init repo

pkg/utils/katex.go

@@ -0,0 +1,115 @@
+package utils
+
+import (
+	"regexp"
+	"strings"
+)
+
+// Helper function to change patterns
+func changeAll(text, oldIns, newIns, leftDelim, rightDelim, newLeftDelim, newRightDelim string) string {
+	pattern := regexp.MustCompile(regexp.QuoteMeta(oldIns) + `\s*` + regexp.QuoteMeta(leftDelim) + `(.*?)` + regexp.QuoteMeta(rightDelim))
+	return pattern.ReplaceAllString(text, newIns+newLeftDelim+"$1"+newRightDelim)
+}
+
+// Helper function to remove dollar surroundings
+func rmDollarSurr(text string) string {
+	if strings.HasPrefix(text, "$") && strings.HasSuffix(text, "$") {
+		return text[1 : len(text)-1]
+	}
+	return text
+}
+
+// ToKatex converts LaTeX formula to KaTeX compatible format
+func ToKatex(formula string) string {
+	res := formula
+
+	// Remove mbox surrounding
+	res = changeAll(res, `\mbox `, " ", "{", "}", "", "")
+	res = changeAll(res, `\mbox`, " ", "{", "}", "", "")
+
+	// Remove hbox surrounding
+	hboxPattern := regexp.MustCompile(`\\hbox to ?-? ?\d+\.\d+(pt)?\{`)
+	res = hboxPattern.ReplaceAllString(res, `\hbox{`)
+	res = changeAll(res, `\hbox`, " ", "{", "}", "", " ")
+
+	// Remove raise surrounding
+	raisePattern := regexp.MustCompile(`\\raise ?-? ?\d+\.\d+(pt)?`)
+	res = raisePattern.ReplaceAllString(res, " ")
+
+	// Remove makebox
+	makeboxPattern := regexp.MustCompile(`\\makebox ?\[\d+\.\d+(pt)?\]\{`)
+	res = makeboxPattern.ReplaceAllString(res, `\makebox{`)
+	res = changeAll(res, `\makebox`, " ", "{", "}", "", " ")
+
+	// Remove vbox, scalebox, raisebox surrounding
+	raisebox := regexp.MustCompile(`\\raisebox\{-? ?\d+\.\d+(pt)?\}\{`)
+	scalebox := regexp.MustCompile(`\\scalebox\{-? ?\d+\.\d+(pt)?\}\{`)
+	res = raisebox.ReplaceAllString(res, `\raisebox{`)
+	res = scalebox.ReplaceAllString(res, `\scalebox{`)
+	res = changeAll(res, `\scalebox`, " ", "{", "}", "", " ")
+	res = changeAll(res, `\raisebox`, " ", "{", "}", "", " ")
+	res = changeAll(res, `\vbox`, " ", "{", "}", "", " ")
+
+	// Handle size instructions
+	sizeInstructions := []string{
+		`\Huge`, `\huge`, `\LARGE`, `\Large`, `\large`,
+		`\normalsize`, `\small`, `\footnotesize`, `\tiny`,
+	}
+	for _, ins := range sizeInstructions {
+		res = changeAll(res, ins, ins, "$", "$", "{", "}")
+	}
+
+	// Handle boldmath
+	res = changeAll(res, `\boldmath `, `\bm`, "{", "}", "{", "}")
+	res = changeAll(res, `\boldmath`, `\bm`, "{", "}", "{", "}")
+	res = changeAll(res, `\boldmath `, `\bm`, "$", "$", "{", "}")
+	res = changeAll(res, `\boldmath`, `\bm`, "$", "$", "{", "}")
+
+	// Handle other instructions
+	res = changeAll(res, `\scriptsize`, `\scriptsize`, "$", "$", "{", "}")
+	res = changeAll(res, `\emph`, `\textit`, "{", "}", "{", "}")
+	res = changeAll(res, `\emph `, `\textit`, "{", "}", "{", "}")
+
+	// Handle math delimiters
+	delimiters := []string{
+		`\left`, `\middle`, `\right`, `\big`, `\Big`, `\bigg`, `\Bigg`,
+		`\bigl`, `\Bigl`, `\biggl`, `\Biggl`, `\bigm`, `\Bigm`, `\biggm`,
+		`\Biggm`, `\bigr`, `\Bigr`, `\biggr`, `\Biggr`,
+	}
+	for _, delim := range delimiters {
+		res = changeAll(res, delim, delim, "{", "}", "", "")
+	}
+
+	// Handle display math
+	displayMath := regexp.MustCompile(`\\\[(.*?)\\\]`)
+	res = displayMath.ReplaceAllString(res, "$1\\newline")
+
+	res = strings.TrimSuffix(res, `\newline`)
+
+	// Remove multiple spaces
+	spaces := regexp.MustCompile(`(\\,){1,}`)
+	res = spaces.ReplaceAllString(res, " ")
+	res = regexp.MustCompile(`(\\!){1,}`).ReplaceAllString(res, " ")
+	res = regexp.MustCompile(`(\\;){1,}`).ReplaceAllString(res, " ")
+	res = regexp.MustCompile(`(\\:){1,}`).ReplaceAllString(res, " ")
+	res = regexp.MustCompile(`\\vspace\{.*?}`).ReplaceAllString(res, "")
+
+	// Merge consecutive text
+	textPattern := regexp.MustCompile(`(\\text\{[^}]*\}\s*){2,}`)
+	res = textPattern.ReplaceAllStringFunc(res, func(match string) string {
+		texts := regexp.MustCompile(`\\text\{([^}]*)\}`).FindAllStringSubmatch(match, -1)
+		var merged strings.Builder
+		for _, t := range texts {
+			merged.WriteString(t[1])
+		}
+		return `\text{` + merged.String() + "}"
+	})
+
+	res = strings.ReplaceAll(res, `\bf `, "")
+	res = rmDollarSurr(res)
+
+	// Remove extra spaces
+	res = regexp.MustCompile(` +`).ReplaceAllString(res, " ")
+
+	return strings.TrimSpace(res)
+}