path: root/autobuild.rb

#!/usr/bin/ruby

require 'fileutils'
require 'rubygems'
require 'sqlite3'
require_relative 'bugscan'

def create_tables(db)
    #db.execute "CREATE TABLE IF NOT EXISTS Packages(Id INTEGER PRIMARY KEY, Name TEXT, Version TEXT, Release TEXT, UNIQUE (Name, Version, Release))"
    db.execute "CREATE TABLE IF NOT EXISTS Packages(Id INTEGER PRIMARY KEY, Name TEXT, UNIQUE (Name))"
    db.execute "CREATE TABLE IF NOT EXISTS Runs(Id INTEGER PRIMARY KEY, Start INTEGER, End INTEGER, UNIQUE(Start))"
    db.execute "CREATE TABLE IF NOT EXISTS ResultValues(Id INTEGER PRIMARY KEY, Name TEXT, UNIQUE(Name))"
    db.execute "CREATE TABLE IF NOT EXISTS Results(Id INTEGER PRIMARY KEY, Package INTEGER, Run INTEGER, Result INTEGER,
		FOREIGN KEY(Package) REFERENCES Packages(Id),
		FOREIGN KEY(Run) REFERENCES Run(Id),
		FOREIGN KEY(Result) REFERENCES ResultValues(Id),
                UNIQUE(Package, Run, Result))"
    db.execute "CREATE TABLE IF NOT EXISTS Diffs(Run INTEGER PRIMARY KEY, New INTEGER, Fixed INTEGER, NotFixed INTEGER)"
    db.execute "CREATE TABLE IF NOT EXISTS Attributes(Id INTEGER PRIMARY KEY, Package INTEGER, Run INTEGER, Attr TEXT,
        FOREIGN KEY(Package) REFERENCES Packages(Id),
        FOREIGN KEY(Run) REFERENCES Run(Id),
        UNIQUE(Package, Run))"
end

def get_or_add_resultvalue(db, result)
    db.execute "INSERT OR IGNORE INTO ResultValues(Name) VALUES('#{result}')"
    return db.execute("SELECT Id FROM ResultValues WHERE Name = '#{result}'")[0][0]
end

def get_or_add_package(db, name, version, release)
    #db.execute "INSERT OR IGNORE INTO Packages(Name, Version, Release) VALUES('#{name}','#{version}','#{release}')"
    #return db.execute("SELECT Id FROM Packages WHERE Name = '#{name}' AND Version = '#{version}' AND Release = '#{release}'")[0][0]
    db.execute "INSERT OR IGNORE INTO Packages(Name) VALUES('#{name}')"
    return db.execute("SELECT Id FROM Packages WHERE Name = '#{name}'")[0][0]
end

# Find interesting attributes about the package. Right now, this is just interesting
# dependencies.
def find_attributes(build_dir)
    attr = {}
    rpm_list_files = Dir.glob(build_dir + '/rpm_qa.*.log')
    rpm_list_file = rpm_list_files[0]
    if rpm_list_file then
        File.open(rpm_list_file, 'r') {|f|
            f.each_line{|l|
                if l !~ /^(.*)-([^-]*)-([^-]*)/ then
                    puts "Skipping " + l
                    next
                end
                name = $1
                if name =~ /^(bwbasic|yabasic|mono-basic)$/ then
                    attr[:lang_basic] = true
                end
                if name =~ /^mono-core$/ then
                    attr[:lang_dotnet] = true
                end
                if name =~ /^erlang-compiler$/ then
                    attr[:lang_erlang] = true
                end
                if name =~ /^gcc-gfortran$/ then
                    attr[:lang_fortran] = true
                end
                if name =~ /^golang$/ then
                    attr[:lang_go] = true
                end
                if name =~ /^groovy$/ then
                    attr[:lang_groovy] = true
                end
                if name =~ /^ghc$/ then
                    attr[:lang_haskell] = true
                end
                if name =~ /^(java|gcc-java)/ then
                    attr[:lang_java] = true
                end
                if name =~ /^clisp-devel$/ then
                    attr[:lang_lisp] = true
                end
                if name =~ /^lib(64)?lua-devel$/ then
                    attr[:lang_lua] = true
                end
                if name =~ /^gcc-objc$/ then
                    attr[:lang_objc] = true
                end
                if name =~ /^(ocaml)-compiler$/ then
                    attr[:lang_ocaml] = true
                end
                if name =~ /^(nodejs|uglify-js|slimit|yuicompressor)$/ then
                    attr[:lang_javascript] = true
                end
                if name =~ /^fpc(-src)?$/ then
                    attr[:lang_pascal] = true
                end
                if name =~ /^(perl)/ then
                    # A bunch of perl modules are installed in every installation,
                    # so ignore those.
                    # A few are also pretty common and are ignored because of
                    # too many false positives:
                    #   perl-libintl-perl
                    #   perl-WWW-Curl (required by openssl)
                    if name !~ /^(perl|perl-base|perl-Config-IniFiles|perl-Exporter-Tiny|perl-File-Slurp|perl-File-Sync|perl-Filesys-Df|perl-IO-stringy|perl-JSON|perl-List-MoreUtils|perl-List-MoreUtils-XS|perl-Locale-gettext|perl-MDK-Common|perl-MDV-Distribconf|perl-MDV-Packdrakeng|perl-SGMLSpm|perl-srpm-macros|perl-Time-ZoneInfo|perl-URPM|perl-XML-LibXML|perl-XML-NamespaceSupport|perl-XML-SAX|perl-XML-SAX-Base|perl-YAML|perl-YAML-Tiny)$/ then
                        attr[:lang_perl] = true
                    end
                end
                if name =~ /^(php-devel|lib(64)?php_common)/ then
                    attr[:lang_php] = true
                end
                if name =~ /^lib(64)?python3?-devel$/ then
                    attr[:lang_python] = true
                end
                if name =~ /^(ruby-devel|lib(64)?ruby[0-9])/ then
                    attr[:lang_ruby] = true
                end
                if name =~ /^rust$/ then
                    attr[:lang_rust] = true
                end
                if name =~ /^tcl$/ then
                    attr[:lang_tcl] = true
                end
                if name =~ /^vala$/ then
                    attr[:lang_vala] = true
                end

                # Check for a few build systems, too
                if name =~ /^ant$/ then
                    attr[:build_ant] = true
                end
                if name =~ /^bazel$/ then
                    attr[:build_bazel] = true
                end
                if name =~ /^cmake$/ then
                    attr[:build_cmake] = true
                end
                if name =~ /^maven$/ then
                    attr[:build_maven] = true
                end
                if name =~ /^sbt$/ then
                    attr[:build_sbt] = true
                end
                if name =~ /^scons$/ then
                    attr[:build_scons] = true
                end
            }
        }
    end

    attr_line = ''
    attr.keys.map{|key| key.to_s}.sort.each do |key|
        attr_line = attr_line + ' ' + key
    end
    attr_line = attr_line.strip()
    return attr_line
end

# Find interesting things about the build, indicating a known bug
def find_bug_matches(build_dir)
    build_files = Dir.glob(build_dir + '/build.*.log')
    build_file = build_files[0]
    if build_file then
        # Big regex that matches everything we're looking for
        # Regexp.union doesn't work as it doesn't bracket each regex
        any_bug_re = Regexp.new('(' + $bug_matches.keys.join(')|(') + ')')
        regex_list = $bug_matches.keys.map{|key| key.to_s}

        # loop over file
        File.open(build_file).each_line do |li|
            begin
                if any_bug_re.match(li) then
                    # loop over individual regexes to find the matching one
                    regex_list.each do |regex|
                       return $bug_matches[regex] if (/#{regex}/.match(li))
                    end
                end
            rescue ArgumentError => e
              # Some lines have invalid UTF-8, just ignore them
            end
        end
        return ''
    end
end

# Find to what stage the build made it
def find_final_stage(build_dir)
    final_stage = 'stage_nobuild'
    build_files = Dir.glob(build_dir + '/build.*.log')
    build_file = build_files[0]
    if build_file then
        stage_re = Regexp.new('^Executing\(%(\w+)\)')
        final_stage = 'stage_preprep'

        # loop over file
        File.open(build_file).each_line do |li|
            begin
                match = stage_re.match(li)
                if match then
                  final_stage = 'stage_' + match[1]
                end
            rescue ArgumentError => e
              # Some lines have invalid UTF-8, just ignore them
            end
        end
    end
    return final_stage
end

def insert_run(db, status_file)
    t_start = Date.parse(File.basename(File.dirname(status_file))).strftime('%s')
    puts "Inserting data for run #{t_start} (#{status_file})"
    t_end = File.mtime(status_file).to_i
    db.execute "INSERT INTO Runs(Start, End) VALUES(#{t_start}, #{t_end})"
    run_id = db.last_insert_row_id
    File.open(status_file, 'r') {|f|
        db.transaction
        f.each_line{|l|
            if l !~ /^(.*)-([^-]*)-([^-]*).src.rpm: (.*)$/ then
                puts l
                next
            end
            name = $1
            version = $2
            release = $3
            result = $4
            if result == 'rejected' then
                next
            end
            result_id = get_or_add_resultvalue(db, result)
            package_id = get_or_add_package(db, name, version, release)
            begin
                db.execute "INSERT INTO Results(Package, Run, Result) VALUES(#{package_id}, #{run_id}, #{result_id})"
            rescue Exception => e
                puts name
            end
            # Store some interesting attributes about the build
            log_dir = File.dirname(status_file) + '/' + name + '-' + version + '-' + release + '.src.rpm'
            attr = find_attributes(log_dir)
            if result != 'ok' then
                bug_attr = find_bug_matches(log_dir)
                attr = attr + ' ' + bug_attr if bug_attr
                attr = attr + ' ' + find_final_stage(log_dir)
            end

            #puts "\t\t\tAttributes: " + attr
            if attr then
                begin
                    db.execute "INSERT INTO Attributes(Package, Run, Attr) VALUES(#{package_id}, #{run_id}, '#{attr.strip}')"
                rescue Exception => e
                    puts name
                end
            end
        }
        db.commit
    }
end

def drop_run(db, r)
    t_start = Date.parse(r).strftime('%s')
    run_id = db.execute("SELECT Id FROM Runs WHERE Start='#{t_start}'")[0][0]
    puts "Dropping run #{t_start}"
    db.execute "DELETE FROM Attributes WHERE Run = #{run_id}"
    db.execute "DELETE FROM Diffs WHERE Run = #{run_id}"
    db.execute "DELETE FROM Results WHERE Run = #{run_id}"
    db.execute "DELETE FROM Runs WHERE Id = #{run_id}"
end

def diff_runs(db, r1, r2)
    (newly_broken, fixed, still_broken) = (db.execute "SELECT New, Fixed, NotFixed FROM DIffs WHERE Run = #{r2}")[0]
    if newly_broken.nil? then
	    ok_id = get_or_add_resultvalue(db, 'ok')
	    not_on_this_arch_id = get_or_add_resultvalue(db, 'not_on_this_arch')
	    failure_query = "Result NOT IN (#{ok_id}, #{not_on_this_arch_id})"
	    newly_broken = (db.execute "SELECT count(Id) FROM Results WHERE Run = #{r2} AND #{failure_query} AND Package NOT IN (SELECT Package FROM Results WHERE Run = #{r1} AND #{failure_query})")[0][0]
	    fixed = (db.execute "SELECT count(Results.Id) FROM Results,Packages WHERE Run = #{r2} AND Package = Packages.Id AND (Result = #{ok_id} OR Result = #{not_on_this_arch_id}) AND Name IN (SELECT Name FROM Results,Packages WHERE Run = #{r1} AND #{failure_query} AND Package = Packages.Id)")[0][0]
	    still_broken = (db.execute "SELECT count(Id) FROM Results WHERE Run = #{r2} AND #{failure_query} AND Package IN (SELECT Package FROM Results WHERE Run = #{r1} AND #{failure_query})")[0][0]
	    db.execute "INSERT INTO Diffs(Run, New, Fixed, NotFixed) VALUES(#{r2},#{newly_broken},#{fixed},#{still_broken})"
    end
    { 'new' => newly_broken, 'fixed' => fixed, 'not fixed' => still_broken }
end

def report(db)
    resultvalues = db.execute("SELECT Name FROM ResultValues").flatten
    runs = db.execute("SELECT Id, Start FROM Runs ORDER BY Start ASC")
    results = {}
    prev_run = 0
    runs.each{|r|
	run_id = r[0]
	results[run_id] = {}
        data = db.execute "SELECT count(Results.Id), ResultValues.Name FROM Results, ResultValues WHERE Results.Result = ResultValues.Id and Run = #{run_id} GROUP BY ResultValues.Name"
	total = 0
	success = 0
	failure = 0
	data.each{|d|
            results[run_id][d[1]] = d[0]
	    total += d[0]
	    if (d[1] == "ok")
                success += d[0]
	    elsif (d[1] == "build_failure" || d[1] == "missing_dep" || d[1] == "install_deps_failure" || d[1] == "recreate_srpm_failure")
		failure += d[0]
	    end
	}
	results[run_id]['success'] = success
	results[run_id]['failure'] = failure
	results[run_id]['total'] = total
	results[run_id]['success_rate'] = (1000*success/(success+failure)).round.to_f/10
	results[run_id].merge!(diff_runs(db, prev_run, run_id))
	prev_run = run_id
    }
    results
end

def export_data(db, target_dir)	
    resultvalues = db.execute("SELECT Name FROM ResultValues").flatten
    runs = db.execute("SELECT Id, Start FROM Runs ORDER BY Start DESC")
    results = report(db)
    File.open(target_dir + "/data.js", 'w') {|df|    
    df.puts "var data = { cols: [ {id: 'A', label: 'Date', type: 'date'},"
    col = 'A'
    resultvalues.each{|v| df.puts "{id: '#{col.next!}', label: '#{v}', type: 'number'}," }
    df.puts "{id: '#{col.next!}', label: 'Total', type: 'number'},"
    df.puts "{id: '#{col.next!}', label: 'Success Rate', type: 'number'},"
    df.puts "{id: '#{col.next!}', label: 'Newly Broken', type: 'number'},"
    df.puts "{id: '#{col.next!}', label: 'Fixed', type: 'number'},"
    df.puts "{id: '#{col.next!}', label: 'Not Fixed', type: 'number'},"
    df.puts  "],"
    df.puts  "rows: ["
    runs.each{|r|
	result = results[r[0]]
	date_val = "new Date(#{DateTime.strptime(r[1].to_s, '%s').strftime('%Y, %m-1, %d')}, 0, 0, 0)"
	date_str = DateTime.strptime(r[1].to_s, '%s').strftime('%Y-%m-%d')
	df.print "{c:[{v: #{date_val}, f: '<a href=\"results.php?run=#{date_str}\">#{date_str}</a>'},"
        resultvalues.each{|v| df.print "{v: #{result[v] || 0}},"}
	df.print "{v: #{result['total']}},"
	df.print "{v: #{result['success_rate']}, f: '#{result['success_rate']}%'},"
	df.print "{v: #{result['new']}},"
	df.print "{v: #{result['fixed']}},"
	df.print "{v: #{result['not fixed']}},"
	df.puts "]},"
    }
    df.puts "]}"
    }
    FileUtils.cp("autobuild.db", target_dir)
end

def finish
	# get date of finished one
	# olddate = readlink current latest
	# change link to point to newdate
	# cleanlogs.sh olddate newdate
	# insert newdate
end

begin 
    db = SQLite3::Database.open "autobuild.db"
    create_tables(db)
    if (ARGV[0] == 'insert') then
        insert_run(db, ARGV[1])
    elsif (ARGV[0] == 'report')
        export_data(db, ARGV[1])
    elsif (ARGV[0] == 'drop')
        drop_run(db, ARGV[1])
    end
rescue SQLite3::Exception => e 
    
    puts "Exception occured"
    puts e
    
ensure
    db.close if db
end