From 41700ad112d7ee7d7ddb5579929ddfaf91bbcaef Mon Sep 17 00:00:00 2001
From: Kevin Ernst <kevin.ernst@cchmc.org>
Date: Sun, 12 Nov 2023 09:09:54 -0500
Subject: [PATCH 1/2] Support reading from stdin with `-f -` (or no args)

---
 webtable_to_text.rb | 23 +++++++++++++++++------
 1 file changed, 17 insertions(+), 6 deletions(-)

diff --git a/webtable_to_text.rb b/webtable_to_text.rb
index ef90d52..dde87dd 100755
--- a/webtable_to_text.rb
+++ b/webtable_to_text.rb
@@ -11,12 +11,19 @@
 
 options = {}
 OptionParser.new do |opts|
-  opts.banner = "  Usage: webtable_to_text.rb [options]"
+  opts.banner = <<BANNER
+ usage:
+   webtable_to_text.rb [options] [-f FILE | -u URL]
+   curl [options] URL | webtable_to_text.rb
+   wget -O- [options] URL | webtable_to_text.rb
+
+ options:
+BANNER
 
   opts.on("-A", "--all", "Print all tables found on the specified page") { options[:all] = true }
   opts.on("-a", "--asciidoc", "Output in asciidoc/asciidoctor format") { options[:asciidoc] = true }
   opts.on("-c", "--csv", "Output in CSV / comma separated values format") { options[:csv] = true }
-  opts.on("-f", "--file FILE", "Specify HTML input file as source for extracting tables") { |v| options[:file] = v }
+  opts.on("-f", "--file FILE", "Specify HTML input file as source for extracting tables; use '-' to read from stdin") { |v| options[:file] = v }
   opts.on("-i", "--interactive", "Interactive mode") { options[:interactive] = true }
   opts.on("-m", "--markdown", "Output in markdown format") { options[:markdown] = true }
   opts.on("-n", "--number NUM", "Print specific table number only; separate multiple numbers with commas") { |v| options[:number] = v }
@@ -27,8 +34,6 @@
 
 end.parse!
 
-source = ""
-
 url = options[:url]
 file = options[:file]
 
@@ -38,9 +43,15 @@
   source_content = URI.open(escaped).read
 elsif file
   source_location = file
-  source_content = File.read(file)
+  if file == "-"
+    # h/t: https://stackoverflow.com/a/273841
+    source_content = ARGF.read
+  else
+    source_content = File.read(file)
+  end
 else
-  abort("  Please provide a source file or URL as input")
+  #abort("  Please provide a source file or URL as input")
+  source_content = ARGF.read
 end
 
 doc = Nokogiri::HTML(source_content)

From 3e354ae4a5354addff13310f61efea611a13e864 Mon Sep 17 00:00:00 2001
From: Kevin Ernst <kevin.ernst@cchmc.org>
Date: Sun, 12 Nov 2023 09:19:33 -0500
Subject: [PATCH 2/2] Print help if no filename and stdin isn't a pipe

---
 webtable_to_text.rb | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/webtable_to_text.rb b/webtable_to_text.rb
index dde87dd..d23b74a 100755
--- a/webtable_to_text.rb
+++ b/webtable_to_text.rb
@@ -50,7 +50,13 @@
     source_content = File.read(file)
   end
 else
-  #abort("  Please provide a source file or URL as input")
+  if $stdin.isatty
+    abort("  Please provide a source file or URL as input. See '--help'.")
+  end
+  if options[:interactive]
+    abort("  Interactive mode not supported when reading from a pipe.")
+  end
+  source_location = '(stdin)'
   source_content = ARGF.read
 end
 
@@ -60,7 +66,7 @@
 len = tables.length
 
 if len < 1
-  abort("  No tables found in page at #{source_location}")
+  abort("  No tables found in page at #{source_location}.")
 end
 
 numstring = "all"