tests/gedcoms/TGC551.ged tests/gedcoms/TGC551LF.ged tests/gedcoms/TGC55C.ged tests/gedcoms/TGC55CLF.ged tests/gedcoms/george_ross.ged tests/gedcoms/mini1-ansel.ged tests/gedcoms/mini1-ansi.ged tests/gedcoms/mini1-unicode.ged tests/gedcoms/mini1-utf8.ged tests/gedcoms/simple.ged tests/parser_spec.rb @@ -17,131 +17,109 @@ # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA # ------------------------------------------------------------------------- -#require '_gedcom' require 'gedcom_date' module GEDCOM # Possibly a better way to do this? - VERSION = "0.0.1" + VERSION = "0.2.1" class Parser - def defaultHandler( data, cookie, parm ) + def initialize &block + @before = {} + @after = {} + @ctxStack = [] + @dataStack = [] + @curlvl = -1 + instance_eval(&block) if block_given? end - def initialize( cookie = nil ) - @cookie = cookie - @pre_handler = Hash.new( [ method( "defaultHandler" ), nil ] ) - @post_handler = Hash.new( [ method( "defaultHandler" ), nil ] ) + def before tag, proc=nil, &block + proc = check_proc_or_block proc, &block + @before[[tag].flatten] = proc end - def setPreHandler( context, func, parm = nil ) - @pre_handler[ context ] = [ func, parm ] + def after tag, proc=nil, &block + proc = check_proc_or_block proc, &block + @after[[tag].flatten] = proc end - def setPostHandler( context, func, parm = nil ) - @post_handler[ context ] = [ func, parm ] - end - - def callPreHandler( context, data, cookie ) - func, parm = @pre_handler[ context ] - func.call( data, cookie, parm ) + def parse( file ) + case file + when String + parse_file(file) + when IO + parse_io(file) + else + raise ArgumentError.new("requires a String or IO") + end end - def callPostHandler( context, data, cookie ) - func, parm = @post_handler[ context ] - func.call( data, cookie, parm ) + def context + @ctxStack end - # The parser is based on a stack. Every time a new level is found that is - # greater than the level of the previously seen item, it is pushed onto the - # stack. If the next item seen is of a lower level than previously seen - # items, those previously seen items are popped off the stack and their post - # handlers are called. - def parse( file ) - ctxStack = [] - dataStack = [] - curlvl = -1 - File.open( file, "r" ) do |f| - f.each_line do |line| - level, tag, rest = line.chop.split( ' ', 3 ) - while level.to_i <= curlvl - callPostHandler( ctxStack, dataStack.last, @cookie ) - ctxStack.pop - dataStack.pop - curlvl -= 1 - end - - tag, rest = rest, tag if tag =~ /@.*@/ - - ctxStack.push tag - dataStack.push rest - curlvl = level.to_i - - callPreHandler( ctxStack, dataStack.last, @cookie ) - end + protected + + def check_proc_or_block proc, &block + unless proc or block_given? + raise ArgumentError.new("proc or block required") end + proc = method(proc) if proc.kind_of? Symbol + proc ||= Proc.new(&block) end - end - - class DatePart - def <=>( dp ) - return -1 if has_year? and !dp.has_year? - return 1 if !has_year? and dp.has_year? - if has_year? and dp.has_year? - rc = ( year <=> dp.year ) - return rc unless rc == 0 + def parse_file(file) + File.open( file, "r" ) do |io| + parse_io(io) end + end - return -1 if dp.has_month? and !dp.has_month? - return 1 if !dp.has_month? and dp.has_month? + def parse_io(io) + io.each_line do |line| + level, tag, rest = line.chop.split( ' ', 3 ) + level = level.to_i + unwind_to level - if has_month? and dp.has_month? - rc = ( month <=> dp.month ) - return rc unless rc == 0 - end + tag, rest = rest, tag if tag =~ /@.*@/ - return -1 if dp.has_day? and !dp.has_day? - return 1 if !dp.has_day? and dp.has_day? + @ctxStack.push tag + @dataStack.push rest + @curlvl = level - if has_day? and dp.has_day? - rc = ( day <=> dp.day ) - return rc unless rc == 0 + do_before @ctxStack, rest end - - return 0 + unwind_to -1 end - end - - class Date - def Date.safe_new( parm ) - Date.new( parm ) { |errmsg| } + + def unwind_to level + while level <= @curlvl + do_after @ctxStack, @dataStack.last + @ctxStack.pop + @dataStack.pop + @curlvl -= 1 + end end - def <=>( d ) - if is_date? and d.is_date? - rc = ( first <=> d.first ) - return rc unless rc == 0 - - if is_range? and d.is_range? - return ( last <=> d.last ) - elsif is_range? - return 1 - elsif d.is_range? - return -1 - end - - return 0 - elsif is_date? - return -1 - elsif d.is_date? - return 1 + def do_before tag, data + if proc = @before[tag] + proc.call data + elsif proc = @before[ANY] + proc.call tag, data end + end - return format <=> d.format + def do_after tag, data + if proc = @after[tag] + proc.call data + elsif proc = @after[ANY] + proc.call tag, data + end end - end -end + + ANY = [:any] + end #/ Parser + +end #/ GEDCOM lib/gedcom.rb @@ -19,6 +19,7 @@ # require 'gedcom_date_parser' module GEDCOM + class DatePart < GEDCOM_DATE_PARSER::GEDDate # Flags @@ -97,7 +98,34 @@ module GEDCOM GEDCOM_DATE_PARSER::DateParser.build_gedcom_date_part_string( self ) end - end + def <=>( dp ) + return -1 if has_year? and !dp.has_year? + return 1 if !has_year? and dp.has_year? + + if has_year? and dp.has_year? + rc = ( year <=> dp.year ) + return rc unless rc == 0 + end + + return -1 if dp.has_month? and !dp.has_month? + return 1 if !dp.has_month? and dp.has_month? + + if has_month? and dp.has_month? + rc = ( month <=> dp.month ) + return rc unless rc == 0 + end + + return -1 if dp.has_day? and !dp.has_day? + return 1 if !dp.has_day? and dp.has_day? + + if has_day? and dp.has_day? + rc = ( day <=> dp.day ) + return rc unless rc == 0 + end + + return 0 + end + end #/ DatePart class Date < GEDCOM_DATE_PARSER::GEDDateValue # Calendar types @@ -126,6 +154,10 @@ module GEDCOM DNSCAN = GEDCOM_DATE_PARSER::GCDNSCAN DEAD = GEDCOM_DATE_PARSER::GCDEAD + def Date.safe_new( parm ) + Date.new( parm ) { |errmsg| } + end + def initialize ( date_str, calendar=DateType::DEFAULT ) begin @date1 = DatePart.new @@ -173,6 +205,28 @@ module GEDCOM (@flags & (BETWEEN | FROMTO)) != 0 ? true : false end + def <=>( d ) + if is_date? and d.is_date? + rc = ( first <=> d.first ) + return rc unless rc == 0 + + if is_range? and d.is_range? + return ( last <=> d.last ) + elsif is_range? + return 1 + elsif d.is_range? + return -1 + end + + return 0 + elsif is_date? + return -1 + elsif d.is_date? + return 1 + end + + return format <=> d.format + end end class DateType @@ -188,4 +242,4 @@ module GEDCOM class DateFormatException < Exception end -end \ No newline at end of file +end lib/gedcom_date.rb @@ -39,39 +39,33 @@ class Birthday end -class BirthdayExtractor < GEDCOM::Parser +class BirthdayExtracter < GEDCOM::Parser def initialize super - - setPreHandler [ "INDI" ], method( :startPerson ) - setPreHandler [ "INDI", "NAME" ], method( :registerName ) - setPreHandler [ "INDI", "BIRT", "DATE" ], method( :registerBirthdate ) - setPostHandler [ "INDI" ], method( :endPerson ) - @currentPerson = nil @allBirthdays = [] - end - - def startPerson( data, state, parm ) - @currentPerson = Birthday.new - end + + before %(INDI) do + @currentPerson = Birthday.new + end - def registerName( data, state, parm ) - @currentPerson.name = data if @currentPerson.name == nil - end + before %w(INDI NAME) do |data| + @currentPerson.name = data if @currentPerson.name == nil + end - def registerBirthdate( data, state, parm ) - if @currentPerson.date == nil - d = GEDCOM::Date.safe_new( data ) - if d.is_date? and d.first.has_year? and d.first.has_month? - @currentPerson.date = d + before %w(INDI BIRT DATE) do |data| + if @currentPerson.date == nil + d = GEDCOM::Date.safe_new( data ) + if d.is_date? and d.first.has_year? and d.first.has_month? + @currentPerson.date = d + end end end - end - def endPerson( data, state, parm ) - @allBirthdays.push @currentPerson if @currentPerson.date != nil - @currentPerson = nil + after %w(INDI) do + @allBirthdays.push @currentPerson if @currentPerson.date != nil + @currentPerson = nil + end end def showPeopleBornIn( month ) @@ -88,7 +82,9 @@ class BirthdayExtractor < GEDCOM::Parser def showPeopleBornOn( month, day ) count = 0 @allBirthdays.sort.each do |ind| - if ind.date.first.month == month && ind.date.first.has_day? && ind.date.first.day == day + if ind.date.first.month == month && + ind.date.first.has_day? && + ind.date.first.day == day count += 1 showPerson( ind ) end @@ -110,7 +106,7 @@ end puts "Parsing '#{ARGV[0]}'..." -parser = BirthdayExtractor.new +parser = BirthdayExtracter.new parser.parse ARGV[0] now = Time.now samples/birthdays.rb @@ -24,37 +24,26 @@ require 'gedcom' -class IndividualCounter < GEDCOM::Parser - attr_reader :individuals - attr_reader :families - - def initialize - super +if ARGV.length < 1 + puts "Please specify the name of a GEDCOM file." + exit(0) +end - @individuals = 0 - @families = 0 +puts "Parsing '#{ARGV[0]}'..." - setPreHandler [ "INDI" ], method( :countPerson ) - setPreHandler [ "FAM" ], method( :countFamily ) - end +individuals = 0 +families = 0 - def countPerson( data, state, parm ) - @individuals += 1 +parser = GEDCOM::Parser.new do + before "INDI" do + individuals += 1 end - def countFamily( data, state, parm ) - @families += 1 + before "FAM" do + families += 1 end end -if ARGV.length < 1 - puts "Please specify the name of a GEDCOM file." - exit(0) -end - -puts "Parsing '#{ARGV[0]}'..." - -parser = IndividualCounter.new parser.parse ARGV[0] -puts "There are #{parser.individuals} individuals and #{parser.families} families in '#{ARGV[0]}'." +puts "There are #{individuals} individuals and #{families} families in '#{ARGV[0]}'." samples/count.rb 43148e8d41bd80482c7c50aba4bbd897d8789f00 Jeremy Slade jeremy@jkslade.net http://github.com/timshadel/gedcom-ruby/commit/cca43d79679d8dce3c10137972cb13deb909ec9c cca43d79679d8dce3c10137972cb13deb909ec9c 2008-12-10T22:34:00-08:00 2008-12-10T22:34:00-08:00 Replaced setPreHandler and setPostHandler with before/after; allow before/after to be defined with blocks; reworked sample scripts to match new API; added specs for GEDCOM::Parser, including some test gedcoms 0f141d537f641298eec9ed80e40a5105e752af73 Jeremy Slade jeremy@jkslade.net