oddb2xml 3.0.19 → 3.0.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: '081c86c751d4c29fa2ce616abdb6043707e50fab79e3b74a67aab568b2084519'
-  data.tar.gz: e3c91770aa8ccd4de0714c45f5a20eb7f87fc989ea84932a8e4cf1a5a3813a7c
+  metadata.gz: fe41f3e5cfb0e100f111a3071f32ae8dd4b4103d3f5726bf4eab061f80389323
+  data.tar.gz: 785dee3549e33031069aceac4f9be31cc844de603d3b183d94b55f976b0c482c
 SHA512:
-  metadata.gz: b8a928d127496fb06cba79dc630c19bd5edc3b09dd3df0c0dda6ca058cd1897643818d4a8aa9e2ab125023d0ebb319dddee2ab1be0fcdabc8e4708299a136dee
-  data.tar.gz: 27c6ef5bcb9fa0df5b7861496468a88b416209a8b83e7c08e3a995407a75944f0d81849d75399254b7ee404787128ed9e4742ef5f6a78107d90f895845afe707
+  metadata.gz: b385af5fc0643374e878da0cd6e1b262e4d7521c5ac5dd3a3670971f8f130ca3492b4aa296410c5e9e6627021bcbb86b508dbd3e80bf1fdd044c61851df69664
+  data.tar.gz: 40a8baa557cdb31905f8711702bbe52f8d968b67488ae88f293233ea3bd1db689c9ea75d27b4b17019cf2369fb593eed3268c3867277daf5cb73c763b6a74791

data/History.txt

@@ -1,3 +1,6 @@
+=== 3.0.20 / 09.06.2026
+* Bugfix/robustness: handle truncated or failed Swissmedic .xlsx downloads instead of crashing later in the parser (issue #121). A download through a scanning/allow-list proxy can return an incomplete .xlsx -- a valid ZIP/Excel header (so `file` still reports "Microsoft Excel 2007+") but a missing end-of-central-directory record -- which made RubyXL die deep in rubyzip with the cryptic "Zip end of central directory signature not found". New Oddb2xml.valid_zip? verifies an archive is complete (PK header + EOCD signature in the tail). SwissmedicDownloader now validates the downloaded file and automatically re-fetches it (up to the normal retry count) when it is empty or truncated, and no longer reuses a cached broken file under --skip-download. As a last resort SwissmedicExtractor raises a clear, actionable error (pointing at connectivity / "oddb2xml --proxy-check") instead of the rubyzip backtrace.
+
 === 3.0.19 / 09.06.2026
 * New option --proxy-check: probe connectivity/proxy reachability for every host oddb2xml could need, print a full OK/BLOCKED/UNREACHABLE report (honouring http(s)_proxy) and exit without downloading or building. Exits 0 if all hosts are reachable, 1 otherwise — handy for cron/deploy preflight on allow-list proxies (e.g. "oddb2xml --proxy-check"). Reuses the 3.0.18 proxy checker; checks run concurrently.
 

data/lib/oddb2xml/downloader.rb

@@ -279,13 +279,21 @@ module Oddb2xml
     def download
       @file2save = File.join(DOWNLOADS, "swissmedic_#{@type}.xlsx")
       report_download(@url, @file2save)
-      if @options[:calc] && @options[:skip_download] && File.exist?(@file2save) && ((Time.now - File.ctime(@file2save)).to_i < 24 * 60 * 60)
+      if @options[:calc] && @options[:skip_download] && File.exist?(@file2save) && ((Time.now - File.ctime(@file2save)).to_i < 24 * 60 * 60) && Oddb2xml.valid_zip?(@file2save)
         Oddb2xml.log "SwissmedicDownloader #{__LINE__}: Skip downloading #{@file2save} #{File.size(@file2save)} bytes"
         return File.expand_path(@file2save)
       end
       begin
         @url = @direct_url_link
         download_as(@file2save, "w+")
+        # The Swissmedic file is an .xlsx (a ZIP). Downloads through scanning
+        # proxies are sometimes truncated (valid header, missing EOCD), which
+        # would later crash RubyXL with a cryptic rubyzip error. Verify the
+        # archive is complete and just fetch it again if not. (issue #121)
+        unless Oddb2xml.valid_zip?(@file2save)
+          raise Oddb2xml::IncompleteDownloadError,
+            "Swissmedic #{@type} xlsx is empty or truncated (#{File.size(@file2save)} bytes)"
+        end
         if @options[:artikelstamm]
           # ssconvert is in the package gnumeric (Debian)
           cmd = "ssconvert '#{@file2save}' '#{File.join(DOWNLOADS, File.basename(@file2save).sub(/\.xls.*/, ".csv"))}' 2> /dev/null"
@@ -293,8 +301,12 @@ module Oddb2xml
           system(cmd)
         end
         return File.expand_path(@file2save)
-      rescue Timeout::Error, Errno::ETIMEDOUT
-        retrievable? ? retry : raise
+      rescue Timeout::Error, Errno::ETIMEDOUT, Oddb2xml::IncompleteDownloadError => error
+        if retrievable?
+          Oddb2xml.log("Retrying Swissmedic #{@type} download: #{error.message}")
+          retry
+        end
+        raise
       ensure
         Oddb2xml.download_finished(@file2save, false)
       end

data/lib/oddb2xml/extractor.rb

@@ -260,6 +260,14 @@ module Oddb2xml
       @type = type
       Oddb2xml.log("SwissmedicExtractor #{@filename} #{File.size(@filename)} bytes")
       return unless File.exist?(@filename)
+      unless Oddb2xml.valid_zip?(@filename)
+        raise "SwissmedicExtractor: '#{@filename}' is not a usable .xlsx " \
+          "(#{File.size(@filename)} bytes). The Swissmedic '#{@type}' download failed or was " \
+          "truncated -- the file is empty, an HTML error/proxy page, or an incomplete ZIP " \
+          "(valid header but missing end-of-central-directory). Check connectivity to " \
+          "www.swissmedic.ch (run 'oddb2xml --proxy-check'); a too-small file usually means " \
+          "the proxy cut off a large download."
+      end
       @sheet = RubyXL::Parser.parse(File.expand_path(@filename)).worksheets[0]
     end
 

data/lib/oddb2xml/util.rb

@@ -3,10 +3,29 @@ require "htmlentities"
 
 module Oddb2xml
   FAKE_GTIN_START = "999999"
+
+  # Raised when a downloaded archive (zip/xlsx) is empty or truncated, so the
+  # caller can retry the download instead of crashing later in the parser.
+  class IncompleteDownloadError < StandardError; end
+
   def self.gen_prodno(iksnr, seqnr)
     sprintf("%05d", iksnr) + sprintf("%02d", seqnr)
   end
 
+  # True only for a *complete* ZIP/xlsx: a valid local-file header (PK) plus the
+  # End Of Central Directory record (PK\x05\x06) near the end. A truncated
+  # download keeps the header (so `file` still says "Microsoft Excel 2007+") but
+  # loses the EOCD, which makes rubyzip fail with the cryptic "end of central
+  # directory signature not found". See issue #121.
+  def self.valid_zip?(file)
+    return false unless file && File.exist?(file)
+    size = File.size(file)
+    return false unless size > 100
+    return false unless File.binread(file, 2) == "PK"
+    offset = [size - 66_000, 0].max
+    File.binread(file, size - offset, offset).include?("PK\x05\x06".b)
+  end
+
   def self.uri_open(url, max_retries: 3)
     retries = 0
     begin

data/lib/oddb2xml/version.rb

@@ -1,3 +1,3 @@
 module Oddb2xml
-  VERSION = "3.0.19"
+  VERSION = "3.0.20"
 end