Changeset 1376

Timestamp:

Jun 14, 2011, 2:02:11 AM (10 years ago)

Author:

Peter Johansson

Message:

closes #385. implementing cache for svncopyright

Location:

trunk

Files:

• lib/CopyrightStats.cc (modified) (7 diffs)
• lib/CopyrightStats.h (modified) (2 diffs)
• lib/CopyrightVisitor.cc (modified) (4 diffs)
• lib/CopyrightVisitor.h (modified) (3 diffs)
• test/Makefile.am (modified) (1 diff)
• test/copyright_cache_test.sh (modified) (2 diffs)
• yat/Makefile.am (modified) (2 diffs)
• yat/dummie.cc (added)

trunk/lib/CopyrightStats.cc

@@ -25 +25 @@
 #include "SVNinfo.h"
 #include "SVNlog.h"
+#include "utility.h"
+
+#include <yat/utility/utility.h>
 
 #include <subversion-1/svn_types.h>
 
 #include <cassert>
+#include <fstream>
 #include <iostream>
 
@@ -35 +39 @@
 
   CopyrightStats::CopyrightStats(const std::string& path, bool ignore_cache,
-                                 std::map<std::string, Alias>& author2alias,
                                  const std::map<int, svn_revnum_t>& year2rev)
-    : author2alias_(author2alias), path_(path)
-  {
+    : path_(path)
+  {
+    cache_file_ = concatenate_path(concatenate_path(directory_name(path), 
+                                                    ".svndigest"),
+                                   file_name(path)+".svncopyright-cache");
     init(ignore_cache, year2rev);
   }
@@ -48 +54 @@
     svn_revnum_t cache_rev = 0;
     if (!ignore_cache)
-      cache_rev = load_cache("");
+      cache_rev = load_cache();
     SVNinfo info(path_);
     if (cache_rev >= info.last_changed_rev())
@@ -58 +64 @@
 
     parse(cache_rev+1, year2rev);
-    write_cache("");
-  }
-
-
-  svn_revnum_t CopyrightStats::load_cache(const std::string& filename)
-  {
-    // FIXME
+    write_cache();
+  }
+
+
+  svn_revnum_t CopyrightStats::load_cache(void)
+  {
+    std::ifstream is(cache_file_.c_str());
+    assert(is.good());
+
+    std::string line;
+    getline(is, line);
+    if (line!="SVNCOPYRIGHT CACHE VERSION 1")
+      return 0;
+    getline(is, line);
+    if (line!=config_) {
+      std::cout << "cache file is for different configuration.\n"
+                << "config code: '" << config_ << "'\n"
+                << "config code in cache file: '" << line << "'\n"
+                << "retrieving statistics from repository.\n";
+      return 0;
+    }
+
+    svn_revnum_t rev = 0;
+    try {
+      getline(is, line);
+      size_t nof_years = yat::utility::convert<size_t>(line);
+      for (size_t i=0; i<nof_years; ++i) {
+        getline(is, line);
+        int year = yat::utility::convert<size_t>(line);
+        std::set<std::string>& users = year2user_[year];
+        getline(is, line);
+        size_t nof_users = yat::utility::convert<size_t>(line);
+        for (size_t i=0; i<nof_users; ++i) {
+          getline(is, line);
+          users.insert(line);
+        }
+      }
+      getline(is, line);
+      rev = yat::utility::convert<svn_revnum_t>(line);
+      getline(is, line);
+      if (line!="SVNCOPYRIGHT CACHE")
+        return 0;
+      return rev;
+    }
+    catch (yat::utility::runtime_error e) {
+      return 0;
+    }
     return 0;
   }
 
 
-  const std::map<int, std::set<Alias> > CopyrightStats::map(void) const
-  {
-    return year2alias_;
+  const std::map<int, std::set<std::string> >& CopyrightStats::map(void) const
+  {
+    return year2user_;
   }
 
@@ -97 +143 @@
       
       const std::string& name = commit->author();
-      // find username in map of aliases
-      std::map<std::string, Alias>::iterator alias = 
-        author2alias_.lower_bound(name);
-      
-      // if alias not found for author
-      if (alias == author2alias_.end() || alias->first!=name) {
-        std::cerr << "svncopyright: warning: no copyright alias found for `" 
-                  << name << "'\n";
-        // insert alias to avoid multiple warnings.
-        Alias a(name, author2alias_.size());
-        // FIXME use insert with hint
-        author2alias_[name] = a;
-        alias = author2alias_.find(name);
-      }
-      assert(alias!=author2alias_.end());
-      
       // skip if alias already has copyright for this year.
-      std::map<int, std::set<Alias> >::const_iterator year_aliases = 
-        year2alias_.find(yearrev->first);
-      if (year_aliases!=year2alias_.end() 
-          && year_aliases->second.count(alias->second))
+      std::map<int, std::set<std::string> >::const_iterator year_users = 
+        year2user_.find(yearrev->first);
+      if (year_users!=year2user_.end() 
+          && year_users->second.count(name))
         continue;
       
@@ -128 +158 @@
         if ((lt==LineTypeParser::code || lt==LineTypeParser::comment) && 
             svn_blame.revision()==commit->revision()) {
-          year2alias_[yearrev->first].insert(alias->second);
+          year2user_[yearrev->first].insert(name);
           break;
         }
@@ -139 +169 @@
   void CopyrightStats::reset(void)
   {
-    year2alias_.clear();
-  }
-
-
-  void CopyrightStats::write_cache(const std::string& filename)
-  {
-    // FIXME
+    year2user_.clear();
+  }
+
+
+  void CopyrightStats::write_cache(void)
+  {
+    mkdir_p(directory_name(cache_file_));
+    std::ofstream os(cache_file_.c_str());
+    assert(os.good());
+
+    os << "SVNCOPYRIGHT CACHE VERSION 1\n";
+    os << config_ << "\n";
+    os << year2user_.size() << "\n";
+    using std::map;
+    using std::set;
+    using std::string;
+    for (map<int, set<string> >::const_iterator i=year2user_.begin();
+         i!=year2user_.end(); ++i) {
+      os << i->first << "\n";
+      os << i->second.size() << "\n";
+      for (set<string>::const_iterator j=i->second.begin(); 
+           j!=i->second.end(); ++j) {
+        os << *j << "\n";
+      }
+    }
+
+    SVNinfo info(path_);
+    os << info.last_changed_rev() << "\n";
+    os << "SVNCOPYRIGHT CACHE\n";
+    os.close();
   }
 

trunk/lib/CopyrightStats.h

@@ -43 +43 @@
      */
     CopyrightStats(const std::string& path, bool ignore_cache,
-                   std::map<std::string, Alias>& author2alias,
                    const std::map<int, svn_revnum_t>& year2rev);
 
-    const std::map<int, std::set<Alias> > map(void) const;
+    const std::map<int, std::set<std::string> >& map(void) const;
   private:
     /**
@@ -55 +54 @@
 
     /// return 0 if load failed; otherwise return rev cache represents
-    svn_revnum_t load_cache(const std::string& filename);
+    svn_revnum_t load_cache(void);
 
     void parse(svn_revnum_t rev, const std::map<int, svn_revnum_t>& year2rev);
     void reset(void);
-    void write_cache(const std::string& filename);
+    void write_cache(void);
 
-    std::map<std::string, Alias>& author2alias_;
+    std::string cache_file_;
+    std::string config_;
     std::string path_;
-    std::map<int, std::set<Alias> > year2alias_;
+    std::map<int, std::set<std::string> > year2user_;
   };
 }} // end of namespace svndigest and namespace theplu

trunk/lib/CopyrightVisitor.cc

@@ -75 +75 @@
     }
     return ss.str();
-  }
-
-
-  void CopyrightVisitor::create_year2alias(std::map<int, std::set<Alias> >& m,
-                                           const File& file)
-  {
-    using namespace std;
-    const Stats& stats = file.stats()["add"];
-
-    // loop over all years
-    for (std::map<int, svn_revnum_t>::const_iterator rev_iter=year2rev_.begin();
-         rev_iter!=year2rev_.end(); ++rev_iter) {
-
-      svn_revnum_t last_rev_this_year = rev_iter->second;
-      svn_revnum_t last_rev_last_year = 0;
-      if (rev_iter != year2rev_.begin()) {
-        last_rev_last_year = (--rev_iter)->second;
-        ++rev_iter;
-      }
-      // do not go beyond BASE rev of file
-      last_rev_this_year = std::min(last_rev_this_year,file.last_changed_rev());
-      last_rev_last_year = std::min(last_rev_last_year,file.last_changed_rev());
-      // loop over authors
-      for (std::set<std::string>::const_iterator a_iter=stats.authors().begin();
-           a_iter!=stats.authors().end(); ++a_iter) {
-
-        // check if anything has been added since last year
-        if ( (stats(LineTypeParser::code, *a_iter, last_rev_this_year) >
-              stats(LineTypeParser::code, *a_iter, last_rev_last_year)) || 
-             (stats(LineTypeParser::comment, *a_iter, last_rev_this_year) >
-              stats(LineTypeParser::comment, *a_iter, last_rev_last_year)) ) {
-        
-        
-          // find username in map of aliases
-          std::map<string,Alias>::iterator name(alias_.lower_bound(*a_iter));
-          
-          // if alias exist insert alias
-          if (name != alias_.end() && name->first==*a_iter)
-            m[rev_iter->first].insert(name->second);
-          else {
-            // else insert user name
-            Alias a(*a_iter,alias_.size());
-            m[rev_iter->first].insert(a);
-            std::cerr << "svncopyright: warning: no copyright alias found for `" 
-                      << *a_iter << "'\n";
-            // insert alias to avoid multiple warnings.
-            alias_.insert(name, std::make_pair(*a_iter, a));
-          }
-        }
-      }
-    }
   }
 
@@ -178 +127 @@
     if (verbose_)
       std::cout << "Parsing '" << file.path() << "'\n";
-    CopyrightStats stats(file.path(), ignore_cache_, alias_, year2rev_);
-    const std::map<int, std::set<Alias> >& map = stats.map();
-    assert(!map.empty());
-    std::string new_block = copyright_block(map, prefix);
+    CopyrightStats stats(file.path(), ignore_cache_, year2rev_);
+    const std::map<int, std::set<std::string> >& year2users = stats.map();
+    assert(!year2users.empty());
+    std::map<int, std::set<Alias> > year2alias;
+    translate(year2users, year2alias);
+    std::string new_block = copyright_block(year2alias, prefix);
     if (old_block==new_block)
       return;
@@ -231 +182 @@
   }
 
+
   void CopyrightVisitor::visit(File& file)
   {
@@ -242 +194 @@
   }
 
+
+  void CopyrightVisitor::translate(const std::set<std::string>& users, 
+                                   std::set<Alias>& aliases)
+  {
+    for (std::set<std::string>::const_iterator user=users.begin(); 
+         user!=users.end(); ++user) {
+      std::map<std::string, Alias>::const_iterator i = alias_.find(*user);
+      // if alias not found for author
+      if (i==alias_.end()) {
+        std::cerr << "svncopyright: warning: no copyright alias found for `" 
+                  << *user << "'\n";
+        // insert alias to avoid multiple warnings.
+        Alias a(*user, alias_.size());
+        alias_[*user] = a;
+      }
+      else {
+        // FIXME: perhaps use hint
+        aliases.insert(i->second);
+      }
+    }
+  }
+
+
+  void 
+  CopyrightVisitor::translate(const std::map<int, std::set<std::string> >& y2u, 
+                              std::map<int, std::set<Alias> >& y2a)
+  {
+    using std::map;
+    using std::set;
+    using std::string;
+    for (map<int, set<string> >::const_iterator yu=y2u.begin(); 
+         yu!=y2u.end();++yu) {
+      set<Alias>& alias = y2a[yu->first];
+      translate(yu->second, alias);
+    }
+  }
+      
 }} // end of namespace svndigest and namespace theplu

trunk/lib/CopyrightVisitor.h

@@ -31 +31 @@
 #include <set>
 #include <string>
+#include <vector>
 
 namespace theplu{
@@ -78 +79 @@
                                 const std::string& prefix) const;
 
-    /**
-       Create a map from year to set of authors.
-     */
-    void create_year2alias(std::map<int, std::set<Alias> >&, const File& file);
 
     /**
-       Look from copyright block in file \a path.
+       Look for copyright block in file \a path.
 
        \param path file to look for copyright
@@ -113 +110 @@
     void update_copyright(const std::string& path, const std::string& block,
                           size_t start_at_line, size_t end_at_line) const;
+
+
+    /**
+       Translating a set of users to a set of aliases using mapping in alias_
+    */
+    void translate(const std::set<std::string>&, std::set<Alias>&);
+
+    /**
+       Translating each year
+     */
+    void translate(const std::map<int, std::set<std::string> >& year2user,
+                   std::map<int, std::set<Alias> >& year2alias);
+
+
   };
 }} // end of namespace svndigest and namespace theplu

trunk/test/Makefile.am

@@ -58 +58 @@
 
 # tests not yet passing are listed here
-XFAIL_TESTS = copyright_cache_test.sh
+XFAIL_TESTS = 
 
 noinst_HEADERS = Suite.h

trunk/test/copyright_cache_test.sh

@@ -56 +56 @@
 # modify the cache file to allow us to detect that it is used
 sed -i 's/peter/winston/' $cache_file
+cat $cache_file
 # create a config reflecting this name change
 sed 's/peter/winston/' config > config2
@@ -66 +67 @@
 diff -u correct.txt copyright.txt || exit_fail
 
-diff -u cache.txt $cache_file || exit_fail
-
 exit_success;

trunk/yat/Makefile.am

@@ -3 +3 @@
 ## $Id$
 
-# Copyright (C) 2009, 2010 Peter Johansson
+# Copyright (C) 2009, 2010, 2011 Peter Johansson
 #
 # This file is part of svndigest, http://dev.thep.lu.se/svndigest
@@ -43 +43 @@
 libyat_a_SOURCES += ColumnStream.cc
 libyat_a_SOURCES += CommandLine.cc
+libyat_a_SOURCES += dummie.cc
 libyat_a_SOURCES += Exception.cc
 libyat_a_SOURCES += Option.cc