test/integration/test-ubuntu-bug-346386-apt-get-update-paywall: use downloadfile()

[apt.git] / methods / mirror.cc

diff --git a/methods/mirror.cc b/methods/mirror.cc
index 428726a3d197cb42ea314adb7eedbb39654de73c..d3aef91bcd3d68054a96754152f418b463d09d8f 100644 (file)
--- a/methods/mirror.cc
+++ b/methods/mirror.cc
@@ -3,62 +3,64 @@
 // $Id: mirror.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
 /* ######################################################################
 
 // $Id: mirror.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
 /* ######################################################################
 

-   Mirror Aquire Method - This is the Mirror aquire method for APT.
+   Mirror Acquire Method - This is the Mirror acquire method for APT.

    
    ##################################################################### */
                                                                        /*}}}*/
 // Include Files                                                       /*{{{*/
    
    ##################################################################### */
                                                                        /*}}}*/
 // Include Files                                                       /*{{{*/

+#include <config.h>
+
+#include <apt-pkg/aptconfiguration.h>

 #include <apt-pkg/fileutl.h>
 #include <apt-pkg/acquire-method.h>
 #include <apt-pkg/acquire-item.h>
 #include <apt-pkg/acquire.h>
 #include <apt-pkg/error.h>
 #include <apt-pkg/fileutl.h>
 #include <apt-pkg/acquire-method.h>
 #include <apt-pkg/acquire-item.h>
 #include <apt-pkg/acquire.h>
 #include <apt-pkg/error.h>

-#include <apt-pkg/hashes.h>

 #include <apt-pkg/sourcelist.h>
 #include <apt-pkg/sourcelist.h>

-
-#include <fstream>
+#include <apt-pkg/configuration.h>
+#include <apt-pkg/metaindex.h>
+#include <apt-pkg/strutl.h>
+
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <algorithm>

 #include <iostream>
 #include <iostream>

-#include <stdarg.h>
+#include <fstream>

 #include <sys/stat.h>
 #include <sys/stat.h>

-#include <sys/types.h>
+#include <sys/utsname.h>

 #include <dirent.h>
 
 using namespace std;
 
 #include <dirent.h>
 
 using namespace std;
 

+#include<sstream>
+

 #include "mirror.h"
 #include "http.h"
 #include "mirror.h"
 #include "http.h"

-#include "apti18n.h"
+#include <apti18n.h>

                                                                        /*}}}*/
 
                                                                        /*}}}*/
 

-/* 
+/* Done:
+ * - works with http (only!)
+ * - always picks the first mirror from the list
+ * - call out to problem reporting script
+ * - supports "deb mirror://host/path/to/mirror-list/// dist component"
+ * - uses pkgAcqMethod::FailReason() to have a string representation
+ *   of the failure that is also send to LP
+ * 

  * TODO: 
  * TODO: 

- * - what about gpgv  failures? better standard format for errors
-     to send back to LP
-   #OR#
- * - implement it at the pkgAcquire::Item::Failed() level but then
-     we need to send back what uri exactly was failing 
-     [mvo: the problem with this approach is ::Failed() is not really
-           called for all failures :/ e.g. md5sum mismatch in a archive
-           is not]
- * - deal with runing as non-root because we can't write to the lists 
+ * - deal with running as non-root because we can't write to the lists

      dir then -> use the cached mirror file
  * - better method to download than having a pkgAcquire interface here
      dir then -> use the cached mirror file
  * - better method to download than having a pkgAcquire interface here

- * - magicmarker is (a bit) evil, maybe just use a similar approach as in
-     clean and read the sources.list and use the GetURI() method to find
-     the prefix?
+ *   and better error handling there!
+ * - support more than http

  * - testing :)
  */
 
 MirrorMethod::MirrorMethod()
  * - testing :)
  */
 
 MirrorMethod::MirrorMethod()

-   : HttpMethod(), HasMirrorFile(false)
+   : HttpMethod(), DownloadedMirrorFile(false), Debug(false)

 {
 {

-#if 0
-   HasMirrorFile=true;
-   BaseUri="mirror://people.ubuntu.com/~mvo/mirror/mirrors";
-   MirrorFile="/var/lib/apt/lists/people.ubuntu.com_%7emvo_apt_mirror_mirrors";
-   Mirror="http://de.archive.ubuntu.com/ubuntu/";
-#endif
-};
+}

 
 // HttpMethod::Configuration - Handle a configuration message          /*{{{*/
 // ---------------------------------------------------------------------
 
 // HttpMethod::Configuration - Handle a configuration message          /*{{{*/
 // ---------------------------------------------------------------------


@@ -81,21 +83,24 @@ bool MirrorMethod::Clean(string Dir)
    if(Debug)
       clog << "MirrorMethod::Clean(): " << Dir << endl;
 
    if(Debug)
       clog << "MirrorMethod::Clean(): " << Dir << endl;
 

+   if(Dir == "/")
+      return _error->Error("will not clean: '/'");
+

    // read sources.list
    pkgSourceList list;
    list.ReadMainList();
 
    // read sources.list
    pkgSourceList list;
    list.ReadMainList();
 

-   DIR *D = opendir(Dir.c_str());   
+   DIR *D = opendir(Dir.c_str());

    if (D == 0)
       return _error->Errno("opendir",_("Unable to read %s"),Dir.c_str());
    if (D == 0)
       return _error->Errno("opendir",_("Unable to read %s"),Dir.c_str());

-   
+

    string StartDir = SafeGetCWD();
    if (chdir(Dir.c_str()) != 0)
    {
       closedir(D);
       return _error->Errno("chdir",_("Unable to change to %s"),Dir.c_str());
    }
    string StartDir = SafeGetCWD();
    if (chdir(Dir.c_str()) != 0)
    {
       closedir(D);
       return _error->Errno("chdir",_("Unable to change to %s"),Dir.c_str());
    }

-   
+

    for (struct dirent *Dir = readdir(D); Dir != 0; Dir = readdir(D))
    {
       // Skip some files..
    for (struct dirent *Dir = readdir(D); Dir != 0; Dir = readdir(D))
    {
       // Skip some files..


@@ -106,148 +111,356 @@ bool MirrorMethod::Clean(string Dir)
         continue;
 
       // see if we have that uri
         continue;
 
       // see if we have that uri

-      for(I=list.begin(); I != list.end(); I++)
+      for(I=list.begin(); I != list.end(); ++I)

       {
         string uri = (*I)->GetURI();
       {
         string uri = (*I)->GetURI();

-        if(uri.substr(0,strlen("mirror://")) != string("mirror://"))
+        if(uri.compare(0, strlen("mirror://"), "mirror://") != 0)

            continue;
            continue;

-        string Marker = _config->Find("Acquire::Mirror::MagicMarker","///");
-        string BaseUri = uri.substr(0,uri.find(Marker));
+        string BaseUri = uri.substr(0,uri.size()-1);

         if (URItoFileName(BaseUri) == Dir->d_name)
            break;
       }
       // nothing found, nuke it
       if (I == list.end())
         unlink(Dir->d_name);
         if (URItoFileName(BaseUri) == Dir->d_name)
            break;
       }
       // nothing found, nuke it
       if (I == list.end())
         unlink(Dir->d_name);

-   };
-   
-   chdir(StartDir.c_str());
+   }
+

    closedir(D);
    closedir(D);

-   return true;   
+   if (chdir(StartDir.c_str()) != 0)
+      return _error->Errno("chdir",_("Unable to change to %s"),StartDir.c_str());
+   return true;

 }
 
 
 }
 
 

-bool MirrorMethod::GetMirrorFile(string uri)
+bool MirrorMethod::DownloadMirrorFile(string /*mirror_uri_str*/)

 {
 {

-   string Marker = _config->Find("Acquire::Mirror::MagicMarker","///");
-   BaseUri = uri.substr(0,uri.find(Marker));
-
+   // not that great to use pkgAcquire here, but we do not have
+   // any other way right now

    string fetch = BaseUri;
    fetch.replace(0,strlen("mirror://"),"http://");
 
    string fetch = BaseUri;
    fetch.replace(0,strlen("mirror://"),"http://");
 

-   // get new file
-   MirrorFile = _config->FindDir("Dir::State::mirrors") + URItoFileName(BaseUri);
+#if 0 // no need for this, the getArchitectures() will also include the main
+      // arch
+   // append main architecture
+   fetch += "?arch=" + _config->Find("Apt::Architecture");
+#endif

 
 

-   if(Debug) 
-   {
-      cerr << "base-uri: " << BaseUri << endl;
-      cerr << "mirror-file: " << MirrorFile << endl;
-   }
+   // append all architectures
+   std::vector<std::string> vec = APT::Configuration::getArchitectures();
+   for (std::vector<std::string>::const_iterator I = vec.begin();
+        I != vec.end(); ++I)
+      if (I == vec.begin())
+         fetch += "?arch=" + (*I);
+      else
+         fetch += "&arch=" + (*I);

 
 

-   // check the file, if it is not older than RefreshInterval just use it
-   // otherwise try to get a new one
-   if(FileExists(MirrorFile)) 
-   {
-      struct stat buf;
-      time_t t,now,refresh;
-      if(stat(MirrorFile.c_str(), &buf) != 0)
-        return false;
-      t = std::max(buf.st_mtime, buf.st_ctime);
-      now = time(NULL);
-      refresh = 60*_config->FindI("Acquire::Mirror::RefreshInterval",360);
-      if(t + refresh > now)
-      {
-        if(Debug)
-           clog << "Mirror file is in RefreshInterval" << endl;
-        HasMirrorFile = true;
-        return true;
-      }
-      if(Debug)
-        clog << "Mirror file " << MirrorFile << " older than " << refresh << "min, re-download it" << endl;
-   }
+   // append the dist as a query string
+   if (Dist != "")
+      fetch += "&dist=" + Dist;
+
+   if(Debug)
+      clog << "MirrorMethod::DownloadMirrorFile(): '" << fetch << "'"
+           << " to " << MirrorFile << endl;

 
 

-   // not that great to use pkgAcquire here, but we do not have 
-   // any other way right now

    pkgAcquire Fetcher;
    new pkgAcqFile(&Fetcher, fetch, "", 0, "", "", "", MirrorFile);
    bool res = (Fetcher.Run() == pkgAcquire::Continue);
    pkgAcquire Fetcher;
    new pkgAcqFile(&Fetcher, fetch, "", 0, "", "", "", MirrorFile);
    bool res = (Fetcher.Run() == pkgAcquire::Continue);

-   if(res)
-      HasMirrorFile = true;
+   if(res) {
+      DownloadedMirrorFile = true;
+      chmod(MirrorFile.c_str(), 0644);
+   }

    Fetcher.Shutdown();
    Fetcher.Shutdown();

+
+   if(Debug)
+      clog << "MirrorMethod::DownloadMirrorFile() success: " << res << endl;
+

    return res;
 }
 
    return res;
 }
 

-bool MirrorMethod::SelectMirror()
+// Randomizes the lines in the mirror file, this is used so that
+// we spread the load on the mirrors evenly
+bool MirrorMethod::RandomizeMirrorFile(string mirror_file)
+{
+   vector<string> content;
+   string line;
+
+   if (!FileExists(mirror_file))
+      return false;
+
+   // read
+   ifstream in(mirror_file.c_str());
+   while ( !in.eof() ) {
+      getline(in, line);
+      content.push_back(line);
+   }
+
+   // we want the file to be random for each different machine, but also
+   // "stable" on the same machine. this is to avoid running into out-of-sync
+   // issues (i.e. Release/Release.gpg different on each mirror)
+   struct utsname buf;
+   int seed=1;
+   if(uname(&buf) == 0) {
+      for(int i=0,seed=1; buf.nodename[i] != 0; ++i) {
+         seed = seed * 31 + buf.nodename[i];
+      }
+   }
+   srand( seed );
+   random_shuffle(content.begin(), content.end());
+
+   // write
+   ofstream out(mirror_file.c_str());
+   while ( !content.empty()) {
+      line = content.back();
+      content.pop_back();
+      out << line << "\n";
+   }
+
+   return true;
+}
+
+/* convert a the Queue->Uri back to the mirror base uri and look
+ * at all mirrors we have for this, this is needed as queue->uri
+ * may point to different mirrors (if TryNextMirror() was run)
+ */
+void MirrorMethod::CurrentQueueUriToMirror()
+{
+   // already in mirror:// style so nothing to do
+   if(Queue->Uri.find("mirror://") == 0)
+      return;
+
+   // find current mirror and select next one
+   for (vector<string>::const_iterator mirror = AllMirrors.begin();
+       mirror != AllMirrors.end(); ++mirror)
+   {
+      if (Queue->Uri.find(*mirror) == 0)
+      {
+        Queue->Uri.replace(0, mirror->length(), BaseUri);
+        return;
+      }
+   }
+   _error->Error("Internal error: Failed to convert %s back to %s",
+                Queue->Uri.c_str(), BaseUri.c_str());
+}
+
+bool MirrorMethod::TryNextMirror()

 {
 {

+   // find current mirror and select next one
+   for (vector<string>::const_iterator mirror = AllMirrors.begin();
+       mirror != AllMirrors.end(); ++mirror)
+   {
+      if (Queue->Uri.find(*mirror) != 0)
+        continue;
+
+      vector<string>::const_iterator nextmirror = mirror + 1;
+      if (nextmirror == AllMirrors.end())
+        break;
+      Queue->Uri.replace(0, mirror->length(), *nextmirror);
+      if (Debug)
+        clog << "TryNextMirror: " << Queue->Uri << endl;
+
+      // inform parent
+      UsedMirror = *nextmirror;
+      Log("Switching mirror");
+      return true;
+   }
+
+   if (Debug)
+      clog << "TryNextMirror could not find another mirror to try" << endl;
+
+   return false;
+}
+
+bool MirrorMethod::InitMirrors()
+{
+   // if we do not have a MirrorFile, fallback
+   if(!FileExists(MirrorFile))
+   {
+      // FIXME: fallback to a default mirror here instead 
+      //        and provide a config option to define that default
+      return _error->Error(_("No mirror file '%s' found "), MirrorFile.c_str());
+   }
+
+   if (access(MirrorFile.c_str(), R_OK) != 0)
+   {
+      // FIXME: fallback to a default mirror here instead 
+      //        and provide a config option to define that default
+      return _error->Error(_("Can not read mirror file '%s'"), MirrorFile.c_str());
+   }  
+

    // FIXME: make the mirror selection more clever, do not 
    //        just use the first one!
    // FIXME: make the mirror selection more clever, do not 
    //        just use the first one!

+   // BUT: we can not make this random, the mirror has to be
+   //      stable across session, because otherwise we can
+   //      get into sync issues (got indexfiles from mirror A,
+   //      but packages from mirror B - one might be out of date etc)

    ifstream in(MirrorFile.c_str());
    ifstream in(MirrorFile.c_str());

-   getline(in, Mirror);
-   if(Debug)
-      cerr << "Using mirror: " << Mirror << endl;
+   string s;
+   while (!in.eof()) 
+   {
+      getline(in, s);
+
+      // ignore lines that start with #
+      if (s.find("#") == 0)
+         continue;
+      // ignore empty lines
+      if (s.size() == 0)
+         continue;
+      // ignore non http lines
+      if (s.compare(0, strlen("http://"), "http://") != 0)
+         continue;
+
+      AllMirrors.push_back(s);
+   }
+   if (AllMirrors.empty()) {
+       return _error->Error(_("No entry found in mirror file '%s'"), MirrorFile.c_str());
+   }
+   Mirror = AllMirrors[0];
+   UsedMirror = Mirror;

    return true;
 }
 
    return true;
 }
 

+string MirrorMethod::GetMirrorFileName(string mirror_uri_str)
+{
+   /* 
+    - a mirror_uri_str looks like this:
+    mirror://people.ubuntu.com/~mvo/apt/mirror/mirrors/dists/feisty/Release.gpg
+   
+    - the matching source.list entry
+    deb mirror://people.ubuntu.com/~mvo/apt/mirror/mirrors feisty main
+   
+    - we actually want to go after:
+    http://people.ubuntu.com/~mvo/apt/mirror/mirrors
+
+    And we need to save the BaseUri for later:
+    - mirror://people.ubuntu.com/~mvo/apt/mirror/mirrors
+
+   FIXME: what if we have two similar prefixes?
+     mirror://people.ubuntu.com/~mvo/mirror
+     mirror://people.ubuntu.com/~mvo/mirror2
+   then mirror_uri_str looks like:
+     mirror://people.ubuntu.com/~mvo/apt/mirror/dists/feisty/Release.gpg
+     mirror://people.ubuntu.com/~mvo/apt/mirror2/dists/feisty/Release.gpg
+   we search sources.list and find:
+     mirror://people.ubuntu.com/~mvo/apt/mirror
+   in both cases! So we need to apply some domain knowledge here :( and
+   check for /dists/ or /Release.gpg as suffixes
+   */
+   string name;
+   if(Debug)
+      std::cerr << "GetMirrorFileName: " << mirror_uri_str << std::endl;
+
+   // read sources.list and find match
+   vector<metaIndex *>::const_iterator I;
+   pkgSourceList list;
+   list.ReadMainList();
+   for(I=list.begin(); I != list.end(); ++I)
+   {
+      string uristr = (*I)->GetURI();
+      if(Debug)
+        std::cerr << "Checking: " << uristr << std::endl;
+      if(uristr.substr(0,strlen("mirror://")) != string("mirror://"))
+        continue;
+      // find matching uri in sources.list
+      if(mirror_uri_str.substr(0,uristr.size()) == uristr)
+      {
+        if(Debug)
+           std::cerr << "found BaseURI: " << uristr << std::endl;
+        BaseUri = uristr.substr(0,uristr.size()-1);
+         Dist = (*I)->GetDist();
+      }
+   }
+   // get new file
+   name = _config->FindDir("Dir::State::mirrors") + URItoFileName(BaseUri);
+
+   if(Debug) 
+   {
+      cerr << "base-uri: " << BaseUri << endl;
+      cerr << "mirror-file: " << name << endl;
+   }
+   return name;
+}
+

 // MirrorMethod::Fetch - Fetch an item                                 /*{{{*/
 // ---------------------------------------------------------------------
 /* This adds an item to the pipeline. We keep the pipeline at a fixed
    depth. */
 bool MirrorMethod::Fetch(FetchItem *Itm)
 {
 // MirrorMethod::Fetch - Fetch an item                                 /*{{{*/
 // ---------------------------------------------------------------------
 /* This adds an item to the pipeline. We keep the pipeline at a fixed
    depth. */
 bool MirrorMethod::Fetch(FetchItem *Itm)
 {

-   // select mirror only once per session
-   if(!HasMirrorFile)
+   if(Debug)
+      clog << "MirrorMethod::Fetch()" << endl;
+
+   // the http method uses Fetch(0) as a way to update the pipeline,
+   // just let it do its work in this case - Fetch() with a valid
+   // Itm will always run before the first Fetch(0)
+   if(Itm == NULL) 
+      return HttpMethod::Fetch(Itm);
+
+   // if we don't have the name of the mirror file on disk yet,
+   // calculate it now (can be derived from the uri)
+   if(MirrorFile.empty())
+      MirrorFile = GetMirrorFileName(Itm->Uri);
+
+  // download mirror file once (if we are after index files)
+   if(Itm->IndexFile && !DownloadedMirrorFile)

    {
       Clean(_config->FindDir("Dir::State::mirrors"));
    {
       Clean(_config->FindDir("Dir::State::mirrors"));

-      GetMirrorFile(Itm->Uri);
-      SelectMirror();
+      if (DownloadMirrorFile(Itm->Uri))
+         RandomizeMirrorFile(MirrorFile);

    }
 
    }
 

-   for (FetchItem *I = Queue; I != 0; I = I->Next)
-   {
-      if(I->Uri.find("mirror://") != string::npos)
-        I->Uri.replace(0,BaseUri.size(),Mirror);
+   if(AllMirrors.empty()) {
+      if(!InitMirrors()) {
+        // no valid mirror selected, something went wrong downloading
+        // from the master mirror site most likely and there is
+        // no old mirror file availalbe
+        return false;
+      }

    }
 
    }
 

+   if(Itm->Uri.find("mirror://") != string::npos)
+      Itm->Uri.replace(0,BaseUri.size(), Mirror);
+
+   if(Debug)
+      clog << "Fetch: " << Itm->Uri << endl << endl;
+

    // now run the real fetcher
    return HttpMethod::Fetch(Itm);
    // now run the real fetcher
    return HttpMethod::Fetch(Itm);

-};
+}

 
 void MirrorMethod::Fail(string Err,bool Transient)
 {
 
 void MirrorMethod::Fail(string Err,bool Transient)
 {

-   // FIXME: queue next mirror?
-   ReportMirrorFailure(Err);
+   // FIXME: TryNextMirror is not ideal for indexfile as we may
+   //        run into auth issues
+
+   if (Debug)
+      clog << "Failure to get " << Queue->Uri << endl;
+
+   // try the next mirror on fail (if its not a expected failure,
+   // e.g. translations are ok to ignore)
+   if (!Queue->FailIgnore && TryNextMirror())
+      return;
+
+   // all mirrors failed, so bail out
+   string s;
+   strprintf(s, _("[Mirror: %s]"), Mirror.c_str());
+   SetIP(s);

 
 

-   if(Queue->Uri.find("http://") != string::npos)
-      Queue->Uri.replace(0,Mirror.size(), BaseUri);
+   CurrentQueueUriToMirror();

    pkgAcqMethod::Fail(Err, Transient);
 }
 
 void MirrorMethod::URIStart(FetchResult &Res)
 {
    pkgAcqMethod::Fail(Err, Transient);
 }
 
 void MirrorMethod::URIStart(FetchResult &Res)
 {

-   if(Queue->Uri.find("http://") != string::npos)
-      Queue->Uri.replace(0,Mirror.size(), BaseUri);
+   CurrentQueueUriToMirror();

    pkgAcqMethod::URIStart(Res);
 }
 
 void MirrorMethod::URIDone(FetchResult &Res,FetchResult *Alt)
 {
    pkgAcqMethod::URIStart(Res);
 }
 
 void MirrorMethod::URIDone(FetchResult &Res,FetchResult *Alt)
 {

-   // FIXME: queue next mirror?
-   if(Queue->ExpectedMD5 != "" && Res.MD5Sum != Queue->ExpectedMD5)
-      ReportMirrorFailure("499 Hash mismatch");
-
-   if(Queue->Uri.find("http://") != string::npos)
-      Queue->Uri.replace(0,Mirror.size(), BaseUri);
+   CurrentQueueUriToMirror();

    pkgAcqMethod::URIDone(Res, Alt);
 }
 
    pkgAcqMethod::URIDone(Res, Alt);
 }
 

-void MirrorMethod::ReportMirrorFailure(string FailCode)
-{
-   // report that Queue->Uri failed
-   std::cerr << "\nReportMirrorFailure: " 
-            << Queue->Uri
-            << " FailCode: " 
-            << FailCode << std::endl;
-}

 
 int main()
 {
 
 int main()
 {