src/ActiveDoc/Parse.pm

#
# Parse.pm
#
# Originally Written by Christopher Williams
#
# Description
# -----------
# maintain parse configurations
#
# Interface
# ---------
# new()                 : A new Parse object
# addtag(name,start,text,end,$object)   : Add a new tag
# addgrouptags()        : add <Group> tag functionality
# addignoretags()       : add <ignore> tag functionality
# parse(filename,[streamhandle], [streamexcludetag]) : 
#                               parse the given file - turn on the stream
#                                  function of the switcher if a filehandle
#                                  supplied as a second argument
# line()                : return the current linenumber in the file
# tagstartline()        : return the linenumber of the last tag opening 
# includeparse(Parse)   : include the settings from another parse object
# tags()                : return list of defined tags
# cleartags()           : clear of all tags
# opencontext(name)     : open a parse context
# closecontext(name)    : close a parse context
# includecontext(name)  : Process when in a given context
# excludecontext(name)  : No Processing when given context
# contexttag(tagname)   : Register the tagname as one able to change context
#                         if not registerd - the close tag will be ignored
#                         too if outside of the specified context!

package ActiveDoc::Parse;
require 5.004;
use XML::Parser;

sub new()
   {
   my $class=shift;
   $self={};
   bless $self, $class;
   my ($dataclass, $parse_style)=@_;

   $self->{xmlparser} = new XML::Parser (
                                         Style => $parse_style,
                                         ParseParamEnt => 1,
                                         ErrorContext => 3,
                                         Pkg   => $dataclass);   
   return $self;
   }

sub parsefilelist()
   {
   my $self=shift;
   my ($files)=@_;
   print __PACKAGE__."::parsefilelist(): Not used?\n";
   }

sub filehead ()
   {
   my $self=shift;
   my $data=@_;
   if (@_)
     {
     $self->{filehead}=shift;
     return;
     }
   return $self->{filehead} || "";
   }

sub filetail ()
   {
   my $self=shift;
   if (@_)
     {
     $self->{filetail}=shift;
     return;
     }
   return $self->{filetail} || "";
   }

sub parse()
   {
   my $self=shift;
   my ($file)=@_;
   eval 
   {
     $self->{data} = $self->{xmlparser}->parse($self->getfilestring_($file));
   };
   if ($@)
      {
      print STDERR "**** ERROR: Failed parsing file: $file\n$@\n";
      }
   return $self;
   }

sub getfilestring_()
   {
   my $self=shift;
   my ($file)=@_;
   my $filestring="";
   my $read=0;
   if (($file!~/\.xml$/) && ($file!~/\/\.SCRAM\/InstalledTools\/[^\/]+$/))
      {
      eval("use SCRAM::Doc2XML");
      if (!$@)
         {
         my $xmlconvertor = SCRAM::Doc2XML->new();
         my $xml=$xmlconvertor->convert($file);
         $filestring = join("",@$xml);
         $xmlconvertor->clean();
         $read=1;
         }
      }
   if (!$read)
   {
     open (IN, "< $file") or die __PACKAGE__.": Cannot read file $file: $!\n";
     $filestring = join("", <IN>);
     close (IN) or die __PACKAGE__.": Cannot read file $file: $!\n";
   }
   $filestring = $self->filehead().$filestring.$self->filetail();
   # Strip spaces at the beginning and end of the line:
   $filestring =~ s/^\s+//g;
   $filestring =~ s/\s+$//g;
   # Finally strip the newlines:
   #$filestring =~ s/\n//g;
   # Strip out spaces in between tags:
   #$filestring =~ s/>\s+</></g;
   $self->{filestring}=$filestring;
   return $filestring;
   }

sub data()
   {
   my $self=shift;
   return $self->{data}->[0];
   }

sub includeparse
   {
   my $self=shift;
   my $obj=shift;
   my $tag;
   
   # copy the tags from  the remote parse object
   foreach $tag ( $obj->tags() )
      {
      $self->addtag($tag,$obj->{tags}->tagsettings($tag));
      }
   }

sub addtag
   {
   my $self=shift;
   $self->{tags}->addtag(@_);
   }

1;
Revision:	1.11
Committed:	Tue Nov 6 14:13:48 2007 UTC (17 years, 6 months ago) by muzaffar
Content type:	text/plain
Branch:	MAIN
CVS Tags:	V1_1_7, V1_1_6, V1_1_5, V1_1_4, V1_1_3, V1_1_2, V1_1_0_reltag8, V1_1_0_reltag7, V1_1_0_reltag6, V1_1_1, V1_1_0_reltag5, V1_1_0_reltag4, V1_1_0_reltag3, V1_1_0_reltag2, V1_1_0_reltag1, V1_1_0_reltag, V1_1_0_cand3, V1_1_0_cand2, V1_1_0_cand1, HEAD_SM_071214, v103_xml_071106
Branch point for:	forBinLess_SCRAM, HEAD_BRANCH_SM_071214
Changes since 1.10:	+55 -6 lines
Log Message:	xml scram with changes to make it fast
#	User	Rev	Content
1	williamc	1.1	#
2			# Parse.pm
3			#
4			# Originally Written by Christopher Williams
5			#
6			# Description
7			# -----------
8			# maintain parse configurations
9			#
10			# Interface
11			# ---------
12	sashby	1.6	# new() : A new Parse object
13	williamc	1.1	# addtag(name,start,text,end,$object) : Add a new tag
14	sashby	1.6	# addgrouptags() : add <Group> tag functionality
15			# addignoretags() : add <ignore> tag functionality
16	williamc	1.1	# parse(filename,[streamhandle], [streamexcludetag]) :
17			# parse the given file - turn on the stream
18			# function of the switcher if a filehandle
19			# supplied as a second argument
20	sashby	1.6	# line() : return the current linenumber in the file
21	williamc	1.1	# tagstartline() : return the linenumber of the last tag opening
22	sashby	1.6	# includeparse(Parse) : include the settings from another parse object
23	williamc	1.2	# tags() : return list of defined tags
24	williamc	1.3	# cleartags() : clear of all tags
25	sashby	1.6	# opencontext(name) : open a parse context
26	williamc	1.4	# closecontext(name) : close a parse context
27			# includecontext(name) : Process when in a given context
28			# excludecontext(name) : No Processing when given context
29			# contexttag(tagname) : Register the tagname as one able to change context
30			# if not registerd - the close tag will be ignored
31			# too if outside of the specified context!
32	williamc	1.1
33			package ActiveDoc::Parse;
34			require 5.004;
35	sashby	1.9	use XML::Parser;
36	williamc	1.1
37	sashby	1.9	sub new()
38	sashby	1.6	{
39			my $class=shift;
40			$self={};
41			bless $self, $class;
42	sashby	1.9	my ($dataclass, $parse_style)=@_;
43
44			$self->{xmlparser} = new XML::Parser (
45			Style => $parse_style,
46			ParseParamEnt => 1,
47			ErrorContext => 3,
48			Pkg => $dataclass);
49	sashby	1.6	return $self;
50			}
51
52	sashby	1.9	sub parsefilelist()
53	sashby	1.6	{
54			my $self=shift;
55	sashby	1.9	my ($files)=@_;
56	muzaffar	1.11	print __PACKAGE__."::parsefilelist(): Not used?\n";
57			}
58
59			sub filehead ()
60			{
61			my $self=shift;
62			my $data=@_;
63			if (@_)
64			{
65			$self->{filehead}=shift;
66			return;
67			}
68			return $self->{filehead} \|\| "";
69			}
70
71			sub filetail ()
72			{
73			my $self=shift;
74			if (@_)
75			{
76			$self->{filetail}=shift;
77			return;
78			}
79			return $self->{filetail} \|\| "";
80	sashby	1.6	}
81
82	sashby	1.9	sub parse()
83	sashby	1.6	{
84			my $self=shift;
85	sashby	1.9	my ($file)=@_;
86	muzaffar	1.11	eval
87			{
88			$self->{data} = $self->{xmlparser}->parse($self->getfilestring_($file));
89			};
90			if ($@)
91			{
92			print STDERR "**** ERROR: Failed parsing file: $file\n$@\n";
93			}
94	sashby	1.9	return $self;
95	sashby	1.6	}
96
97	sashby	1.9	sub getfilestring_()
98	sashby	1.7	{
99			my $self=shift;
100	sashby	1.9	my ($file)=@_;
101	muzaffar	1.11	my $filestring="";
102			my $read=0;
103			if (($file!~/\.xml$/) && ($file!~/\/\.SCRAM\/InstalledTools\/[^\/]+$/))
104			{
105			eval("use SCRAM::Doc2XML");
106			if (!$@)
107			{
108			my $xmlconvertor = SCRAM::Doc2XML->new();
109			my $xml=$xmlconvertor->convert($file);
110			$filestring = join("",@$xml);
111			$xmlconvertor->clean();
112			$read=1;
113			}
114			}
115			if (!$read)
116			{
117			open (IN, "< $file") or die __PACKAGE__.": Cannot read file $file: $!\n";
118			$filestring = join("", <IN>);
119			close (IN) or die __PACKAGE__.": Cannot read file $file: $!\n";
120			}
121			$filestring = $self->filehead().$filestring.$self->filetail();
122	sashby	1.9	# Strip spaces at the beginning and end of the line:
123			$filestring =~ s/^\s+//g;
124			$filestring =~ s/\s+$//g;
125			# Finally strip the newlines:
126	muzaffar	1.11	#$filestring =~ s/\n//g;
127	sashby	1.9	# Strip out spaces in between tags:
128	muzaffar	1.11	#$filestring =~ s/>\s+</></g;
129	sashby	1.9	$self->{filestring}=$filestring;
130			return $filestring;
131	sashby	1.7	}
132
133	sashby	1.9	sub data()
134	sashby	1.6	{
135			my $self=shift;
136	sashby	1.9	return $self->{data}->[0];
137	sashby	1.6	}
138
139			sub includeparse
140			{
141			my $self=shift;
142			my $obj=shift;
143			my $tag;
144
145			# copy the tags from the remote parse object
146			foreach $tag ( $obj->tags() )
147			{
148			$self->addtag($tag,$obj->{tags}->tagsettings($tag));
149			}
150			}
151
152			sub addtag
153			{
154			my $self=shift;
155			$self->{tags}->addtag(@_);
156			}
157
158	sashby	1.9	1;