bin/html-table2csv

diff --git a/bin/html-table2csv b/bin/html-table2csv
index fc3689f9e45a318ece4e385c8bb1f7f786580ab0..29fd490041f2e3c16b8ed7ac4be38d693132da4c 100755 (executable)
--- a/bin/html-table2csv
+++ b/bin/html-table2csv
@@ -28,7 +28,10 @@ my $VERBOSE = $main::VERBOSE = 1;
 my $DEBUG = $main::DEBUG = 0;
 
 my %Opt = (
-  'table' => undef,
+  'nume' => undef,
+  'tag' => 'table',
+  'sep' => "\t",
+  'html' => 0,
 );
 get_options(\%Opt);
 help() unless @ARGV;
@@ -41,26 +44,42 @@ $main::_DATA_ = undef;
 #################################################################################
 @ARGV = map {m,^\w+://, ? "curl -s '$_' |" : $_} @ARGV if @ARGV;
 my $html = join('',<>);
+$html =~ s/[\r\n]+//g;
 $html =~ s/\s+/ /g;
 
+my %TAGS = (
+  'table' => [
+    'table',
+    'tr',
+    't[dh]',
+  ],
+  'dl' => [
+    'dl',
+    '',
+    'd[td]',
+  ],
+);
+my ($T1,$T2,$T3) = @{ $TAGS{$Opt{tag}} };
 my $table_num = 0;
 
-for my $table ($html =~ m,<table[^>]*>(.*?)<\s*/\s*table\s*>,gi) {
+for my $table ($html =~ m,<${T1}[^>]*>(.*?)<\s*/\s*${T1}\s*>,gi) {
   $table_num++;
-  next if defined $Opt{'table'} and $Opt{table} and $Opt{table} != $table_num;
+  #warn $table_num;
+  next if defined $Opt{'num'} and $Opt{num} and $Opt{num} != $table_num;
 
-  for my $tr ($table =~ m,<tr[^>]*>(.*?)<\s*/\s*tr\s*>,gi) {
+  $table = "<>$table</>" unless $T2;
+  for my $tr ($table =~ m,<${T2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi) {
     my @col;
 
-    for my $td ($tr =~ m,<t[dh][^>]*>(.*?)<\s*/\s*t[dh]\s*>,gi) {
-      $td = html2txt($td);
+    for my $td ($tr =~ m,<${T3}[^>]*>(.*?)<\s*/\s*${T3}\s*>,gi) {
+      $td = html2txt($td) unless $Opt{html};
       $_ = chr(194).chr(160); $td =~ s/$_/ /g;
       $td =~ s/\s+/ /g;
       $td = str_trim($td);
       push(@col,$td);
     }
 
-    print join("\t",@col)."\n";
+    print join($Opt{sep},@col)."\n";
 
   }
 
@@ -176,7 +195,10 @@ Quick usage:
 
 =head1 OPTIONS
 
- -option[table|t=i]   Only dump table number
+ -option[tag|T=s]    Default: table
+ -option[sep|s=s]    Default: tab
+ -option[html!]      Escape html (default: yes)
+ -option[num|n=i]    Only dump table number
  -option[verbose|v+] Verbose mode: increase the verbosity level.
  -option[debug+]     Debug mode: increase the verbosity level.
  -option[version|V]  Print version (default: $VERSION)
@@ -185,14 +207,6 @@ Quick usage:
 
 =cut
 
-=head1 EXAMPLES
-
-...
-
-=head1 REQUIRES
-
-Getopt::Std, Pod::Usage
-
 =head1 COPYRIGHT AND LICENSE
 
 Copyright (C) 2017 Nicolas Boisselier