move ByHash into its own function

[apt.git] / apt-pkg / tagfile.h

diff --git a/apt-pkg/tagfile.h b/apt-pkg/tagfile.h
index 87d070d28e47d7ea876e5df6cdfedfd823ea043b..db43bfbf961da1d4983595b9edc851f4161f243a 100644 (file)
--- a/apt-pkg/tagfile.h
+++ b/apt-pkg/tagfile.h
@@ -20,30 +20,37 @@
 #ifndef PKGLIB_TAGFILE_H
 #define PKGLIB_TAGFILE_H
 
+#include <apt-pkg/macros.h>
 
-#include <apt-pkg/fileutl.h>
 #include <stdio.h>
 
+#include <string>
+#include <vector>
+#include <list>
+
+#ifndef APT_8_CLEANER_HEADERS
+#include <apt-pkg/fileutl.h>
+#endif
+
+class FileFd;
+
 class pkgTagSection
 {
    const char *Section;
-   // We have a limit of 256 tags per section.
-   unsigned int Indexes[256];
-   unsigned int AlphaIndexes[0x100];
-   unsigned int TagCount;
+   struct TagData {
+      unsigned int StartTag;
+      unsigned int EndTag;
+      unsigned int StartValue;
+      unsigned int NextInBucket;
+
+      TagData(unsigned int const StartTag) : StartTag(StartTag), NextInBucket(0) {}
+   };
+   std::vector<TagData> Tags;
+   unsigned int LookupTable[0x100];
+
    // dpointer placeholder (for later in case we need it)
    void *d;
 
-   /* This very simple hash function for the last 8 letters gives
-      very good performance on the debian package files */
-   inline static unsigned long AlphaHash(const char *Text, const char *End = 0)
-   {
-      unsigned long Res = 0;
-      for (; Text != End && *Text != ':' && *Text != 0; Text++)
-        Res = ((unsigned long)(*Text) & 0xDF) ^ (Res << 1);
-      return Res & 0xFF;
-   }
-
    protected:
    const char *Stop;
 
@@ -53,32 +60,55 @@ class pkgTagSection
    inline bool operator !=(const pkgTagSection &rhs) {return Section != rhs.Section;};
    
    bool Find(const char *Tag,const char *&Start, const char *&End) const;
-   bool Find(const char *Tag,unsigned &Pos) const;
-   string FindS(const char *Tag) const;
-   signed int FindI(const char *Tag,signed long Default = 0) const ;
+   bool Find(const char *Tag,unsigned int &Pos) const;
+   std::string FindS(const char *Tag) const;
+   signed int FindI(const char *Tag,signed long Default = 0) const;
+   bool FindB(const char *Tag, bool const &Default = false) const;
    unsigned long long FindULL(const char *Tag, unsigned long long const &Default = 0) const;
    bool FindFlag(const char *Tag,unsigned long &Flags,
                 unsigned long Flag) const;
-   bool static const FindFlag(unsigned long &Flags, unsigned long Flag,
+   bool static FindFlag(unsigned long &Flags, unsigned long Flag,
                                const char* Start, const char* Stop);
-   bool Scan(const char *Start,unsigned long MaxLength);
+
+   /** \brief searches the boundaries of the current section
+    *
+    * While parameter Start marks the beginning of the section, this method
+    * will search for the first double newline in the data stream which marks
+    * the end of the section. It also does a first pass over the content of
+    * the section parsing it as encountered for processing later on by Find
+    *
+    * @param Start is the beginning of the section
+    * @param MaxLength is the size of valid data in the stream pointed to by Start
+    * @param Restart if enabled internal state will be cleared, otherwise it is
+    *  assumed that now more data is available in the stream and the parsing will
+    *  start were it encountered insufficent data the last time.
+    *
+    * @return \b true if section end was found, \b false otherwise.
+    *  Beware that internal state will be inconsistent if \b false is returned!
+    */
+   APT_MUSTCHECK bool Scan(const char *Start, unsigned long MaxLength, bool const Restart = true);
    inline unsigned long size() const {return Stop - Section;};
    void Trim();
    virtual void TrimRecord(bool BeforeRecord, const char* &End);
-   
-   inline unsigned int Count() const {return TagCount;};
-   inline bool Exists(const char* const Tag) {return AlphaIndexes[AlphaHash(Tag)] != 0;}
- 
+
+   /** \brief amount of Tags in the current section
+    *
+    * Note: if a Tag is mentioned repeatly it will be counted multiple
+    * times, but only the last occurance is available via Find methods.
+    */
+   unsigned int Count() const;
+   bool Exists(const char* const Tag) const;
+
    inline void Get(const char *&Start,const char *&Stop,unsigned int I) const
-                   {Start = Section + Indexes[I]; Stop = Section + Indexes[I+1];}
-           
+                   {Start = Section + Tags[I].StartTag; Stop = Section + Tags[I+1].StartTag;}
+
    inline void GetSection(const char *&Start,const char *&Stop) const
    {
       Start = Section;
       Stop = this->Stop;
    };
    
-   pkgTagSection() : Section(0), Stop(0) {};
+   pkgTagSection();
    virtual ~pkgTagSection() {};
 };
 
@@ -87,8 +117,9 @@ class pkgTagFile
 {
    pkgTagFilePrivate *d;
 
-   bool Fill();
-   bool Resize();
+   APT_HIDDEN bool Fill();
+   APT_HIDDEN bool Resize();
+   APT_HIDDEN bool Resize(unsigned long long const newSize);
 
    public:
 
@@ -96,6 +127,8 @@ class pkgTagFile
    unsigned long Offset();
    bool Jump(pkgTagSection &Tag,unsigned long long Offset);
 
+   void Init(FileFd *F,unsigned long long Size = 32*1024);
+
    pkgTagFile(FileFd *F,unsigned long long Size = 32*1024);
    virtual ~pkgTagFile();
 };