madlisp/src/Tokenizer.php

97 lines
3.0 KiB
PHP
Raw Normal View History

2020-05-28 10:10:00 +00:00
<?php
namespace MadLisp;
class Tokenizer
{
public function tokenize(string $a): array
{
$tokens = [];
$current = '';
2020-05-31 02:56:23 +00:00
$isString = false;
$isComment = false;
$parens = [0, 0, 0];
$parenIndexes = ['(' => 0, ')' => 0, '[' => 1, ']' => 1, '{' => 2, '}' => 2];
2020-05-28 10:10:00 +00:00
$addCurrent = function () use (&$tokens, &$current) {
if ($current !== '') {
$tokens[] = $current;
$current = '';
}
};
for ($i = 0; $i < strlen($a); $i++) {
$c = substr($a, $i, 1);
2020-05-31 02:56:23 +00:00
if ($isString) {
2020-05-28 10:10:00 +00:00
// Inside string, add all characters
$current .= $c;
2020-05-31 04:34:24 +00:00
// Stop at first double quote
2020-05-28 10:10:00 +00:00
if ($c == '"') {
$addCurrent();
2020-05-31 02:56:23 +00:00
$isString = false;
}
} elseif ($isComment) {
// Comments stop at first newline
if ($c == "\n" || $c == "\r") {
$isComment = false;
2020-05-28 10:10:00 +00:00
}
} else {
2020-05-31 02:56:23 +00:00
// Not inside string or comment
2020-05-28 10:10:00 +00:00
if ($c == '"') {
// Start of string
$addCurrent();
$current .= $c;
2020-05-31 02:56:23 +00:00
$isString = true;
} elseif ($c == ';') {
// Start of comment
$addCurrent();
$isComment = true;
2020-05-28 10:10:00 +00:00
} elseif ($c == ' ' || $c == "\t" || $c == "\n" || $c == "\r") {
// Whitespace is ignored
$addCurrent();
2020-05-31 02:56:23 +00:00
} elseif ($c == '(' || $c == '[' || $c == '{') {
// Start of collection
2020-05-28 10:10:00 +00:00
$addCurrent();
2020-05-31 02:56:23 +00:00
$tokens[] = $c;
$parens[$parenIndexes[$c]]++;
} elseif ($c == ')' || $c == ']' || $c == '}') {
// End of collection
if ($parens[$parenIndexes[$c]] == 0) {
throw new MadLispException("unexpected closing $c");
2020-05-28 10:10:00 +00:00
}
$addCurrent();
2020-05-31 02:56:23 +00:00
$tokens[] = $c;
$parens[$parenIndexes[$c]]--;
} elseif ($c == "'") {
// Other special characters
$addCurrent();
$tokens[] = $c;
2020-05-28 10:10:00 +00:00
} else {
// All other characters
$current .= $c;
}
}
}
2020-05-31 02:56:23 +00:00
// Add last token
2020-05-28 10:10:00 +00:00
$addCurrent();
// Check for errors
2020-05-31 02:56:23 +00:00
if ($parens[0] != 0) {
throw new MadLispException("missing closing )");
} elseif ($parens[1] != 0) {
throw new MadLispException("missing closing ]");
} elseif ($parens[2] != 0) {
throw new MadLispException("missing closing }");
} elseif ($isString) {
2020-05-28 10:10:00 +00:00
throw new MadLispException("unterminated string");
}
return $tokens;
}
}