From d5f3384a13efe44b8407cbe55537e0bdc36a093c Mon Sep 17 00:00:00 2001
From: leandrofroes <leandro08pf@outlook.com>
Date: Mon, 13 Feb 2023 17:14:27 -0300
Subject: [PATCH] Fix MS URL search bug; Add func argc counter to the output

---
 pkg/cache/cache.go      |  2 ++
 pkg/scrapy/function.go  | 10 ++++++----
 pkg/scrapy/google.go    |  2 +-
 pkg/scrapy/structure.go |  5 ++---
 pkg/scrapy/type.go      |  3 +--
 pkg/utils/utils.go      |  3 +++
 6 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/pkg/cache/cache.go b/pkg/cache/cache.go
index 7bde537..4cb8300 100644
--- a/pkg/cache/cache.go
+++ b/pkg/cache/cache.go
@@ -20,6 +20,8 @@ func addFunctionCache(search, cachePath string, api *utils.API) (entry string){
   if api.DLL != ""{
     f.WriteString("Exported by: " + api.DLL + "\n\n")
   }
+  
+  f.WriteString("Number of arguments: " + api.Argc + "\n\n")
 
   f.WriteString(api.Description + "\n\n")
   f.WriteString(api.CodeA + "\n")
diff --git a/pkg/scrapy/function.go b/pkg/scrapy/function.go
index 26d25a5..b808789 100644
--- a/pkg/scrapy/function.go
+++ b/pkg/scrapy/function.go
@@ -4,6 +4,7 @@ import (
   "log"
   "regexp"
   "strings"
+  "strconv"
 
   "github.com/leandrofroes/manw/pkg/utils"
   "github.com/leandrofroes/manw/pkg/cache"
@@ -15,7 +16,6 @@ func ParseMSDNFunction(search, url string) *utils.API{
   api := utils.API{}
 
   collector := colly.NewCollector(
-    colly.AllowedDomains("docs.microsoft.com"),
     colly.UserAgent("Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:47.0) Gecko/20100101 Firefox/47.0"),
   )
 
@@ -58,6 +58,8 @@ func ParseMSDNFunction(search, url string) *utils.API{
   collector.OnHTML("pre", func(e *colly.HTMLElement){
     if e.Index == 0 {
       api.CodeA = e.Text
+      api.Argc = strconv.Itoa(len(strings.Split(e.Text, "\n")) - 3)
+
       return
     }
     if e.Index == 1{
@@ -91,10 +93,10 @@ func RunFunctionScraper(search, cachePath string){
 
   if cachePath != ""{
     if !cache.CheckCache(search, cachePath){
-      searchAux := "+api+function+msdn"
+      searchAux := "+function+msdn"
     
       url := GoogleMSDNSearch(search, searchAux)
-    
+
       if url == ""{
         utils.Warning("Unable to find this Windows function.")
       }
@@ -104,7 +106,7 @@ func RunFunctionScraper(search, cachePath string){
       cache.RunFunctionCache(search, cachePath, api)
     } 
   } else {
-    searchAux := "+api+function+msdn"
+    searchAux := "+function+msdn"
 
     url := GoogleMSDNSearch(search, searchAux)
   
diff --git a/pkg/scrapy/google.go b/pkg/scrapy/google.go
index 352376e..bcef3e0 100644
--- a/pkg/scrapy/google.go
+++ b/pkg/scrapy/google.go
@@ -26,7 +26,7 @@ func GoogleMSDNSearch(search, searchAux string) string{
       item := sellector.Eq(node)
       link, _ := item.Attr("href")
 
-      re, err := regexp.Compile("https://docs.microsoft.com/en-us/+")
+      re, err := regexp.Compile(".microsoft.com/en-us/+")
       utils.CheckError(err)
 
       if re.MatchString(link) {
diff --git a/pkg/scrapy/structure.go b/pkg/scrapy/structure.go
index 4b63303..d4e2875 100644
--- a/pkg/scrapy/structure.go
+++ b/pkg/scrapy/structure.go
@@ -15,7 +15,6 @@ func ParseMSDNStructure(search, url string) *utils.API{
   api := utils.API{}
 
   collector := colly.NewCollector(
-    colly.AllowedDomains("docs.microsoft.com"),
     colly.UserAgent("Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:47.0) Gecko/20100101 Firefox/47.0"),
   )
 
@@ -80,7 +79,7 @@ func RunStructureScraper(search, cachePath string){
 
   if cachePath != ""{
     if !cache.CheckCache(search, cachePath){
-      searchAux := "+api+msdn"
+      searchAux := "+msdn"
 
       url := GoogleMSDNSearch(search, searchAux)
     
@@ -105,4 +104,4 @@ func RunStructureScraper(search, cachePath string){
 
     utils.PrintMSDNStructure(api)
   }
-}
+}
\ No newline at end of file
diff --git a/pkg/scrapy/type.go b/pkg/scrapy/type.go
index b684729..7934605 100644
--- a/pkg/scrapy/type.go
+++ b/pkg/scrapy/type.go
@@ -14,7 +14,6 @@ import(
 func parseMSDNDataType(search, url string) string{
   var dataTypeInfo string
   collector := colly.NewCollector(
-    colly.AllowedDomains("docs.microsoft.com"),
     colly.UserAgent("Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:47.0) Gecko/20100101 Firefox/47.0"),
   )
 
@@ -53,7 +52,7 @@ func RunTypeScraper(search, cachePath string){
       searchAux := "+windows+data+type+msdn"
 
       url := GoogleMSDNSearch(search, searchAux)
-    
+
       if url == ""{
         utils.Warning("Unable to find this Windows data type.")
       }
diff --git a/pkg/utils/utils.go b/pkg/utils/utils.go
index d245368..08edc1d 100644
--- a/pkg/utils/utils.go
+++ b/pkg/utils/utils.go
@@ -16,6 +16,7 @@ type API struct {
   Return      string
   Source      string
   DLL         string
+  Argc        string
 }
 
 func CheckError(err error){
@@ -44,6 +45,8 @@ func PrintMSDNFunc(api *API){
   if api.DLL != ""{
     fmt.Printf("Exported by: " + api.DLL + "\n\n")
   }
+  
+  fmt.Printf("Number of arguments: " + api.Argc + "\n\n")
 
   fmt.Printf(api.Description + "\n\n")
   fmt.Printf(api.CodeA + "\n")