 /**
  * Wrap the given array of nodes in an <mrow> node if needed, i.e.,
  * unless the array has length 1.  Always returns a single node.
@@ -1997,6 +2041,8 @@ const buildExpression = function(expression, style, isOrdgroup) {
     return [group];
   }
 
+  consolidateNumbers(expression);
+
   const groups = [];
   for (let i = 0; i < expression.length; i++) {
     const group = buildGroup$1(expression[i], style);
@@ -2096,18 +2142,6 @@ function buildMathML(tree, texExpression, style, settings) {
     wrapper = new mathMLTree.MathNode("semantics", [wrapper, annotation]);
   }
 
-  if (wrap !== "none" && wrapper.children.length > 1) {
-    const maths = [];
-    for (let i = 0; i < wrapper.children.length; i++) {
-      const math = new mathMLTree.MathNode("math", [wrapper.children[i]]);
-      if (settings.xml) {
-        math.setAttribute("xmlns", "http://www.w3.org/1998/Math/MathML");
-      }
-      maths.push(math);
-    }
-    return mathMLTree.newDocumentFragment(maths)
-  }
-
   const math = new mathMLTree.MathNode("math", [wrapper]);
 
   if (settings.xml) {
@@ -2115,6 +2149,9 @@ function buildMathML(tree, texExpression, style, settings) {
   }
   if (settings.displayMode) {
     math.setAttribute("display", "block");
+    math.style.display = math.children.length === 1 && math.children[0].type === "mtable"
+      ? "inline"
+      : "inline-block";
   }
   return math;
 }
@@ -7648,7 +7685,7 @@ const smallCaps = Object.freeze({
 // "mathord" and "textord" ParseNodes created in Parser.js from symbol Groups in
 // src/symbols.js.
 
-const numberRegEx$1 = /^\d(?:[\d,.]*\d)?$/;  // Keep in sync with numberRegEx in Parser.js
+const numberRegEx = /^\d(?:[\d,.]*\d)?$/;
 const latinRegEx = /[A-Ba-z]/;
 
 const italicNumber = (text, variant, tag) => {
@@ -7702,7 +7739,7 @@ defineFunctionBuilders({
     const variant = getVariant(group, style) || "normal";
 
     let node;
-    if (numberRegEx$1.test(group.text)) {
+    if (numberRegEx.test(group.text)) {
       const tag = group.mode === "text" ? "mtext" : "mn";
       if (variant === "italic" || variant === "bold-italic") {
         return italicNumber(text, variant, tag)
@@ -8015,8 +8052,7 @@ const combiningDiacriticalMarksEndRegex = new RegExp(`${combiningDiacriticalMark
 const tokenRegexString =
   `(${spaceRegexString}+)|` + // whitespace
   `${controlSpaceRegexString}|` +  // whitespace
-  "(number" +         // numbers (in non-strict mode)
-  "|[!-\\[\\]-\u2027\u202A-\uD7FF\uF900-\uFFFF]" + // single codepoint
+  "([!-\\[\\]-\u2027\u202A-\uD7FF\uF900-\uFFFF]" + // single codepoint
   `${combiningDiacriticalMarkString}*` + // ...plus accents
   "|[\uD800-\uDBFF][\uDC00-\uDFFF]" + // surrogate pair
   `${combiningDiacriticalMarkString}*` + // ...plus accents
@@ -8031,12 +8067,7 @@ class Lexer {
     // Separate accents from characters
     this.input = input;
     this.settings = settings;
-    this.tokenRegex = new RegExp(
-      // Strict Temml, like TeX, lexes one numeral at a time.
-      // Default Temml lexes contiguous numerals into a single <mn> element.
-      tokenRegexString.replace("number|", settings.strict ? "" : "\\d(?:[\\d,.]*\\d)?|"),
-      "g"
-    );
+    this.tokenRegex = new RegExp(tokenRegexString, 'g');
     // Category codes. The lexer only supports comment characters (14) for now.
     // MacroExpander additionally distinguishes active (13).
     this.catcodes = {
@@ -11804,8 +11835,6 @@ var unicodeSymbols = {
 
 /* eslint no-constant-condition:0 */
 
-const numberRegEx = /^\d(?:[\d,.]*\d)?$/;  // Keep in sync with numberRegEx in symbolsOrd.js
-
 /**
  * This file contains the parser used to parse out a TeX expression from the
  * input. Since TeX isn't context-free, standard parsers don't work particularly
@@ -12727,15 +12756,6 @@ class Parser {
         };
       }
       symbol = s;
-    } else if (!this.strict && numberRegEx.test(text)) {
-      // A number. Wrap in a <mn> if in math mode; <mtext> otherwise.
-      this.consume();
-      return {
-        type: "textord",
-        mode: this.mode,
-        loc: SourceLocation.range(nucleus),
-        text
-      }
     } else if (text.charCodeAt(0) >= 0x80) {
       // no symbol for e.g. ^
       if (this.settings.strict) {
@@ -12975,7 +12995,7 @@ class Style {
  * https://mit-license.org/
  */
 
-const version = "0.10.2";
+const version = "0.10.4";
 
 function postProcess(block) {
   const labelMap = {};

package/dist/temmlPostProcess.js

@@ -14,7 +14,7 @@
    * https://mit-license.org/
    */
 
-  const version = "0.10.2";
+  const version = "0.10.4";
 
   function postProcess(block) {
     const labelMap = {};

package/package.json

@@ -1,8 +1,14 @@
 {
   "name": "temml",
-  "version": "0.10.2",
+  "version": "0.10.4",
   "description": "TeX to MathML conversion in JavaScript.",
   "main": "dist/temml.js",
+  "exports": {
+    ".": {
+      "require": "./dist/temml.cjs"
+    },
+    "./*": "./*"
+  },
   "homepage": "https://temml.org",
   "repository": {
     "type": "git",
@@ -24,7 +30,7 @@
   },
   "scripts": {
     "lint": "eslint temml.js src",
-    "unit-test": "node -r esm ./test/unit-test.js",
+    "unit-test": "node ./test/unit-test.cjs",
     "visual-test": "node utils/buildTests.js",
     "test": "yarn lint && node utils/buildTests.js && yarn unit-test",
     "minify": "terser test/temml.js -o site/assets/temml.min.js -c -m && terser contrib/mhchem/mhchem.js -o site/assets/mhchem.min.js -c -m",

package/src/Lexer.js

@@ -49,8 +49,7 @@ export const combiningDiacriticalMarksEndRegex = new RegExp(`${combiningDiacriti
 const tokenRegexString =
   `(${spaceRegexString}+)|` + // whitespace
   `${controlSpaceRegexString}|` +  // whitespace
-  "(number" +         // numbers (in non-strict mode)
-  "|[!-\\[\\]-\u2027\u202A-\uD7FF\uF900-\uFFFF]" + // single codepoint
+  "([!-\\[\\]-\u2027\u202A-\uD7FF\uF900-\uFFFF]" + // single codepoint
   `${combiningDiacriticalMarkString}*` + // ...plus accents
   "|[\uD800-\uDBFF][\uDC00-\uDFFF]" + // surrogate pair
   `${combiningDiacriticalMarkString}*` + // ...plus accents
@@ -65,12 +64,7 @@ export default class Lexer {
     // Separate accents from characters
     this.input = input;
     this.settings = settings;
-    this.tokenRegex = new RegExp(
-      // Strict Temml, like TeX, lexes one numeral at a time.
-      // Default Temml lexes contiguous numerals into a single <mn> element.
-      tokenRegexString.replace("number|", settings.strict ? "" : "\\d(?:[\\d,.]*\\d)?|"),
-      "g"
-    );
+    this.tokenRegex = new RegExp(tokenRegexString, 'g');
     // Category codes. The lexer only supports comment characters (14) for now.
     // MacroExpander additionally distinguishes active (13).
     this.catcodes = {

package/src/Parser.js

@@ -16,8 +16,6 @@ import { isDelimiter } from "./functions/delimsizing"
 import unicodeAccents from /*preval*/ "./unicodeAccents";
 import unicodeSymbols from /*preval*/ "./unicodeSymbols";
 
-const numberRegEx = /^\d(?:[\d,.]*\d)?$/  // Keep in sync with numberRegEx in symbolsOrd.js
-
 /**
  * This file contains the parser used to parse out a TeX expression from the
  * input. Since TeX isn't context-free, standard parsers don't work particularly
@@ -939,15 +937,6 @@ export default class Parser {
         };
       }
       symbol = s;
-    } else if (!this.strict && numberRegEx.test(text)) {
-      // A number. Wrap in a <mn> if in math mode; <mtext> otherwise.
-      this.consume()
-      return {
-        type: "textord",
-        mode: this.mode,
-        loc: SourceLocation.range(nucleus),
-        text
-      }
     } else if (text.charCodeAt(0) >= 0x80) {
       // no symbol for e.g. ^
       if (this.settings.strict) {

package/src/Settings.js

@@ -17,7 +17,7 @@ export default class Settings {
     this.leqno = utils.deflt(options.leqno, false);                // boolean
     this.errorColor = utils.deflt(options.errorColor, "#b22222");  // string
     this.macros = options.macros || {};
-    this.wrap = utils.deflt(options.wrap, "none")         // "none" | "tex" | "="
+    this.wrap = utils.deflt(options.wrap, "tex")                    // "tex" | "="
     this.xml = utils.deflt(options.xml, false);                     // boolean
     this.colorIsTextColor = utils.deflt(options.colorIsTextColor, false);  // booelean
     this.strict = utils.deflt(options.strict, false);    // boolean

package/src/buildMathML.js

@@ -75,6 +75,48 @@ export const consolidateText = mrow => {
   return mtext
 }
 
+const numberRegEx = /^[0-9]$/
+const isCommaOrDot = node => {
+  return (node.type === "atom" && node.text === ",") ||
+         (node.type === "textord" && node.text === ".")
+}
+const consolidateNumbers = expression => {
+  // Consolidate adjacent numbers. We want to return <mn>1,506.3</mn>,
+  // not <mn>1</mn><mo>,</mo><mn>5</mn><mn>0</mn><mn>6</mn><mi>.</mi><mn>3</mn>
+  if (expression.length < 2) { return }
+  const nums = [];
+  let inNum = false
+  // Find adjacent numerals
+  for (let i = 0; i < expression.length; i++) {
+    const node = expression[i];
+    if (node.type === "textord" && numberRegEx.test(node.text)) {
+      if (!inNum) { nums.push({ start: i }) }
+      inNum = true
+    } else {
+      if (inNum) { nums[nums.length - 1].end = i - 1 }
+      inNum = false
+    }
+  }
+  if (inNum) { nums[nums.length - 1].end = expression.length - 1 }
+
+  // Determine if numeral groups are separated by a comma or dot.
+  for (let i = nums.length - 1; i > 0; i--) {
+    if (nums[i - 1].end === nums[i].start - 2 && isCommaOrDot(expression[nums[i].start - 1])) {
+      // Merge the two groups.
+      nums[i - 1].end = nums[i].end
+      nums.splice(i, 1)
+    }
+  }
+
+  // Consolidate the number nodes
+  for (let i = nums.length - 1; i >= 0; i--) {
+    for (let j = nums[i].start + 1; j <= nums[i].end; j++) {
+      expression[nums[i].start].text += expression[j].text
+    }
+    expression.splice(nums[i].start + 1, nums[i].end - nums[i].start)
+  }
+}
+
 /**
  * Wrap the given array of nodes in an <mrow> node if needed, i.e.,
  * unless the array has length 1.  Always returns a single node.
@@ -110,6 +152,8 @@ export const buildExpression = function(expression, style, isOrdgroup) {
     return [group];
   }
 
+  consolidateNumbers(expression)
+
   const groups = [];
   for (let i = 0; i < expression.length; i++) {
     const group = buildGroup(expression[i], style);
@@ -209,18 +253,6 @@ export default function buildMathML(tree, texExpression, style, settings) {
     wrapper = new mathMLTree.MathNode("semantics", [wrapper, annotation]);
   }
 
-  if (wrap !== "none" && wrapper.children.length > 1) {
-    const maths = []
-    for (let i = 0; i < wrapper.children.length; i++) {
-      const math = new mathMLTree.MathNode("math", [wrapper.children[i]])
-      if (settings.xml) {
-        math.setAttribute("xmlns", "http://www.w3.org/1998/Math/MathML")
-      }
-      maths.push(math)
-    }
-    return mathMLTree.newDocumentFragment(maths)
-  }
-
   const math = new mathMLTree.MathNode("math", [wrapper])
 
   if (settings.xml) {
@@ -228,6 +260,9 @@ export default function buildMathML(tree, texExpression, style, settings) {
   }
   if (settings.displayMode) {
     math.setAttribute("display", "block");
+    math.style.display = math.children.length === 1 && math.children[0].type === "mtable"
+      ? "inline"
+      : "inline-block"
   }
   return math;
 }

package/src/functions/symbolsOrd.js

@@ -7,7 +7,7 @@ import * as mml from "../buildMathML"
 // "mathord" and "textord" ParseNodes created in Parser.js from symbol Groups in
 // src/symbols.js.
 
-const numberRegEx = /^\d(?:[\d,.]*\d)?$/  // Keep in sync with numberRegEx in Parser.js
+const numberRegEx = /^\d(?:[\d,.]*\d)?$/
 const latinRegEx = /[A-Ba-z]/
 
 const italicNumber = (text, variant, tag) => {

package/src/linebreaking.js

@@ -13,10 +13,12 @@ import mathMLTree from "./mathMLTree"
  * Then the top level of a <math> element can be occupied by <mrow> elements, and the browser
  * will break after a <mrow> if the expression extends beyond the container limit.
  *
- * We want the expression to render with soft line breaks after each top-level binary or
+ * The default is for soft line breaks after each top-level binary or
  * relational operator, per TeXbook p. 173. So we gather the expression into <mrow>s so that
  * each <mrow> ends in a binary or relational operator.
  *
+ * An option is for soft line breaks before an "=" sign. That changes the <mrow>s.
+ *
  * Soft line breaks will not work in Chromium and Safari, only Firefox.
  *
  * Hopefully browsers will someday do their own linebreaking and we will be able to delete

package/src/postProcess.js

@@ -8,7 +8,7 @@
  * https://mit-license.org/
  */
 
-export const version = "0.10.2";
+export const version = "0.10.4";
 
 export function postProcess(block) {
   const labelMap = {}