From b603f85bb6a373cb171ae4b581ad5642a1ef29f2 Mon Sep 17 00:00:00 2001
From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Mon, 21 May 2018 15:53:41 +0100
Subject: [PATCH] bin/html2csv

---
 bin/{html-table2csv => html2csv} | 20 +++++++++++++++-----
 1 file changed, 15 insertions(+), 5 deletions(-)
 rename bin/{html-table2csv => html2csv} (88%)

diff --git a/bin/html-table2csv b/bin/html2csv
similarity index 88%
rename from bin/html-table2csv
rename to bin/html2csv
index 29fd4900..0e280c5d 100755
--- a/bin/html-table2csv
+++ b/bin/html2csv
@@ -28,8 +28,11 @@ my $VERBOSE = $main::VERBOSE = 1;
 my $DEBUG = $main::DEBUG = 0;
 
 my %Opt = (
-  'nume' => undef,
+  'num' => undef,
   'tag' => 'table',
+  'exp1'=> '',
+  'exp2'=> '',
+  'exp3'=> '',
   'sep' => "\t",
   'html' => 0,
 );
@@ -59,19 +62,23 @@ my %TAGS = (
     'd[td]',
   ],
 );
+exists $TAGS{$Opt{tag}} or die "$NAME: Unknown tag '$Opt{tag}'\n";
 my ($T1,$T2,$T3) = @{ $TAGS{$Opt{tag}} };
 my $table_num = 0;
 
-for my $table ($html =~ m,<${T1}[^>]*>(.*?)<\s*/\s*${T1}\s*>,gi) {
+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,<${T2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi) {
+  for my $tr ($table =~ m,<${T2}${exp2}[^>]*>(.*?)<\s*/\s*${T2}\s*>,gi) {
     my @col;
 
-    for my $td ($tr =~ m,<${T3}[^>]*>(.*?)<\s*/\s*${T3}\s*>,gi) {
+    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;
@@ -196,9 +203,12 @@ Quick usage:
 =head1 OPTIONS
 
  -option[tag|T=s]    Default: table
- -option[sep|s=s]    Default: tab
+ -option[sep|s=s]    Default: tab (\t)
  -option[html!]      Escape html (default: yes)
  -option[num|n=i]    Only dump table number
+ -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
  -option[verbose|v+] Verbose mode: increase the verbosity level.
  -option[debug+]     Debug mode: increase the verbosity level.
  -option[version|V]  Print version (default: $VERSION)
-- 
2.47.3