// check is an non-exported function for better error handling.
func (p *Page) check() (err error) {
if settings.Verbose {
fmt.Println("[/] Downloading:", p.ReqUrl.String())
}
// Retrieve result from download or return timeout error.
var r struct {
*html.Node
error
}
select {
case r = <-errWrapDownload(p):
if r.error != nil {
return errutil.Err(r.error)
}
case <-time.After(settings.TimeoutDuration):
return errutil.NewNoPosf("timeout: %s", p.ReqUrl.String())
}
// Extract selection from downloaded source.
selection, err := p.makeSelection(r.Node)
if err != nil {
return errutil.Err(err)
}
// Filename is the URL encoded and the protocol is stripped.
linuxPath, err := filename.Encode(p.UrlAsFilename())
if err != nil {
return errutil.Err(err)
}
// Debug - no selection.
debug, err := htmlutil.RenderClean(r.Node)
if err != nil {
return errutil.Err(err)
}
// Update the debug comparison file.
debugCachePathName := settings.DebugCacheRoot + linuxPath + ".htm"
err = ioutil.WriteFile(debugCachePathName, []byte(debug), settings.Global.FilePerms)
if err != nil {
return errutil.Err(err)
}
// If the selection is empty, the CSS selection is probably wrong so we will
// alert the user about this problem.
if len(selection) == 0 {
return errutil.NewNoPosf("Update was empty. URL: %s", p.ReqUrl)
}
cachePathName := settings.CacheRoot + linuxPath + ".htm"
// Read in comparison.
buf, err := ioutil.ReadFile(cachePathName)
if err != nil {
if !os.IsNotExist(err) {
return errutil.Err(err)
}
// If the page hasn't been checked before, create a new comparison file.
err = ioutil.WriteFile(
cachePathName,
[]byte(selection),
settings.Global.FilePerms,
)
if err != nil {
return errutil.Err(err)
}
readPathName := settings.ReadRoot + linuxPath + ".htm"
// If the page hasn't been checked before, create a new comparison file.
err = ioutil.WriteFile(
readPathName,
[]byte(selection),
settings.Global.FilePerms,
)
if err != nil {
return errutil.Err(err)
}
debugReadPathName := settings.DebugReadRoot + linuxPath + ".htm"
// Update the debug prev file.
err = ioutil.WriteFile(debugReadPathName, []byte(debug), settings.Global.FilePerms)
if err != nil {
return errutil.Err(err)
}
if settings.Verbose {
fmt.Println("[+] New site added:", p.ReqUrl.String())
}
return nil
}
// The distance between to strings in percentage.
dist := distance.Approx(string(buf), selection)
// If the distance is within the threshold level, i.e if the check was a
// match.
if dist > p.Settings.Threshold {
u := p.ReqUrl.String()
settings.Updates[u] = true
if settings.Verbose {
fmt.Println("[!] Updated:", p.ReqUrl.String())
}
// If the page has a mail and all compulsory global mail settings are
// set, send a mail to notify the user about an update.
if p.Settings.RecvMail != "" &&
settings.Global.SenderMail.AuthServer != "" &&
settings.Global.SenderMail.OutServer != "" &&
settings.Global.SenderMail.Address != "" {
// Mail the selection without the stripping functions, since their
// only purpose is to remove false-positives. It will make the
// output look better.
mailPage := Page{p.ReqUrl, p.Settings}
mailPage.Settings.StripFuncs = nil
mailPage.Settings.Regexp = ""
sel, err := mailPage.makeSelection(r.Node)
if err != nil {
return errutil.Err(err)
}
err = mail.Send(p.ReqUrl, p.Settings.RecvMail, sel)
if err != nil {
return errutil.Err(err)
}
delete(settings.Updates, u)
}
// Save updates to file.
err = settings.SaveUpdates()
if err != nil {
return errutil.Err(err)
}
// Update the comparison file.
err = ioutil.WriteFile(cachePathName, []byte(selection), settings.Global.FilePerms)
if err != nil {
return errutil.Err(err)
}
} else {
if settings.Verbose {
fmt.Println("[-] No update:", p.ReqUrl.String())
}
}
return nil
}
func main() {
app := cli.NewApp()
app.Name = "stardust"
app.Usage = "String similarity measures for tab separated values."
app.Author = "Martin Czygan"
app.Email = "*****@*****.**"
app.Version = "0.1.1"
app.Flags = []cli.Flag{
cli.StringFlag{
Name: "fields, f",
Value: "1,2",
Usage: "c1,c2 the two columns to use for the comparison",
},
cli.StringFlag{
Name: "delimiter, d",
Value: "\t",
Usage: "column delimiter (defaults to tab)",
},
}
app.Commands = []cli.Command{
{
Name: "adhoc",
Usage: "Adhoc distance",
Description: "Ad-hoc percentage difference found on the web (https://godoc.org/github.com/karlek/nyfiken/distance).",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure := distance.Approx(r.Left(), r.Right())
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "cosine",
Usage: "Cosine word-wise",
Description: "A a measure of similarity between two vectors. The bigger the return value is, the more similar the two texts are.",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure := similar.Cosine(r.Left(), r.Right())
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "coslev",
Usage: "Cosine word-wise and levenshtein combined",
Description: "Experimenal.",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure := similar.Get(r.Left(), r.Right())
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "dice",
Usage: "Sørensen–Dice coefficient",
Description: "Semimetric version of the Jaccard index.",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, _ := stardust.SorensenDiceDistance(r.Left(), r.Right())
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "hamming",
Usage: "Hamming distance",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, err := stardust.HammingDistance(r.Left(), r.Right())
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "jaro",
Usage: "Jaro distance",
Description: "Similar to Ngram, but faster.",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, err := stardust.JaroDistance(r.Left(), r.Right())
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "jaro-winkler",
Usage: "Jaro-Winkler distance",
Description: "It is a variant of the Jaro distance metric.",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, err := stardust.JaroWinklerDistance(r.Left(), r.Right(), c.Float64("boost"), c.Int("size"))
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
Flags: []cli.Flag{
cli.Float64Flag{
Name: "boost, b",
Value: 0.5,
Usage: "boost factor",
},
cli.IntFlag{
Name: "size, p",
Value: 3,
Usage: "prefix size",
},
},
},
{
Name: "levenshtein",
Usage: "Levenshtein distance",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, err := stardust.LevenshteinDistance(r.Left(), r.Right())
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
},
{
Name: "ngram",
Usage: "Ngram distance",
Description: "Compute Ngram distance, which lies between 0 and 1 (equal).",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
measure, err := stardust.NgramDistanceSize(r.Left(), r.Right(), c.Int("size"))
if err != nil {
log.Fatal(err)
}
fmt.Printf("%s\t%v\n", strings.Join(r.Fields, "\t"), measure)
}
},
Flags: []cli.Flag{
cli.IntFlag{
Name: "size, s",
Value: 3,
Usage: "value of n",
},
},
},
{
Name: "plain",
Usage: "Plain passthrough (for IO benchmarks)",
Action: func(c *cli.Context) {
records := stardust.RecordGenerator(c)
for r := range records {
fmt.Printf("%s\n", strings.Join(r.Fields, "\t"))
}
},
},
}
app.Run(os.Args)
}
