PureBasic

Code : Tout sélectionner

Global host$ = "<nom_du_host>"     ; www.purebasic.com par exemple
Global site$ = "<adresse_du_site>" ; http://www.purebasic.com par exemple
Global page$ = "<nom_de_la_page>"  ; index.php par exemple

Procedure checkURL(link$, code$ = "200 OK")
  If link$ = #Null$
    Debug "no link"
    ProcedureReturn #False
  EndIf
  Protected header$
  Protected result$
  If InitNetwork()
    header$ = GetHTTPHeader(link$)
    result$ = StringField(header$, 1, #LF$)
    If FindString(result$, code$, 1, #PB_String_NoCase)
      ProcedureReturn #True
    Else
      ProcedureReturn #False
    EndIf
  Else
    ProcedureReturn #False
  EndIf
EndProcedure

Procedure getURL(page$)
  Protected retval
  Protected URL$, header$, result$
  ;;;
  Protected Cookie$         = "<cookie>"
  Protected Accept$         = "Accept: text/html, application/xhtml+xml, image/jxr, */*"
  Protected AcceptLanguage$ = "Accept-Language: fr-FR"
  Protected UserAgent$      = "User-Agent: <user_agent>"
  Protected Referer$        = "Referer: " + site$
  Protected Host$           = "Host: " + host$
  Protected param$
  param$ = " --header=" + Chr(34) + Cookie$ + Chr(34)
  param$ + " --header=" + Chr(34) + Accept$ + Chr(34)
  param$ + " --header=" + Chr(34) + AcceptLanguage$ + Chr(34)
  param$ + " --header=" + Chr(34) + UserAgent$ + Chr(34)
  param$ + " --header=" + Chr(34) + Referer$ + Chr(34)
  param$ + " --header=" + Chr(34) + Host$ + Chr(34)
  ;;;
  SetFileAttributes(page$, #PB_FileSystem_Normal)
  DeleteFile(page$, #PB_FileSystem_Force)
  URL$ = site$ + "/" + page$
  If checkURL(URL$)
;;;;La ligne qui suit ne fonctionne pas a cause de l'absence des headers
;;;;retval = ReceiveHTTPFile(URL$, page$, #PB_HTTP_NoRedirect)
    retval = RunProgram("wget.exe", param$ + " " + URL$, AppPath$, #PB_Program_Wait|#PB_Program_Hide)
    If retval
      ProcedureReturn #True
    Else
      DeleteFile(page$, #PB_FileSystem_Force)
      ProcedureReturn #False
    EndIf
  EndIf
EndProcedure

getURL(page$)

Code : Tout sélectionner

*Buffer = ReceiveHTTPMemory(url$)

Code : Tout sélectionner

InitNetwork()

url$= "http://www.purebasic.com/index.php"
Header$ = GetHTTPHeader(url$)

Debug GetURLPart(url$, #PB_URL_Site)
Debug GetURLPart(url$, #PB_URL_Path)

Debug ""

Repeat
  Index+1
  Line$ = StringField(Header$, Index, #LF$)
  Debug Line$
Until Line$ = ""

Debug ""

*Buffer = ReceiveHTTPMemory(url$)
If *Buffer
  Size = MemorySize(*Buffer)
  Debug "Content: " + PeekS(*Buffer, Size, #PB_UTF8|#PB_ByteLength)
  FreeMemory(*Buffer)
Else
  Debug "Failed"
EndIf

Code : Tout sélectionner

; working with static libcurl
InitNetwork()
IncludeFile "libcurl.pbi"

curl      = curl_easy_init()
url.s     = str2curl("http://une-de-mes-url-qui-pose-probleme/une-page")

header.s  = str2curl("Cookie: récupéré avec Firefox")
header.s  = str2curl("Accept: text/html, application/xhtml+xml, image/jxr, */*")
header.s  = str2curl("Accept-Language: fr-FR")
header.s  = str2curl("User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2704.79 Safari/537.36 Edge/14.14393")
header.s  = str2curl("Referer: récupéré avec Firefox")
header.s  = str2curl("Host: récupéré avec Firefox")

If curl
  curl_easy_setopt(curl,#CURLOPT_URL,@url)
  curl_easy_setopt(curl,#CURLOPT_IPRESOLVE,#CURL_IPRESOLVE_V4)
  curl_easy_setopt(curl,#CURLOPT_TIMEOUT,30)
  *header = curl_slist_append(0,header)
  curl_easy_setopt(curl,#CURLOPT_HTTPHEADER,*header)
  curl_easy_setopt(curl,#CURLOPT_WRITEFUNCTION,@curlWriteData())
  res = curl_easy_perform(curl)
  resData.s = curlGetData()
  curl_easy_getinfo(curl,#CURLINFO_RESPONSE_CODE,@resHTTP)
  Debug "result: " + Str(res)
  If Not res
    Debug "HTTP code: " + Str(resHTTP)
    Debug "HTTP data: " + #CRLF$ + resData
  EndIf
  curl_easy_cleanup(curl)
Else
  Debug "can't init curl!"
EndIf

End