Removes dependencies on cgo (flex + bison)

3 files changed, 24 insertions, 102 deletions

diff --git a/scanner.go b/scanner.go
index 9b35de3..b972b70 100644
--- a/scanner.go
+++ b/scanner.go
@@ -1,14 +1,11 @@
 package main
 
-/*
-#include <stdio.h>
-#include "urlscanner.h"
-*/
-import "C"
 import (
 	"html/template"
 	"log"
 	"strings"
+
+	"github.com/PuerkitoBio/goquery"
 )
 
 type Link struct {
@@ -18,39 +15,34 @@ type Link struct {
 }
 
 func getToken(input string) []Link {
-	var token C.int
-	var ls []Link
-	var l Link
-
-	C.scan_string(C.CString(input))
-
-	for token = C.yylex(); token != C.MYEOF; token = C.yylex() {
 
-		if token == C.TOKEN_URL {
-			// flex reads the href attr
+	var links []Link
+	var link Link
 
-			l.Url = template.URL(C.GoString(C.yylval))
+	html_code := strings.NewReader(input)
+	doc, err := goquery.NewDocumentFromReader(html_code)
+	if err != nil {
+		log.Fatal(err)
+	}
 
-			if strings.HasSuffix(C.GoString(C.yylval), "/") {
-				l.IsDir = true
-			}
+	doc.Find("a").Each(func(i int, s *goquery.Selection) {
 
-		} else if token == C.TOKEN_TEXT {
-			// flex reads the link description
+		url, ok := s.Attr("href")
+		if !ok {
+			log.Fatal("Fehler beim URL parsen")
+		}
+		link.Url = template.URL(url)
 
-			l.Text = template.HTML(C.GoString(C.yylval))
-			ls = append(ls, l)
-			l = Link{}
+		if strings.HasSuffix(url, "/") {
+			link.IsDir = true
 		} else {
-			// lexical error
-
-			l = Link{}
-			log.Printf("Lexical Error on line %d \n", C.yylineno)
-			continue
+			link.IsDir = false
 		}
-	}
 
-	return ls
-}
+		link.Text = template.HTML(s.Text())
 
-// sort
+		links = append(links, link)
+	})
+
+	return links
+}
diff --git a/urlscanner.h b/urlscanner.h
deleted file mode 100644
index f815c41..0000000
--- a/urlscanner.h
+++ /dev/null
@@ -1,18 +0,0 @@
-#ifndef URLSCANNER_H
-#define URLSCANNER_H
-
-#define MYEOF      EOF
-#define TOKEN_URL  257
-#define TOKEN_TEXT 258
-
-#ifndef YYSTYPE
-#define YYSTYPE yystype
-typedef char* yystype;
-#endif
-
-extern int yylex();
-extern yystype yylval;
-extern int yylineno;
-extern void scan_string ( const char *str );
-
-#endif
diff --git a/urlscanner.l b/urlscanner.l
deleted file mode 100644
index 79768b2..0000000
--- a/urlscanner.l
+++ /dev/null
@@ -1,52 +0,0 @@
-%option noyywrap
-%option yylineno
-%option nounput
-
-%{
-#include "urlscanner.h"
-#include "string.h"
-yystype yylval;
-#define YY_NO_INPUT
-%}
-
-%x ATAG_HREF
-%x ATAG_BETWEEN
-%x ATAG_DESC
-
-%%
-
-[^<]* {  }
-
-"<a"[^>]*"href=\"" { BEGIN(ATAG_HREF); }
-
-<ATAG_HREF>[^\"]+ {
-  yylval = strdup(yytext);
-  BEGIN(ATAG_BETWEEN);
-  return TOKEN_URL;
-}
-
-<ATAG_HREF>\" {
-  yylval = "";
-  BEGIN(ATAG_BETWEEN);
-  return TOKEN_URL;
-}
-
-<ATAG_BETWEEN>[^>]* { }
-
-<ATAG_BETWEEN>">" { BEGIN(ATAG_DESC); }
-
-<ATAG_DESC>[^<]* {
-  yylval = strdup(yytext);
-  BEGIN(INITIAL);
-  return TOKEN_TEXT;
-}
-
-. { /* skip */ }
-
-<<EOF>>                            { return MYEOF; }
-
-%%
-
-void scan_string(const char* str) {
-    yy_switch_to_buffer(yy_scan_string(str));
-}