Merge pull request #198 from bee-san/bee-set-to-0

Set some regex to 0 to prevent spammy reports

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "pywhat"
-version = "4.3.0"
+version = "4.3.1"
 description = "What is that thing?"
 authors = ["Bee <github@skerritt.blog>"]
 license = "MIT"

pywhat/Data/regex.json

@@ -1763,7 +1763,7 @@
       "Regex": "^([0-9a-fA-F]{24})$",
       "plural_name": false,
       "Description": null,
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "Identifiers",
@@ -1776,7 +1776,7 @@
       "Regex": "^([0-9]{10})$",
       "plural_name": false,
       "Description": "Seconds elapsed since unix epoch: 1970, between year 2001 and 2286",
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "UNIX Timestamp",
@@ -1789,7 +1789,7 @@
       "Regex": "^([0-9]{13})$",
       "plural_name": false,
       "Description": "Milliseconds elapsed since unix epoch: 1970, between year 2001 and 2286",
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "UNIX Timestamp",
@@ -1802,7 +1802,7 @@
       "Regex": "^([0-9]{8,10})$",
       "plural_name": false,
       "Description": "Seconds elapsed since unix epoch: 1970",
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "UNIX Timestamp",
@@ -1815,7 +1815,7 @@
       "Regex": "^([0-9]{11,13})$",
       "plural_name": false,
       "Description": "Milliseconds elapsed since unix epoch: 1970",
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "UNIX Timestamp",
@@ -1828,7 +1828,7 @@
       "Regex": "^([0-9A-HJKMNP-TV-Z]{26})$",
       "plural_name": false,
       "Description": null,
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "Identifiers",
@@ -1840,7 +1840,7 @@
       "Regex": "^((?=.*[A-Z])(?=.*[a-z])[0-9A-Za-z_-]{10}[048AEIMQUYcgkosw]{1})$",
       "plural_name": false,
       "Description": null,
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": "https://www.youtube.com/watch?v=",
       "Tags": [
          "Media",
@@ -1853,7 +1853,7 @@
       "Regex": "^([0-9]{10})$",
       "plural_name": false,
       "Description": null,
-      "Rarity": 0.1,
+      "Rarity": 0,
       "URL": null,
       "Tags": [
          "Credentials",

pywhat/__init__.py

@@ -2,7 +2,7 @@
 from pywhat.helper import AvailableTags, Keys
 from pywhat.identifier import Identifier
 
-__version__ = "4.3.0"
+__version__ = "4.3.1"
 
 tags = AvailableTags().get_tags()
 pywhat_tags = tags  # left for backward compatibility purposes

tests/test_click.py

@@ -573,7 +573,7 @@ def test_pgp_private_key():
 
 def test_file_fixture_turkish_car_plate():
     runner = CliRunner()
-    result = runner.invoke(main, ["fixtures/file"])
+    result = runner.invoke(main, ["--rarity", "0:", "fixtures/file"])
     assert result.exit_code == 0
     assert re.findall("Turkish License Plate Number", str(result.output))
 
@@ -594,7 +594,7 @@ def test_file_fixture_turkish_id_number():
 
 def test_file_fixture_turkish_tax_number():
     runner = CliRunner()
-    result = runner.invoke(main, ["fixtures/file"])
+    result = runner.invoke(main, ["--rarity", "0:", "fixtures/file"])
     assert result.exit_code == 0
     assert re.findall("Turkish Tax Number", str(result.output))
 
@@ -608,14 +608,14 @@ def test_file_fixture_uuid():
 
 def test_file_fixture_objectid():
     runner = CliRunner()
-    result = runner.invoke(main, ["fixtures/file"])
+    result = runner.invoke(main, ["--rarity", "0:", "fixtures/file"])
     assert result.exit_code == 0
     assert re.findall("ObjectID", str(result.output))
 
 
 def test_file_fixture_ulid():
     runner = CliRunner()
-    result = runner.invoke(main, ["fixtures/file"])
+    result = runner.invoke(main, ["--rarity", "0:", "fixtures/file"])
     assert result.exit_code == 0
     assert re.findall("ULID", str(result.output))
 

tests/test_regex_identifier.py

@@ -3,11 +3,14 @@
 import pytest
 
 from pywhat import regex_identifier
-from pywhat.filter import Filter
+from pywhat.filter import Distribution, Filter
 from pywhat.helper import load_regexes
 
 database = load_regexes()
 r = regex_identifier.RegexIdentifier()
+filter1 = Filter({"MinRarity": 0.0})
+d = Distribution(filter1)
+r_rarity_0 = regex_identifier.RegexIdentifier()
 
 
 def _assert_match_first_item(name, res):
@@ -415,12 +418,12 @@ def test_youtube2():
 
 
 def test_youtube_id():
-    res = r.check(["dQw4w9WgXcQ"])
+    res = r.check(["dQw4w9WgXcQ"], dist=d)
     _assert_match_first_item("YouTube Video ID", res)
 
 
 def test_youtube_id2():
-    res = r.check(["078-05-1120"])
+    res = r.check(["078-05-1120"], dist=d)
     assert "YouTube Video ID" not in res
 
 
@@ -533,35 +536,35 @@ def test_arn4():
 
 
 def test_unix_timestamp():
-    res = r.check(["1577836800"])  # 2020-01-01
+    res = r.check(["1577836800"], dist=d)  # 2020-01-01
     keys = [m["Regex Pattern"]["Name"] for m in res]
     assert "Unix Timestamp" in keys
     assert "Recent Unix Timestamp" in keys
 
 
 def test_unix_timestamp2():
-    res = r.check(["94694400"])  # 1973-01-01
+    res = r.check(["94694400"], dist=d)  # 1973-01-01
     keys = [m["Regex Pattern"]["Name"] for m in res]
     assert "Unix Timestamp" in keys
     assert "Recent Unix Timestamp" not in keys
 
 
 def test_unix_timestamp3():
-    res = r.check(["1234567"])  # 7 numbers
+    res = r.check(["1234567"], dist=d)  # 7 numbers
     keys = [m["Regex Pattern"]["Name"] for m in res]
     assert "Unix Timestamp" not in keys
     assert "Recent Unix Timestamp" not in keys
 
 
 def test_unix_timestamp4():
-    res = r.check(["1577836800000"])  # 2020-01-01
+    res = r.check(["1577836800000"], dist=d)  # 2020-01-01
     keys = [m["Regex Pattern"]["Name"] for m in res]
     assert "Unix Millisecond Timestamp" in keys
     assert "Recent Unix Millisecond Timestamp" in keys
 
 
 def test_unix_timestamp5():
-    res = r.check(["94694400000"])  # 1973-01-01
+    res = r.check(["94694400000"], dist=d)  # 1973-01-01
     keys = [m["Regex Pattern"]["Name"] for m in res]
     assert "Unix Millisecond Timestamp" in keys
     assert "Recent Unix Millisecond Timestamp" not in keys
@@ -1006,7 +1009,7 @@ def test_turkish_id_number2():
 
 
 def test_turkish_tax_number():
-    res = r.check(["1234567890"])
+    res = r.check(["1234567890"], dist=d)
     assert "Turkish Tax Number" in str(res)
 
 
@@ -1016,12 +1019,12 @@ def test_uuid():
 
 
 def test_objectid():
-    res = r.check(["5fc7c33a7ef88b139122a38a"])
+    res = r_rarity_0.check(["5fc7c33a7ef88b139122a38a"], dist=d)
     assert "ObjectID" in str(res)
 
 
 def test_ulid():
-    res = r.check(["01ERJ58HMWDN3VTRRHZQV2T5R5"])
+    res = r_rarity_0.check(["01ERJ58HMWDN3VTRRHZQV2T5R5"], dist=d)
     assert "ULID" in str(res)