openphar 0.1.0

data/lib/openphar.rb

@@ -0,0 +1,119 @@
+# frozen_string_literal: true
+
+# Openphar - Machine-readable pharmacopoeia data library
+#
+# A model-driven library for parsing, transforming, and exporting
+# pharmacopoeia data from multiple publishers (JP, Ph.Int., etc.)
+#
+# @example Basic usage
+#   require 'openphar'
+#
+#   # Parse pharmacopoeia data
+#   parser = Openphar::Parsers::PhIntJsonParser.new
+#   monographs = parser.parse_directory('./data')
+#
+#   # Export to Neo4j
+#   exporter = Openphar::Exporters::Neo4jExporter.new
+#   exporter.export_monographs('./data/jp', './output/neo4j')
+
+module Openphar
+  # Base error class for all Openphar errors
+  class Error < StandardError; end
+
+  class << self
+    # Returns the global configuration
+    #
+    # @return [Configuration]
+    def configuration
+      @configuration ||= Configuration.new
+    end
+
+    # Yields the configuration for setup
+    #
+    # @yield [Configuration]
+    def configure
+      yield(configuration)
+    end
+
+    # Returns the path to the ontology files
+    #
+    # @return [String]
+    def ontology_path
+      @ontology_path ||= File.expand_path('../ontology', __dir__)
+    end
+
+    # Returns the path to the data files
+    #
+    # @return [String]
+    def data_path
+      @data_path ||= File.expand_path('../data', __dir__)
+    end
+
+    # Returns the base IRI for the ontology
+    #
+    # @return [String]
+    def base_iri
+      'https://www.openphar.org'
+    end
+
+    # Returns the Herbapedia base IRI
+    #
+    # @return [String]
+    def herbapedia_iri
+      'https://www.herbapedia.org'
+    end
+
+    # Convenience method to access the monograph repository
+    #
+    # @return [Repositories::MonographRepository]
+    def repository
+      Repositories::MonographRepository.instance
+    end
+  end
+
+  # Configuration class for library settings
+  #
+  # @example
+  #   Openphar.configure do |config|
+  #     config.data_path = '/path/to/data'
+  #     config.logger = Logger.new(STDOUT)
+  #   end
+  class Configuration
+    attr_accessor :ontology_path, :data_path, :logger
+
+    def initialize
+      @ontology_path = Openphar.ontology_path
+      @data_path = Openphar.data_path
+      @logger = Logger.new($stdout, level: :warn)
+    end
+  end
+end
+
+# Load dependencies (external gems)
+require 'lutaml/model'
+require 'rdf'
+require 'rdf/turtle'
+require 'rdf/json'
+require 'json/ld'
+require 'nokogiri'
+require 'thor'
+
+# Load version (direct require, no autoload for constants)
+require_relative 'openphar/version'
+
+# Autoload direct children modules (one level deep only)
+# Each module file defines autoload for its own children
+module Openphar
+  autoload :Core, 'openphar/core'
+  autoload :Models, 'openphar/models'
+  autoload :Parsers, 'openphar/parsers'
+  autoload :Repositories, 'openphar/repositories'
+  autoload :Transformers, 'openphar/transformers'
+  autoload :Linkers, 'openphar/linkers'
+  autoload :Migrators, 'openphar/migrators'
+  autoload :Exporters, 'openphar/exporters'
+  autoload :Cli, 'openphar/cli'
+  autoload :Errors, 'openphar/errors'
+  autoload :MonographMerger, 'openphar/monograph_merger'
+  autoload :Registry, 'openphar/registry'
+end

data/ontology/bibliographic/edition.ttl

@@ -0,0 +1,267 @@
+# Open Pharmacopoeia - Bibliographic Ontology
+#
+# This module defines classes and properties for tracking pharmacopoeia
+# editions, supplements, and version history.
+#
+# Classes:
+#   - Edition: A specific edition of a pharmacopoeia publication
+#   - Supplement: A supplement to a pharmacopoeia edition
+#   - VersionHistory: Version tracking for monographs
+#
+@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix owl: <http://www.w3.org/2002/07/owl#> .
+@prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
+@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
+@prefix dct: <http://purl.org/dc/terms/> .
+@prefix op: <https://www.openphar.org/ontology/core/> .
+@prefix bib: <https://www.openphar.org/ontology/bibliographic/> .
+
+<https://www.openphar.org/ontology/bibliographic/>
+    a owl:Ontology ;
+    rdfs:label "Open Pharmacopoeia Bibliographic Ontology"@en ;
+    rdfs:comment "Ontology for tracking pharmacopoeia editions, supplements, and version history."@en ;
+    owl:imports <https://www.openphar.org/ontology/core/> ;
+    dct:created "2025-02-25"^^xsd:date ;
+    dct:modified "2025-02-25"^^xsd:date .
+
+# =============================================================================
+# EDITION CLASS
+# =============================================================================
+
+bib:Edition
+    a owl:Class ;
+    rdfs:label "Pharmacopoeia Edition"@en ;
+    rdfs:comment "A specific edition of a pharmacopoeia publication, such as JP18 or Ph.Int.13."@en ;
+    rdfs:subClassOf [
+        a owl:Restriction ;
+        owl:onProperty bib:editionNumber ;
+        owl:minCardinality 1
+    ] ;
+    rdfs:subClassOf [
+        a owl:Restriction ;
+        owl:onProperty bib:hasPublisher ;
+        owl:cardinality 1
+    ] .
+
+# Edition properties
+
+bib:editionNumber
+    a owl:DatatypeProperty ;
+    rdfs:label "edition number"@en ;
+    rdfs:comment "The edition number as a string (e.g., '18', '13')."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:string .
+
+bib:editionLabel
+    a owl:DatatypeProperty ;
+    rdfs:label "edition label"@en ;
+    rdfs:comment "A human-readable label for the edition (e.g., 'Japanese Pharmacopoeia 18th Edition')."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range rdf:langString .
+
+bib:publicationDate
+    a owl:DatatypeProperty ;
+    rdfs:label "publication date"@en ;
+    rdfs:comment "The date when the edition was officially published."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:date .
+
+bib:effectiveDate
+    a owl:DatatypeProperty ;
+    rdfs:label "effective date"@en ;
+    rdfs:comment "The date when the edition becomes officially effective."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:date .
+
+bib:hasPublisher
+    a owl:ObjectProperty ;
+    rdfs:label "has publisher"@en ;
+    rdfs:comment "The pharmacopoeia publisher of this edition."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range op:Publisher .
+
+bib:hasSupplement
+    a owl:ObjectProperty ;
+    rdfs:label "has supplement"@en ;
+    rdfs:comment "A supplement to this edition."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range bib:Supplement .
+
+bib:containsMonograph
+    a owl:ObjectProperty ;
+    rdfs:label "contains monograph"@en ;
+    rdfs:comment "A monograph included in this edition."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range op:PharmacopoeiaMonograph .
+
+bib:monographCount
+    a owl:DatatypeProperty ;
+    rdfs:label "monograph count"@en ;
+    rdfs:comment "The total number of monographs in this edition."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:integer .
+
+bib:officialLanguage
+    a owl:DatatypeProperty ;
+    rdfs:label "official language"@en ;
+    rdfs:comment "The official language(s) of the edition."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:string .
+
+bib:editionUrl
+    a owl:DatatypeProperty ;
+    rdfs:label "edition URL"@en ;
+    rdfs:comment "The official URL where the edition can be accessed."@en ;
+    rdfs:domain bib:Edition ;
+    rdfs:range xsd:anyURI .
+
+# =============================================================================
+# SUPPLEMENT CLASS
+# =============================================================================
+
+bib:Supplement
+    a owl:Class ;
+    rdfs:label "Pharmacopoeia Supplement"@en ;
+    rdfs:comment "A supplement to a pharmacopoeia edition containing additions, revisions, or deletions."@en ;
+    rdfs:subClassOf [
+        a owl:Restriction ;
+        owl:onProperty bib:supplementNumber ;
+        owl:minCardinality 1
+    ] ;
+    rdfs:subClassOf [
+        a owl:Restriction ;
+        owl:onProperty bib:supplementTo ;
+        owl:cardinality 1
+    ] .
+
+# Supplement properties
+
+bib:supplementNumber
+    a owl:DatatypeProperty ;
+    rdfs:label "supplement number"@en ;
+    rdfs:comment "The supplement number (e.g., '1', '2')."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range xsd:string .
+
+bib:supplementTo
+    a owl:ObjectProperty ;
+    rdfs:label "supplement to"@en ;
+    rdfs:comment "The edition this supplement applies to."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range bib:Edition .
+
+bib:supplementPublicationDate
+    a owl:DatatypeProperty ;
+    rdfs:label "supplement publication date"@en ;
+    rdfs:comment "The date when the supplement was published."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range xsd:date .
+
+bib:supplementEffectiveDate
+    a owl:DatatypeProperty ;
+    rdfs:label "supplement effective date"@en ;
+    rdfs:comment "The date when the supplement becomes effective."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range xsd:date .
+
+bib:addsMonograph
+    a owl:ObjectProperty ;
+    rdfs:label "adds monograph"@en ;
+    rdfs:comment "A monograph added by this supplement."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range op:PharmacopoeiaMonograph .
+
+bib:revisesMonograph
+    a owl:ObjectProperty ;
+    rdfs:label "revises monograph"@en ;
+    rdfs:comment "A monograph revised by this supplement."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range op:PharmacopoeiaMonograph .
+
+bib:deletesMonograph
+    a owl:ObjectProperty ;
+    rdfs:label "deletes monograph"@en ;
+    rdfs:comment "A monograph deleted by this supplement."@en ;
+    rdfs:domain bib:Supplement ;
+    rdfs:range op:PharmacopoeiaMonograph .
+
+# =============================================================================
+# VERSION HISTORY CLASS
+# =============================================================================
+
+bib:VersionHistory
+    a owl:Class ;
+    rdfs:label "Version History"@en ;
+    rdfs:comment "A record of version changes for a monograph across editions and supplements."@en .
+
+# Version history properties
+
+bib:versionFor
+    a owl:ObjectProperty ;
+    rdfs:label "version for"@en ;
+    rdfs:comment "The monograph this version history tracks."@en ;
+    rdfs:domain bib:VersionHistory ;
+    rdfs:range op:PharmacopoeiaMonograph .
+
+bib:hasVersion
+    a owl:ObjectProperty ;
+    rdfs:label "has version"@en ;
+    rdfs:comment "A specific version entry in the history."@en ;
+    rdfs:domain bib:VersionHistory ;
+    rdfs:range bib:MonographVersion .
+
+bib:MonographVersion
+    a owl:Class ;
+    rdfs:label "Monograph Version"@en ;
+    rdfs:comment "A specific version of a monograph in an edition or supplement."@en .
+
+bib:inEdition
+    a owl:ObjectProperty ;
+    rdfs:label "in edition"@en ;
+    rdfs:comment "The edition or supplement containing this version."@en ;
+    rdfs:domain bib:MonographVersion ;
+    rdfs:range bib:Edition .
+
+bib:changeType
+    a owl:ObjectProperty ;
+    rdfs:label "change type"@en ;
+    rdfs:comment "The type of change (new, revised, deleted, unchanged)."@en ;
+    rdfs:domain bib:MonographVersion ;
+    rdfs:range bib:ChangeType .
+
+bib:ChangeType
+    a owl:Class ;
+    rdfs:label "Change Type"@en ;
+    rdfs:comment "The type of change for a monograph version."@en .
+
+bib:NewMonograph
+    a bib:ChangeType ;
+    rdfs:label "New Monograph"@en ;
+    skos:prefLabel "New"@en .
+
+bib:RevisedMonograph
+    a bib:ChangeType ;
+    rdfs:label "Revised Monograph"@en ;
+    skos:prefLabel "Revised"@en .
+
+bib:DeletedMonograph
+    a bib:ChangeType ;
+    rdfs:label "Deleted Monograph"@en ;
+    skos:prefLabel "Deleted"@en .
+
+bib:UnchangedMonograph
+    a bib:ChangeType ;
+    rdfs:label "Unchanged Monograph"@en ;
+    skos:prefLabel "Unchanged"@en .
+
+# =============================================================================
+# MONOGRAPH TO EDITION LINKING PROPERTY
+# =============================================================================
+
+op:belongsToEdition
+    a owl:ObjectProperty ;
+    rdfs:label "belongs to edition"@en ;
+    rdfs:comment "The pharmacopoeia edition this monograph belongs to."@en ;
+    rdfs:domain op:PharmacopoeiaMonograph ;
+    rdfs:range bib:Edition .

data/ontology/context/chp.jsonld

@@ -0,0 +1,69 @@
+{
+  "@context": {
+    "@version": 1.1,
+    "@base": "https://www.openphar.org/data/chp/",
+    "@vocab": "https://www.openphar.org/ontology/publisher/chp/",
+
+    "op": "https://www.openphar.org/ontology/core/",
+    "chp": "https://www.openphar.org/ontology/publisher/chp/",
+    "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
+    "rdfs": "http://www.w3.org/2000/01/rdf-schema#",
+    "owl": "http://www.w3.org/2002/07/owl#",
+    "xsd": "http://www.w3.org/2001/XMLSchema#",
+    "skos": "http://www.w3.org/2004/02/skos/core#",
+    "dct": "http://purl.org/dc/terms/",
+
+    "Monograph": { "@id": "chp:Monograph" },
+    "TcmCrudeDrug": { "@id": "chp:TcmCrudeDrug" },
+    "TcmExtract": { "@id": "chp:TcmExtract" },
+    "TcmFormulation": { "@id": "chp:TcmFormulation" },
+    "ChemicalSubstance": { "@id": "chp:ChemicalSubstance" },
+    "ChemicalPreparation": { "@id": "chp:ChemicalPreparation" },
+    "Biologic": { "@id": "chp:Biologic" },
+    "GeneralChapter": { "@id": "chp:GeneralChapter" },
+    "Section": { "@id": "chp:Section" },
+
+    "entryId":       { "@id": "chp:entryId",       "@type": "xsd:integer" },
+    "bookId":        { "@id": "chp:bookId",        "@type": "xsd:integer" },
+    "bookName":      { "@id": "chp:bookName",      "@type": "xsd:string"  },
+    "directoryId":   { "@id": "chp:directoryId",   "@type": "xsd:integer" },
+    "directoryTitle":{ "@id": "chp:directoryTitle","@type": "xsd:string"  },
+    "pinyinTitle":   { "@id": "chp:pinyinTitle",   "@type": "xsd:string"  },
+    "eTitle":        { "@id": "chp:eTitle",        "@type": "xsd:string"  },
+    "pageNum":       { "@id": "chp:pageNum",       "@type": "xsd:integer" },
+    "htmlContent":   { "@id": "chp:htmlContent",   "@type": "xsd:string"  },
+    "partTitles":    { "@id": "chp:partTitle",     "@type": "xsd:string", "@container": "@set" },
+    "visitCount":    { "@id": "chp:visitCount",    "@type": "xsd:integer" },
+    "fetchedAt":     { "@id": "dct:issued",        "@type": "xsd:dateTime" },
+
+    "section": {
+      "@id": "chp:hasSection",
+      "@type": "@id",
+      "@container": "@set",
+      "@context": {
+        "@version": 1.1,
+        "@vocab": "https://www.openphar.org/ontology/publisher/chp/",
+        "sectionName":  { "@id": "chp:sectionName",  "@type": "xsd:string"  },
+        "sequence":     { "@id": "chp:sequence",     "@type": "xsd:integer" },
+        "htmlContent":  { "@id": "chp:htmlContent",  "@type": "xsd:string"  },
+        "textContent":  { "@id": "chp:textContent",  "@type": "xsd:string"  }
+      }
+    },
+
+    "slug":          { "@id": "chp:slug",          "@type": "xsd:string"  },
+    "wireKey":       { "@id": "chp:wireKey",       "@type": "xsd:string"  },
+
+    "title":         "dct:title",
+    "prefLabel":     "skos:prefLabel",
+    "publisher":     "dct:publisher",
+    "edition":       { "@id": "op:editionLabel", "@type": "xsd:string" },
+    "monographId":   { "@id": "op:monographId",   "@type": "xsd:string" },
+    "belongsToEdition": { "@id": "op:belongsToEdition", "@type": "@id" },
+    "definition":    "skos:definition",
+    "status":        { "@id": "op:status", "@type": "xsd:string" },
+    "sameSubstanceAs":  { "@id": "op:sameSubstanceAs",  "@type": "@id", "@container": "@set" },
+
+    "id":   "@id",
+    "type": "@type"
+  }
+}

data/ontology/context/crude-drugs.jsonld

@@ -0,0 +1,56 @@
+{
+  "@context": {
+    "@import": "https://www.openphar.org/ontology/context/pharmacopoeia.jsonld",
+    "@base": "https://www.openphar.org/data/monograph/",
+
+    "CrudeDrugMonograph": {
+      "@id": "op:CrudeDrugMonograph",
+      "@context": {
+        "botanicalSource": {
+          "@id": "op:botanicalSource",
+          "@type": "@id",
+          "@context": {
+            "referencesPlantSpecies": {
+              "@id": "op:referencesPlantSpecies",
+              "@type": "@id"
+            },
+            "plantPartUsed": { "@id": "op:plantPartUsed", "@container": "@language" },
+            "collectionPeriod": { "@id": "op:collectionPeriod", "@container": "@language" }
+          }
+        },
+        "macroscopicDescription": {
+          "@id": "op:macroscopicDescription",
+          "@type": "@id",
+          "@context": {
+            "color": { "@id": "op:color", "@container": "@language" },
+            "shape": { "@id": "op:shape", "@container": "@language" },
+            "size": { "@id": "op:size", "@container": "@language" },
+            "texture": { "@id": "op:texture", "@container": "@language" },
+            "odor": { "@id": "op:odor", "@container": "@language" },
+            "taste": { "@id": "op:taste", "@container": "@language" },
+            "fracture": { "@id": "op:fracture", "@container": "@language" }
+          }
+        },
+        "microscopicDescription": {
+          "@id": "op:microscopicDescription",
+          "@type": "@id",
+          "@context": {
+            "transverseSection": { "@id": "op:transverseSection", "@container": "@language" },
+            "powderCharacteristics": { "@id": "op:powderCharacteristics", "@container": "@language" },
+            "diagnosticFeatures": { "@id": "op:diagnosticFeatures", "@container": "@language" }
+          }
+        }
+      }
+    },
+
+    "extractiveTests": {
+      "@id": "op:extractiveTests",
+      "@container": "@set"
+    },
+
+    "AlcoholSolubleExtractive": { "@id": "method:AlcoholSolubleExtractive", "@type": "@id" },
+    "WaterSolubleExtractive": { "@id": "method:WaterSolubleExtractive", "@type": "@id" },
+    "DiluteAlcoholSolubleExtractive": { "@id": "method:DiluteAlcoholSolubleExtractive", "@type": "@id" },
+    "EtherSolubleExtractive": { "@id": "method:EtherSolubleExtractive", "@type": "@id" }
+  }
+}