merge from the lp:~donkult/apt/sid branch

7 files changed, 158 insertions, 42 deletions

diff --git a/apt-pkg/aptconfiguration.cc b/apt-pkg/aptconfiguration.cc
index 899004d9f..b5f29472d 100644
--- a/apt-pkg/aptconfiguration.cc
+++ b/apt-pkg/aptconfiguration.cc
@@ -8,13 +8,18 @@
    ##################################################################### */
 									/*}}}*/
 // Include Files							/*{{{*/
-#include <apt-pkg/fileutl.h>
 #include <apt-pkg/aptconfiguration.h>
 #include <apt-pkg/configuration.h>
+#include <apt-pkg/fileutl.h>
+#include <apt-pkg/macros.h>
+#include <apt-pkg/strutl.h>
+
+#include <sys/types.h>
+#include <dirent.h>
 
-#include <vector>
-#include <string>
 #include <algorithm>
+#include <string>
+#include <vector>
 									/*}}}*/
 namespace APT {
 // getCompressionTypes - Return Vector of usbale compressiontypes	/*{{{*/
@@ -96,7 +101,7 @@ const Configuration::getCompressionTypes(bool const &Cached) {
    will result in "de_DE, de, en".
    The special word "none" is the stopcode for the not-All code vector */
 std::vector<std::string> const Configuration::getLanguages(bool const &All,
-				bool const &Cached, char const * const Locale) {
+				bool const &Cached, char const ** const Locale) {
 	using std::string;
 
 	// The detection is boring and has a lot of cornercases,
@@ -117,27 +122,64 @@ std::vector<std::string> const Configuration::getLanguages(bool const &All,
 		}
 	}
 
-	// get the environment language code
+	// Include all Language codes we have a Translation file for in /var/lib/apt/lists
+	// so they will be all included in the Cache.
+	std::vector<string> builtin;
+	DIR *D = opendir(_config->FindDir("Dir::State::lists").c_str());
+	if (D != 0) {
+		builtin.push_back("none");
+		for (struct dirent *Ent = readdir(D); Ent != 0; Ent = readdir(D)) {
+			string const name = Ent->d_name;
+			size_t const foundDash = name.rfind("-");
+			size_t const foundUnderscore = name.rfind("_");
+			if (foundDash == string::npos || foundUnderscore == string::npos ||
+			    foundDash <= foundUnderscore ||
+			    name.substr(foundUnderscore+1, foundDash-(foundUnderscore+1)) != "Translation")
+				continue;
+			string const c = name.substr(foundDash+1);
+			if (unlikely(c.empty() == true) || c == "en")
+				continue;
+			// Skip unusual files, like backups or that alike
+			string::const_iterator s = c.begin();
+			for (;s != c.end(); ++s) {
+				if (isalpha(*s) == 0)
+					break;
+			}
+			if (s != c.end())
+				continue;
+			if (std::find(builtin.begin(), builtin.end(), c) != builtin.end())
+				continue;
+			builtin.push_back(c);
+		}
+	}
+
+	// get the environment language codes: LC_MESSAGES (and later LANGUAGE)
 	// we extract both, a long and a short code and then we will
 	// check if we actually need both (rare) or if the short is enough
-	string const envMsg = string(Locale == 0 ? std::setlocale(LC_MESSAGES, NULL) : Locale);
+	string const envMsg = string(Locale == 0 ? std::setlocale(LC_MESSAGES, NULL) : *Locale);
 	size_t const lenShort = (envMsg.find('_') != string::npos) ? envMsg.find('_') : 2;
-	size_t const lenLong = (envMsg.find('.') != string::npos) ? envMsg.find('.') : (lenShort + 3);
+	size_t const lenLong = (envMsg.find_first_of(".@") != string::npos) ? envMsg.find_first_of(".@") : (lenShort + 3);
 
 	string envLong = envMsg.substr(0,lenLong);
 	string const envShort = envLong.substr(0,lenShort);
-	bool envLongIncluded = true, envShortIncluded = false;
+	bool envLongIncluded = true;
 
 	// first cornercase: LANG=C, so we use only "en" Translation
 	if (envLong == "C") {
 		codes.push_back("en");
-		return codes;
+		allCodes = codes;
+		allCodes.insert(allCodes.end(), builtin.begin(), builtin.end());
+		if (All == true)
+			return allCodes;
+		else
+			return codes;
 	}
 
+	// to save the servers from unneeded queries, we only try also long codes
+	// for languages it is realistic to have a long code translation file…
+	// TODO: Improve translation acquire system to drop them dynamic
+	char const *needLong[] = { "cs", "en", "pt", "sv", "zh", NULL };
 	if (envLong != envShort) {
-		// to save the servers from unneeded queries, we only try also long codes
-		// for languages it is realistic to have a long code translation file...
-		char const *needLong[] = { "cs", "en", "pt", "sv", "zh", NULL };
 		for (char const **l = needLong; *l != NULL; l++)
 			if (envShort.compare(*l) == 0) {
 				envLongIncluded = false;
@@ -155,7 +197,51 @@ std::vector<std::string> const Configuration::getLanguages(bool const &All,
 	if (oldAcquire.empty() == false && oldAcquire != "environment") {
 		if (oldAcquire != "none")
 			codes.push_back(oldAcquire);
-		return codes;
+		codes.push_back("en");
+		allCodes = codes;
+		for (std::vector<string>::const_iterator b = builtin.begin();
+		     b != builtin.end(); ++b)
+			if (std::find(allCodes.begin(), allCodes.end(), *b) == allCodes.end())
+				allCodes.push_back(*b);
+		if (All == true)
+			return allCodes;
+		else
+			return codes;
+	}
+
+	// It is very likely we will need to environment codes later,
+	// so let us generate them now from LC_MESSAGES and LANGUAGE
+	std::vector<string> environment;
+	// take care of LC_MESSAGES
+	if (envLongIncluded == false)
+		environment.push_back(envLong);
+	environment.push_back(envShort);
+	// take care of LANGUAGE
+	string envLang = Locale == 0 ? getenv("LANGUAGE") : *(Locale+1);
+	if (envLang.empty() == false) {
+		std::vector<string> env = ExplodeString(envLang,':');
+		short addedLangs = 0; // add a maximum of 3 fallbacks from the environment
+		for (std::vector<string>::const_iterator e = env.begin();
+		     e != env.end() && addedLangs < 3; ++e) {
+			if (unlikely(e->empty() == true) || *e == "en")
+				continue;
+			if (*e == envLong || *e == envShort)
+				continue;
+			if (std::find(environment.begin(), environment.end(), *e) != environment.end())
+				continue;
+			if (e->find('_') != string::npos) {
+				// Drop LongCodes here - ShortCodes are also included
+				string const shorty = e->substr(0, e->find('_'));
+				char const **n = needLong;
+				for (; *n != NULL; ++n)
+					if (shorty == *n)
+						break;
+				if (*n == NULL)
+					continue;
+			}
+			++addedLangs;
+			environment.push_back(*e);
+		}
 	}
 
 	// Support settings like Acquire::Translation=none on the command line to
@@ -163,26 +249,35 @@ std::vector<std::string> const Configuration::getLanguages(bool const &All,
 	string const forceLang = _config->Find("Acquire::Languages","");
 	if (forceLang.empty() == false) {
 		if (forceLang == "environment") {
-			if (envLongIncluded == false)
-				codes.push_back(envLong);
-			if (envShortIncluded == false)
-				codes.push_back(envShort);
-			return codes;
+			codes = environment;
 		} else if (forceLang != "none")
 			codes.push_back(forceLang);
-		return codes;
+		allCodes = codes;
+		for (std::vector<string>::const_iterator b = builtin.begin();
+		     b != builtin.end(); ++b)
+			if (std::find(allCodes.begin(), allCodes.end(), *b) == allCodes.end())
+				allCodes.push_back(*b);
+		if (All == true)
+			return allCodes;
+		else
+			return codes;
 	}
 
 	std::vector<string> const lang = _config->FindVector("Acquire::Languages");
 	// the default setting -> "environment, en"
 	if (lang.empty() == true) {
-		if (envLongIncluded == false)
-			codes.push_back(envLong);
-		if (envShortIncluded == false)
-			codes.push_back(envShort);
+		codes = environment;
 		if (envShort != "en")
 			codes.push_back("en");
-		return codes;
+		allCodes = codes;
+		for (std::vector<string>::const_iterator b = builtin.begin();
+		     b != builtin.end(); ++b)
+			if (std::find(allCodes.begin(), allCodes.end(), *b) == allCodes.end())
+				allCodes.push_back(*b);
+		if (All == true)
+			return allCodes;
+		else
+			return codes;
 	}
 
 	// the configs define the order, so add the environment
@@ -191,32 +286,31 @@ std::vector<std::string> const Configuration::getLanguages(bool const &All,
 	for (std::vector<string>::const_iterator l = lang.begin();
 	     l != lang.end(); l++) {
 		if (*l == "environment") {
-			if (envLongIncluded == true && envShortIncluded == true)
-				continue;
-			if (envLongIncluded == false) {
-				envLongIncluded = true;
-				if (noneSeen == false)
-					codes.push_back(envLong);
-				allCodes.push_back(envLong);
-			}
-			if (envShortIncluded == false) {
-				envShortIncluded = true;
+			for (std::vector<string>::const_iterator e = environment.begin();
+			     e != environment.end(); ++e) {
+				if (std::find(allCodes.begin(), allCodes.end(), *e) != allCodes.end())
+					continue;
 				if (noneSeen == false)
-					codes.push_back(envShort);
-				allCodes.push_back(envShort);
+					codes.push_back(*e);
+				allCodes.push_back(*e);
 			}
 			continue;
 		} else if (*l == "none") {
 			noneSeen = true;
 			continue;
-		} else if ((envLongIncluded == true && *l == envLong) ||
-		         (envShortIncluded == true && *l == envShort))
+		} else if (std::find(allCodes.begin(), allCodes.end(), *l) != allCodes.end())
 			continue;
 
 		if (noneSeen == false)
 			codes.push_back(*l);
 		allCodes.push_back(*l);
 	}
+
+	for (std::vector<string>::const_iterator b = builtin.begin();
+	     b != builtin.end(); ++b)
+		if (std::find(allCodes.begin(), allCodes.end(), *b) == allCodes.end())
+			allCodes.push_back(*b);
+
 	if (All == true)
 		return allCodes;
 	else
diff --git a/apt-pkg/aptconfiguration.h b/apt-pkg/aptconfiguration.h
index f2f04a39b..2ba1b3825 100644
--- a/apt-pkg/aptconfiguration.h
+++ b/apt-pkg/aptconfiguration.h
@@ -64,7 +64,7 @@ public:									/*{{{*/
 	 *  \return a vector of (all) Language Codes in the prefered usage order
 	 */
 	std::vector<std::string> static const getLanguages(bool const &All = false,
-			bool const &Cached = true, char const * const Locale = 0);
+			bool const &Cached = true, char const ** const Locale = 0);
 
 									/*}}}*/
 };
diff --git a/apt-pkg/contrib/strutl.cc b/apt-pkg/contrib/strutl.cc
index 8e8820949..1b9922a31 100644
--- a/apt-pkg/contrib/strutl.cc
+++ b/apt-pkg/contrib/strutl.cc
@@ -1000,6 +1000,24 @@ bool TokSplitString(char Tok,char *Input,char **List,
    return true;
 }
 									/*}}}*/
+// ExplodeString - Split a string up into a vector			/*{{{*/
+// ---------------------------------------------------------------------
+/* This can be used to split a given string up into a vector, so the
+   propose is the same as in the method above and this one is a bit slower
+   also, but the advantage is that we an iteratable vector */
+vector<string> ExplodeString(string const &haystack, char const &split)
+{
+   string::const_iterator start = haystack.begin();
+   string::const_iterator end = start;
+   vector<string> exploded;
+   do {
+      for (; end != haystack.end() && *end != split; ++end);
+      exploded.push_back(string(start, end));
+      start = end + 1;
+   } while (end != haystack.end() && (++end) != haystack.end());
+   return exploded;
+}
+									/*}}}*/
 // RegexChoice - Simple regex list/list matcher				/*{{{*/
 // ---------------------------------------------------------------------
 /* */
diff --git a/apt-pkg/contrib/strutl.h b/apt-pkg/contrib/strutl.h
index a1e7f3934..e72288f4c 100644
--- a/apt-pkg/contrib/strutl.h
+++ b/apt-pkg/contrib/strutl.h
@@ -59,6 +59,7 @@ bool StrToNum(const char *Str,unsigned long &Res,unsigned Len,unsigned Base = 0)
 bool Hex2Num(const string &Str,unsigned char *Num,unsigned int Length);
 bool TokSplitString(char Tok,char *Input,char **List,
 		    unsigned long ListMax);
+vector<string> ExplodeString(string const &haystack, char const &split);
 void ioprintf(ostream &out,const char *format,...) APT_FORMAT2;
 void strprintf(string &out,const char *format,...) APT_FORMAT2;
 char *safe_snprintf(char *Buffer,char *End,const char *Format,...) APT_FORMAT3;
diff --git a/apt-pkg/deb/debmetaindex.cc b/apt-pkg/deb/debmetaindex.cc
index 8f28f053b..520e94a80 100644
--- a/apt-pkg/deb/debmetaindex.cc
+++ b/apt-pkg/deb/debmetaindex.cc
@@ -181,6 +181,7 @@ bool debReleaseIndex::GetIndexes(pkgAcquire *Owner, bool GetAll) const
       for (vector<string>::const_iterator l = lang.begin();
 		l != lang.end(); l++)
       {
+	if (*l == "none") continue;
 	debTranslationsIndex i = debTranslationsIndex(URI,Dist,(*I)->Section,(*l).c_str());
 	i.GetIndexes(Owner);
       }
@@ -219,8 +220,10 @@ vector <pkgIndexFile *> *debReleaseIndex::GetIndexFiles()
          Indexes->push_back(new debPackagesIndex (URI, Dist, (*I)->Section, IsTrusted()));
 
 	 for (vector<string>::const_iterator l = lang.begin();
-		l != lang.end(); l++)
+		l != lang.end(); l++) {
+	    if (*l == "none") continue;
 	    Indexes->push_back(new debTranslationsIndex(URI,Dist,(*I)->Section,(*l).c_str()));
+	 }
       }
    }
 
diff --git a/apt-pkg/deb/dpkgpm.cc b/apt-pkg/deb/dpkgpm.cc
index 9d144a227..fb6054f79 100644
--- a/apt-pkg/deb/dpkgpm.cc
+++ b/apt-pkg/deb/dpkgpm.cc
@@ -1063,7 +1063,7 @@ bool pkgDPkgPM::Go(int OutStatusFd)
 
 	 // wait for input or output here
 	 FD_ZERO(&rfds);
-	 if (!stdin_is_dev_null)
+	 if (master >= 0 && !stdin_is_dev_null)
 	    FD_SET(0, &rfds); 
 	 FD_SET(_dpkgin, &rfds);
 	 if(master >= 0)
diff --git a/apt-pkg/init.h b/apt-pkg/init.h
index f0757f644..b3e4b147f 100644
--- a/apt-pkg/init.h
+++ b/apt-pkg/init.h
@@ -22,7 +22,7 @@
 // Non-ABI-Breaks should only increase RELEASE number.
 // See also buildlib/libversion.mak
 #define APT_PKG_MAJOR 4
-#define APT_PKG_MINOR 8
+#define APT_PKG_MINOR 9
 #define APT_PKG_RELEASE 0
     
 extern const char *pkgVersion;