fix: distinguish tweets with post id

DimensionDev · Nov 27, 2019 · 863ebe9 · 863ebe9
1 parent b36fd65
commit 863ebe9
Show file tree

Hide file tree

Showing 2 changed files with 63 additions and 48 deletions.
diff --git a/src/social-network-provider/twitter.com/ui/fetch.ts b/src/social-network-provider/twitter.com/ui/fetch.ts
@@ -8,7 +8,7 @@ import {
 } from '../../../social-network/ui'
 import { deconstructPayload } from '../../../utils/type-transform/Payload'
 import { instanceOfTwitterUI } from './index'
-import { bioCardParser, postParser, postImageParser } from '../utils/fetch'
+import { bioCardParser, postParser, postImageParser, postIdParser } from '../utils/fetch'
 import { isNil } from 'lodash-es'
 import Services from '../../../extension/service'
 import { twitterUrl } from '../utils/url'
@@ -86,15 +86,19 @@ const registerUserCollector = () => {
 }
 
 const registerPostCollector = (self: SocialNetworkUI) => {
+    const getTweetNode = (node: HTMLElement) => {
+        return node.closest<HTMLDivElement>(
+            [
+                '.tweet', // timeline page for legacy twitter
+                '.main-tweet', // detail page for legacy twitter
+                'article > div', // new twitter
+            ].join(),
+        )
+    }
+
     new MutationObserverWatcher(postsContentSelector())
         .useForeach((node, _, proxy) => {
-            const tweetNode = node.closest<HTMLDivElement>(
-                [
-                    '.tweet', // timeline page for legacy twitter
-                    '.main-tweet', // detail page for legacy twitter
-                    'article > div', // new twitter
-                ].join(),
-            )
+            const tweetNode = getTweetNode(node)
             if (!tweetNode) return
             // noinspection JSUnnecessarySemicolon
             const info = getEmptyPostInfoByElement({
@@ -142,7 +146,10 @@ const registerPostCollector = (self: SocialNetworkUI) => {
             }
         })
         .setDOMProxyOption({ afterShadowRootInit: { mode: 'closed' } })
-        .assignKeys(node => node.innerHTML)
+        .assignKeys(node => {
+            const tweetNode = getTweetNode(node)
+            return tweetNode ? `${postIdParser(tweetNode)}${node.innerHTML}` : node.innerHTML
+        })
         .startWatch({
             childList: true,
             subtree: true,

diff --git a/src/social-network-provider/twitter.com/utils/fetch.ts b/src/social-network-provider/twitter.com/utils/fetch.ts
@@ -22,10 +22,14 @@ const parseNameArea = (t: string) => {
     }
 }
 
-const parsePid = (t: string) => {
+const parseId = (t: string) => {
     return regexMatch(t, /status\/(\d+)/, 1)!
 }
 
+const isMobilePost = (node: HTMLElement) => {
+    return node.classList.contains('tweet') ?? node.classList.contains('main-tweet')
+}
+
 export const bioCardParser = (cardNode: HTMLDivElement) => {
     if (cardNode.classList.contains('profile')) {
         const avatarElement = cardNode.querySelector<HTMLImageElement>('.avatar img')
@@ -67,8 +71,41 @@ export const bioCardParser = (cardNode: HTMLDivElement) => {
     }
 }
 
+export const postIdParser = (node: HTMLElement) => {
+    if (isMobilePost(node)) {
+        const idNode = node.querySelector<HTMLAnchorElement>('.tweet-text')
+        return idNode ? idNode.getAttribute('data-id') ?? undefined : undefined
+    } else {
+        const idNode = defaultTo(
+            node.children[1].querySelector<HTMLAnchorElement>('a[href*="status"]'),
+            node.parentElement!.querySelector<HTMLAnchorElement>('a[href*="status"]'),
+        )
+        return idNode ? parseId(idNode.href) : undefined
+    }
+}
+
+export const postNameParser = (node: HTMLElement) => {
+    if (isMobilePost(node)) {
+        return parseNameArea(notNullable(node.querySelector<HTMLTableCellElement>('.user-info')).innerText)
+    } else {
+        const tweetElement = node.querySelector('[data-testid="tweet"]') ?? node
+        return parseNameArea(notNullable(tweetElement.children[1].querySelector<HTMLAnchorElement>('a')).innerText)
+    }
+}
+
+export const postAvatarParser = (node: HTMLElement) => {
+    if (isMobilePost(node)) {
+        const avatarElement = node.querySelector<HTMLImageElement>('.avatar img')
+        return avatarElement ? avatarElement.src : undefined
+    } else {
+        const tweetElement = node.querySelector('[data-testid="tweet"]') ?? node
+        const avatarElement = tweetElement.children[0].querySelector<HTMLImageElement>(`img[src*="twimg.com"]`)
+        return avatarElement ? avatarElement.src : undefined
+    }
+}
+
 export const postContentParser = (node: HTMLElement) => {
-    if (node.classList.contains('tweet') || node.classList.contains('main-tweet')) {
+    if (isMobilePost(node)) {
         const containerNode = node.querySelector('.tweet-text > div')
         if (!containerNode) {
             return ''
@@ -96,18 +133,15 @@ export const postContentParser = (node: HTMLElement) => {
 }
 
 export const postImageParser = async (node: HTMLElement) => {
-    if (node.classList.contains('tweet') || node.classList.contains('main-tweet')) {
+    if (isMobilePost(node)) {
         // TODO: Support steganography in legacy twitter
         return ''
     } else {
         const imgNodes = node.querySelectorAll<HTMLImageElement>('img[src*="twimg.com/media"]')
         if (!imgNodes.length) return ''
-        const imgUrls = Array.from(imgNodes).map(node => node.getAttribute('src') || '')
+        const imgUrls = Array.from(imgNodes).map(node => node.getAttribute('src') ?? '')
         if (!imgUrls.length) return ''
-        const tweetElement = node.querySelector('[data-testid="tweet"]') || node
-        const { handle } = parseNameArea(
-            notNullable(tweetElement.children[1].querySelector<HTMLAnchorElement>('a')).innerText,
-        )
+        const { handle } = postNameParser(node)
         const posterIdentity = new PersonIdentifier(twitterUrl.hostIdentifier, handle)
         return (
             await Promise.all(
@@ -126,36 +160,10 @@ export const postImageParser = async (node: HTMLElement) => {
 }
 
 export const postParser = (node: HTMLElement) => {
-    if (node.classList.contains('tweet') || node.classList.contains('main-tweet')) {
-        const { name, handle } = parseNameArea(
-            notNullable(node.querySelector<HTMLTableCellElement>('.user-info')).innerText,
-        )
-        const avatarElement = node.querySelector<HTMLImageElement>('.avatar img')
-        const pidLocation = node.querySelector<HTMLAnchorElement>('.tweet-text')
-        return {
-            name,
-            handle,
-            pid: pidLocation ? pidLocation.getAttribute('data-id') : undefined,
-            avatar: avatarElement ? avatarElement.src : undefined,
-            content: postContentParser(node),
-        }
-    } else {
-        const tweetElement = node.querySelector('[data-testid="tweet"]') || node
-        const { name, handle } = parseNameArea(
-            notNullable(tweetElement.children[1].querySelector<HTMLAnchorElement>('a')).innerText,
-        )
-        const avatarElement = tweetElement.children[0].querySelector<HTMLImageElement>(`img[src*="twimg.com"]`)
-        const pidLocation = defaultTo(
-            node.children[1].querySelector<HTMLAnchorElement>('a[href*="status"]'),
-            node.parentElement!.querySelector<HTMLAnchorElement>('a[href*="status"]'),
-        )
-        return {
-            name,
-            handle,
-            // pid may not available at promoted tweet
-            pid: pidLocation ? parsePid(pidLocation.href) : undefined,
-            avatar: avatarElement ? avatarElement.src : undefined,
-            content: postContentParser(node),
-        }
+    return {
+        ...postNameParser(node),
+        avatar: postAvatarParser(node),
+        pid: postIdParser(node),
+        content: postContentParser(node),
     }
 }