// load expressions, returning list and augmented parse tree list
func load(filename string) []*RegEx {
elist := make([]*RegEx, 0) // expression structure list
rx.LoadExpressions(filename, func(l *rx.RegExParsed) {
if l.Tree != nil { // if a real expression
e := &RegEx{}
e.index = len(elist)
e.cnum = len(elist)
e.RegExParsed = *l // save parse tree
e.examples = genex(l.Tree) // gen examples unaugmented
t := rx.Augment(l.Tree, e.index)
e.dfa = rx.BuildDFA(t) // build DFA
if verbose {
fmt.Println()
showexpr(e) // show details of expr
showexamples(e) // show its examples
}
elist = append(elist, e)
}
})
if verbose {
fmt.Printf("\nloaded %d expression(s), ignored %d more\n",
rx.InputRegExCount, rx.InputErrorCount)
}
return elist
}
// details responds to an inspection request for a single expression
func details(w http.ResponseWriter, r *http.Request) {
field := fmt.Sprintf("v%d", baseExpr)
expr := r.FormValue(field) // must read before any writing
expr = strings.TrimSpace(expr) // trim leading/trailing blanks
putheader(w, r, "Inspect Expression")
tree, err := rx.Parse(expr)
if err == nil {
fmt.Fprintf(w, "<P>Regular Expression: %s\n", hx(expr))
// must print (or at least stringize) tree before augmenting
fmt.Fprintf(w, "<P>Initial Parse Tree: %s\n", hx(tree))
augt := rx.Augment(tree, 0)
dfa := rx.BuildDFA(augt)
dmin := dfa.Minimize()
fmt.Fprintf(w, "<P>Augmented Tree: %s\n", hx(augt))
fmt.Fprintf(w, "<h2>Examples</h2>\n<P>")
genexamples(w, tree, 0)
genexamples(w, tree, 1)
genexamples(w, tree, 2)
genexamples(w, tree, 3)
genexamples(w, tree, 5)
genexamples(w, tree, 8)
showaut(w, dmin, []string{expr})
} else {
fmt.Fprint(w, "<P>")
showerror(w, err)
}
fmt.Fprint(w, "<h2>Try another?</h2>")
putform(w, "/details", "Enter a regular expression:",
1, []string{expr}, 0, nil)
putfooter(w, r)
}
// individual handles separate processing of each input regex
func individual(l *rx.RegExParsed, cx int, i int, image string, augt rx.Node) {
babble("tree: %s\n", image)
babble("augmnt: %v\n", augt)
x := augt.MaxLen()
if x >= 0 {
babble("length: %d to %d\n", augt.MinLen(), x)
} else {
babble("length: %d to *\n", augt.MinLen())
}
babble("cplxty: %d\n", cx)
dfa := rx.BuildDFA(augt)
if *opt['R'] {
rand.Seed(int64(seedvalue))
}
if *opt['g'] {
rx.ShowLabel(os.Stdout, "Examples")
examples(dfa, l.Tree, 0) // gen and test w/ max repl of 0
examples(dfa, l.Tree, 1) // ... and 1
examples(dfa, l.Tree, 2) // ... and 2
examples(dfa, l.Tree, 3) // ... and 3
examples(dfa, l.Tree, 5) // ... and 5
examples(dfa, l.Tree, 8) // ... and 8
}
showDFA(dfa, "Annotated Tree", false)
}
// main control
func main() {
filename := cmdline() // process command line
maxEx := *numElem * 4
exprs, trees := load(filename) // load expressions
qns := make([]string, 0, maxEx) //create list of question words
answers := new(rx.BitSet) //create list of answers
dfaList := make([]*rx.DFA, 0, len(trees)) //list of candidate DFAs
ind := make([]int, 0, len(dfaList)) //index array to track id of each regex
a := 0
init := 0
if *mode == 1 { //run with examples
//ask for examples
fmt.Println("Enter some examples: ")
//read examples
r := bufio.NewReader(os.Stdin)
line, err := r.ReadString(delim)
if err != nil {
fmt.Println(err)
os.Exit(1)
}
for line != string('\n') {
line = line[:len(line)-1]
qns = append(qns, line)
answers.Set(a)
a++
line, err = r.ReadString(delim)
if err != nil {
fmt.Println(err)
os.Exit(1)
}
}
//ask for counter examples
fmt.Println("\nEnter some counter-examples: ")
rr := bufio.NewReader(os.Stdin)
//read counter examples
line, err = rr.ReadString(delim)
if err != nil {
fmt.Println(err)
os.Exit(1)
}
for line != string('\n') {
line = line[:len(line)-1]
qns = append(qns, line)
line, err = rr.ReadString(delim)
if err != nil {
fmt.Println(err)
os.Exit(1)
}
}
fmt.Printf("Processing ...\n")
//build the DFAs of all candidates
for i := 0; i < len(trees); i++ {
aDfa := rx.BuildDFA(trees[i])
dfaList = append(dfaList, aDfa)
//fmt.Printf("rx { %d } %s\n", i, exprs[i].Expr)
}
fmt.Printf("\n")
//refine the candidate set based on examples and counter examples
dfaList, exprs, ind = refine(answers, qns, dfaList, exprs, ind)
init++
fmt.Printf("Length of dfaList: %d\n", len(dfaList))
for i := 0; i < len(ind); i++ {
fmt.Printf("rx { %d } %s\n", ind[i], exprs[i].Expr)
}
} else if *mode == 2 { //partition
numPerGroup := *numElem
//build DFAs of all candidates
for i := 0; i < len(trees); i++ {
aDfa := rx.BuildDFA(trees[i])
dfaList = append(dfaList, aDfa)
//fmt.Printf("rx { %d } %s\n", i, exprs[i].Expr)
}
times := 0
for len(dfaList) > 50 {
permArray := rand.Perm(len(dfaList)) //randomly permute candidate set
qns = make([]string, 0, maxEx) //initialize new set of questions
answers = new(rx.BitSet) //initialize new set of answers
//ask best question in each partition. Return questions and answers
answers, qns = askBulk(answers, qns, permArray, dfaList, exprs, numPerGroup)
fmt.Println("Here length of dfaList =", len(dfaList))
//refine the candidate set based on questions ans answers
dfaList, exprs, ind = refine(answers, qns, dfaList, exprs, ind)
init++
fmt.Printf("Length of dfaList: %d\n", len(dfaList))
times++
}
} else if *mode == 3 { //ask all DFAs
numPerGroup := *numElem
//build DFA's of all candidates
for i := 0; i < len(trees); i++ {
aDfa := rx.BuildDFA(trees[i])
dfaList = append(dfaList, aDfa)
fmt.Printf("rx { %d } %s\n", i, exprs[i].Expr)
}
times := 0
for len(dfaList) > 50 {
//fmt.Println("iteration = ", times)
//default partition is 20 per group
permArray := rand.Perm(len(dfaList)) //randomly permute candidate set
qns = make([]string, 0, maxEx) //initialize new set of questions
answers = new(rx.BitSet) //initialize new set of answers
//ask best question based on number of DFAs that accept question word
answers, qns = askBulk2(answers, qns, permArray, dfaList, exprs, numPerGroup)
fmt.Println("Here length of dfaList =", len(dfaList))
//fmt.Printf("\n")
//refine candidate set based on questions and answers
dfaList, exprs, ind = refine(answers, qns, dfaList, exprs, ind)
init++
fmt.Printf("Length of dfaList: %d\n", len(dfaList))
times++
}
} else { //mode = 0, run basic algorithm
//build DFA's of all candidates
for i := 0; i < len(trees); i++ {
aDfa := rx.BuildDFA(trees[i])
dfaList = append(dfaList, aDfa)
//fmt.Printf("rx { %d } %s\n", i, exprs[i].Expr)
}
}
param := 1
value := *numElem
tempLen := 0
done := false
numPerGroup := 5
for len(dfaList) > param && !done {
subjlist := make([]int, 0, *group)
subjtrees := make([]rx.Node, 0, *group)
expressions := make([]*RegEx, 0, len(dfaList))
if init == 0 { //if this is the first iteration, mode = 0
//pick a random subset
for i := 0; i < *group && i < len(dfaList); i++ {
j := rand.Intn(len(dfaList)) // naive; can duplicate
subjlist = append(subjlist, j)
subjtrees = append(subjtrees, dfaList[j].Tree)
expressions = append(expressions, &RegEx{j, exprs[j].Expr})
fmt.Printf("rx { %d } %s\n", j, exprs[j].Expr)
}
init++
} else if len(dfaList) < value { //if the number of dfa's is small enough, run qns on all remaining
for i := 0; i < len(dfaList); i++ {
//j := rand.Intn(len(dfaList)) // naive; can duplicate
subjlist = append(subjlist, i)
subjtrees = append(subjtrees, dfaList[i].Tree)
expressions = append(expressions, &RegEx{ind[i], exprs[i].Expr})
fmt.Printf("rx { %d } %s\n", ind[i], exprs[i].Expr)
}
} else { //if number of dfa's is too large, do nothing
}
qns = make([]string, 0, maxEx) //initialize new set of questions
answers = new(rx.BitSet) //initialize new set of answers
permArray := rand.Perm(len(dfaList)) //randomly permute candidate DFAs
fmt.Println("len dfa list = ", len(dfaList))
//ask best question based on number of candidates that accept question word
answers, qns = askBulk2(answers, qns, permArray, dfaList, exprs, numPerGroup)
fmt.Printf("\n")
fmt.Println("before length of dfaList = ", len(dfaList))
tempLen = len(dfaList) //store current length
//refine the candidate set based on questions and answers
dfaList, exprs, ind = refine(answers, qns, dfaList, exprs, ind)
fmt.Println("length of dfaList = ", len(dfaList))
//if refinement did not change candidate list then we have equivalent reg exp
if tempLen == len(dfaList) {
done = true
fmt.Println("Main: There are ", len(dfaList), " reg exs that match your query and they are equivalent:")
for i := 0; i < len(dfaList); i++ {
fmt.Printf("rx { %d } %s\n ", ind[i], exprs[i].Expr)
}
}
}
//if one remaining reg ex, print it
if len(dfaList) == 1 {
fmt.Printf("\nMain: The reg ex you are looking for is: rx { %d } %s\n\nMETA DATA: \n", ind[0], exprs[0].Expr)
// print accumulated metadata
exprs[0].ShowMeta(os.Stdout, "")
fmt.Printf("\n")
}
//if no remaining reg exs, no match in our library
if len(dfaList) == 0 {
fmt.Printf("Main: No reg ex in our library matches your query.\n")
}
}