Create a KaTeX macro (small subset version) parser and expander

This commit is contained in:
naskya 2023-03-26 07:31:12 +09:00
parent 450f798f12
commit da17e8978a
No known key found for this signature in database
GPG Key ID: 164DFF24E2D40139
1 changed files with 280 additions and 0 deletions

View File

@ -0,0 +1,280 @@
type KaTeXCommand = {
args: number;
rule: (string | number)[];
};
function parseSingleKaTeXCommand(src: string): [string, KaTeXCommand] {
const invalid: [string, KaTeXCommand] = ["", { args: 0, rule: [] }];
const skipSpaces = (pos: number): number => {
while (src[pos] === " ")
++pos;
return pos;
};
if (!src.startsWith("\\newcommand") || src.slice(-1) !== "}")
return invalid;
// current index we are checking (= "\\newcommand".length)
let currentPos: number = 11;
currentPos = skipSpaces(currentPos);
// parse {\name}, (\name), or [\name]
let bracket: string;
if (src[currentPos] === "{")
bracket = "{}";
else if (src[currentPos] === "(")
bracket = "()";
else if (src[currentPos] === "[")
bracket = "[]";
else
return invalid;
++currentPos;
currentPos = skipSpaces(currentPos);
if (src[currentPos] !== "\\")
return invalid;
const closeNameBracketPos: number = src.indexOf(bracket[1], currentPos);
if (closeNameBracketPos === -1)
return invalid;
const name: string = src.slice(currentPos + 1, closeNameBracketPos).trim();
if (!/^[a-zA-Z]+$/.test(name))
return invalid;
currentPos = skipSpaces(closeNameBracketPos + 1);
let command: KaTeXCommand = { args: 0, rule: [] };
// parse [number of arguments] (optional)
if (src[currentPos] === "[") {
const closeArgsBracketPos: number = src.indexOf("]", currentPos);
command.args = Number(src.slice(currentPos + 1, closeArgsBracketPos).trim());
currentPos = closeArgsBracketPos + 1;
if (Number.isNaN(command.args) || command.args < 0)
return invalid;
} else if (src[currentPos] === "{") {
command.args = 0;
} else {
return invalid;
}
currentPos = skipSpaces(currentPos);
// parse {rule}
if (src[currentPos] !== "{")
return invalid;
++currentPos;
currentPos = skipSpaces(currentPos);
while (currentPos < src.length - 1) {
let numbersignPos: number = -1;
let isEscaped: boolean = false;
for (let i = currentPos; i < src.length - 1; ++i) {
if (src[i] !== "\\" && src[i] !== "#") {
isEscaped = false;
continue;
}
if (src[i] === "\\") {
isEscaped = !isEscaped;
continue;
}
if (!isEscaped && src[i] === "#") {
numbersignPos = i;
break;
}
}
if (numbersignPos === -1) {
command.rule.push(src.slice(currentPos, -1));
break;
}
const argIndexEndPos = src.slice(numbersignPos + 1).search(/[^\d]/) + numbersignPos;
const argIndex: number = Number(src.slice(numbersignPos + 1, argIndexEndPos + 1));
if (Number.isNaN(argIndex) || argIndex < 1 || command.args < argIndex)
return invalid;
if (currentPos !== numbersignPos)
command.rule.push(src.slice(currentPos, numbersignPos));
command.rule.push(argIndex);
currentPos = argIndexEndPos + 1;
}
if (command.args === 0)
return [name, command];
else
return [name + bracket[0], command];
}
export function parseKaTeXCommands(src: string): string {
let result: { [name: string]: KaTeXCommand } = {};
for (const s of src.split("\n")) {
const [name, command]: [string, KaTeXCommand] = parseSingleKaTeXCommand(s.trim());
if (name !== "")
result[name] = command;
}
return JSON.stringify(result);
}
export function expandKaTeXCommand(src: string, commandsAsJsonString: string): string {
const commands = JSON.parse(commandsAsJsonString);
const bracketKinds = 3;
const openBracketId: { [bracket: string]: number } = {"(": 0, "{": 1, "[": 2};
const closeBracketId: { [bracket: string]: number } = {")": 0, "}": 1, "]": 2};
const openBracketFromId = ["(", "{", "["];
const closeBracketFromId = [")", "}", "]"];
// mappings from open brackets to their corresponding close brackets
type BracketMapping = { [openBracketPos: number]: number };
const bracketMapping = ((): BracketMapping => {
let result: BracketMapping = {};
const n = src.length;
let depths = new Array<number>(bracketKinds).fill(0); // current bracket depth for "()", "{}", and "[]"
let buffer = Array.from(Array<number[]>(bracketKinds), () => Array<number>(n));
let isEscaped = false;
for (let i = 0; i < n; ++i) {
if (!isEscaped && src[i] === "\\" && i + 1 < n && ["{", "}", "\\"].includes(src[i+1])) {
isEscaped = true;
continue;
}
if (isEscaped
|| (src[i] !== "\\"
&& !openBracketFromId.includes(src[i])
&& !closeBracketFromId.includes(src[i])))
{
isEscaped = false;
continue;
}
isEscaped = false;
if (openBracketFromId.includes(src[i])) {
const id: number = openBracketId[src[i]];
buffer[id][depths[id]] = i;
++depths[id];
} else if (closeBracketFromId.includes(src[i])) {
const id: number = closeBracketId[src[i]];
--depths[id];
result[buffer[id][depths[id]]] = i;
}
}
return result;
})();
function expandSingleKaTeXCommand(expandedArgs: string[], commandName: string): string {
let result = "";
for (const block of commands[commandName].rule) {
if (typeof block === "string")
result += block;
else
result += expandedArgs[block - 1];
}
return result;
}
let numberOfExpansions = 0;
const maxNumberOfExpansions = 200; // to prevent infinite expansion loop
// only expand src.slice(beginPos, endPos)
function expandKaTeXCommandImpl(beginPos: number, endPos: number): string {
if (endPos <= beginPos)
return "";
const raw: string = src.slice(beginPos, endPos + 1);
const fallback: string = raw; // returned for invalid inputs
if (maxNumberOfExpansions <= numberOfExpansions)
return fallback;
++numberOfExpansions;
// search for a custom command
let checkedPos = beginPos - 1;
let commandName = "";
let commandBackslashPos = 0;
// for commands w/o args: unused
// w/ args: the first open bracket ("(", "{", or "[") after cmd name
let commandArgBeginPos = 0;
// for commands w/o args: the end of cmd name
// w/ args: the closing bracket of the last arg
let commandArgEndPos = 0;
while (checkedPos < endPos) {
checkedPos = src.indexOf("\\", checkedPos + 1);
// there is no command to expand
if (checkedPos === -1)
return raw;
// is it a custom command?
let nonAlphaPos = src.slice(checkedPos + 1).search(/[^A-Za-z]/) + checkedPos + 1;
let commandNameCandidate = src.slice(checkedPos + 1, nonAlphaPos);
if (commands.hasOwnProperty(commandNameCandidate)) {
// this is a custom command without args
commandBackslashPos = checkedPos;
commandArgEndPos = nonAlphaPos - 1;
commandName = commandNameCandidate;
break;
}
let nextOpenBracketPos = endPos;
for (let i = 0; i < bracketKinds; ++i) {
const pos = src.indexOf(openBracketFromId[i], checkedPos + 1);
if (pos !== -1 && pos < nextOpenBracketPos)
nextOpenBracketPos = pos;
}
if (nextOpenBracketPos === endPos)
return fallback; // there is no open bracket
commandNameCandidate += src[nextOpenBracketPos];
if (commands.hasOwnProperty(commandNameCandidate)) {
commandBackslashPos = checkedPos;
commandArgBeginPos = nextOpenBracketPos;
commandArgEndPos = nextOpenBracketPos; // to search the first arg from here
commandName = commandNameCandidate;
break;
}
}
const numArgs: number = commands[commandName].args;
const openBracket: string = commandName.slice(-1);
let expandedArgs = new Array<string>(numArgs);
for (let i = 0; i < numArgs; ++i) {
// find the first open bracket after what we've searched
const nextOpenBracketPos = src.indexOf(openBracket, commandArgEndPos);
if (nextOpenBracketPos === -1)
return fallback; // not enough arguments are provided
if (!bracketMapping[nextOpenBracketPos])
return fallback; // found open bracket doesn't correspond to any close bracket
commandArgEndPos = bracketMapping[nextOpenBracketPos];
expandedArgs[i] = expandKaTeXCommandImpl(nextOpenBracketPos + 1, commandArgEndPos);
}
return src.slice(beginPos, commandBackslashPos)
+ expandSingleKaTeXCommand(expandedArgs, commandName)
+ expandKaTeXCommandImpl(commandArgEndPos + 1, endPos);
}
return expandKaTeXCommandImpl(0, src.length);
}