Merge pull request #47 from nixys/feat/45

feat(#45): Generating correct values for types
nixys · Sep 18, 2024 · 3b480fa · 3b480fa
2 parents c14e6e2 + ddc3389
commit 3b480fa
Show file tree

Hide file tree

Showing 3 changed files with 118 additions and 86 deletions.
diff --git a/README.md b/README.md
@@ -355,9 +355,9 @@ _Values to masquerade a columns in accordance with the types see below._
 | `numeric`     | `0.0` |
 | `real`        | `0.0` |
 | `double`      | `0.0` |
-| `character`   | `randomized string data` |
-| `bpchar`      | `randomized string data` |
-| `text`        | `randomized string data` |
+| `character`   | `randomized character data"` |
+| `bpchar`      | `randomized bpchar data` |
+| `text`        | `randomized text data` |
 
 **MySQL:**
 
@@ -372,25 +372,25 @@ _Values to masquerade a columns in accordance with the types see below._
 | `int` |              `0` |
 | `integer` |          `0` |
 | `bigint` |           `0` |
-| `float` |            `0` |
-| `double` |           `0` |
-| `double precision` | `0` |
-| `decimal` |          `0` |
-| `dec` |              `0` |
-| `char` |       `randomized string data` | 
-| `varchar` |    `randomized string data` | 
-| `tinytext` |   `randomized string data` | 
-| `text` |       `randomized string data` | 
-| `mediumtext` | `randomized string data` | 
-| `longtext` |   `randomized string data` | 
-| `enum` |       `randomized string data` | 
-| `set` |        `randomized string data` | 
-| `date` |       `randomized string data` | 
-| `datetime` |   `randomized string data` | 
-| `timestamp` |  `randomized string data` | 
-| `time` |       `randomized string data` | 
-| `year` |       `randomized string data` | 
-| `json` |       `randomized string data` |
+| `float` |            `0.0` |
+| `double` |           `0.0` |
+| `double precision` | `0.0` |
+| `decimal` |          `0.0` |
+| `dec` |              `0.0` |
+| `char` |       `randomized char` (String will be truncated to "COLUMN_SIZE" length.)| 
+| `varchar` |    `randomized varchar` (String will be truncated to "COLUMN_SIZE" length.) | 
+| `tinytext` |   `randomized tinytext` | 
+| `text` |       `randomized text` | 
+| `mediumtext` | `randomized mediumtext` | 
+| `longtext` |   `randomized longtext` | 
+| `enum` |       Last value from `enum` | 
+| `set` |        Last value from `set` | 
+| `date` |       `2024-01-01` | 
+| `datetime` |   `2024-01-01 00:00:00` | 
+| `timestamp` |  `2024-01-01 00:00:00` | 
+| `time` |       `00:00:00` | 
+| `year` |       `2024` | 
+| `json` |       `{"randomized": "json_data"}` |
 | `binary` |     `cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=` |
 | `varbinary` |  `cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=` |
 | `tinyblob` |   `cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=` |

diff --git a/modules/anonymizers/mysql/security_types.go b/modules/anonymizers/mysql/security_types.go
@@ -5,124 +5,118 @@ import (
 	"github.com/nixys/nxs-data-anonymizer/modules/filters/relfilter"
 )
 
-const (
-	securityTypeString = "randomized string data"
-	securityTypeNum    = "0"
-	securityTypeBinary = "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo="
-)
-
 var typeRuleDefault = []relfilter.TypeRuleOpts{
 
 	// String
 	{
-		Selector: "(?i)^char\\((\\d+)\\)|^char ",
+		Selector: "(?i)^char\\((\\d+)\\)|^char",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "{{ trunc (index .ColumnTypeGroups 0 1 | int ) \"randomized char\" }}",
 			Unique: false,
 		},
 	},
 	{
-		Selector: "(?i)^varchar",
+		Selector: "(?i)^varchar\\((\\d+)\\)|^varchar",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "{{ trunc (index .ColumnTypeGroups 0 1 | int ) \"randomized varchar\" }}",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^tinytext",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "randomized tinytext",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^text",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "randomized text",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^mediumtext",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "randomized mediumtext",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^longtext",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "randomized longtext",
 			Unique: false,
 		},
 	},
 	{
-		Selector: "(?i)^enum",
+		Selector: "(?i)^enum\\(.*'(.*)'.*\\)",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "{{ index .ColumnTypeGroups 0 1 }}",
 			Unique: false,
 		},
 	},
 	{
-		Selector: "(?i)^set",
+		Selector: "(?i)^set\\(.*'(.*)'.*\\)",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "{{ index .ColumnTypeGroups 0 1 }}",
 			Unique: false,
 		},
 	},
 	{
-		Selector: "(?i)^date",
+		Selector: "(?i)^datetime",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "2024-01-01 00:00:00",
 			Unique: false,
 		},
 	},
 	{
-		Selector: "(?i)^datetime",
+		Selector: "(?i)^date",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "2024-01-01",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^timestamp",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "2024-01-01 00:00:00",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^time",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "00:00:00",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^year",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "2024",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^json",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeString,
+			Value:  "{\"randomized\": \"json_data\"}",
 			Unique: false,
 		},
 	},
@@ -132,103 +126,103 @@ var typeRuleDefault = []relfilter.TypeRuleOpts{
 		Selector: "(?i)^bit",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^bool",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^boolean",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^tinyint",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^smallint",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^mediumint",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^int",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^integer",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^bigint",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^float",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0.0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^double",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0.0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^decimal",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0.0",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^dec",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeNum,
+			Value:  "0.0",
 			Unique: false,
 		},
 	},
@@ -238,45 +232,45 @@ var typeRuleDefault = []relfilter.TypeRuleOpts{
 		Selector: "(?i)^binary",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^varbinary",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	}, {
 		Selector: "(?i)^tinyblob",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^blob",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	},
 	{
 		Selector: "(?i)^mediumblob",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	}, {
 		Selector: "(?i)^longblob",
 		Rule: relfilter.ColumnRuleOpts{
 			Type:   misc.ValueTypeTemplate,
-			Value:  securityTypeBinary,
+			Value:  "cmFuZG9taXplZCBiaW5hcnkgZGF0YQo=",
 			Unique: false,
 		},
 	},