feat: Enhance ingredient parsing to support mixed fractions and add description extraction in recipe parsers

This commit is contained in:
Nils-Johan Gynther
2026-04-12 10:50:59 +02:00
parent 9ca7fcce96
commit 3d4994f24d
5 changed files with 314 additions and 218 deletions
@@ -59,15 +59,25 @@ export abstract class RecipeParser {
cleaned = cleaned.replace(/\s*\([^)]*\)/, '').trim();
}
// Hantera bråkdelar: "1/2" eller "1 / 2"
const fractionMatch = cleaned.match(/^([\d.]+)\s*\/\s*([\d.]+)/);
// Hantera bråkdelar: "1/2" eller "1 1/2" eller "1 1 / 2"
// Regex: (optional whole)? numerator / denominator
const fractionMatch = cleaned.match(/^(\d+)?\s*(\d+)\s*\/\s*([\d.]+)/);
let quantity = 0;
let remainingText = cleaned;
if (fractionMatch) {
const numerator = parseFloat(fractionMatch[1]);
const denominator = parseFloat(fractionMatch[2]);
quantity = numerator / denominator;
if (fractionMatch[1]) {
// Heltal + bråk: "1 1/2"
const whole = parseFloat(fractionMatch[1]);
const numerator = parseFloat(fractionMatch[2]);
const denominator = parseFloat(fractionMatch[3]);
quantity = whole + (numerator / denominator);
} else {
// Bara bråk: "1/2"
const numerator = parseFloat(fractionMatch[2]);
const denominator = parseFloat(fractionMatch[3]);
quantity = numerator / denominator;
}
remainingText = cleaned.substring(fractionMatch[0].length).trim();
} else {
const numberMatch = remainingText.match(/^([\d.,]+)/);
@@ -42,6 +42,7 @@ export class GenericRecipeParser extends RecipeParser {
private extractFromJsonLd(recipe: any): ParsedRecipe {
const name = recipe.name || '';
const description = recipe.description || '';
const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = [];
if (recipe.recipeIngredient && Array.isArray(recipe.recipeIngredient)) {
@@ -71,6 +72,7 @@ export class GenericRecipeParser extends RecipeParser {
return {
name,
description,
ingredients,
instructions,
};
@@ -90,6 +92,15 @@ export class GenericRecipeParser extends RecipeParser {
name = titleMatch[1].trim();
}
// Försöka extrahera beskrivning från meta-taggar
let description = '';
const descMatch = html.match(
/<meta\s+name="description"\s+content="([^"]+)"/i
);
if (descMatch) {
description = descMatch[1].trim();
}
// Försöka extrahera ingredienser från vanliga strukturer
const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = [];
@@ -129,6 +140,7 @@ export class GenericRecipeParser extends RecipeParser {
return {
name,
description,
ingredients,
instructions,
};
@@ -45,6 +45,9 @@ export class IcaRecipeParser extends RecipeParser {
// Extrahera titel
const name = recipe.name || '';
// Extrahera beskrivning
const description = recipe.description || '';
// Extrahera ingredienser
const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = [];
if (recipe.recipeIngredient && Array.isArray(recipe.recipeIngredient)) {
@@ -75,6 +78,7 @@ export class IcaRecipeParser extends RecipeParser {
return {
name,
description,
ingredients,
instructions,
};
@@ -96,6 +100,15 @@ export class IcaRecipeParser extends RecipeParser {
}
}
// Extrahera beskrivning från meta-taggar
let description = '';
const descMatch = html.match(
/<meta\s+name="description"\s+content="([^"]+)"/i
);
if (descMatch) {
description = descMatch[1].trim();
}
const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = [];
const ingredientRegex =
/<li[^>]*class="[^"]*ingredient[^"]*"[^>]*>([^<]+)<\/li>/gi;
@@ -117,6 +130,7 @@ export class IcaRecipeParser extends RecipeParser {
return {
name,
description,
ingredients,
instructions,
};
+43 -3
View File
@@ -529,7 +529,7 @@ function parseRecipeMarkdown(markdown: string): ParsedRecipe {
const heading = trimmed.replace(/^##\s+/, '').trim().toLowerCase();
if (/ingrediens/.test(heading)) {
currentSection = 'ingredients';
} else if (/instruktion|tillagning|gör så här|steg/.test(heading)) {
} else if (/instruktion|tillagning|gör så här|steg|tillväg|metod/.test(heading)) {
currentSection = 'instructions';
} else {
currentSection = 'none';
@@ -570,12 +570,21 @@ function parseRecipeMarkdown(markdown: string): ParsedRecipe {
* Parsar en ingrediensrad, t.ex.:
* "400 g kycklingfilé"
* "2 dl grädde (eller crème fraiche)"
* "1 1/2 dl crème fraiche"
* "1 polka- eller gulbeta"
* "1 kruka basilika"
* "salt"
*/
function parseIngredientLine(text: string): ParsedIngredient {
const trimmed = text.trim();
// Kända enheter
const knownUnits = [
'g', 'kg', 'hg', 'mg', 'ml', 'dl', 'l', 'tl',
'st', 'tsk', 'msk', 'krm', 'matsled', 'tesled',
'pris', 'portion', 'burk', 'förp', 'paket',
];
// Extrahera eventuell parentes-not i slutet
let note: string | null = null;
let main = trimmed;
@@ -585,13 +594,44 @@ function parseIngredientLine(text: string): ParsedIngredient {
main = trimmed.slice(0, parenMatch.index).trim();
}
// Försök matcha bråk först: "1 1/2 dl crème fraiche" eller "1/2 dl"
const fractionMatch = main.match(/^(\d+)?\s*(\d+)\s*\/\s*([\d.]+)\s+(\S+)\s+(.*)$/);
if (fractionMatch) {
let quantity = 0;
if (fractionMatch[1]) {
quantity = parseFloat(fractionMatch[1]) + parseFloat(fractionMatch[2]) / parseFloat(fractionMatch[3]);
} else {
quantity = parseFloat(fractionMatch[2]) / parseFloat(fractionMatch[3]);
}
const candidateUnit = fractionMatch[4].toLowerCase();
if (knownUnits.includes(candidateUnit)) {
return {
quantity,
unit: candidateUnit,
rawName: fractionMatch[5].trim(),
note,
};
}
}
// Försök matcha "kvantitet enhet namn" — t.ex. "400 g kycklingfilé" eller "2.5 dl grädde"
const fullMatch = main.match(/^(\d+(?:[.,]\d+)?)\s+(\S+)\s+(.+)$/);
if (fullMatch) {
const candidateUnit = fullMatch[2].toLowerCase();
// Validera att det andra ordet är en känd enhet
if (knownUnits.includes(candidateUnit)) {
return {
quantity: parseNumber(fullMatch[1]),
unit: candidateUnit,
rawName: fullMatch[3].trim(),
note,
};
}
// Om inte känd enhet, behandla som "kvantitet namn" utan enhet
return {
quantity: parseNumber(fullMatch[1]),
unit: fullMatch[2],
rawName: fullMatch[3].trim(),
unit: 'st',
rawName: fullMatch[2] + ' ' + fullMatch[3],
note,
};
}