openc_bot 0.0.11

data/openc_bot.gemspec

@@ -0,0 +1,35 @@
+# -*- encoding: utf-8 -*-
+lib = File.expand_path('../lib', __FILE__)
+$LOAD_PATH.unshift(lib) unless $LOAD_PATH.include?(lib)
+require 'openc_bot/version'
+
+Gem::Specification.new do |gem|
+  gem.name          = "openc_bot"
+  gem.version       = OpencBot::VERSION
+  gem.authors       = ["Chris Taggart"]
+  gem.email         = ["info@opencorporates.com"]
+  gem.description   = %q{This gem is to make the writing and running of bots for OpenCorporates quick and easy}
+  gem.summary       = %q{Helper gem for writing external bots for OpenCorporates}
+  gem.homepage      = ""
+
+  gem.files         = `git ls-files`.split($/)
+
+  gem.executables   = ['openc_bot']
+
+  gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
+  gem.require_paths = ["lib",'lib/openc_bot/helpers']
+
+  gem.add_dependency "rake"
+  gem.add_dependency "activesupport", "3.2.17"
+  gem.add_dependency "nokogiri"
+  # gem.add_dependency "sqlite3"
+  gem.add_dependency "json"
+  gem.add_dependency "json-schema"
+  gem.add_dependency "httpclient"
+  gem.add_dependency "backports"
+  gem.add_dependency "scraperwiki", "3.0.2"
+
+  gem.add_development_dependency "perftools.rb"
+  gem.add_development_dependency "debugger"
+  gem.add_development_dependency "rspec"
+end

data/schemas/company-schema.json

@@ -0,0 +1,112 @@
+{
+  "$schema": "http://json-schema.org/draft-04/schema#",
+  "title": "Company Schema",
+  "type": "object",
+  "description": "A company in OpenCorporates",
+  "properties": {
+    "company_number": {
+      "type": "string",
+      "minLength": 1
+    },
+    "name": {
+      "type": "string",
+      "minLength": 1
+    },
+    "jurisdiction_code": {
+      "type": "string",
+      "minLength": 2,
+      "maxLength": 5
+    },
+    "incorporation_date": {
+      "type": "date"
+    },
+    "dissolution_date": {
+      "type": "date"
+    },
+    "retrieved_at": {
+      "type": "date-time"
+    },
+    "current_status": {
+      "type": "string"
+    },
+    "company_type": {
+      "type": "string"
+    },
+    "registry_url": {
+      "type": "string"
+    },
+    "registered_address": {
+      "$ref": "includes/address.json"
+    },
+    "officers": {
+      "type": "array",
+      "items": {
+        "$ref": "includes/officer.json"
+      }
+    },
+    "share_parcels": {
+      "type": "array",
+      "items": {
+        "$ref": "includes/share-parcel.json"
+      }
+    },
+    "total_shares": {
+      "$ref": "includes/total-shares.json"
+    },
+    "filings": {
+      "type": "array",
+      "items": {
+        "$ref": "includes/filing.json"
+      }
+    },
+    "previous_names": {
+      "type": "array",
+      "items": {
+        "$ref": "includes/previous_name.json"
+      }
+    },
+    "branch": {
+      "type": ["string","null"],
+      "description": "A flag to denote whether a company is a branch entity. This should only be set if the company is a type of branch (otherwise should be null). In general the only option here is 'F' for a 'Foreign' branch, i.e. an out-of-jurisdiction entity that has registered as having a presence in the jurisdiction. In the US this is sometimes called a Foreign Corporation",
+      "enum": ["F", "L",null]
+    },
+    "all_attributes": {
+      "type": "object",
+      "description": "Other arbitrary attributes for a given company",
+      "properties": {
+        "jurisdiction_of_origin": {
+          "type": ["string","null"],
+          "description": "The jurisdiction of the 'home' company if this is a branch",
+          "minLength": 1
+        },
+        "home_company_number": {
+          "type": ["string","null"],
+          "description": "If the entity is a 'branch', this is the company_number of the 'home' company in the home company's jurisdiction",
+          "minLength": 1
+        },
+        "home_legal_name": {
+          "type": ["string","null"],
+          "description": "The legal name of the 'home' company in its jurisdiction if this is a branch, and the name is different from the legal name of the branch",
+          "minLength": 1
+        },
+        "registered_agent_address": {
+          "type": ["string","null"],
+          "description": "The address of the 'Agent', a public address to which legal papers can be served",
+          "minLength": 1
+        },
+        "registered_agent_name": {
+          "type": ["string","null"],
+          "description": "The 'Agent' of the company \u2013 a person or entity that is empowered to accept service for the company",
+          "minLength": 1
+        },
+        "number_of_employees": {
+          "type": "number",
+          "description": "The number of employees",
+          "minimum": 0
+        }
+      }
+    }
+  },
+  "additionalProperties": false,
+  "required": ["company_number", "name", "jurisdiction_code"]
+}

data/schemas/includes/address.json

@@ -0,0 +1,23 @@
+{ "oneOf":
+  [
+    {
+      "type": ["string","null"],
+      "minLength": 2
+    },
+    {
+      "name": "Address",
+      "description": "An address object",
+      "type": "object",
+      "properties": { "street_address": { "type": "string" },
+                      "locality": { "type": "date" },
+                      "region":   { "type": "string" },
+                      "postal_code":   { "type": "string" },
+                      "country":   { "type": "string" }
+      },
+      "anyOf": [ { "required": ["street_address"] },
+                 { "required": ["postal_code"] }
+               ]
+    }
+  ]
+}
+

data/schemas/includes/base-statement.json

@@ -0,0 +1,27 @@
+{
+  "name": "Base statement",
+  "description": "A base statement which must be extended by a schema to be used in OpenCorporates",
+  "type": "object",
+  "properties": {
+    "source_url": {
+      "description": "Place where this fact can be verified",
+      "type": "string"
+    },
+    "source_jurisdiction": {
+      "description": "Jurisdiction of the source of the data",
+      "type": "string"
+    },
+    "sample_date": {
+      "description": "Date on which this fact was known to be true",
+      "type": "string"
+    },
+    "company": {
+      "$ref" : "../includes/company.json"
+    },
+    "data": {
+      "type": "array",
+      "additionalItems": false
+    }
+  },
+  "required": ["source_url", "data", "sample_date", "company"]
+}

data/schemas/includes/company.json

@@ -0,0 +1,14 @@
+{
+  "name": "Company",
+  "description": "A company which is the subject of a statement",
+  "type": "object",
+  "properties": {
+    "name": { "type": "string" },
+    "jurisdiction": { "type": "string" },
+    "identifier": {
+      "type": "string",
+      "pattern": "^[^/]+/[^/]+$"
+    }
+  },
+  "required": [ "name", "jurisdiction" ]
+}

data/schemas/includes/filing.json

@@ -0,0 +1,20 @@
+{
+  "name": "Filing",
+  "description": "A statutory filing",
+  "type": "object",
+  "properties": { "title": { "type": "string" },
+                  "date": { "type": "date" },
+                  "description":   { "type": "string" },
+                  "uid":   { "type": "string" },
+                  "url":   { "type": "string" },
+                  "filing_type_code":   { "type": "string" },
+                  "filing_type_name":   { "type": "string" },
+                  "other_attributes": { "type": "object" }
+  },
+  "required": [ "date" ],
+  "anyOf": [ { "required": ["title"] },
+             { "required": ["description"] },
+             { "required": ["filing_type_name"] }
+           ]
+}
+

data/schemas/includes/license-data.json

@@ -0,0 +1,27 @@
+{
+  "name": "License Data",
+  "type": "object",
+  "properties": {
+    "data_type": { "enum": [ "licence" ] },
+    "properties": {
+      "type": "object",
+      "properties": {
+        "regulator": {
+          "description": "The regulating body that issued the licence",
+          "type": "string"},
+        "jurisdiction_code": {
+          "description": "The jurisdiction for which licence was issued",
+          "type": "string"},
+        "licence_number": {
+          "type": "string"},
+        "jurisdiction_classification": {
+          "type": "array",
+          "minItems": 1},
+        "oc_classification": {
+          "type": "array"}
+      },
+      "required": [ "jurisdiction_code", "jurisdiction_classification" ]
+    }
+  },
+  "required": [ "data_type", "properties" ]
+}

data/schemas/includes/officer.json

@@ -0,0 +1,14 @@
+{
+  "name": "Officer",
+  "description": "An officer (director, senior executive) of a company",
+  "type": "object",
+  "properties": { "name": { "type": "string", "minLength": 1 },
+                  "start_date": { "type": "date" },
+                  "end_date":   { "type": "date" },
+                  "position":   { "type": "string" },
+                  "other_attributes": { "type": "object" },
+                  "uid":   { "type": "string" }
+  },
+  "required": [ "name" ]
+}
+

data/schemas/includes/previous_name.json

@@ -0,0 +1,11 @@
+{
+  "name": "PreviousName",
+  "description": "A previous name of a company",
+  "type": "object",
+  "properties": { "company_name": { "type": "string", "minLength": 1 },
+                  "con_date": { "type": "date" },
+                  "start_date": { "type": "date" }
+  },
+  "required": [ "company_name" ]
+}
+

data/schemas/includes/share-parcel-data.json

@@ -0,0 +1,67 @@
+{
+  "name": "Share Parcel Data",
+  "type": "object",
+  "properties": {
+    "data_type": { "enum": [ "share_parcel" ] },
+    "properties": {
+      "type": "object",
+      "required": ["shareholders"],
+      "properties": {
+        "number_of_shares": {
+          "description": "Number of shares, if known",
+          "type": "integer"},
+        "percentage_of_shares": {
+          "description": "Percentage of shares, if known",
+          "type": "number",
+          "maximum": 100,
+          "minimum": 0
+        },
+        "shareholders": {
+          "description": "Legal persons who own this share parcel",
+          "type": "array",
+          "minItems": 1,
+          "items": {
+            "anyOf": [
+              {"type": "object",
+               "name": "person",
+               "required": ["name"],
+               "properties": {
+                 "name": {
+                   "description": "Name of natural person or company",
+                   "type": "string"
+                 },
+                 "jurisdiction": {
+                   "description": "Jurisdiction of registration, if company",
+                   "type": "string"
+                 },
+                 "company_number": {
+                   "description": "Company number, if company and known",
+                   "type": "string"
+                 },
+                 "identifier": {
+                   "description": "Unique identifier of person",
+                   "type": "string"
+                 },
+                 "type": {
+                   "description": "Type of person if known (company or natural person)",
+                   "enum" : ["Company", "Person"]
+                 },
+                 "address": {
+                   "description": "Address given for owner of parcel",
+                   "type": "string"
+                 },
+                 "address_country": {
+                   "description": "Country part of owner's address",
+                   "type": "string"
+                 }
+
+               }
+              }
+            ]
+          }
+        }
+      }
+    }
+  },
+  "required": [ "data_type", "properties" ]
+}

data/schemas/includes/share-parcel.json

@@ -0,0 +1,60 @@
+{
+  "name": "Share Parcel",
+  "type": "object",
+  "description": "A parcel of shares in a company",
+    "required": ["shareholders"],
+    "properties": {
+      "number_of_shares": {
+        "description": "Number of shares, if known",
+        "type": "integer"},
+      "percentage_of_shares": {
+        "description": "Percentage of shares, if known",
+        "type": "number",
+        "maximum": 100,
+        "minimum": 0
+      },
+      "shareholders": {
+        "description": "Legal persons who own this share parcel",
+        "type": "array",
+        "minItems": 1,
+        "items": {
+          "type": "object",
+          "name": "shareholder",
+          "required": ["name"],
+          "properties": {
+            "name": {
+              "description": "Name of natural person or company",
+              "type": "string",
+              "minLength": 1
+            },
+            "jurisdiction": {
+              "description": "Jurisdiction of registration, if company",
+              "type": "string"
+            },
+            "company_number": {
+              "description": "Company number, if company and known",
+              "type": "string"
+            },
+            "identifier": {
+              "description": "Unique identifier of person",
+              "type": "string"
+            },
+            "type": {
+              "description": "Type of person if known (company or natural person)",
+              "enum" : ["Company", "Person"]
+            },
+            "address": {
+              "description": "Address given for owner of parcel",
+              "type": "string"
+            },
+            "address_country": {
+              "description": "Country part of owner's address",
+              "type": "string"
+            }
+
+          }
+        }
+    }
+  },
+  "required": [ ]
+}

data/schemas/includes/subsidiary-relationship-data.json

@@ -0,0 +1,52 @@
+{
+  "name": "Share Parcel Data",
+  "type": "object",
+  "properties": {
+    "data_type": { "enum": [ "subsidiary_relationship" ] },
+    "properties": {
+      "type": "object",
+      "required": ["subsidiary"],
+      "additionalProperties": false,
+      "properties": {
+        "direct": {
+          "description": "If the control is direct (if via an intermediary, this value should be false; if unknown, left blank)",
+          "type": "boolean"},
+        "significant": {
+          "description": "Does the source define the control as somehow significant?",
+          "type": "boolean"},
+        "subsidiary": {
+          "description": "Company that is controlled",
+          "type": "object",
+          "required": ["name"],
+          "additionalProperties": false,
+          "properties": {
+            "name": {
+              "description": "Name of company",
+              "type": "string"
+            },
+            "jurisdiction": {
+              "description": "Jurisdiction of registration",
+              "type": "string"
+            },
+            "company_number": {
+              "description": "Company number, if company and known",
+              "type": "string"
+            },
+            "identifier": {
+              "description": "Unique identifier of person",
+              "type": "string"
+            },
+            "address": {
+              "description": "Address given for owner of parcel",
+              "type": "string"
+            },
+            "address_country": {
+              "description": "Country part of owner's address",
+              "type": "string"
+            }
+          }
+        }
+      }
+    }
+  }
+}