Generate budget brand matches in pattern.js, fix json download

This commit is contained in:
Mario Zechner 2023-06-26 02:46:22 +02:00
parent 930fe762d3
commit ebca44dc4e
2 changed files with 74 additions and 51 deletions

View File

@ -2,33 +2,18 @@ const fs = require("fs");
const analysis = require("./analysis");
const knn = require("./site/js/knn");
if (!fs.existsSync("patterns")) fs.mkdirSync("patterns");
if (!fs.existsSync("patterns/sorted-billa-spar.json")) {
const items = analysis.readJSON("data/latest-canonical.json.br");
const billaSparItem = items
.filter((item) => item.store == "billa" || item.store == "spar")
.filter((item) => {
return !(
item.name.includes("Clever") ||
item.name.includes("S-BUDGET") ||
item.name.includes("Ja! Natürlich") ||
item.name.includes("SPAR") ||
item.name.includes("BILLA")
);
});
knn.vectorizeItems(billaSparItem);
const billaItems = billaSparItem.filter((item) => item.store == "billa");
const sparItems = billaSparItem.filter((item) => item.store == "spar");
console.log(billaItems.length + " " + sparItems.length);
function similaritySort(items, filter, filterA, filterB) {
const filteredItems = items.filter(filter);
knn.vectorizeItems(filteredItems);
const itemsA = filteredItems.filter(filterA);
const itemsB = filteredItems.filter(filterB);
const sortedItems = [];
sparItems.forEach((item) => (item.sorted = false));
const total = billaItems.length;
while (billaItems.length > 0) {
const refItem = billaItems.shift();
const similar = knn.findMostSimilarItem(refItem, sparItems);
itemsB.forEach((item) => (item.sorted = false));
const total = itemsA.length;
while (itemsA.length > 0) {
const refItem = itemsA.shift();
const similar = knn.findMostSimilarItem(refItem, itemsB);
if (similar.item != null) {
sortedItems.push(refItem);
sortedItems.push(similar.item);
@ -37,32 +22,70 @@ if (!fs.existsSync("patterns/sorted-billa-spar.json")) {
}
if (sortedItems.length % 100 == 0) console.log(sortedItems.length / 2 + "/" + total);
}
return sortedItems;
}
function filterSimilarItems(items) {
const filteredItems = [];
for (let i = 0; i < items.length; i += 2) {
const a = items[i];
const b = items[i + 1];
if (b.priceHistory.some((price) => price.price == a.price) && a.quantity == b.quantity) {
filteredItems.push(a);
filteredItems.push(b);
}
}
return filteredItems;
}
if (!fs.existsSync("patterns")) fs.mkdirSync("patterns");
if (!fs.existsSync("patterns/sorted-billa-spar.json")) {
const items = analysis.readJSON("data/latest-canonical.json.br");
const sortedItems = similaritySort(
items,
(item) => {
if (!(item.store == "billa" || item.store == "spar")) return false;
return !["Clever", "S-BUDGET", "Ja! Natürlich", "SPAR", "BILLA"].some((str) => item.name.includes(str));
},
(item) => item.store === "billa",
(item) => item.store === "spar"
);
analysis.writeJSON("patterns/sorted-billa-spar.json", sortedItems);
}
const sortedItems = analysis.readJSON("patterns/sorted-billa-spar.json");
const filteredItems = [];
let result = "";
for (let i = 0; i < sortedItems.length; i += 2) {
const a = sortedItems[i];
const b = sortedItems[i + 1];
const similarity = knn.dotProduct(a.vector, b.vector);
result += "billa " + a.name + "\n";
result += "spar " + b.name + " " + similarity.toFixed(5) + "\n\n";
if (b.priceHistory.some((price) => price.price == a.price) && a.quantity == b.quantity) {
filteredItems.push(a);
filteredItems.push(b);
}
{
const sortedItems = analysis.readJSON("patterns/sorted-billa-spar.json");
const filteredItems = filterSimilarItems(sortedItems);
analysis.writeJSON("site/data/billa-spar-cart.json", {
name: "Markenprodukte Billa/Spar",
items: filteredItems.map((item) => {
return { store: item.store, id: item.id };
}),
});
}
if (!fs.existsSync("patterns/sorted-budget.json")) {
const items = analysis.readJSON("data/latest-canonical.json.br");
const sortedItems = similaritySort(
items,
(item) => {
if (!(item.store == "billa" || item.store == "spar")) return false;
return ["Clever", "S-BUDGET"].some((str) => item.name.includes(str));
},
(item) => item.store === "billa",
(item) => item.store === "spar"
);
analysis.writeJSON("patterns/sorted-budget.json", sortedItems);
}
{
const sortedItems = analysis.readJSON("patterns/sorted-budget.json");
const filteredItems = filterSimilarItems(sortedItems);
analysis.writeJSON("site/data/budget-cart.json", {
name: "Diskont-Marken Produkte Billa/Spar",
items: filteredItems.map((item) => {
return { store: item.store, id: item.id };
}),
});
}
console.log("Sorted: " + sortedItems.length);
analysis.writeJSON("patterns/sorted-billa-spar-cart.json", { name: "Billa Spar Sortiert", items: sortedItems });
console.log("Filtered: " + filteredItems.length);
analysis.writeJSON("patterns/sorted-billa-spar-filtered-cart.json", { name: "Billa Spar Sortiert Gefiltert", items: filteredItems });
analysis.writeJSON("site/data/billa-spar-cart.json", {
name: "Markenprodukte Billa/Spar",
items: filteredItems.map((item) => {
return { store: item.store, id: item.id };
}),
});
fs.writeFileSync("patterns/sorted-billa-spar.txt", result, "utf-8");

View File

@ -88,7 +88,7 @@ class ItemsList extends View {
elements.json.addEventListener("click", (event) => {
event.preventDefault();
if (!this.model) return;
this.download(this.model.filteredItems);
this.download(this.model.filteredItems, true);
});
elements.csv.addEventListener("click", (event) => {