bin/html2csv

diff --git a/bin/html2csv b/bin/html2csv
index 0e280c5d1e38975ecffbd48836472dbb48449c81..f525c13e75f89f39febee355551650e500cd8e09 100755 (executable)
--- a/bin/html2csv
+++ b/bin/html2csv
@@ -29,6 +29,7 @@ my $DEBUG = $main::DEBUG = 0;
 
 my %Opt = (
   'num' => undef,
+  'print_num' => 0,
   'tag' => 'table',
   'exp1'=> '',
   'exp2'=> '',
@@ -48,6 +49,7 @@ $main::_DATA_ = undef;
 @ARGV = map {m,^\w+://, ? "curl -s '$_' |" : $_} @ARGV if @ARGV;
 my $html = join('',<>);
 $html =~ s/[\r\n]+//g;
+$_ = $Opt{sep}; $html =~ s/$_/ /g;
 $html =~ s/\s+/ /g;
 
 my %TAGS = (
@@ -60,35 +62,48 @@ my %TAGS = (
     'dl',
     '',
     'd[td]',
+    2,
   ],
 );
 exists $TAGS{$Opt{tag}} or die "$NAME: Unknown tag '$Opt{tag}'\n";
-my ($T1,$T2,$T3) = @{ $TAGS{$Opt{tag}} };
+my ($T1,$T2,$T3,$COUNT) = @{ $TAGS{$Opt{tag}} };
+$COUNT ||= 0;
 my $table_num = 0;
 
 my $exp1 = $Opt{exp1}; $exp1 and $exp1 = "[^>]*$exp1";
 my $exp2 = $Opt{exp2}; $exp2 and $exp2 = "[^>]*$exp2";
 my $exp3 = $Opt{exp3}; $exp3 and $exp3 = "[^>]*$exp3";
+
 for my $table ($html =~ m,<${T1}${exp1}[^>]*>(.*?)<\s*/\s*${T1}\s*>,gi) {
   $table_num++;
   #warn $table_num;
   next if defined $Opt{'num'} and $Opt{num} and $Opt{num} != $table_num;
 
   $table = "<>$table</>" unless $T2;
-  for my $tr ($table =~ m,<${T2}${exp2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi) {
+  for my $tr ($table =~ m,<(${T2})${exp2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi) {
     my @col;
+    print "$table_num " if $Opt{print_num};
 
+       $tr = "<>$tr</>" unless $T3;
+       my $count = 0;
     for my $td ($tr =~ m,<${T3}${exp3}[^>]*>(.*?)<\s*/\s*${T3}\s*>,gi) {
       $td = html2txt($td) unless $Opt{html};
       $_ = chr(194).chr(160); $td =~ s/$_/ /g;
       $td =~ s/\s+/ /g;
       $td = str_trim($td);
+      $count++;
       push(@col,$td);
+      if (1 and $COUNT and $count > $COUNT) {
+               print join($Opt{sep},@col)."\n";
+               @col = ();
+      }
     }
 
-    print join($Opt{sep},@col)."\n";
+    print join($Opt{sep},@col)."\n" if @col;
+    #print "\n" unless $T3;
 
   }
+  #print "\n" unless $T2;
 
 }
 
@@ -206,6 +221,7 @@ Quick usage:
  -option[sep|s=s]    Default: tab (\t)
  -option[html!]      Escape html (default: yes)
  -option[num|n=i]    Only dump table number
+ -option[print_num|pn!] Print num tag preffix
  -option[exp1|E1=s]  Regexp filter on tag 1
  -option[exp2|E2=s]  Regexp filter on tag 2
  -option[exp3|E3=s]  Regexp filter on tag 3