How to modify a slice within for loop?

Question

I've got a slice of articles in my reading list. Each Article has the attribute "FeedURL" that has the URL of the feed the article came from. When I unsubscribe from a feed, I want to be able to remove every Article that contains that Feed's URL.

type Article struct {
    FeedURL string
    URL     string // should be unique
    // ... more data
}

func unsubscribe(articleList []Article, url string) []Article {
   // how do I remove every Article from articleList that contains url?
}

func main() {
    myArticleList := []Article{
        Article{"http://blog.golang.org/feed.atom", "http://blog.golang.org/race-detector"},
        Article{"http://planet.python.org/rss20.xml", "http://archlinux.me/dusty/2013/06/29/creating-an-application-in-kivy-part-3/"},
        Article{"http://planet.python.org/rss20.xml", "http://feedproxy.google.com/~r/cubicweborg/~3/BncbP-ap0n0/2957378"},
        // ... much more examples
    }

    myArticleList = unsubscribe(myArticleList, "http://planet.python.org/rss20.xml")

    fmt.Printf("%+v", myArticleList)
}

What is the efficient way of solving this problem?

At first my code looked like this for unsubscribe:

func unsubscribe(articleList []Article, url string) []Article {
    for _, article := range articleList {
        if article.FeedURL == url {
            articleList = append(articleList[:i], articleList[i+1:]...)
        }
    }
    return articleList
}

But then I realized that this would change the slice and make the for loop unpredictable.

What is an efficient and pretty way to accomplish this?

Answer 1

To be efficient:

• Use a slice of pointers to Articles, then we will be moving pointers to structures instead of structure values.
• If the order of the Articles in the list is not important, use the unordered algorithm; it reduces pointer movement. Otherwise, use the ordered algorithm. In any case, minimize pointer movement.
• Don't leave dangling pointers at the end of the list. The garbage collector will think they are still in use; it looks at the slice capacity not the slice length.
• Minimize memory allocations.

For example,

package main

import "fmt"

type Article struct {
    FeedURL string
    URL     string // should be unique
    // ... more data
}

// Remove every Article from an articleList that contains url without preserving order.
func unsubscribeUnordered(a []*Article, url string) []*Article {
    for i := 0; i < len(a); i++ {
        if a[i].FeedURL == url {
            a[len(a)-1], a[i], a = nil, a[len(a)-1], a[:len(a)-1]
            i--
        }
    }
    return a
}

// Remove every Article from an articleList that contains url while preserving order.
func unsubscribeOrdered(a []*Article, url string) []*Article {
    j := 0
    for i := 0; i < len(a); i++ {
        if a[i].FeedURL == url {
            continue
        }
        if i != j {
            a[j] = a[i]
        }
        j++
    }
    for k := j; k < len(a); k++ {
        a[k] = nil
    }
    return a[:j]
}

func NewArticleList() []*Article {
    return []*Article{
        &Article{"http://blog.golang.org/feed.atom", "http://blog.golang.org/race-detector"},
        &Article{"http://planet.python.org/rss20.xml", "http://archlinux.me/dusty/2013/06/29/creating-an-application-in-kivy-part-3/"},
        &Article{"http://planet.python.org/rss20.xml", "http://feedproxy.google.com/~r/cubicweborg/~3/BncbP-ap0n0/2957378"},
        // ... much more examples
    }
}

func PrintArticleList(a []*Article) {
    fmt.Print("[")
    for _, e := range a {
        fmt.Printf("%+v", *e)
    }
    fmt.Println("]")
}

func main() {
    PrintArticleList(NewArticleList())
    ao := unsubscribeOrdered(NewArticleList(), "http://planet.python.org/rss20.xml")
    PrintArticleList(ao)
    auo := unsubscribeUnordered(NewArticleList(), "http://planet.python.org/rss20.xml")
    PrintArticleList(auo)
}

Output:

[{FeedURL:http://blog.golang.org/feed.atom URL:http://blog.golang.org/race-detector}{FeedURL:http://planet.python.org/rss20.xml URL:http://archlinux.me/dusty/2013/06/29/creating-an-application-in-kivy-part-3/}{FeedURL:http://planet.python.org/rss20.xml URL:http://feedproxy.google.com/~r/cubicweborg/~3/BncbP-ap0n0/2957378}]

[{FeedURL:http://blog.golang.org/feed.atom URL:http://blog.golang.org/race-detector}]
[{FeedURL:http://blog.golang.org/feed.atom URL:http://blog.golang.org/race-detector}]

Answer 2

PeterSO's answer is gets the job done, and with efficiency.

But, I might go with something simple like this

func unsubscribe(articleList []Article, url string) (filtered []Article) {
    filtered = articleList[:0] // optional.  reuses already-allocated memory.
    for _, article := range articleList {
        if article.FeedURL != url {
            filtered = append(filtered, article)
        }
    }
    return
}

which only takes about a two seconds to read, and comprehend.

The idea works fine with pointers to articles too and, like PeterSO said, if your Article struct is big, that may be a good thing to do.