Merge pull request #52 from shinspiegel/multipart_module

feat: created new multiparser

api.ts

@@ -1,7 +1,7 @@
 import { compress as brotli } from 'https://deno.land/x/[email protected]/mod.ts'
-import { FormDataReader } from 'https://deno.land/x/[email protected]/multipart.ts'
 import { gzipEncode } from 'https://deno.land/x/[email protected]/mod.ts'
 import log from './log.ts'
+import { multiParser } from './multiparser.ts'
 import { ServerRequest } from './std.ts'
 import type { APIRequest, FormDataBody } from './types.ts'
 
@@ -90,43 +90,22 @@ export class Request extends ServerRequest implements APIRequest {
     async decodeBody(type: "form-data"): Promise<FormDataBody>
     async decodeBody(type: string): Promise<any> {
         if (type === "text") {
-            try {
-                const buff: Uint8Array = await Deno.readAll(this.body);
-                const encoded = new TextDecoder("utf-8").decode(buff);
-                return encoded;
-            } catch (err) {
-                console.error("Failed to parse the request body.", err);
-            }
+            const buff: Uint8Array = await Deno.readAll(this.body);
+            const encoded = new TextDecoder("utf-8").decode(buff);
+            return encoded;
         }
 
         if (type === "json") {
-            try {
-                const buff: Uint8Array = await Deno.readAll(this.body);
-                const encoded = new TextDecoder("utf-8").decode(buff);
-                const json = JSON.parse(encoded);
-                return json;
-            } catch (err) {
-                console.error("Failed to parse the request body.", err);
-            }
+            const buff: Uint8Array = await Deno.readAll(this.body);
+            const encoded = new TextDecoder("utf-8").decode(buff);
+            const json = JSON.parse(encoded);
+            return json;
         }
 
         if (type === "form-data") {
-            try {
-                const boundary = this.headers.get("content-type");
-
-                if (!boundary) throw new Error("Failed to get the content-type")
-
-                const reader = new FormDataReader(boundary, this.body);
-                const { fields, files } = await reader.read({ maxSize: 1024 * 1024 * 10 });
-
-                return {
-                    get: (key: string) => fields[key],
-                    getFile: (key: string) => files?.find(i => i.name === key)
-                }
-
-            } catch (err) {
-                console.error("Failed to parse the request form-data", err)
-            }
+            const contentType = this.headers.get("content-type") as string
+            const form = await multiParser(this.body, contentType);
+            return form;
         }
     }
 

multiparser.ts

@@ -0,0 +1,198 @@
+import { bytes } from "./std.ts";
+import { FormDataBody, FormFile } from "./types.ts";
+
+const encoder = new TextEncoder();
+const decoder = new TextDecoder();
+
+const encode = {
+    contentType: encoder.encode("Content-Type"),
+    filename: encoder.encode("filename"),
+    name: encoder.encode("name"),
+    dashdash: encoder.encode("--"),
+    boundaryEqual: encoder.encode("boundary="),
+    returnNewline2: encoder.encode("\r\n\r\n"),
+    carriageReturn: encoder.encode("\r"),
+};
+
+export async function multiParser(
+    body: Deno.Reader,
+    contentType: string
+): Promise<FormDataBody> {
+    let buf = await Deno.readAll(body);
+    let boundaryByte = getBoundary(contentType);
+
+    if (!boundaryByte) {
+        throw new Error("No boundary data information");
+    }
+
+    // Generate an array of Uint8Array
+    const pieces = getFieldPieces(buf, boundaryByte!);
+
+    // Set all the pieces into one single object
+    const form = getForm(pieces);
+
+    return form;
+}
+
+function createFormData(): FormDataBody {
+    return {
+        fields: {},
+        files: [],
+        getFile(key: string) {
+            return this.files.find((i) => i.name === key);
+        },
+        get(key: string) {
+            return this.fields[key];
+        },
+    };
+}
+
+function getForm(pieces: Uint8Array[]) {
+    let form: FormDataBody = createFormData();
+
+    for (let piece of pieces) {
+        const { headerByte, contentByte } = splitPiece(piece);
+        const headers = getHeaders(headerByte);
+
+        // it's a string field
+        if (typeof headers === "string") {
+            // empty content, discard it
+            if (contentByte.byteLength === 1 && contentByte[0] === 13) {
+                continue;
+            }
+
+            // headers = "field1"
+            else {
+                form.fields[headers] = decoder.decode(contentByte);
+            }
+        }
+
+        // it's a file field
+        else {
+            let file: FormFile = {
+                name: headers.name,
+                filename: headers.filename,
+                contentType: headers.contentType,
+                size: contentByte.byteLength,
+                content: contentByte,
+            };
+
+            form.files.push(file);
+        }
+    }
+    return form;
+}
+
+function getHeaders(headerByte: Uint8Array) {
+    let contentTypeIndex = bytes.findIndex(headerByte, encode.contentType);
+
+    // no contentType, it may be a string field, return name only
+    if (contentTypeIndex < 0) {
+        return getNameOnly(headerByte);
+    }
+
+    // file field, return with name, filename and contentType
+    else {
+        return getHeaderNContentType(headerByte, contentTypeIndex);
+    }
+}
+
+function getHeaderNContentType(
+    headerByte: Uint8Array,
+    contentTypeIndex: number,
+) {
+    let headers: Record<string, string> = {};
+
+    let contentDispositionByte = headerByte.slice(0, contentTypeIndex - 2);
+    headers = getHeaderOnly(contentDispositionByte);
+
+    // jump over <Content-Type: >
+    let contentTypeByte = headerByte.slice(
+        contentTypeIndex + encode.contentType.byteLength + 2,
+    );
+
+    headers.contentType = decoder.decode(contentTypeByte);
+    return headers;
+}
+
+function getHeaderOnly(headerLineByte: Uint8Array) {
+    let headers: Record<string, string> = {};
+
+    let filenameIndex = bytes.findIndex(headerLineByte, encode.filename);
+    if (filenameIndex < 0) {
+        headers.name = getNameOnly(headerLineByte);
+    } else {
+        headers = getNameNFilename(headerLineByte, filenameIndex);
+    }
+    return headers;
+}
+
+function getNameNFilename(headerLineByte: Uint8Array, filenameIndex: number) {
+    // fetch filename first
+    let nameByte = headerLineByte.slice(0, filenameIndex - 2);
+    let filenameByte = headerLineByte.slice(
+        filenameIndex + encode.filename.byteLength + 2,
+        headerLineByte.byteLength - 1,
+    );
+
+    let name = getNameOnly(nameByte);
+    let filename = decoder.decode(filenameByte);
+    return { name, filename };
+}
+
+function getNameOnly(headerLineByte: Uint8Array) {
+    let nameIndex = bytes.findIndex(headerLineByte, encode.name);
+
+    // jump <name="> and get string inside double quote => "string"
+    let nameByte = headerLineByte.slice(
+        nameIndex + encode.name.byteLength + 2,
+        headerLineByte.byteLength - 1,
+    );
+
+    return decoder.decode(nameByte);
+}
+
+function splitPiece(piece: Uint8Array) {
+    const contentIndex = bytes.findIndex(piece, encode.returnNewline2);
+    const headerByte = piece.slice(0, contentIndex);
+    const contentByte = piece.slice(contentIndex + 4);
+
+    return { headerByte, contentByte };
+}
+
+function getFieldPieces(
+    buf: Uint8Array,
+    boundaryByte: Uint8Array,
+): Uint8Array[] {
+    const startBoundaryByte = bytes.concat(encode.dashdash, boundaryByte);
+    const endBoundaryByte = bytes.concat(startBoundaryByte, encode.dashdash);
+
+    const pieces = [];
+
+    while (!bytes.hasPrefix(buf, endBoundaryByte)) {
+        // jump over boundary + '\r\n'
+        buf = buf.slice(startBoundaryByte.byteLength + 2);
+        let boundaryIndex = bytes.findIndex(buf, startBoundaryByte);
+
+        // get field content piece
+        pieces.push(buf.slice(0, boundaryIndex - 1));
+        buf = buf.slice(boundaryIndex);
+    }
+
+    return pieces;
+}
+
+function getBoundary(contentType: string): Uint8Array | undefined {
+    let contentTypeByte = encoder.encode(contentType);
+    let boundaryIndex = bytes.findIndex(contentTypeByte, encode.boundaryEqual);
+
+    if (boundaryIndex >= 0) {
+        // jump over 'boundary=' to get the real boundary
+        let boundary = contentTypeByte.slice(
+            boundaryIndex + encode.boundaryEqual.byteLength,
+        );
+        return boundary;
+    } else {
+        return undefined;
+    }
+}

multiparser_test.ts

@@ -0,0 +1,34 @@
+import { assertEquals } from "https://deno.land/std/testing/asserts.ts";
+import { multiParser } from "./multiparser.ts";
+
+const encoder = new TextEncoder();
+
+const contentType = "multipart/form-data; boundary=ALEPH-BOUNDARY";
+const simpleString = '--ALEPH-BOUNDARY\rContent-Disposition: form-data; name="string_1"\r\n\r\nsimple string here\r--ALEPH-BOUNDARY--';
+const complexString = 'some text to be ignored\r\r--ALEPH-BOUNDARY\rContent-Disposition: form-data; name="id"\r\n\r\n666\r--ALEPH-BOUNDARY\rContent-Disposition: form-data; name="title"\r\n\r\nHello World\r--ALEPH-BOUNDARY\rContent-Disposition: form-data; name="multiline"\r\n\r\nworld,\n hello\r--ALEPH-BOUNDARY\rContent-Disposition: form-data; name="file1"; filename="file_name.ext"\rContent-Type: video/mp2t\r\n\r\nsome random data\r--ALEPH-BOUNDARY--\rmore text to be ignored to be ignored\r';
+
+Deno.test(`basic multiparser string`, async () => {
+    const buff = new Deno.Buffer(encoder.encode(simpleString));
+    const multiForm = await multiParser(buff, contentType);
+
+    assertEquals(multiForm.get("string_1"), "simple string here");
+});
+
+Deno.test(`complex multiparser string`, async () => {
+    const buff = new Deno.Buffer(encoder.encode(complexString));
+    const multiFrom = await multiParser(buff, contentType);
+
+    // Asseting multiple string values
+    assertEquals(multiFrom.get("id"), "666");
+    assertEquals(multiFrom.get("title"), "Hello World");
+    assertEquals(multiFrom.get("multiline"), "world,\n hello");
+
+    // Asserting a file information
+    const file = multiFrom.getFile("file1");
+
+    if (!file) { return }
+
+    assertEquals(file.name, "file1");
+    assertEquals(file.contentType, "video/mp2t");
+    assertEquals(file.size, 16);
+});

std.ts

@@ -1,3 +1,4 @@
+export * as bytes from 'https://deno.land/[email protected]/bytes/mod.ts'
 export { Untar } from 'https://deno.land/[email protected]/archive/tar.ts'
 export * as colors from 'https://deno.land/[email protected]/fmt/colors.ts'
 export { ensureDir } from 'https://deno.land/[email protected]/fs/ensure_dir.ts'

types.ts

@@ -114,8 +114,10 @@ export interface RouterURL {
  * The form data body
  */
 export interface FormDataBody {
-    get(key: string): string
-    getFile(key: string): FormFile
+    fields: Record<string, string>;
+    files: FormFile[];
+    get(key: string): string | undefined;
+    getFile(key: string): FormFile | undefined;
 }
 
 /**
@@ -126,5 +128,5 @@ export interface FormFile {
     content: Uint8Array
     contentType: string
     filename: string
-    originalName: string
-}
+    size: number
+}