Salastil/selenium-twitter-scraper
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: 0 count for no reply, like, retweet, or analytics

Browse Source
This commit is contained in:
Jarrian
2023-09-23 10:06:39 +08:00
parent 407a717438
commit 33dc3403d3
2 changed files with 24 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
main.ipynb
View File

@@ -194,6 +194,9 @@
" self.reply_cnt = card.find_element(\n", " self.reply_cnt = card.find_element(\n",
" \"xpath\", './/div[@data-testid=\"reply\"]//span'\n", " \"xpath\", './/div[@data-testid=\"reply\"]//span'\n",
" ).text\n", " ).text\n",
" \n",
" if self.reply_cnt == \"\":\n",
" self.reply_cnt = \"0\"\n",
" except NoSuchElementException:\n", " except NoSuchElementException:\n",
" self.reply_cnt = \"0\"\n", " self.reply_cnt = \"0\"\n",
"\n", "\n",
@@ -201,6 +204,9 @@
" self.retweet_cnt = card.find_element(\n", " self.retweet_cnt = card.find_element(\n",
" \"xpath\", './/div[@data-testid=\"retweet\"]//span'\n", " \"xpath\", './/div[@data-testid=\"retweet\"]//span'\n",
" ).text\n", " ).text\n",
" \n",
" if self.retweet_cnt == \"\":\n",
" self.retweet_cnt = \"0\"\n",
" except NoSuchElementException:\n", " except NoSuchElementException:\n",
" self.retweet_cnt = \"0\"\n", " self.retweet_cnt = \"0\"\n",
"\n", "\n",
@@ -208,6 +214,9 @@
" self.like_cnt = card.find_element(\n", " self.like_cnt = card.find_element(\n",
" \"xpath\", './/div[@data-testid=\"like\"]//span'\n", " \"xpath\", './/div[@data-testid=\"like\"]//span'\n",
" ).text\n", " ).text\n",
" \n",
" if self.like_cnt == \"\":\n",
" self.like_cnt = \"0\"\n",
" except NoSuchElementException:\n", " except NoSuchElementException:\n",
" self.like_cnt = \"0\"\n", " self.like_cnt = \"0\"\n",
"\n", "\n",
@@ -215,6 +224,9 @@
" self.analytics_cnt = card.find_element(\n", " self.analytics_cnt = card.find_element(\n",
" \"xpath\", './/a[contains(@href, \"/analytics\")]//span'\n", " \"xpath\", './/a[contains(@href, \"/analytics\")]//span'\n",
" ).text\n", " ).text\n",
" \n",
" if self.analytics_cnt == \"\":\n",
" self.analytics_cnt = \"0\"\n",
" except NoSuchElementException:\n", " except NoSuchElementException:\n",
" self.analytics_cnt = \"0\"\n", " self.analytics_cnt = \"0\"\n",
"\n", "\n",

12
scraper/tweet.py
View File

@@ -53,6 +53,9 @@ class Tweet:
self.reply_cnt = card.find_element( self.reply_cnt = card.find_element(
"xpath", './/div[@data-testid="reply"]//span' "xpath", './/div[@data-testid="reply"]//span'
).text ).text
if self.reply_cnt == "":
self.reply_cnt = "0"
except NoSuchElementException: except NoSuchElementException:
self.reply_cnt = "0" self.reply_cnt = "0"
@@ -60,6 +63,9 @@ class Tweet:
self.retweet_cnt = card.find_element( self.retweet_cnt = card.find_element(
"xpath", './/div[@data-testid="retweet"]//span' "xpath", './/div[@data-testid="retweet"]//span'
).text ).text
if self.retweet_cnt == "":
self.retweet_cnt = "0"
except NoSuchElementException: except NoSuchElementException:
self.retweet_cnt = "0" self.retweet_cnt = "0"
@@ -67,6 +73,9 @@ class Tweet:
self.like_cnt = card.find_element( self.like_cnt = card.find_element(
"xpath", './/div[@data-testid="like"]//span' "xpath", './/div[@data-testid="like"]//span'
).text ).text
if self.like_cnt == "":
self.like_cnt = "0"
except NoSuchElementException: except NoSuchElementException:
self.like_cnt = "0" self.like_cnt = "0"
@@ -74,6 +83,9 @@ class Tweet:
self.analytics_cnt = card.find_element( self.analytics_cnt = card.find_element(
"xpath", './/a[contains(@href, "/analytics")]//span' "xpath", './/a[contains(@href, "/analytics")]//span'
).text ).text
if self.analytics_cnt == "":
self.analytics_cnt = "0"
except NoSuchElementException: except NoSuchElementException:
self.analytics_cnt = "0" self.analytics_cnt = "0"