@mradex77/google-play-scraper 0.1.1 → 0.2.0

package/dist/index.d.cts

@@ -252,8 +252,8 @@ declare const searchOptionsSchema: z.ZodObject<{
   term: z.ZodString;
   num: z.ZodDefault<z.ZodNumber>;
   price: z.ZodDefault<z.ZodEnum<{
-    all: "all";
     free: "free";
+    all: "all";
     paid: "paid";
   }>>;
   fullDetail: z.ZodDefault<z.ZodBoolean>;

package/dist/index.d.ts

@@ -252,8 +252,8 @@ declare const searchOptionsSchema: z.ZodObject<{
   term: z.ZodString;
   num: z.ZodDefault<z.ZodNumber>;
   price: z.ZodDefault<z.ZodEnum<{
-    all: "all";
     free: "free";
+    all: "all";
     paid: "paid";
   }>>;
   fullDetail: z.ZodDefault<z.ZodBoolean>;

package/dist/index.js

@@ -495,6 +495,36 @@ const appSchema = z.object({
 	url: z.string()
 });
 //#endregion
+//#region src/core/text.ts
+const TAB = 9;
+const LINE_FEED = 10;
+const CARRIAGE_RETURN = 13;
+const UNIT_SEPARATOR = 31;
+const DELETE = 127;
+const C1_END = 159;
+const SURROGATE_START = 55296;
+const SURROGATE_END = 57343;
+function isPreservedWhitespace(code) {
+	return code === TAB || code === LINE_FEED || code === CARRIAGE_RETURN;
+}
+function isControlCharacter(code) {
+	if (isPreservedWhitespace(code)) return false;
+	return code <= UNIT_SEPARATOR || code >= DELETE && code <= C1_END;
+}
+function isLoneSurrogate(code) {
+	return code >= SURROGATE_START && code <= SURROGATE_END;
+}
+function sanitizeText(value) {
+	if (typeof value !== "string") return;
+	let result = "";
+	for (const character of value) {
+		const code = character.codePointAt(0) ?? 0;
+		if (isControlCharacter(code) || isLoneSurrogate(code)) continue;
+		result += character;
+	}
+	return result;
+}
+//#endregion
 //#region src/features/app/transforms.ts
 const COMMENT_ROOTS = ["ds:8", "ds:9"];
 const MAX_COMMENTS = 5;
@@ -511,12 +541,11 @@ function descriptionHtmlLocalized(value) {
 		0,
 		1
 	]);
-	const resolved = typeof translated === "string" && translated.length > 0 ? translated : original;
-	return typeof resolved === "string" ? resolved : void 0;
+	return sanitizeText(typeof translated === "string" && translated.length > 0 ? translated : original);
 }
 function descriptionText(html) {
 	if (typeof html !== "string") return;
-	return cheerio.load(`<div>${html.replace(/<br>/g, "\r\n")}</div>`)("div").text();
+	return sanitizeText(cheerio.load(`<div>${html.replace(/<br>/g, "\r\n")}</div>`)("div").text());
 }
 function priceText(value) {
 	return typeof value === "string" && value.length > 0 ? value : "Free";
@@ -1295,7 +1324,8 @@ const appSpecs = {
 			1,
 			1
 		]],
-		schema: shape$6.recentChanges
+		schema: shape$6.recentChanges,
+		transform: sanitizeText
 	},
 	comments: {
 		paths: [[]],
@@ -1853,6 +1883,17 @@ function priceGoogleValue(value) {
 		default: return 0;
 	}
 }
+function matchesPriceFilter(free, filter) {
+	switch (filter) {
+		case "free": return free;
+		case "paid": return !free;
+		default: return true;
+	}
+}
+function filterByPrice(items, filter) {
+	if (filter === "all") return [...items];
+	return items.filter((item) => matchesPriceFilter(item.free, filter));
+}
 //#endregion
 //#region src/features/search/search.ts
 const searchOptionsSchema = baseOptionsSchema.extend({
@@ -1913,7 +1954,7 @@ function createSearch(getApp) {
 		});
 		const client = clientFromOptions(parsed);
 		const page = firstPage(parseScriptData(await client.request({ url: `${SEARCH_URL}?${params.toString()}` })));
-		const sliced = (await fetchClusterApps({
+		const sliced = filterByPrice(await fetchClusterApps({
 			client,
 			lang: parsed.lang,
 			country: parsed.country,
@@ -1924,7 +1965,7 @@ function createSearch(getApp) {
 			appsPath: CLUSTER_MAPPINGS$1.apps,
 			tokenPath: CLUSTER_MAPPINGS$1.token,
 			context: SEARCH_CONTEXT
-		})).slice(0, parsed.num);
+		}), parsed.price).slice(0, parsed.num);
 		if (parsed.fullDetail) return resolveFullDetail(sliced, parsed, getApp);
 		return z.array(searchResultSchema).parse(sliced);
 	};
@@ -2547,17 +2588,35 @@ const developerOptionsSchema = baseOptionsSchema.extend({
 	fullDetail: z.boolean().default(false)
 });
 const DEVELOPER_CONTEXT = "developer";
-function extractInitial(blocks, numeric) {
-	const mappings = numeric ? NUMERIC_INITIAL_MAPPINGS : NAME_INITIAL_MAPPINGS;
-	const itemSpecs = numeric ? numericItemSpecs : nameItemSpecs;
-	const appsData = getPath(blocks, mappings.apps);
-	const apps = Array.isArray(appsData) ? appsData.map((item) => extract(item, itemSpecs, DEVELOPER_CONTEXT)) : [];
-	const token = getPath(blocks, mappings.token);
+const NUMERIC_LAYOUT = {
+	mappings: NUMERIC_INITIAL_MAPPINGS,
+	itemSpecs: numericItemSpecs
+};
+const NAME_LAYOUT = {
+	mappings: NAME_INITIAL_MAPPINGS,
+	itemSpecs: nameItemSpecs
+};
+function extractLayout(blocks, layout) {
+	const appsData = getPath(blocks, layout.mappings.apps);
+	if (!Array.isArray(appsData) || appsData.length === 0) return;
+	const apps = appsData.map((item) => extract(item, layout.itemSpecs, DEVELOPER_CONTEXT));
+	const token = getPath(blocks, layout.mappings.token);
 	return {
 		apps,
 		token: typeof token === "string" ? token : void 0
 	};
 }
+function extractInitial(blocks, numeric) {
+	const ordered = numeric ? [NUMERIC_LAYOUT, NAME_LAYOUT] : [NAME_LAYOUT, NUMERIC_LAYOUT];
+	for (const layout of ordered) {
+		const extracted = extractLayout(blocks, layout);
+		if (extracted !== void 0) return extracted;
+	}
+	return {
+		apps: [],
+		token: void 0
+	};
+}
 function createDeveloper(getApp) {
 	return async function developer(options) {
 		const parsed = parseOptions(developerOptionsSchema, options, DEVELOPER_CONTEXT);
@@ -2841,6 +2900,9 @@ function alwaysNull() {
 function emptyToUndefined(value) {
 	return typeof value === "string" && value.length > 0 ? value : void 0;
 }
+function cleanReplyText(value) {
+	return emptyToUndefined(sanitizeText(value));
+}
 function buildCriteria(entry) {
 	if (!Array.isArray(entry)) return {
 		criteria: void 0,
@@ -2892,7 +2954,8 @@ const reviewItemSpecs = {
 	},
 	text: {
 		paths: [[4]],
-		schema: shape.text
+		schema: shape.text,
+		transform: sanitizeText
 	},
 	replyDate: {
 		paths: [[7, 2]],
@@ -2902,7 +2965,7 @@ const reviewItemSpecs = {
 	replyText: {
 		paths: [[7, 1]],
 		schema: shape.replyText,
-		transform: emptyToUndefined
+		transform: cleanReplyText
 	},
 	version: {
 		paths: [[10]],