fix: unit parsing edge cases
This commit is contained in:
parent
f5009d09bb
commit
3787756137
2 changed files with 66 additions and 2 deletions
|
|
@ -16,8 +16,10 @@
|
|||
const NUM = '\\d+(?:\\.\\d+)?';
|
||||
const FRAC = '\\d+\\s*/\\s*\\d+';
|
||||
const MIXED = '\\d+\\s+\\d+\\s*/\\s*\\d+';
|
||||
const UNICODE_FRAC = '[\u00BC-\u00BE\u2150-\u215E]';
|
||||
const MIXED_UNI = `\\d+\\s*${UNICODE_FRAC}`;
|
||||
const APPROX_PREFIX = '~\\s*';
|
||||
const SINGLE_AMOUNT = `(?:${APPROX_PREFIX})?(?:${MIXED}|${FRAC}|${NUM})`;
|
||||
const SINGLE_AMOUNT = `(?:${APPROX_PREFIX})?(?:${MIXED}|${FRAC}|${MIXED_UNI}|${UNICODE_FRAC}|${NUM})`;
|
||||
const RANGE_SEP = '\\s*(?:-|to)\\s*';
|
||||
const AMOUNT = `(?:${SINGLE_AMOUNT}${RANGE_SEP}${SINGLE_AMOUNT}|${SINGLE_AMOUNT})`;
|
||||
|
||||
|
|
@ -39,6 +41,27 @@ const DIM_UNIT = '(?:inch(?:es)?|in\\.?|cm|mm)';
|
|||
* @param {string} str — e.g. "200", "1.5", "1/2", "1 1/2", "~250"
|
||||
* @returns {{ value: number, approximate: boolean }}
|
||||
*/
|
||||
const UNICODE_FRAC_MAP = {
|
||||
'\u00BC': 0.25, // ¼
|
||||
'\u00BD': 0.5, // ½
|
||||
'\u00BE': 0.75, // ¾
|
||||
'\u2150': 1/7, // ⅐
|
||||
'\u2151': 1/9, // ⅑
|
||||
'\u2152': 1/10, // ⅒
|
||||
'\u2153': 1/3, // ⅓
|
||||
'\u2154': 2/3, // ⅔
|
||||
'\u2155': 0.2, // ⅕
|
||||
'\u2156': 0.4, // ⅖
|
||||
'\u2157': 0.6, // ⅗
|
||||
'\u2158': 0.8, // ⅘
|
||||
'\u2159': 1/6, // ⅙
|
||||
'\u215A': 5/6, // ⅚
|
||||
'\u215B': 0.125, // ⅛
|
||||
'\u215C': 0.375, // ⅜
|
||||
'\u215D': 0.625, // ⅝
|
||||
'\u215E': 0.875, // ⅞
|
||||
};
|
||||
|
||||
function parseSingleAmount(str) {
|
||||
str = str.trim();
|
||||
const approximate = str.startsWith('~');
|
||||
|
|
@ -46,6 +69,20 @@ function parseSingleAmount(str) {
|
|||
str = str.replace(/^~\s*/, '');
|
||||
}
|
||||
|
||||
// Unicode mixed number: "1½", "1 ½"
|
||||
const uniMixedMatch = str.match(/^(\d+)\s*([\u00BC-\u00BE\u2150-\u215E])$/);
|
||||
if (uniMixedMatch) {
|
||||
const whole = parseInt(uniMixedMatch[1], 10);
|
||||
const frac = UNICODE_FRAC_MAP[uniMixedMatch[2]] || 0;
|
||||
return { value: whole + frac, approximate };
|
||||
}
|
||||
|
||||
// Standalone Unicode fraction: "½", "¾"
|
||||
const uniFracMatch = str.match(/^([\u00BC-\u00BE\u2150-\u215E])$/);
|
||||
if (uniFracMatch) {
|
||||
return { value: UNICODE_FRAC_MAP[uniFracMatch[1]] || 0, approximate };
|
||||
}
|
||||
|
||||
// Mixed number: "1 1/2"
|
||||
const mixedMatch = str.match(/^(\d+)\s+(\d+)\s*\/\s*(\d+)$/);
|
||||
if (mixedMatch) {
|
||||
|
|
@ -299,6 +336,33 @@ function findDimensions(text) {
|
|||
});
|
||||
}
|
||||
|
||||
// Standalone AMOUNT + dimension unit (e.g. "1 inch", "0.5-1cm")
|
||||
const standaloneDimRe = new RegExp(
|
||||
`(${AMOUNT})\\s*(${DIM_UNIT})\\b`,
|
||||
'gi'
|
||||
);
|
||||
|
||||
while ((m = standaloneDimRe.exec(text)) !== null) {
|
||||
// Skip if overlapping with an NxN match already found
|
||||
const alreadyMatched = results.some(
|
||||
r => m.index >= r.index && m.index < r.index + r.match.length
|
||||
);
|
||||
if (alreadyMatched) continue;
|
||||
|
||||
const amount = parseAmount(m[1]);
|
||||
const unit = normalizeUnit(m[2]);
|
||||
|
||||
results.push({
|
||||
match: m[0],
|
||||
index: m.index,
|
||||
type: 'dimension',
|
||||
amount,
|
||||
unit,
|
||||
approximate: typeof amount.approximate === 'boolean' ? amount.approximate : false,
|
||||
alt: null,
|
||||
});
|
||||
}
|
||||
|
||||
return results;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -38,7 +38,7 @@ makes: 1 9x13 pan or 4 large rectangular pyrex containers
|
|||
## Tools
|
||||
- pan
|
||||
- mixing bowl
|
||||
- 9 x 13 inch glass baking dish/4 8x6x2 inch pyrex dishes
|
||||
- 9 x 13 inch glass baking dish/8x6x2 inch pyrex dishes 4
|
||||
## Steps
|
||||
- preheat oven to 350 °F
|
||||
- dice onion
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue