me/offline-twitter
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

REFACTOR: create 'GetMore' function for paginated queries with cursors

Browse Source
This commit is contained in:
Alessio 2023-12-26 11:42:02 -06:00
parent cec8f4d124
commit f3f2550d85
4 changed files with 66 additions and 128 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

166
pkg/scraper/api_types_v2.go
View File

@ -806,6 +806,39 @@ func (r APIV2Response) ToTweetTroveAsLikes() (TweetTrove, error) {
return ret, err
}
type PaginatedQuery interface {
NextPage(api *API, cursor string) (APIV2Response, error)
}
func (api *API) GetMore(pq PaginatedQuery, response *APIV2Response, count int) error {
last_response := response
for last_response.GetCursorBottom() != "" && len(response.GetMainInstruction().Entries) < count {
fresh_response, err := pq.NextPage(api, last_response.GetCursorBottom())
if err != nil {
return fmt.Errorf("error getting next page for %#v: %w", pq, err)
}
if fresh_response.GetCursorBottom() == last_response.GetCursorBottom() && len(fresh_response.GetMainInstruction().Entries) == 0 {
// Empty response, cursor same as previous: end of feed has been reached
return END_OF_FEED
}
if fresh_response.IsEmpty() {
// Response has a pinned tweet, but no other content: end of feed has been reached
return END_OF_FEED // TODO: check that there actually is a pinned tweet and the request didn't just fail lol
}
last_response = &fresh_response
// Copy over the entries
response.GetMainInstruction().Entries = append(
response.GetMainInstruction().Entries,
last_response.GetMainInstruction().Entries...)
fmt.Printf("Have %d entries so far\n", len(response.GetMainInstruction().Entries))
}
return nil
}
// Get a User feed using the new GraphQL twitter api
func (api *API) GetGraphqlFeedFor(user_id UserID, cursor string) (APIV2Response, error) {
url, err := url.Parse(GraphqlURL{
@ -837,43 +870,12 @@ func (api *API) GetGraphqlFeedFor(user_id UserID, cursor string) (APIV2Response,
return response, err
}
/**
* Resend the request to get more tweets if necessary
*
* args:
* - user_id: the user's UserID
* - response: an "out" parameter; the APIV2Response that tweets, RTs and users will be appended to
* - min_tweets: the desired minimum amount of tweets to get
*/
func (api *API) GetMoreTweetsFromGraphqlFeed(user_id UserID, response *APIV2Response, min_tweets int) error {
// TODO user-feed-infinite-fetch: what if you reach the end of the user's timeline? Might loop
// forever getting no new tweets
last_response := response
for last_response.GetCursorBottom() != "" && len(response.GetMainInstruction().Entries) < min_tweets {
fresh_response, err := api.GetGraphqlFeedFor(user_id, last_response.GetCursorBottom())
if err != nil {
return err
type PaginatedUserFeed struct {
user_id UserID
}
if fresh_response.GetCursorBottom() == last_response.GetCursorBottom() && len(fresh_response.GetMainInstruction().Entries) == 0 {
// Empty response, cursor same as previous: end of feed has been reached
return END_OF_FEED
}
if fresh_response.IsEmpty() {
// Response has a pinned tweet, but no other content: end of feed has been reached
return END_OF_FEED // TODO: check that there actually is a pinned tweet and the request didn't just fail lol
}
last_response = &fresh_response
// Copy over the entries
response.GetMainInstruction().Entries = append(
response.GetMainInstruction().Entries,
last_response.GetMainInstruction().Entries...)
fmt.Printf("Have %d entries so far\n", len(response.GetMainInstruction().Entries))
}
return nil
func (p PaginatedUserFeed) NextPage(api *API, cursor string) (APIV2Response, error) {
return api.GetGraphqlFeedFor(p.user_id, cursor)
}
func (api *API) GetTweetDetail(tweet_id TweetID, cursor string) (APIV2Response, error) {
@ -921,40 +923,12 @@ func (api *API) GetTweetDetail(tweet_id TweetID, cursor string) (APIV2Response,
return response, err
}
func (api *API) GetMoreTweetReplies(tweet_id TweetID, response *APIV2Response, min_tweets int) error {
if len(response.Errors) != 0 {
if response.Errors[0].Message == "_Missing: No status found with that ID." {
return ErrTweetNotFound
}
panic(fmt.Sprintf("Unknown error: %s", response.Errors[0].Message))
type PaginatedTweetReplies struct {
tweet_id TweetID
}
last_response := response
for last_response.GetCursorBottom() != "" && len(response.GetMainInstruction().Entries) < min_tweets {
fresh_response, err := api.GetTweetDetail(tweet_id, last_response.GetCursorBottom())
if err != nil {
return err
}
if fresh_response.GetCursorBottom() == last_response.GetCursorBottom() && len(fresh_response.GetMainInstruction().Entries) == 0 {
// Empty response, cursor same as previous: end of feed has been reached
return END_OF_FEED
}
if fresh_response.IsEmpty() {
// Response has a pinned tweet, but no other content: end of feed has been reached
return END_OF_FEED // TODO: check that there actually is a pinned tweet and the request didn't just fail lol
}
last_response = &fresh_response
// Copy over the entries
response.GetMainInstruction().Entries = append(
response.GetMainInstruction().Entries,
last_response.GetMainInstruction().Entries...)
fmt.Printf("Have %d entries so far\n", len(response.GetMainInstruction().Entries))
}
return nil
func (p PaginatedTweetReplies) NextPage(api *API, cursor string) (APIV2Response, error) {
return api.GetTweetDetail(p.tweet_id, cursor)
}
func (api *API) GetUserLikes(user_id UserID, cursor string) (APIV2Response, error) {
@ -1003,32 +977,12 @@ func (api *API) GetUserLikes(user_id UserID, cursor string) (APIV2Response, erro
return response, nil
}
func (api *API) GetMoreUserLikes(user_id UserID, response *APIV2Response, max_results int) error {
last_response := response
for last_response.GetCursorBottom() != "" && len(response.GetMainInstruction().Entries) < max_results {
fresh_response, err := api.GetUserLikes(user_id, last_response.GetCursorBottom())
if err != nil {
return err
}
if fresh_response.GetCursorBottom() == last_response.GetCursorBottom() || len(fresh_response.GetMainInstruction().Entries) == 0 {
// Empty response, cursor same as previous: end of feed has been reached
return END_OF_FEED
type PaginatedUserLikes struct {
user_id UserID
}
last_response = &fresh_response
// Copy the results over
// Copy over the entries
response.GetMainInstruction().Entries = append(
response.GetMainInstruction().Entries,
last_response.GetMainInstruction().Entries...)
fmt.Printf("Have %d tweets\n", len(response.GetMainInstruction().Entries))
}
fmt.Println()
fmt.Printf("Cursor bottom: %q\n", last_response.GetCursorBottom())
fmt.Printf("Entries count: %d\n", len(response.GetMainInstruction().Entries))
return nil
func (p PaginatedUserLikes) NextPage(api *API, cursor string) (APIV2Response, error) {
return api.GetUserLikes(p.user_id, cursor)
}
func GetUserLikes(user_id UserID, how_many int) (TweetTrove, error) {
@ -1038,7 +992,7 @@ func GetUserLikes(user_id UserID, how_many int) (TweetTrove, error) {
}
if len(response.GetMainInstruction().Entries) < how_many && response.GetCursorBottom() != "" {
err = the_api.GetMoreUserLikes(user_id, &response, how_many)
err = the_api.GetMore(PaginatedUserLikes{user_id}, &response, how_many)
if errors.Is(err, END_OF_FEED) {
println("End of feed!")
} else if err != nil {
@ -1214,30 +1168,10 @@ func (api *API) Search(query string, cursor string) (APIV2Response, error) {
return result, err
}
func (api *API) GetMoreTweetsFromSearch(query string, response *APIV2Response, max_results int) error {
last_response := response
for last_response.GetCursorBottom() != "" && len(response.GetMainInstruction().Entries) < max_results {
fresh_response, err := api.Search(query, last_response.GetCursorBottom())
if err != nil {
return err
}
if fresh_response.GetCursorBottom() == last_response.GetCursorBottom() || len(fresh_response.GetMainInstruction().Entries) == 0 {
// Empty response, cursor same as previous: end of feed has been reached
return END_OF_FEED
type PaginatedSearch struct {
query string
}
last_response = &fresh_response
// Copy the results over
// Copy over the entries
response.GetMainInstruction().Entries = append(
response.GetMainInstruction().Entries,
last_response.GetMainInstruction().Entries...)
fmt.Printf("Have %d tweets\n", len(response.GetMainInstruction().Entries))
}
fmt.Println()
fmt.Printf("Cursor bottom: %q\n", last_response.GetCursorBottom())
fmt.Printf("Entries count: %d\n", len(response.GetMainInstruction().Entries))
return nil
func (p PaginatedSearch) NextPage(api *API, cursor string) (APIV2Response, error) {
return api.Search(p.query, cursor)
}

2
pkg/scraper/search.go
View File

@ -24,7 +24,7 @@ func Search(query string, min_results int) (trove TweetTrove, err error) {
}
if len(api_response.GetMainInstruction().Entries) < min_results && api_response.GetCursorBottom() != "" {
err = the_api.GetMoreTweetsFromSearch(query, &api_response, min_results)
err = the_api.GetMore(PaginatedSearch{query}, &api_response, min_results)
if errors.Is(err, END_OF_FEED) {
println("End of feed!")
} else if err != nil {

8
pkg/scraper/tweet.go
View File

@ -333,12 +333,16 @@ func GetTweetFullAPIV2(id TweetID, how_many int) (trove TweetTrove, err error) {
err = fmt.Errorf("Error getting tweet detail: %d\n %w", id, err)
return
}
err = the_api.GetMoreTweetReplies(id, &resp, how_many)
if errors.Is(err, ErrTweetNotFound) {
if len(resp.Errors) != 0 {
if resp.Errors[0].Message == "_Missing: No status found with that ID." {
trove := NewTweetTrove()
trove.Tweets[id] = Tweet{ID: id, TombstoneType: "deleted", IsConversationScraped: true}
return trove, nil
}
panic(fmt.Sprintf("Unknown error: %s", resp.Errors[0].Message))
}
err = the_api.GetMore(PaginatedTweetReplies{id}, &resp, how_many)
if err != nil && !errors.Is(err, END_OF_FEED) {
err = fmt.Errorf("Error getting more replies in tweet detail: %d\n %w", id, err)
return

2
pkg/scraper/user_feed.go
View File

@ -40,7 +40,7 @@ func GetUserFeedGraphqlFor(user_id UserID, min_tweets int) (trove TweetTrove, er
}
if len(api_response.GetMainInstruction().Entries) < min_tweets && api_response.GetCursorBottom() != "" {
err = the_api.GetMoreTweetsFromGraphqlFeed(user_id, &api_response, min_tweets)
err = the_api.GetMore(PaginatedUserFeed{user_id}, &api_response, min_tweets)
if err != nil && !errors.Is(err, END_OF_FEED) {
return
}