From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Wed, 25 Jan 2017 15:28:20 +0000 (+0000)
Subject: bin/html-table2csv
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=79aa64982e4ec27f5f15cf0840e8dd9fb2927d7e;p=nb.git

bin/html-table2csv
---

diff --git a/bin/html-table2csv b/bin/html-table2csv
index 470e9381..e46f2fdf 100755
--- a/bin/html-table2csv
+++ b/bin/html-table2csv
@@ -39,7 +39,7 @@ $main::_DATA_ = undef;
 # BEGIN
 #
 #################################################################################
-@ARGV = map {m,^\w+://, ? "curl -s '$_'" : $_} @ARGV if @ARGV;
+@ARGV = map {m,^\w+://, ? "curl -s '$_' |" : $_} @ARGV if @ARGV;
 my $html = join('',<>);
 $html =~ s/\s+/ /g;
 
@@ -54,6 +54,7 @@ for my $table ($html =~ m,<table[^>]*>(.*?)<\s*/\s*table\s*>,gi) {
 
     for my $td ($tr =~ m,<t[dh][^>]*>(.*?)<\s*/\s*t[dh]\s*>,gi) {
       $td = html2txt($td);
+      $_ = chr(194).chr(160); $td =~ s/$_/ /g;
       $td =~ s/\s+/ /g;
       $td = str_trim($td);
       push(@col,$td);