diff --git a/backend/src/quick-import/parsers/base.parser.ts b/backend/src/quick-import/parsers/base.parser.ts index bfca6265..b448e63f 100644 --- a/backend/src/quick-import/parsers/base.parser.ts +++ b/backend/src/quick-import/parsers/base.parser.ts @@ -9,6 +9,7 @@ export interface ParsedRecipe { quantity: number; unit: string; name: string; + note?: string; }>; instructions?: string; } @@ -29,61 +30,119 @@ export abstract class RecipeParser { * Hanterar format som: * - "3 ägg" * - "150 g lax" + * - "1/2 citron" * - "1 msk senap" * - "salt och peppar" + * - "1 förp handskalade räkor i lake (à 570 g)" */ protected parseIngredientLine(line: string): { quantity: number; unit: string; name: string; + note?: string; } | null { - const cleaned = line.replace(/<[^>]+>/g, '').trim(); + let cleaned = line.replace(/<[^>]+>/g, '').trim(); if (!cleaned) return null; // Kända enheter const knownUnits = [ - 'g', 'kg', 'hg', 'ml', 'dl', 'l', + 'g', 'kg', 'hg', 'mg', 'ml', 'dl', 'l', 'tl', 'st', 'tsk', 'msk', 'krm', 'matsked', 'tesked', 'pris', 'portion', 'burk', 'förp', 'paket', ]; - // Försök extrahera: [quantity] [unit?] [productName] - // Regex: början med valfri mängd, sedan valfritt ord (potentiell enhet), sedan resten - const match = cleaned.match(/^([\d.,]+)?\s*([a-zåäö]*)\s*(.*)$/i); + // Extrahera parentetisk info + let parentheticalText = ''; + const parentheteMatch = cleaned.match(/\s*\(([^)]*)\)/); + if (parentheteMatch) { + parentheticalText = parentheteMatch[1].trim(); + cleaned = cleaned.replace(/\s*\([^)]*\)/, '').trim(); + } - if (!match) { + // Hantera bråkdelar: "1/2" eller "1 / 2" + const fractionMatch = cleaned.match(/^([\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; + remainingText = cleaned.substring(fractionMatch[0].length).trim(); + } else { + const numberMatch = remainingText.match(/^([\d.,]+)/); + if (numberMatch) { + quantity = parseFloat(numberMatch[1].replace(',', '.')); + remainingText = remainingText.substring(numberMatch[0].length).trim(); + } + } + + // Extrahera potentiell enhet + let potentialUnit = ''; + let productName = remainingText; + + if (remainingText) { + const unitMatch = remainingText.match(/^([a-zåäö]+)\b/i); + if (unitMatch) { + const candidateUnit = unitMatch[1].toLowerCase(); + if (knownUnits.includes(candidateUnit)) { + potentialUnit = candidateUnit; + productName = remainingText.substring(candidateUnit.length).trim(); + } + } + } + + // Analysera parenthetical text för måttenhet + let parenthHasUnit = false; + if (parentheticalText) { + for (const unit of knownUnits) { + if (parentheticalText.toLowerCase().includes(unit)) { + parenthHasUnit = true; + break; + } + } + } + + let note: string | undefined = undefined; + + // Om vi hade quantity i huvuddelen och parenthetical innehåller unit + // → spara parenthetical som note + if (quantity > 0 && parenthHasUnit) { + note = parentheticalText; + } + + // Om ingen mängd i huvuddelen men parenthetical hade både mängd och unit + // → parse parenthetical som quantity + unit + if (quantity === 0 && parentheticalText) { + const parenthMatch = parentheticalText.match(/^[\D]*?([\d.,]+)?\s*([a-zåäö]*)?\s*(.*)$/i); + if (parenthMatch) { + let pQuantity = parenthMatch[1] ? parseFloat(parenthMatch[1].replace(',', '.')) : 0; + let pUnit = parenthMatch[2]?.toLowerCase() || ''; + let pRest = parenthMatch[3]?.trim() || ''; + + if (knownUnits.includes(pUnit) && pQuantity > 0) { + quantity = pQuantity; + potentialUnit = pUnit; + note = parentheticalText; + } + } + } + + // Om ingen mängd och enhet, bara returna produktnamnet + if (quantity === 0) { return { quantity: 0, - unit: 'st', + unit: '', name: cleaned, + note: parentheticalText || undefined, }; } - let quantity = match[1] ? parseFloat(match[1].replace(',', '.')) : 0; - let potentialUnit = match[2]?.toLowerCase().trim() || ''; - let productName = match[3]?.trim() || ''; - - // Om potentialUnit är inte en känd enhet, lägg det tillbaka i produktnamnet - if (potentialUnit && !knownUnits.includes(potentialUnit)) { - productName = potentialUnit + (productName ? ' ' + productName : ''); - potentialUnit = ''; - } - - // Om inget produktnamn men vi hade potentialUnit, denna är faktiskt produktnamnet - if (!productName && potentialUnit) { - productName = potentialUnit; - potentialUnit = ''; - } - - // Fallback: om vi har mängd men inget annat, vara produktnamn - if (quantity > 0 && !potentialUnit && !productName) { - return null; // Ogiltigt - } - return { quantity, - unit: potentialUnit || 'st', + unit: potentialUnit, name: productName, + note: note, }; } } diff --git a/backend/src/quick-import/parsers/generic.parser.ts b/backend/src/quick-import/parsers/generic.parser.ts index f83c2bc7..898ff28e 100644 --- a/backend/src/quick-import/parsers/generic.parser.ts +++ b/backend/src/quick-import/parsers/generic.parser.ts @@ -43,7 +43,7 @@ export class GenericRecipeParser extends RecipeParser { private extractFromJsonLd(recipe: any): ParsedRecipe { const name = recipe.name || ''; - const ingredients: Array<{ quantity: number; unit: string; name: string }> = []; + const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = []; if (recipe.recipeIngredient && Array.isArray(recipe.recipeIngredient)) { for (const ing of recipe.recipeIngredient) { const parsed = this.parseIngredientLine(ing); @@ -91,7 +91,7 @@ export class GenericRecipeParser extends RecipeParser { } // Försöka extrahera ingredienser från vanliga strukturer - const ingredients: Array<{ quantity: number; unit: string; name: string }> = []; + const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = []; // Testa olika ingredient-selectors const ingredientPatterns = [ diff --git a/backend/src/quick-import/parsers/ica.parser.ts b/backend/src/quick-import/parsers/ica.parser.ts index dcaf58e3..ae058194 100644 --- a/backend/src/quick-import/parsers/ica.parser.ts +++ b/backend/src/quick-import/parsers/ica.parser.ts @@ -46,7 +46,7 @@ export class IcaRecipeParser extends RecipeParser { const name = recipe.name || ''; // Extrahera ingredienser - const ingredients: Array<{ quantity: number; unit: string; name: string }> = []; + const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = []; if (recipe.recipeIngredient && Array.isArray(recipe.recipeIngredient)) { for (const ing of recipe.recipeIngredient) { const parsed = this.parseIngredientLine(ing); @@ -96,7 +96,7 @@ export class IcaRecipeParser extends RecipeParser { } } - const ingredients: Array<{ quantity: number; unit: string; name: string }> = []; + const ingredients: Array<{ quantity: number; unit: string; name: string; note?: string }> = []; const ingredientRegex = /
+ {markdown}
+
+
+ {JSON.stringify(parsed, null, 2)}
+
+