Taking Heuristics for a spin.

lib/linguist/heuristics.rb

@@ -1,7 +1,7 @@
 module Linguist
   # A collection of simple heuristics that can be used to better analyze languages.
   class Heuristics
-    ACTIVE = false
+    ACTIVE = true
 
     # Public: Given an array of String language names,
     # apply heuristics against the given data and return an array
@@ -13,24 +13,13 @@ module Linguist
     # Returns an array of Languages or []
     def self.find_by_heuristics(data, languages)
       if active?
-        if languages.all? { |l| ["Objective-C", "C++"].include?(l) }
-          disambiguate_c(data, languages)
-        end
         if languages.all? { |l| ["Perl", "Prolog"].include?(l) }
-          disambiguate_pl(data, languages)
+          result = disambiguate_pl(data, languages)
         end
         if languages.all? { |l| ["ECL", "Prolog"].include?(l) }
-          disambiguate_ecl(data, languages)
+          result = disambiguate_ecl(data, languages)
-        end
-        if languages.all? { |l| ["TypeScript", "XML"].include?(l) }
-          disambiguate_ts(data, languages)
-        end
-        if languages.all? { |l| ["Common Lisp", "OpenCL"].include?(l) }
-          disambiguate_cl(data, languages)
-        end
-        if languages.all? { |l| ["Rebol", "R"].include?(l) }
-          disambiguate_r(data, languages)
         end
+        return result
       end
     end
 

lib/linguist/language.rb

@@ -136,7 +136,7 @@ module Linguist
         elsif (determined = Heuristics.find_by_heuristics(data, possible_language_names)) && !determined.empty?
           determined.first
         # Lastly, fall back to the probablistic classifier.
-        elsif classified = Classifier.classify(Samples::DATA, data, possible_language_names ).first
+        elsif classified = Classifier.classify(Samples::DATA, data, possible_language_names).first
           # Return the actual Language object based of the string language name (i.e., first element of `#classify`)
           Language[classified[0]]
         end

lib/linguist/languages.yml

@@ -1202,7 +1202,7 @@ LFE:
 LLVM:
   extensions:
   - .ll
-  
+
 LabVIEW:
   type: programming
   lexer: Text only
@@ -1784,9 +1784,9 @@ Prolog:
   type: programming
   color: "#74283c"
   extensions:
-  - .prolog
-  - .ecl
   - .pl
+  - .ecl
+  - .prolog
 
 Propeller Spin:
   type: programming

lib/linguist/samples.json

@@ -254,6 +254,9 @@
       ".dlm",
       ".pro"
     ],
+    "IGOR Pro": [
+      ".ipf"
+    ],
     "Idris": [
       ".idr"
     ],
@@ -845,8 +848,8 @@
       "exception.zep.php"
     ]
   },
-  "tokens_total": 654541,
+  "tokens_total": 659103,
-  "languages_total": 900,
+  "languages_total": 903,
   "tokens": {
     "ABAP": {
       "*/**": 1,
@@ -28185,6 +28188,52 @@
       "L": 1,
       "example": 1
     },
+    "IGOR Pro": {
+      "#pragma": 2,
+      "rtGlobals": 2,
+      "Function": 6,
+      "FooBar": 1,
+      "(": 10,
+      ")": 10,
+      "return": 7,
+      "End": 7,
+      "FooBarSubType": 1,
+      "ButtonControl": 1,
+      "Function/D": 1,
+      "FooBarVar": 1,
+      "static": 3,
+      "FooBarStatic": 1,
+      "threadsafe": 2,
+      "FooBarStaticThreadsafe": 1,
+      "FooBarThread": 1,
+      "CallOperationsAndBuiltInFuncs": 1,
+      "string": 4,
+      "var": 3,
+      "someDQString": 2,
+      "Make/N": 1,
+      "myWave": 2,
+      "Redimension/N": 1,
+      "-": 3,
+      "print": 1,
+      "strlen": 1,
+      "StrConstant": 1,
+      "myConstString": 1,
+      "constant": 1,
+      "myConst": 1,
+      "Structure": 2,
+      "struct1": 1,
+      "str": 2,
+      "variable": 2,
+      "EndStructure": 2,
+      "struct2": 1,
+      "#include": 1,
+      "#ifdef": 1,
+      "NOT_DEFINED": 1,
+      "//": 1,
+      "conditional": 1,
+      "compilation": 1,
+      "#endif": 1
+    },
     "INI": {
       ";": 1,
       "editorconfig.org": 1,
@@ -55878,54 +55927,434 @@
       "TWO_PI": 1
     },
     "Prolog": {
-      "-": 161,
+      "-": 348,
-      "module": 3,
+      "module": 4,
-      "(": 327,
+      "(": 1156,
+      "cpa_admin": 1,
+      "[": 290,
+      "change_password_form//1": 1,
+      "]": 288,
+      ")": 1153,
+      ".": 230,
+      "use_module": 20,
+      "user": 45,
+      "user_db": 1,
+      "library": 19,
+      "http/http_parameters": 1,
+      "http/http_session": 1,
+      "http/html_write": 1,
+      "http/html_head": 1,
+      "http/mimetype": 1,
+      "http/http_dispatch": 1,
+      "url": 1,
+      "debug": 5,
+      "lists": 1,
+      "option": 8,
+      "http_settings": 1,
+      "http_handler": 20,
+      "cliopatria": 38,
+      "list_users": 13,
+      "create_admin": 3,
+      "add_user_form": 4,
+      "add_openid_server_form": 4,
+      "add_user": 6,
+      "self_register": 4,
+      "add_openid_server": 3,
+      "edit_user_form": 7,
+      "edit_user": 4,
+      "del_user": 5,
+      "edit_openid_server_form": 6,
+      "edit_openid_server": 4,
+      "del_openid_server": 4,
+      "change_password_form": 7,
+      "change_password": 5,
+      "login_form": 4,
+      "user_login": 4,
+      "user_logout": 4,
+      "settings": 4,
+      "save_settings": 3,
+      "%": 193,
+      "+": 47,
+      "Request": 46,
+      "HTTP": 4,
+      "Handler": 1,
+      "listing": 1,
+      "registered": 3,
+      "users.": 2,
+      "_Request": 11,
+      "authorized": 13,
+      "admin": 31,
+      "if_allowed": 4,
+      "edit": 15,
+      "true": 32,
+      "UserOptions": 5,
+      "openid": 5,
+      "OpenIDOptions": 2,
+      "reply_html_page": 15,
+      "default": 16,
+      "title": 13,
+      "h1": 15,
+      "user_table": 3,
+      "p": 11,
+      "action": 18,
+      "location_by_id": 13,
+      "openid_server_table": 4,
+      "Token": 2,
+      "Options": 43,
+      "logged_on": 4,
+      "User": 77,
+      "anonymous": 2,
+      "catch": 2,
+      "check_permission": 1,
+      "_": 62,
+      "fail": 3,
+      "//": 7,
+      "HTML": 4,
+      "component": 3,
+      "generating": 1,
+      "a": 25,
+      "table": 9,
+      "of": 10,
+      "{": 22,
+      "setof": 3,
+      "U": 2,
+      "current_user": 6,
+      "Users": 2,
+      "}": 22,
+      "html": 25,
+      "class": 14,
+      "block": 2,
+      "tr": 16,
+      "th": 12,
+      "|": 44,
+      "T": 4,
+      "user_property": 8,
+      "realname": 18,
+      "Name": 19,
+      "findall": 1,
+      "Idle": 6,
+      "Login": 9,
+      "connection": 1,
+      "Pairs0": 2,
+      "keysort": 1,
+      "Pairs": 4,
+      "OnLine": 4,
+      ";": 37,
+      "length": 9,
+      "N": 7,
+      "online": 3,
+      "td": 20,
+      "on_since": 3,
+      "idle": 3,
+      "edit_user_button": 3,
+      "href": 6,
+      "encode": 4,
+      "_Idle": 1,
+      "_Connections": 2,
+      "format_time": 1,
+      "string": 10,
+      "Date": 2,
+      "_Login": 1,
+      "mmss_duration": 2,
+      "String": 10,
+      "Time": 3,
+      "in": 1,
+      "seconds": 1,
+      "Secs": 4,
+      "is": 24,
+      "round": 1,
+      "Hour": 2,
+      "Min": 4,
+      "mod": 2,
+      "Sec": 2,
+      "format": 9,
+      "Create": 2,
+      "the": 14,
+      "administrator": 1,
+      "login.": 2,
+      "throw": 12,
+      "error": 13,
+      "permission_error": 4,
+      "create": 3,
+      "context": 3,
+      "align": 16,
+      "center": 6,
+      "new_user_form": 3,
+      "real_name": 2,
+      "Form": 3,
+      "to": 19,
+      "register": 3,
+      "user.": 2,
+      "value": 11,
+      "PermUser": 3,
+      "form": 14,
+      "method": 6,
+      "input": 25,
+      "pwd1": 5,
+      "type": 16,
+      "password": 21,
+      "pwd2": 5,
+      "permissions": 5,
+      "buttons": 5,
+      "colspan": 6,
+      "right": 11,
+      "submit": 6,
+      "Label": 10,
+      "name": 8,
+      "size": 5,
+      "Only": 1,
+      "provide": 1,
+      "field": 4,
+      "if": 4,
+      "this": 1,
+      "not": 1,
+      "already": 2,
+      "given.": 1,
+      "This": 2,
+      "because": 1,
+      "firefox": 1,
+      "determines": 1,
+      "login": 3,
+      "from": 3,
+      "text": 6,
+      "immediately": 1,
+      "above": 1,
+      "entry.": 1,
+      "Other": 1,
+      "browsers": 1,
+      "may": 1,
+      "do": 1,
+      "it": 1,
+      "different": 1,
+      "so": 1,
+      "only": 2,
+      "having": 1,
+      "one": 1,
+      "probably": 1,
+      "savest": 1,
+      "solution.": 1,
+      "RealName": 13,
+      "hidden": 7,
+      "_Options": 1,
+      "API": 1,
+      "new": 6,
+      "The": 3,
+      "current": 3,
+      "must": 3,
+      "have": 1,
+      "administrative": 2,
+      "rights": 2,
+      "or": 3,
+      "database": 1,
+      "be": 6,
+      "empty.": 1,
+      "FirstUser": 2,
+      "http_parameters": 10,
+      "Password": 13,
+      "Retype": 4,
+      "read": 13,
+      "Read": 12,
+      "write": 11,
+      "Write": 12,
+      "Admin": 12,
+      "attribute_declarations": 10,
+      "attribute_decl": 21,
+      "password_mismatch": 2,
+      "password_hash": 3,
+      "Hash": 6,
+      "phrase": 6,
+      "allow": 17,
+      "Allow": 11,
+      "user_add": 3,
+      "reply_login": 5,
+      "Self": 1,
+      "and": 6,
+      "enable_self_register": 3,
+      "set": 2,
+      "true.": 1,
+      "limited": 1,
+      "annotate": 2,
+      "access.": 1,
+      "Returns": 1,
+      "forbidden": 3,
+      "false": 5,
+      "exists": 1,
+      "http_location_by_id": 1,
+      "MyUrl": 3,
+      "setting": 1,
+      "http_reply": 3,
+      "properties": 3,
+      "User.": 1,
+      "b": 6,
+      "i": 1,
+      "Term": 16,
+      "..": 11,
+      "Value": 15,
+      "O2": 6,
+      "p_name": 2,
+      "permission_checkbox": 4,
+      "Actions": 3,
+      "openid_server_property": 3,
+      "pterm": 5,
+      "Action": 17,
+      "memberchk": 4,
+      "Opts": 4,
+      "checked": 2,
+      "def_user_permissions": 3,
+      "DefPermissions": 2,
+      "checkbox": 1,
+      "Handle": 3,
+      "reply": 3,
+      "form.": 2,
+      "optional": 6,
+      "description": 18,
+      "modify_user": 2,
+      "modify_permissions": 2,
+      "Property": 6,
+      "_Name": 2,
+      "var": 7,
+      "set_user_property": 3,
+      "Access": 2,
+      "on": 2,
+      "_Access": 1,
+      "off": 5,
+      "_Repositiory": 2,
+      "_Action": 3,
+      "Delete": 2,
+      "delete": 2,
+      "user_del": 1,
+      "change": 2,
+      "context_error": 2,
+      "not_logged_in": 2,
+      "UserID": 1,
+      "that": 2,
+      "shows": 1,
+      "for": 4,
+      "changing": 1,
+      "UserID.": 1,
+      "id": 3,
+      "user_or_old": 3,
+      "pwd0": 2,
+      "handler": 2,
+      "password.": 1,
+      "logged": 1,
+      "on.": 1,
+      "New": 3,
+      "existence_error": 1,
+      "validate_password": 2,
+      "presents": 1,
+      "If": 2,
+      "there": 1,
+      "parameter": 2,
+      "return_to": 3,
+      "openid.return_to": 1,
+      "using": 1,
+      "redirect": 1,
+      "given": 2,
+      "URL.": 1,
+      "Otherwise": 1,
+      "display": 1,
+      "welcome": 1,
+      "page.": 1,
+      "ReturnTo": 6,
+      "Extra": 3,
+      "moved_temporary": 1,
+      "Logout": 1,
+      "logout": 1,
+      "Param": 1,
+      "DeclObtions": 1,
+      "semidet.": 4,
+      "Provide": 1,
+      "reusable": 1,
+      "declarations": 1,
+      "calls": 2,
+      "http_parameters/3.": 1,
+      "openid_server": 9,
+      "bool": 4,
+      "Def": 2,
+      "oneof": 1,
+      "Return": 1,
+      "an": 3,
+      "page": 1,
+      "add": 2,
+      "OpenID": 4,
+      "server.": 1,
+      "new_openid_form": 2,
+      "new_openid_form//": 1,
+      "det.": 5,
+      "Present": 1,
+      "provider.": 1,
+      "openid_description": 1,
+      "code": 2,
+      "canonical_url": 1,
+      "URL0": 2,
+      "URL": 4,
+      "parse_url": 2,
+      "Parts": 2,
+      "Server": 30,
+      "openid_property": 2,
+      "List": 1,
+      "servers": 1,
+      "S": 2,
+      "openid_current_server": 1,
+      "Servers": 2,
+      "openid_list_servers": 4,
+      "H": 2,
+      "openid_list_server": 2,
+      "openid_field": 3,
+      "edit_openid_button": 3,
+      "Field": 2,
+      "server": 1,
+      "Description": 2,
+      "modify_openid": 2,
+      "openid_modify_permissions": 2,
+      "openid_set_property": 2,
+      "openid_del_server": 1,
+      "Show": 1,
+      "settings.": 3,
+      "has": 1,
+      "editing": 1,
+      "edit_settings": 2,
+      "Edit": 4,
+      "http_show_settings": 1,
+      "hide_module": 1,
+      "warn_no_edit": 3,
+      "settings_no_edit": 1,
+      "Save": 1,
+      "modified": 1,
+      "http_apply_settings": 1,
+      "save": 1,
+      "with": 3,
+      "br": 1,
       "format_spec": 12,
-      "[": 87,
       "format_error/2": 1,
       "format_spec/2": 1,
       "format_spec//1": 1,
       "spec_arity/2": 1,
       "spec_types/2": 1,
-      "]": 87,
-      ")": 326,
-      ".": 107,
-      "use_module": 8,
-      "library": 8,
       "dcg/basics": 1,
       "eos//0": 1,
       "integer//1": 1,
       "string_without//2": 1,
-      "error": 6,
       "when": 3,
       "when/2": 1,
-      "%": 71,
       "mavis": 1,
       "format_error": 8,
-      "+": 14,
       "Goal": 29,
       "Error": 25,
-      "string": 8,
-      "is": 12,
       "nondet.": 1,
-      "format": 8,
       "Format": 23,
       "Args": 19,
       "format_error_": 5,
-      "_": 30,
-      "debug": 4,
       "Spec": 10,
       "is_list": 1,
       "spec_types": 8,
       "Types": 16,
       "types_error": 3,
-      "length": 4,
       "TypesLen": 3,
       "ArgsLen": 3,
       "types_error_": 4,
       "Arg": 6,
-      "|": 25,
       "Type": 3,
       "ground": 5,
       "is_of_type": 2,
@@ -55940,13 +56369,9 @@
       "format_fail/3.": 1,
       "prolog_walk_code": 1,
       "module_class": 1,
-      "user": 5,
       "infer_meta_predicates": 1,
-      "false": 2,
       "autoload": 1,
       "format/": 1,
-      "{": 7,
-      "}": 7,
       "are": 3,
       "always": 1,
       "loaded": 1,
@@ -55967,7 +56392,6 @@
       "checker.": 1,
       "succeed": 2,
       "even": 1,
-      "if": 1,
       "no": 1,
       "found": 1,
       "Module": 4,
@@ -55975,12 +56399,10 @@
       "predicate_property": 1,
       "imported_from": 1,
       "Source": 2,
-      "memberchk": 2,
       "system": 1,
       "prolog_debug": 1,
       "can_check": 2,
       "assert": 2,
-      "to": 5,
       "avoid": 1,
       "printing": 1,
       "goals": 1,
@@ -55989,32 +56411,22 @@
       "prolog": 2,
       "message": 1,
       "message_location": 1,
-      "//": 1,
       "eos.": 1,
       "escape": 2,
       "Numeric": 4,
       "Modifier": 2,
-      "Action": 15,
       "Rest": 12,
       "numeric_argument": 5,
       "modifier_argument": 3,
-      "action": 6,
-      "text": 4,
-      "String": 6,
-      ";": 12,
       "Codes": 21,
       "string_codes": 4,
       "string_without": 1,
       "list": 4,
-      "semidet.": 3,
       "text_codes": 6,
-      "phrase": 3,
       "spec_arity": 2,
       "FormatSpec": 2,
       "Arity": 3,
       "positive_integer": 1,
-      "det.": 4,
-      "type": 2,
       "Item": 2,
       "Items": 2,
       "item_types": 3,
@@ -56029,17 +56441,14 @@
       "Text": 1,
       "codes": 5,
       "Var": 5,
-      "var": 4,
       "Atom": 3,
       "atom": 6,
-      "N": 5,
       "integer": 7,
       "C": 5,
       "colon": 1,
       "no_colon": 1,
       "is_action": 4,
       "multi.": 1,
-      "a": 4,
       "d": 3,
       "e": 1,
       "float": 3,
@@ -56047,7 +56456,6 @@
       "G": 2,
       "I": 1,
       "n": 1,
-      "p": 1,
       "any": 3,
       "r": 1,
       "s": 2,
@@ -56056,7 +56464,6 @@
       "func": 13,
       "op": 2,
       "xfy": 2,
-      "of": 8,
       "/2": 3,
       "list_util": 1,
       "xfy_list/3": 1,
@@ -56070,7 +56477,6 @@
       "at": 3,
       "compile": 3,
       "time": 3,
-      "for": 1,
       "macro": 1,
       "expansion.": 1,
       "compile_function/4.": 1,
@@ -56081,19 +56487,14 @@
       "evaluable/1": 1,
       "throws": 1,
       "exception": 1,
-      "with": 2,
       "strings": 1,
       "evaluable": 1,
       "term_variables": 1,
       "F": 26,
       "function_composition_term": 2,
       "Functor": 8,
-      "true": 5,
-      "..": 6,
       "format_template": 7,
       "has_type": 2,
-      "fail": 1,
-      "be": 4,
       "explicit": 1,
       "Dict": 3,
       "is_dict": 1,
@@ -56101,7 +56502,6 @@
       "Function": 5,
       "Argument": 1,
       "Apply": 1,
-      "an": 1,
       "Argument.": 1,
       "A": 1,
       "predicate": 4,
@@ -56110,18 +56510,15 @@
       "argument": 2,
       "generates": 1,
       "output": 1,
-      "and": 2,
       "penultimate": 1,
       "accepts": 1,
       "input.": 1,
-      "This": 1,
       "realized": 1,
       "by": 2,
       "expanding": 1,
       "function": 2,
       "application": 2,
       "chained": 1,
-      "calls": 1,
       "time.": 1,
       "itself": 1,
       "can": 3,
@@ -56130,12 +56527,8 @@
       "reverse": 4,
       "sort": 2,
       "c": 2,
-      "b": 4,
       "meta_predicate": 2,
-      "throw": 1,
-      "permission_error": 1,
       "call": 4,
-      "context": 1,
       "X": 10,
       "Y": 7,
       "defer": 1,
@@ -56143,13 +56536,10 @@
       "run": 1,
       "P": 2,
       "Creates": 1,
-      "new": 2,
       "composing": 1,
       "G.": 1,
-      "The": 1,
       "functions": 2,
       "composed": 1,
-      "create": 1,
       "compiled": 1,
       "which": 1,
       "behaves": 1,
@@ -56164,7 +56554,6 @@
       "syntax": 1,
       "highlighting": 1,
       "functions_to_compose": 2,
-      "Term": 10,
       "Funcs": 7,
       "functor": 1,
       "Op": 3,
@@ -56186,9 +56575,7 @@
       "compile_predicates": 1,
       "Output": 2,
       "compound": 1,
-      "setof": 1,
       "arg": 1,
-      "Name": 2,
       "Args0": 2,
       "nth1": 2,
       "lib": 1,
@@ -56199,11 +56586,9 @@
       "#": 9,
       "labeling": 2,
       "vabsIC": 1,
-      "or": 1,
       "faitListe": 3,
       "First": 2,
       "Taille": 2,
-      "Min": 2,
       "Max": 2,
       "Min..Max": 1,
       "Taille1": 2,
@@ -56258,7 +56643,6 @@
       "Rs1": 2,
       "left": 4,
       "stay": 1,
-      "right": 1,
       "L": 2
     },
     "Propeller Spin": {
@@ -72649,6 +73033,7 @@
     "Haskell": 302,
     "Hy": 155,
     "IDL": 418,
+    "IGOR Pro": 97,
     "INI": 27,
     "Idris": 148,
     "Inform 7": 75,
@@ -72725,7 +73110,7 @@
     "PostScript": 107,
     "PowerShell": 12,
     "Processing": 74,
-    "Prolog": 2420,
+    "Prolog": 6885,
     "Propeller Spin": 13519,
     "Protocol Buffer": 63,
     "PureScript": 1652,
@@ -72862,6 +73247,7 @@
     "Haskell": 3,
     "Hy": 2,
     "IDL": 4,
+    "IGOR Pro": 2,
     "INI": 2,
     "Idris": 1,
     "Inform 7": 2,
@@ -72938,7 +73324,7 @@
     "PostScript": 1,
     "PowerShell": 2,
     "Processing": 1,
-    "Prolog": 5,
+    "Prolog": 6,
     "Propeller Spin": 10,
     "Protocol Buffer": 1,
     "PureScript": 4,
@@ -73005,5 +73391,5 @@
     "fish": 3,
     "wisp": 1
   },
-  "md5": "7a970958bd95602c130be259e8f3fc31"
+  "md5": "1500ca2d5506e7d4e02f8c1d14e348a4"
 }