TagSection: Introduce functions for looking up by key ids

Introduce a new enum class and add functions that can do a lookup
with that enum class. This uses triehash.

4 files changed, 190 insertions, 11 deletions

diff --git a/apt-pkg/CMakeLists.txt b/apt-pkg/CMakeLists.txt
index 1b493c819..25ed13ec3 100644
--- a/apt-pkg/CMakeLists.txt
+++ b/apt-pkg/CMakeLists.txt
@@ -3,6 +3,19 @@ include_directories(${PROJECT_BINARY_DIR}/include/apt-pkg)
 
 add_definitions("-DAPT_PKG_EXPOSE_STRING_VIEW")
 
+file(MAKE_DIRECTORY ${PROJECT_BINARY_DIR}/include/apt-pkg/)
+execute_process(COMMAND ${PROJECT_SOURCE_DIR}/triehash/triehash.pl
+                        --ignore-case
+                         --header ${PROJECT_BINARY_DIR}/include/apt-pkg/tagfile-keys.h
+                         --code ${CMAKE_CURRENT_BINARY_DIR}/tagfile-keys.cc
+                         --enum-class
+                         --enum-name pkgTagSection::Key
+                         --function-name pkgTagHash
+                         --include "<apt-pkg/tagfile.h>"
+                         ${CMAKE_CURRENT_SOURCE_DIR}/tagfile-keys.list)
+set_property(DIRECTORY APPEND PROPERTY CMAKE_CONFIGURE_DEPENDS "tagfile-keys.list")
+
+
 # Set the version of the library
 execute_process(COMMAND awk -v ORS=. "/^\#define APT_PKG_M/ {print \$3}"
                 COMMAND sed "s/\\.\$//"
@@ -17,7 +30,7 @@ message(STATUS "Building libapt-pkg ${MAJOR} (release ${MINOR})")
 set(APT_PKG_MAJOR ${MAJOR} PARENT_SCOPE) # exporting for methods/CMakeLists.txt
 
 # Definition of the C++ files used to build the library
-file(GLOB_RECURSE library "*.cc")
+file(GLOB_RECURSE library "*.cc"  "${CMAKE_CURRENT_BINARY_DIR}/tagfile-keys.cc")
 file(GLOB_RECURSE headers "*.h")
 
 # Create a library using the C++ files
diff --git a/apt-pkg/tagfile-keys.list b/apt-pkg/tagfile-keys.list
new file mode 100644
index 000000000..82f0d8316
--- /dev/null
+++ b/apt-pkg/tagfile-keys.list
@@ -0,0 +1,82 @@
+Architecture
+Binary
+Breaks
+Bugs
+Build-Conflicts
+Build-Conflicts-Arch
+Build-Conflicts-Indep
+Build-Depends
+Build-Depends-Arch
+Build-Depends-Indep
+Built-For-Profiles
+Built-Using
+Checksums-Md5
+Checksums-Sha1
+Checksums-Sha256
+Checksums-Sha512
+Class
+Conffiles
+Config-Version
+Conflicts
+Depends
+Description
+Description-md5
+Directory
+Dm-Upload-Allowed
+Enhances
+Essential
+Filename
+Files
+Format
+Homepage
+Important
+Installed-Size
+Installer-Menu-Item
+Kernel-Version
+Maintainer
+MD5sum
+MSDOS-Filename
+Multi-Arch
+Optional
+Origin
+Original-Maintainer
+Package
+Package-List
+Package_Revision
+Package-Revision
+Package-Type
+Pre-Depends
+Priority
+Provides
+Recommended
+Recommends
+Replaces
+Revision
+Section
+SHA1
+SHA256
+SHA512
+Size
+Source
+Standards-Version
+Status
+Subarchitecture
+Suggests
+Tag
+Task
+Testsuite
+Testsuite-Triggers
+Triggers-Awaited
+Triggers-Pending
+Uploaders
+Vcs-Arch
+Vcs-Browse
+Vcs-Browser
+Vcs-Bzr
+Vcs-Cvs
+Vcs-Darcs
+Vcs-Git
+Vcs-Hg
+Vcs-Mtn
+Vcs-Svn
+Version
diff --git a/apt-pkg/tagfile.cc b/apt-pkg/tagfile.cc
index b81efe42a..200a7f2fd 100644
--- a/apt-pkg/tagfile.cc
+++ b/apt-pkg/tagfile.cc
@@ -14,6 +14,7 @@
 #include<config.h>
 
 #include <apt-pkg/tagfile.h>
+#include <apt-pkg/tagfile-keys.h>
 #include <apt-pkg/error.h>
 #include <apt-pkg/strutl.h>
 #include <apt-pkg/fileutl.h>
@@ -512,7 +513,8 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const R
 
    pkgTagSectionPrivate::TagData lastTagData(0);
    lastTagData.EndTag = 0;
-   unsigned long lastTagHash = 0;
+   Key lastTagKey = Key::Unknown;
+   unsigned int lastTagHash = 0;
    while (Stop < End)
    {
       TrimRecord(true,End);
@@ -528,11 +530,15 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const R
 	 // store the last found tag
 	 if (lastTagData.EndTag != 0)
 	 {
-	    if (BetaIndexes[lastTagHash] != 0)
-	       lastTagData.NextInBucket = BetaIndexes[lastTagHash];
-	    APT_IGNORE_DEPRECATED_PUSH
-	    BetaIndexes[lastTagHash] = TagCount;
-	    APT_IGNORE_DEPRECATED_POP
+	    if (lastTagKey != Key::Unknown) {
+	       AlphaIndexes[static_cast<size_t>(lastTagKey)] = TagCount;
+	    } else {
+	       if (BetaIndexes[lastTagHash] != 0)
+		  lastTagData.NextInBucket = BetaIndexes[lastTagHash];
+	       APT_IGNORE_DEPRECATED_PUSH
+	       BetaIndexes[lastTagHash] = TagCount;
+	       APT_IGNORE_DEPRECATED_POP
+	    }
 	    d->Tags.push_back(lastTagData);
 	 }
 
@@ -549,7 +555,9 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const R
 	    ;
 	 ++EndTag;
 	 lastTagData.EndTag = EndTag - Section;
-	 lastTagHash = BetaHash(Stop, EndTag - Stop);
+	 lastTagKey = pkgTagHash(Stop, EndTag - Stop);
+	 if (lastTagKey == Key::Unknown)
+	    lastTagHash = BetaHash(Stop, EndTag - Stop);
 	 // find the beginning of the value
 	 Stop = Colon + 1;
 	 for (; Stop < End && isspace_ascii(*Stop) != 0; ++Stop)
@@ -574,9 +582,13 @@ bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength, bool const R
       {
 	 if (lastTagData.EndTag != 0)
 	 {
-	    if (BetaIndexes[lastTagHash] != 0)
-	       lastTagData.NextInBucket = BetaIndexes[lastTagHash];
-	    APT_IGNORE_DEPRECATED(BetaIndexes[lastTagHash] = TagCount;)
+	    if (lastTagKey != Key::Unknown) {
+	       AlphaIndexes[static_cast<size_t>(lastTagKey)] = TagCount;
+	    } else {
+	       if (BetaIndexes[lastTagHash] != 0)
+		  lastTagData.NextInBucket = BetaIndexes[lastTagHash];
+	       APT_IGNORE_DEPRECATED(BetaIndexes[lastTagHash] = TagCount;)
+	    }
 	    d->Tags.push_back(lastTagData);
 	 }
 
@@ -620,10 +632,20 @@ bool pkgTagSection::Exists(StringView Tag) const
 // TagSection::Find - Locate a tag					/*{{{*/
 // ---------------------------------------------------------------------
 /* This searches the section for a tag that matches the given string. */
+bool pkgTagSection::Find(Key key,unsigned int &Pos) const
+{
+   auto Bucket = AlphaIndexes[static_cast<size_t>(key)];
+   Pos = Bucket - 1;
+   return Bucket != 0;
+}
 bool pkgTagSection::Find(StringView TagView,unsigned int &Pos) const
 {
    const char * const Tag = TagView.data();
    size_t const Length = TagView.length();
+   auto key = pkgTagHash(Tag, Length);
+   if (key != Key::Unknown)
+      return Find(key, Pos);
+
    unsigned int Bucket = BetaIndexes[BetaHash(Tag, Length)];
    if (Bucket == 0)
       return false;
@@ -664,6 +686,12 @@ bool pkgTagSection::Find(StringView Tag,const char *&Start,
    unsigned int Pos;
    return Find(Tag, Pos) && FindInternal(Pos, Start, End);
 }
+bool pkgTagSection::Find(Key key,const char *&Start,
+		         const char *&End) const
+{
+   unsigned int Pos;
+   return Find(key, Pos) && FindInternal(Pos, Start, End);
+}
 									/*}}}*/
 // TagSection::FindS - Find a string					/*{{{*/
 StringView pkgTagSection::Find(StringView Tag) const
@@ -674,6 +702,14 @@ StringView pkgTagSection::Find(StringView Tag) const
       return StringView();
    return StringView(Start, End - Start);
 }
+StringView pkgTagSection::Find(Key key) const
+{
+   const char *Start;
+   const char *End;
+   if (Find(key,Start,End) == false)
+      return StringView();
+   return StringView(Start, End - Start);
+}
 									/*}}}*/
 // TagSection::FindRawS - Find a string					/*{{{*/
 StringView pkgTagSection::FindRawInternal(unsigned int Pos) const
@@ -693,6 +729,11 @@ StringView pkgTagSection::FindRaw(StringView Tag) const
    unsigned int Pos;
    return Find(Tag, Pos) ? FindRawInternal(Pos) : "";
 }
+StringView pkgTagSection::FindRaw(Key key) const
+{
+   unsigned int Pos;
+   return Find(key, Pos) ? FindRawInternal(Pos) : "";
+}
 									/*}}}*/
 // TagSection::FindI - Find an integer					/*{{{*/
 // ---------------------------------------------------------------------
@@ -723,6 +764,12 @@ signed int pkgTagSection::FindIInternal(unsigned int Pos,signed long Default) co
       return Default;
    return Result;
 }
+signed int pkgTagSection::FindI(Key key,signed long Default) const
+{
+   unsigned int Pos;
+
+   return Find(key, Pos) ? FindIInternal(Pos) : Default;
+}
 signed int pkgTagSection::FindI(StringView Tag,signed long Default) const
 {
    unsigned int Pos;
@@ -753,6 +800,12 @@ unsigned long long pkgTagSection::FindULLInternal(unsigned int Pos, unsigned lon
       return Default;
    return Result;
 }
+unsigned long long pkgTagSection::FindULL(Key key, unsigned long long const &Default) const
+{
+   unsigned int Pos;
+
+   return Find(key, Pos) ? FindULLInternal(Pos, Default) : Default;
+}
 unsigned long long pkgTagSection::FindULL(StringView Tag, unsigned long long const &Default) const
 {
    unsigned int Pos;
@@ -770,6 +823,11 @@ bool pkgTagSection::FindBInternal(unsigned int Pos, bool Default) const
       return Default;
    return StringToBool(string(Start, Stop));
 }
+bool pkgTagSection::FindB(Key key, bool Default) const
+{
+   unsigned int Pos;
+   return Find(key, Pos) ? FindBInternal(Pos, Default): Default;
+}
 bool pkgTagSection::FindB(StringView Tag, bool Default) const
 {
    unsigned int Pos;
@@ -788,6 +846,14 @@ bool pkgTagSection::FindFlagInternal(unsigned int Pos, uint8_t &Flags,
       return true;
    return FindFlag(Flags, Flag, Start, Stop);
 }
+bool pkgTagSection::FindFlag(Key key, uint8_t &Flags,
+			     uint8_t const Flag) const
+{
+   unsigned int Pos;
+   if (Find(key,Pos) == false)
+      return true;
+   return FindFlagInternal(Pos, Flags, Flag);
+}
 bool pkgTagSection::FindFlag(StringView Tag, uint8_t &Flags,
 			     uint8_t const Flag) const
 {
@@ -824,6 +890,12 @@ bool pkgTagSection::FindFlagInternal(unsigned int Pos,unsigned long &Flags,
       return true;
    return FindFlag(Flags, Flag, Start, Stop);
 }
+bool pkgTagSection::FindFlag(Key key,unsigned long &Flags,
+			     unsigned long Flag) const
+{
+   unsigned int Pos;
+   return Find(key, Pos) ? FindFlagInternal(Pos, Flags, Flag) : true;
+}
 bool pkgTagSection::FindFlag(StringView Tag,unsigned long &Flags,
 			     unsigned long Flag) const
 {
diff --git a/apt-pkg/tagfile.h b/apt-pkg/tagfile.h
index cdca54756..3b89d0e45 100644
--- a/apt-pkg/tagfile.h
+++ b/apt-pkg/tagfile.h
@@ -89,7 +89,19 @@ class pkgTagSection
    std::string FindS(const char *Tag) const;
    std::string FindRawS(const char *Tag) const;
 
+   // Functions for lookup with a perfect hash function
+   enum class Key;
+   APT_HIDDEN bool Find(Key key,const char *&Start, const char *&End) const;
+   APT_HIDDEN bool Find(Key key,unsigned int &Pos) const;
+   APT_HIDDEN signed int FindI(Key key,signed long Default = 0) const;
+   APT_HIDDEN bool FindB(Key key, bool Default = false) const;
+   APT_HIDDEN unsigned long long FindULL(Key key, unsigned long long const &Default = 0) const;
+   APT_HIDDEN bool FindFlag(Key key,uint8_t &Flags, uint8_t const Flag) const;
+   APT_HIDDEN bool FindFlag(Key key,unsigned long &Flags, unsigned long Flag) const;
+   APT_HIDDEN bool Exists(Key key) const;
 #ifdef APT_PKG_EXPOSE_STRING_VIEW
+   APT_HIDDEN APT::StringView Find(Key key) const;
+   APT_HIDDEN APT::StringView FindRaw(Key key) const;
    APT_HIDDEN bool Find(APT::StringView Tag,const char *&Start, const char *&End) const;
    APT_HIDDEN bool Find(APT::StringView Tag,unsigned int &Pos) const;
    APT_HIDDEN APT::StringView Find(APT::StringView Tag) const;