oddb2xml 3.0.17 → 3.0.18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: df4185a65a50fc628ba92bc98fb8ac378563d5653882e3e34b54d1c1265e3808
-  data.tar.gz: a62c0d6b0b643458e9885c6ba7d8c1d54c1435696a3c73da2788fd22539a9915
+  metadata.gz: a4aadf4a537b9f625c491c4609116f6aef44c4f558e0a273a0d2da705ad8bfd0
+  data.tar.gz: 946a8f14539626a383dcab8e73cdc5e89410e8df2721a44d0fca50cc868cf11c
 SHA512:
-  metadata.gz: e3718c1c76fbced6bb1ce43350ae13b42fbf747e2dce91980be24df4b9dcea1a78448b20a344ab50685effe3a33341a20ebd520c1a70ed995c996a6fd0dfb0ff
-  data.tar.gz: e0c0dd04ccacc29933a582fd05d19ebee40ce7ae420ca632e4d6f5068ae92de5928ab1fa6daf3b9bd4894f5797dde0090fad74d8bdc25a3610f648b57624aa1c
+  metadata.gz: d91d784a98fd256a0ce056c0cfa36959249c7d4cf6b194a2b97c360569ad6f3e503e57c4395722d8a83b5f163ec3b9f5187b8f28ca193de8cc83eff80cf59c2f
+  data.tar.gz: 75f26b3a4559da9776b77c718a916e7ee76715a49475066b8b1e575ba5aa9790b9c6797245784a799af211bb18c86711600ce7ce8cb1794feae4769bb442eaed

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    oddb2xml (3.0.17)
+    oddb2xml (3.0.18)
       csv
       htmlentities
       httpi

data/History.txt

@@ -1,3 +1,7 @@
+=== 3.0.18 / 09.06.2026
+* Bugfix: stop the FHIR downloader from crashing with "Errno::ENOENT @ rb_file_s_size ... foph-sl-export-latest-de.ndjson" when run with --skip-download (issue #121). FhirDownloader#skip_download? returned true on the bare --skip-download flag and then called File.size on a file that was never downloaded. Each oddb2xml run uses its own ./downloads dir, so deploy scripts that download once and then re-run with --skip-download in a fresh dir hit this every time. skip_download? now requires the target NDJSON to actually exist on disk before honouring the flag; a missing file falls through to a normal download instead of crashing.
+* New: proxy / connectivity preflight check (issue #121). At the very start of a run, oddb2xml now probes the outbound hosts it needs (honouring the http(s)_proxy environment) and prints a loud warning if any host is blocked by the proxy (HTTP 407 on an allow-list proxy such as Aspectra's Skyhigh gateway) or otherwise unreachable — surfacing the cause up front instead of a later empty-output/Errno symptom. The probed host set is option-aware (e.g. id.gs1.ch only with --firstbase, epl.bag.admin.ch only with --fhir). It only warns and never aborts the run; downloads still proceed and fail individually as before. Checks run concurrently (~6s worst case) and are skipped during tests; set ODDB2XML_SKIP_PROXY_CHECK=1 to silence it.
+
 === 3.0.17 / 08.06.2026
 * Compatibility: relax the nokogiri requirement from ">= 1.19.3" to ">= 1.18.10" (lower bound only, no upper cap) so oddb2xml installs on Ruby 3.1 again. nokogiri 1.19.x requires Ruby >= 3.2, which made 3.0.13–3.0.16 uninstallable on Ruby 3.1.0 (the bare "gem install" crashed with the misleading RubyGems resolver error "undefined method `request' for nil:NilClass"). With a lower-bound-only floor, RubyGems still resolves the security-fixed 1.19.3 on Ruby >= 3.2, while Ruby 3.1 installs 1.18.10 (the newest 3.1-compatible release). oddb2xml uses no nokogiri 1.19-only API and runs no CSS/XSLT/C14N over untrusted input, so the 1.19.3 advisories do not apply to its usage. Note: Ruby 3.1 is EOL (since March 2025) — upgrading to Ruby >= 3.2 remains recommended.
 

data/lib/oddb2xml/cli.rb

@@ -3,6 +3,7 @@ require "oddb2xml/downloader"
 require "oddb2xml/extractor"
 require "oddb2xml/compressor"
 require "oddb2xml/options"
+require "oddb2xml/proxy_check"
 require "oddb2xml/util"
 require "rubyXL"
 require "date" # for today
@@ -38,6 +39,7 @@ module Oddb2xml
     def run
       threads = []
       start_time = Time.now
+      ProxyCheck.run(@options)
       files2rm = Dir.glob(File.join(DOWNLOADS, "*"))
       FileUtils.rm_f(files2rm, verbose: true) if (files2rm.size > 0) && !Oddb2xml.skip_download?
       if @options[:calc] && !(@options[:extended] || @options[:firstbase])

data/lib/oddb2xml/fhir_support.rb

@@ -79,7 +79,12 @@ module Oddb2xml
     end
 
     def skip_download?
-        @options[:skip_download] || (File.exist?(@file2save) && file_age_hours(@file2save) < 24)
+      # Only skip when the target file actually exists on disk. The bare
+      # @options[:skip_download] flag is not enough: each oddb2xml run uses its
+      # own ./downloads dir, so a flag-only short-circuit made download_one call
+      # File.size on a missing NDJSON and crash with Errno::ENOENT (issue #121).
+      return false unless File.exist?(@file2save)
+      @options[:skip_download] || file_age_hours(@file2save) < 24
     end
 
     def file_age_hours(file)

data/lib/oddb2xml/proxy_check.rb

@@ -0,0 +1,108 @@
+# frozen_string_literal: true
+
+require "net/http"
+require "uri"
+require "openssl"
+
+module Oddb2xml
+  # Preflight connectivity check. Run once at the very start of a CLI run, it
+  # probes every outbound host oddb2xml needs (honouring the http(s)_proxy
+  # environment) and prints a loud warning if any host is blocked by the proxy
+  # (HTTP 407 on an allow-list proxy such as Aspectra's Skyhigh gateway) or is
+  # otherwise unreachable. It never aborts the run -- downloads still proceed and
+  # fail individually as before; this just surfaces the cause up front instead of
+  # leaving the user to decode a later Errno/empty-output symptom. See issue #121.
+  module ProxyCheck
+    module_function
+
+    # host => human-readable description of what breaks when it is unreachable.
+    # Hosts only needed for certain options are added conditionally (see #hosts_for).
+    BASE_HOSTS = {
+      "files.refdata.ch" => "Refdata articles",
+      "www.swissmedic.ch" => "Swissmedic registrations",
+      "raw.githubusercontent.com" => "ATC codes (cpp2sqlite)"
+    }.freeze
+
+    TIMEOUT = 6 # seconds, per host (open + read); checks run concurrently
+
+    def proxy_uri
+      env = ENV["https_proxy"] || ENV["HTTPS_PROXY"] || ENV["http_proxy"] || ENV["HTTP_PROXY"]
+      return nil if env.nil? || env.empty?
+      env = "http://#{env}" unless env.start_with?("http")
+      URI.parse(env)
+    rescue URI::InvalidURIError
+      nil
+    end
+
+    def hosts_for(options = {})
+      hosts = BASE_HOSTS.dup
+      hosts["epl.bag.admin.ch"] = "BAG FHIR data (--fhir)" if options[:fhir]
+      hosts["id.gs1.ch"] = "GS1 NONPHARMA (--firstbase / -b)" if options[:firstbase]
+      hosts["www.spezialitaetenliste.ch"] = "BAG Spezialitätenliste" unless options[:fhir]
+      hosts["www.medregbm.admin.ch"] = "Medizinalberuferegister (-x address)" if options[:address]
+      hosts
+    end
+
+    # Returns :ok, :blocked (proxy 407) or :unreachable for a single host.
+    def check_host(host, proxy)
+      http =
+        if proxy
+          Net::HTTP.new(host, 443, proxy.host, proxy.port, proxy.user, proxy.password)
+        else
+          Net::HTTP.new(host, 443)
+        end
+      http.use_ssl = true
+      http.verify_mode = OpenSSL::SSL::VERIFY_NONE
+      http.open_timeout = TIMEOUT
+      http.read_timeout = TIMEOUT
+      http.start do |h|
+        res = h.head("/")
+        return :blocked if res.code.to_s == "407"
+        return :ok # any HTTP answer (200/301/403/404/...) means the host is reachable
+      end
+    rescue => error
+      msg = error.message.to_s.downcase
+      return :blocked if msg.include?("407") || msg.include?("authenticationrequired") || msg.include?("proxy")
+      :unreachable
+    end
+
+    # Probe all relevant hosts concurrently and warn about any that fail.
+    def run(options = {})
+      return if defined?(RSpec) || defined?(VCR) # never touch the network in tests
+      return if ENV["ODDB2XML_SKIP_PROXY_CHECK"]
+
+      proxy = proxy_uri
+      hosts = hosts_for(options)
+      results = hosts.map do |host, desc|
+        Thread.new { [host, desc, check_host(host, proxy)] }
+      end.map(&:value)
+
+      problems = results.reject { |(_host, _desc, status)| status == :ok }
+      return if problems.empty?
+
+      warn_about(problems, proxy)
+    end
+
+    def warn_about(problems, proxy)
+      line = "=" * 72
+      $stderr.puts line
+      $stderr.puts " oddb2xml CONNECTIVITY WARNING"
+      $stderr.puts " The following hosts could not be reached -- the corresponding"
+      $stderr.puts " downloads will FAIL or produce incomplete data:"
+      problems.each do |(host, desc, status)|
+        tag = (status == :blocked) ? "BLOCKED by proxy (407)" : "UNREACHABLE          "
+        $stderr.puts format("   [%s] %-26s %s", tag, host, desc)
+      end
+      if proxy
+        $stderr.puts ""
+        $stderr.puts " Proxy in use: #{proxy.host}:#{proxy.port}"
+        if problems.any? { |(_h, _d, s)| s == :blocked }
+          $stderr.puts " This looks like an allow-list proxy. Ask your admin to allow the"
+          $stderr.puts " hosts above (HTTPS/443), or set credentials in http(s)_proxy."
+        end
+      end
+      $stderr.puts " (Set ODDB2XML_SKIP_PROXY_CHECK=1 to silence this check.)"
+      $stderr.puts line
+    end
+  end
+end

data/lib/oddb2xml/version.rb

@@ -1,3 +1,3 @@
 module Oddb2xml
-  VERSION = "3.0.17"
+  VERSION = "3.0.18"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: oddb2xml
 version: !ruby/object:Gem::Version
-  version: 3.0.17
+  version: 3.0.18
 platform: ruby
 authors:
 - Yasuhiro Asaka, Zeno R.R. Davatz, Niklaus Giger
@@ -483,6 +483,7 @@ files:
 - lib/oddb2xml/fhir_support.rb
 - lib/oddb2xml/options.rb
 - lib/oddb2xml/parslet_compositions.rb
+- lib/oddb2xml/proxy_check.rb
 - lib/oddb2xml/refdata_cleanup.rb
 - lib/oddb2xml/semantic_check.rb
 - lib/oddb2xml/util.rb