replace every call to toupper with one to our own tolower_ascii

[apt.git] / apt-pkg / contrib / strutl.cc

diff --git a/apt-pkg/contrib/strutl.cc b/apt-pkg/contrib/strutl.cc
index 4c05f2df880a2200f7872d863fbdeef425f4dd75..ab47cdede0d93abeb305f1c2c8f3438de24af2c7 100644 (file)
--- a/apt-pkg/contrib/strutl.cc
+++ b/apt-pkg/contrib/strutl.cc
@@ -43,9 +43,10 @@ bool UTF8ToCodeset(const char *codeset, const string &orig, string *dest)
 {
   iconv_t cd;
   const char *inbuf;
-  char *inptr, *outbuf, *outptr;
-  size_t insize, outsize;
-  
+  char *inptr, *outbuf;
+  size_t insize, bufsize;
+  dest->clear();
+
   cd = iconv_open(codeset, "UTF-8");
   if (cd == (iconv_t)(-1)) {
      // Something went wrong
@@ -55,33 +56,49 @@ bool UTF8ToCodeset(const char *codeset, const string &orig, string *dest)
      else
        perror("iconv_open");
      
-     // Clean the destination string
-     *dest = "";
-     
      return false;
   }
 
-  insize = outsize = orig.size();
+  insize = bufsize = orig.size();
   inbuf = orig.data();
   inptr = (char *)inbuf;
-  outbuf = new char[insize+1];
-  outptr = outbuf;
+  outbuf = new char[bufsize];
+  size_t lastError = -1;
 
   while (insize != 0)
   {
+     char *outptr = outbuf;
+     size_t outsize = bufsize;
      size_t const err = iconv(cd, &inptr, &insize, &outptr, &outsize);
+     dest->append(outbuf, outptr - outbuf);
      if (err == (size_t)(-1))
      {
-       insize--;
-       outsize++;
-       inptr++;
-       *outptr = '?';
-       outptr++;
+       switch (errno)
+       {
+       case EILSEQ:
+          insize--;
+          inptr++;
+          // replace a series of unknown multibytes with a single "?"
+          if (lastError != insize) {
+             lastError = insize - 1;
+             dest->append("?");
+          }
+          break;
+       case EINVAL:
+          insize = 0;
+          break;
+       case E2BIG:
+          if (outptr == outbuf)
+          {
+             bufsize *= 2;
+             delete[] outbuf;
+             outbuf = new char[bufsize];
+          }
+          break;
+       }
      }
   }
 
-  *outptr = '\0';
-  *dest = outbuf;
   delete[] outbuf;
   
   iconv_close(cd);
@@ -549,7 +566,7 @@ int stringcmp(string::const_iterator A,string::const_iterator AEnd,
 int stringcasecmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
 {
    for (; A != AEnd && B != BEnd; A++, B++)
-      if (toupper(*A) != toupper(*B))
+      if (tolower_ascii(*A) != tolower_ascii(*B))
         break;
 
    if (A == AEnd && B == BEnd)
@@ -558,7 +575,7 @@ int stringcasecmp(const char *A,const char *AEnd,const char *B,const char *BEnd)
       return 1;
    if (B == BEnd)
       return -1;
-   if (toupper(*A) < toupper(*B))
+   if (tolower_ascii(*A) < tolower_ascii(*B))
       return -1;
    return 1;
 }
@@ -567,7 +584,7 @@ int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
                  const char *B,const char *BEnd)
 {
    for (; A != AEnd && B != BEnd; A++, B++)
-      if (toupper(*A) != toupper(*B))
+      if (tolower_ascii(*A) != tolower_ascii(*B))
         break;
 
    if (A == AEnd && B == BEnd)
@@ -576,7 +593,7 @@ int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
       return 1;
    if (B == BEnd)
       return -1;
-   if (toupper(*A) < toupper(*B))
+   if (tolower_ascii(*A) < tolower_ascii(*B))
       return -1;
    return 1;
 }
@@ -584,7 +601,7 @@ int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
                  string::const_iterator B,string::const_iterator BEnd)
 {
    for (; A != AEnd && B != BEnd; A++, B++)
-      if (toupper(*A) != toupper(*B))
+      if (tolower_ascii(*A) != tolower_ascii(*B))
         break;
 
    if (A == AEnd && B == BEnd)
@@ -593,7 +610,7 @@ int stringcasecmp(string::const_iterator A,string::const_iterator AEnd,
       return 1;
    if (B == BEnd)
       return -1;
-   if (toupper(*A) < toupper(*B))
+   if (tolower_ascii(*A) < tolower_ascii(*B))
       return -1;
    return 1;
 }
@@ -772,28 +789,28 @@ bool ReadMessages(int Fd, vector<string> &List)
 // MonthConv - Converts a month string into a number                   /*{{{*/
 // ---------------------------------------------------------------------
 /* This was lifted from the boa webserver which lifted it from 'wn-v1.07'
-   Made it a bit more robust with a few touppers though. */
+   Made it a bit more robust with a few tolower_ascii though. */
 static int MonthConv(char *Month)
 {
-   switch (toupper(*Month)) 
+   switch (tolower_ascii(*Month)) 
    {
-      case 'A':
-      return toupper(Month[1]) == 'P'?3:7;
-      case 'D':
+      case 'a':
+      return tolower_ascii(Month[1]) == 'p'?3:7;
+      case 'd':
       return 11;
-      case 'F':
+      case 'f':
       return 1;
-      case 'J':
-      if (toupper(Month[1]) == 'A')
+      case 'j':
+      if (tolower_ascii(Month[1]) == 'a')
         return 0;
-      return toupper(Month[2]) == 'N'?5:6;
-      case 'M':
-      return toupper(Month[2]) == 'R'?2:4;
-      case 'N':
+      return tolower_ascii(Month[2]) == 'n'?5:6;
+      case 'm':
+      return tolower_ascii(Month[2]) == 'r'?2:4;
+      case 'n':
       return 10;
-      case 'O':
+      case 'o':
       return 9;
-      case 'S':
+      case 's':
       return 8;
 
       // Pretend it is January..
@@ -983,6 +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                                /*{{{*/
 // ---------------------------------------------------------------------
 /* */
@@ -1098,10 +1133,13 @@ char *safe_snprintf(char *Buffer,char *End,const char *Format,...)
 
 // tolower_ascii - tolower() function that ignores the locale          /*{{{*/
 // ---------------------------------------------------------------------
-/* */
-int tolower_ascii(int c)
+/* This little function is the most called method we have and tries
+   therefore to do the absolut minimum - and is noteable faster than
+   standard tolower/toupper and as a bonus avoids problems with different
+   locales - we only operate on ascii chars anyway. */
+int tolower_ascii(int const c)
 {
-   if (c >= 'A' and c <= 'Z')
+   if (c >= 'A' && c <= 'Z')
       return c + 32;
    return c;
 }
@@ -1306,3 +1344,15 @@ string URI::SiteOnly(const string &URI)
    return U;
 }
                                                                        /*}}}*/
+// URI::NoUserPassword - Return the schema, site and path for the URI  /*{{{*/
+// ---------------------------------------------------------------------
+/* */
+string URI::NoUserPassword(const string &URI)
+{
+   ::URI U(URI);
+   U.User.clear();
+   U.Password.clear();
+   U.Port = 0;
+   return U;
+}
+                                                                       /*}}}*/