kataba 1.1.1 → 1.1.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 701caff7f874cbe1f9a02c4d30e58576e3c9970aba61cef220ac96460b729f9c
-  data.tar.gz: 803f46524665ca8f9d0e1ee82dfa154dee798363a2326195feeb54ef412e26b5
+  metadata.gz: 0c1a21b706cc4e327925cd9d936247e6f5ce17ea3964853f1b442e5ee8ecae3b
+  data.tar.gz: 11ad87d5a2baede9403e4b83e2e5db01f5f2b8fe9702371228018cb96534ec22
 SHA512:
-  metadata.gz: 7b323b6a3742abfba45ebf1dd98345c73f5e7b703229b9a7f852069a616e2b7cac26db91e9d23f71b5495878108bad8bb71d0e765ea5a4c7f8ce7b8415349ecb
-  data.tar.gz: e653e7a40a363a9d4b174d84ebc5e3907edc0a05c8c9a13231688d86ed033e6c0c38ff26f84ac8d58c6b97fb4eb22fa70897a2b29aaf774164e20a1a31ac4741
+  metadata.gz: e2bdb712cd9c063469ee35b6dd5c3f1321018fc490ed25cb095c63715aaf63829915db1b99bfac94d87e574c88e301145fde7359cc57139dd1a997d61954bddc
+  data.tar.gz: d21c6a7c83b2a93c89115305856cf2824a62cfa75e7aaa32bda7ea0f4d648bb4d8e6e4db3d4e45ea734a2bc495ea33d3737eb62bc85b3fd076958375d3d0e6fa

data/lib/kataba/fetcher.rb

@@ -32,10 +32,25 @@ module Kataba
       '/mods/xml.xsd' => '/standards/mods/xml.xsd',
     }.freeze
 
+    # Seconds to wait for a schema host to accept the connection, and then to
+    # deliver the body. Net::HTTP's own defaults are 60s each, which is a
+    # fallback rather than a considered number: a schema is a small file from a
+    # public host, and both are paid per redirect hop and again on the
+    # alternate-scheme attempt, so an unbounded fetch can hold a caller's
+    # thread for many minutes.
+    DEFAULT_OPEN_TIMEOUT = 3
+    DEFAULT_READ_TIMEOUT = 10
+
     class FetchError < StandardError; end
 
-    def initialize(uri)
+    # A subclass, so a consumer's `rescue FetchError` covers a timeout while
+    # one that wants to retry a timeout without retrying a 404 still can.
+    class FetchTimeout < FetchError; end
+
+    def initialize(uri, open_timeout: DEFAULT_OPEN_TIMEOUT, read_timeout: DEFAULT_READ_TIMEOUT)
       @original_uri = uri
+      @open_timeout = open_timeout
+      @read_timeout = read_timeout
     end
 
     def fetch
@@ -75,7 +90,23 @@ module Kataba
       parsed = URI.parse(uri)
       http = Net::HTTP.new(parsed.host, parsed.port)
       http.use_ssl = (parsed.scheme == 'https')
-      response = http.get(parsed.request_uri)
+      http.open_timeout = @open_timeout
+      http.read_timeout = @read_timeout
+      # Net::HTTP retries an idempotent request once on a read timeout, so the
+      # ceiling per hop would otherwise be twice read_timeout. Nothing here
+      # reuses a socket, so that retry can only re-wait a timeout that has
+      # already expired.
+      http.max_retries = 0
+
+      response =
+        begin
+          http.get(parsed.request_uri)
+        rescue Timeout::Error => e
+          # Both Net::OpenTimeout and Net::ReadTimeout are Timeout::Errors, and
+          # Timeout::Error is a RuntimeError — so an unwrapped timeout lands in
+          # a consumer's generic rescue rather than its fetch-failure branch.
+          raise FetchTimeout, "timed out fetching #{uri} (#{e.class}: #{e.message})"
+        end
 
       if response.is_a?(Net::HTTPRedirection)
         target = resolve_redirect(parsed, response['location'])

data/lib/kataba.rb

@@ -46,9 +46,23 @@ module Kataba
     #   Kataba.configuration.mirror_list = File.join(Rails.root, 'config', 'mirror.yml')
     attr_accessor :mirror_list
 
+    # Seconds Kataba waits for a schema host to accept a connection, and then
+    # to deliver the body. Raise them for a slow mirror; lower them when the
+    # caller is a request thread that must not be held. Both are paid per
+    # redirect hop and again on the alternate-scheme attempt, so the ceiling
+    # for one schema is a multiple of these, not either on its own.
+    #
+    # Example:
+    #   Kataba.configuration.open_timeout = 5
+    #   Kataba.configuration.read_timeout = 20
+    attr_accessor :open_timeout
+    attr_accessor :read_timeout
+
     # Default configuration values
     def initialize
       @offline_storage = "#{Dir.tmpdir}/kataba"
+      @open_timeout = Kataba::Fetcher::DEFAULT_OPEN_TIMEOUT
+      @read_timeout = Kataba::Fetcher::DEFAULT_READ_TIMEOUT
     end
   end
 
@@ -61,6 +75,14 @@ module Kataba
   # Arguments:
   #   xsd_uri: (String)
 
+  # Dir.chdir is process-wide state, not thread-local: if two threads each
+  # have a chdir block open at once, Ruby raises "conflicting chdir during
+  # another chdir block" rather than let one thread's cwd stomp on the
+  # other's. A shared Mutex serializes just that step across threads in one
+  # process — cheap, since by this point it's a local file read + a Nokogiri
+  # parse, not the network fetch (which stays outside the lock).
+  CHDIR_MUTEX = Mutex.new
+
   def self.fetch_schema(xsd_uri)
     uri_md5 = Digest::MD5.hexdigest(xsd_uri)
     dir_path = "#{self.configuration.offline_storage}"
@@ -77,8 +99,10 @@ module Kataba
       end
 
       # Validate and return Nokogiri schema
-      Dir.chdir(dir_path) do
-        return Nokogiri::XML::Schema(IO.read(xsd_path))
+      CHDIR_MUTEX.synchronize do
+        Dir.chdir(dir_path) do
+          return Nokogiri::XML::Schema(IO.read(xsd_path))
+        end
       end
     rescue Nokogiri::XML::SyntaxError
       # Poisoned cache (e.g. a pre-fix install that stored a bad fetch).
@@ -122,7 +146,9 @@ module Kataba
           fetch_uri = mirror unless mirror.to_s.empty?
         end
 
-        body = Kataba::Fetcher.new(fetch_uri).fetch
+        body = Kataba::Fetcher.new(fetch_uri,
+                                   open_timeout: self.configuration.open_timeout,
+                                   read_timeout: self.configuration.read_timeout).fetch
 
         # Write to a .part file first; only rename to the final cache path
         # after we've confirmed the bytes parse as XML. Without this, a

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: kataba
 version: !ruby/object:Gem::Version
-  version: 1.1.1
+  version: 1.1.3
 platform: ruby
 authors:
 - David Cliff
@@ -14,56 +14,56 @@ dependencies:
   name: nokogiri
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '1.19'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '1.19'
 - !ruby/object:Gem::Dependency
   name: rake
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
 - !ruby/object:Gem::Dependency
   name: minitest
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
 - !ruby/object:Gem::Dependency
   name: webmock
   requirement: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
   type: :development
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
-    - - ">="
+    - - '>='
       - !ruby/object:Gem::Version
         version: '0'
 description: Kataba allows for mirroring and offline storage of XSD files, to enhance
@@ -85,16 +85,16 @@ require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">="
+  - - '>='
     - !ruby/object:Gem::Version
       version: '3.2'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
-  - - ">="
+  - - '>='
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.4.10
+rubygems_version: 3.0.9
 signing_key:
 specification_version: 4
 summary: XML Schema Definition (XSD) mirroring and offline validation for Nokogiri