Add search

cmd/tests.sh

@@ -181,6 +181,10 @@ test $(sqlite3 twitter.db "select is_stub from tweets where id = 145452142414465
 test $(sqlite3 twitter.db "select is_stub from tweets where id = 1454522147750260742") = 1
 
 
+# Test search
+tw search "from:michaelmalice constitution"
+test $(sqlite3 twitter.db "select count(*) from tweets where user_id = 44067298 and text like '%constitution%'") -gt "30"  # Not sure exactly how many
+
 # TODO: Maybe this file should be broken up into multiple test scripts
 
 echo -e "\033[32mAll tests passed.  Finished successfully.\033[0m"

cmd/twitter/helpers.go

@@ -46,6 +46,9 @@ This application downloads tweets from twitter and saves them in a SQLite databa
           <TARGET> is the user handle.
           Gets the most recent ~50 tweets.
           If "get_user_tweets_all" is used, gets up to ~3200 tweets (API limit).
+
+    search
+          <TARGET> is the search query.  Should be wrapped in quotes if it has spaces.
 `
 
 

cmd/twitter/main.go

@@ -61,6 +61,8 @@ func main() {
 		fetch_user_feed(target, 999999999)
 	case "download_tweet_content":
 		download_tweet_content(target)
+	case "search":
+		search(target)
 	default:
 		die("Invalid operation: " + operation, true, 3)
 	}
@@ -238,7 +240,7 @@ func download_tweet_content(tweet_identifier string) {
 
 	tweet, err := profile.GetTweetById(tweet_id)
 	if err != nil {
-		panic("Couldn't get tweet from database: " + err.Error())
+		panic(fmt.Sprintf("Couldn't get tweet (ID %d) from database: %s", tweet_id, err.Error()))
 	}
 	err = profile.DownloadTweetContentFor(&tweet)
 	if err != nil {
@@ -256,3 +258,36 @@ func download_user_content(handle scraper.UserHandle) {
 		panic("Error getting content: " + err.Error())
 	}
 }
+
+
+func search(query string) {
+	tweets, retweets, users, err := scraper.Search(query, 1000);
+	if err != nil {
+		die("Error scraping search results: " + err.Error(), false, -100)
+	}
+
+	for _, u := range users {
+		err = profile.SaveUser(u)
+		if err != nil {
+			die("Error saving user: " + err.Error(), false, 4)
+		}
+		err = profile.DownloadUserContentFor(&u)
+		if err != nil {
+			die("Error getting user content: " + err.Error(), false, 10)
+		}
+	}
+
+	for _, t := range tweets {
+		// fmt.Println(t)
+		err = profile.SaveTweet(t)
+		if err != nil {
+			die("Error saving tweet: " + err.Error(), false, 4)
+		}
+		err = profile.DownloadTweetContentFor(&t)
+		if err != nil {
+			die("Error getting tweet content: " + err.Error(), false, 11)
+		}
+	}
+
+	fmt.Printf("Saved %d tweets, %d retweets and %d users.  Exiting successfully\n", len(tweets), len(retweets), len(users))
+}

doc/curl requests

@@ -26,3 +26,9 @@ curl \
     -H "Authorization: Bearer AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs%3D1Zv7ttfk8LF81IUq16cHjhLTvJu4FA33AGWWjCpTnA" \
     -H "X-Guest-Token: 1396177150890348547" \
     https://twitter.com/i/api/2/timeline/conversation/1395881699142160387.json
+
+
+curl \
+    -H "Authorization: Bearer AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs%3D1Zv7ttfk8LF81IUq16cHjhLTvJu4FA33AGWWjCpTnA" \
+    -H "X-Guest-Token: 1449946080792104970" \
+    "https://twitter.com/i/api/2/search/adaptive.json?count=50&spelling_corrections=1&query_source=typed_query&pc=1&q=potatoes"

scraper/api_request_utils.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"io/ioutil"
 	"net/http"
+	"net/url"
 	"time"
 )
 
@@ -217,6 +218,75 @@ func (api API) GetUser(handle UserHandle) (APIUser, error) {
     return response.ConvertToAPIUser(), err
 }
 
+func (api API) Search(query string, cursor string) (TweetResponse, error) {
+	client := &http.Client{Timeout: 10 * time.Second}
+	req, err := http.NewRequest("GET", "https://twitter.com/i/api/2/search/adaptive.json?count=50&spelling_corrections=1&query_source=typed_query&pc=1&q=" + url.QueryEscape(query), nil)
+	if err != nil {
+		return TweetResponse{}, err
+	}
+
+	err = ApiRequestAddTokens(req)
+	if err != nil {
+		return TweetResponse{}, err
+	}
+
+	ApiRequestAddAllParams(req)
+	if cursor != "" {
+		UpdateQueryCursor(req, cursor, false)
+	}
+
+	fmt.Println(req.URL.String())
+
+	resp, err := client.Do(req)
+	if err != nil {
+		return TweetResponse{}, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		content, _ := ioutil.ReadAll(resp.Body)
+		return TweetResponse{}, fmt.Errorf("Error while searching for %q.  HTTP %s: %s", req.URL, resp.Status, content)
+	}
+
+	body, err := ioutil.ReadAll(resp.Body)
+	if err != nil {
+		return TweetResponse{}, err
+	}
+	// fmt.Println(string(body))
+
+	var response TweetResponse
+	err = json.Unmarshal(body, &response)
+	return response, err
+}
+
+func (api API) GetMoreTweetsFromSearch(query string, response *TweetResponse, max_results int) error {
+	last_response := response
+	for last_response.GetCursor() != "" && len(response.GlobalObjects.Tweets) < max_results {
+		fresh_response, err := api.Search(query, last_response.GetCursor())
+		if err != nil {
+			return err
+		}
+		if fresh_response.GetCursor() == last_response.GetCursor() && len(fresh_response.GlobalObjects.Tweets) == 0 {
+			// Empty response, cursor same as previous: end of feed has been reached
+			return END_OF_FEED
+		}
+
+		last_response = &fresh_response
+
+		// Copy the results over
+		for id, tweet := range last_response.GlobalObjects.Tweets {
+			response.GlobalObjects.Tweets[id] = tweet
+		}
+		for id, user := range last_response.GlobalObjects.Users {
+			response.GlobalObjects.Users[id] = user
+		}
+		fmt.Printf("Have %d tweets\n", len(response.GlobalObjects.Tweets))
+		// fmt.Printf("Cursor: %s\n", last_response.GetCursor())
+	}
+	return nil
+}
+
+
 // Add Bearer token and guest token
 func ApiRequestAddTokens(req *http.Request) error {
 	req.Header.Set("Authorization", "Bearer " + BEARER_TOKEN)

scraper/api_types.go

@@ -308,9 +308,18 @@ func (t *TweetResponse) HandleTombstones() []string {
 
 func (t *TweetResponse) GetCursor() string {
 	entries := t.Timeline.Instructions[0].AddEntries.Entries
-	last_entry := entries[len(entries) - 1]
-	if strings.Contains(last_entry.EntryID, "cursor") {
-		return last_entry.Content.Operation.Cursor.Value
+	if len(entries) > 0 {
+		last_entry := entries[len(entries) - 1]
+		if strings.Contains(last_entry.EntryID, "cursor") {
+			return last_entry.Content.Operation.Cursor.Value
+		}
+	}
+
+	// Next, try the other format ("replaceEntry")
+	instructions := t.Timeline.Instructions
+	last_replace_entry := instructions[len(instructions) - 1].ReplaceEntry.Entry
+	if strings.Contains(last_replace_entry.EntryID, "cursor") {
+		return last_replace_entry.Content.Operation.Cursor.Value
 	}
 	return ""
 }