jamra/gocleo

A golang implementation of the Cleo search

gocleo's Issues

You shouldn't run a standalone webserver

After having played a while with your lib, it turns out that I like it but I think running a standalone web server is not a good idea. Because the most common use case is using your lib as part of a web applicaion. So you have to launch two web servers and patch your lib in order to allow cross domain AJAX requests (see issue #3), e.g. :

func main() {
  go cleo.InitAndRun("liste.txt", "7777", nil)  //go routine needed here to allow main to carry on
  http.HandleFunc("/home", homeHandler) //Without patching, my AJAX requests fail
  err := http.ListenAndServe(":9999", nil)
  if err != nil {
    panic(err)
  }
}

I think that you shouldn't start a standalone web server but offer additionnal HTTP handlers in the manner of expvar package (it adds its handlers by side effect) :

import _ "expvar"

The difference will be that the developper will have the responsability to run its own web server. Once started, the application importing gocleo will have gocleo's handlers.

In the lib, you should add an init function (called when gocleo package is imported) so as to register your handler(s):

func init() {
    http.HandleFunc("/cleo", searchHandler)
}

You should limit the dependencies of your lib (I love Gorilla toolkit but it's a bit underexploited here). And handle parameters (whether they are coming from a POST or a GET request) the classic way :

func Search(w http.ResponseWriter, r *http.Request) {
        query := r.FormValue("query")
        ...

Instead of :

func Search(w http.ResponseWriter, r *http.Request) {
        vars := mux.Vars(r)
        query := vars["query"]
        ...

Then, my application becomes :

import (
 "net/http"
 "github.com/jamra/gocleo"    //No side effect by import, because i still need to init the indexes
)

func main() {
  cleo.BuildIndexes("liste.txt", nil)   //No need to create a goroutine now (I renamed your func)
  http.HandleFunc("/home", homeHandler) //my AJAX requests work now
  err := http.ListenAndServe(":9999", nil)
  if err != nil {
    panic(err)
  }
}

I can now use the following URLs (with only one webserver) :
http://localhost:9999/home
http://localhost:9999/cleo?query=wonderful

By doing so, you don't need to solve the issue #3 anymore because the two web pages belong now to the same domain.

Levenshtein distance typo

Disregard

Is it supposed to work ?

Hi,

Is it supposed to work ? Or is it still under development ? Because all my tests fail.

Remove debug messages

Your code is fast but you spoil it by printing debug messages to the console. Please remove these two lines (they are good for you when you debug, but useless in production):

fmt.Println("Query:", query)
...
fmt.Printf("The call took %v to run.\n", t1.Sub(t0))

[git] Ignore "examples" executable

Please add a rule to ignore the "examples" executable file (see : https://help.github.com/articles/ignoring-files) which is into "examples" folder.

Allow cross domain AJAX Requests

If you want to distribute your lib as a standalone web server (A), then you have to take into account that other developpers using it might have their own web server (B) in their go app. Please patch you lib to allow web pages served by B to request server A.

func Search(w http.ResponseWriter, r *http.Request) {
    vars := mux.Vars(r)
    query := vars["query"]

    searchResult := CleoSearch(m.iIndex, m.fIndex, query)
    sort.Sort(ByScore{searchResult})
    myJson, _ := json.Marshal(searchResult)

    // allow cross domain AJAX requests
        w.Header().Set("Access-Control-Allow-Origin", "*")
    fmt.Fprintf(w, string(myJson))
}

Improve error handling

At key moments of the code, you should panic or log.Fatal, for example:

file, _ := os.Open(corpusPath)

You shouldn't ignore the error because the entire lib will not work from this point.