Additional pg_dump args (#78)

CHANGELOG.md

@@ -6,6 +6,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 ### 🚀 Added
+- Additional arguments for pg_dump [#78](https://github.com/datanymizer/datanymizer/pull/78)
+  ([@evgeniy-r](https://github.com/evgeniy-r))
 - `base64_token` and `base64url_token` transformers. [#77](https://github.com/datanymizer/datanymizer/pull/77)
   ([@evgeniy-r](https://github.com/evgeniy-r))
 - Key-value store that allows to share information between template transformers

cli/pg_datanymizer/src/app.rs

@@ -42,6 +42,7 @@ impl App {
             engine,
             self.options.pg_dump_location.clone(),
             self.options.file.clone(),
+            self.options.pg_dump_args.clone(),
         )
     }
 
@@ -78,10 +79,10 @@ impl App {
         } else {
             let mut builder = TlsConnector::builder();
 
-            if self.options.accept_invalid_hostnames.unwrap_or(false) {
+            if self.options.accept_invalid_hostnames {
                 builder.danger_accept_invalid_hostnames(true);
             }
-            if self.options.accept_invalid_certs.unwrap_or(false) {
+            if self.options.accept_invalid_certs {
                 builder.danger_accept_invalid_certs(true);
             }
 

cli/pg_datanymizer/src/options.rs

@@ -9,48 +9,46 @@ pub struct Options {
     database: String,
 
     #[structopt(
-        short = "c",
-        long = "config",
+        short,
+        long,
         help = "Path to config file",
         default_value = "./config.yml"
     )]
     pub config: String,
+
     #[structopt(
-        short = "f",
-        long = "file",
+        short,
+        long,
         name = "FILE",
         help = "Path to dump file, example: /tmp/dump.sql"
     )]
     pub file: Option<String>,
 
     #[structopt(
-        short = "d",
+        short,
         long = "dbname",
         help = "database to dump",
         default_value = "postgres"
     )]
     pub db_name: String,
+
     #[structopt(
-        short = "h",
-        long = "host",
+        short,
+        long,
         help = "Database server host or socket directory",
         default_value = "localhost"
     )]
     pub host: String,
-    #[structopt(
-        short = "p",
-        long = "port",
-        help = "Database server port number [default: 5432]"
-    )]
+
+    #[structopt(short, long, help = "Database server port number [default: 5432]")]
     pub port: Option<u16>,
-    #[structopt(
-        short = "U",
-        long = "username",
-        help = "Connect as specified database user"
-    )]
+
+    #[structopt(short = "U", long, help = "Connect as specified database user")]
     pub username: Option<String>,
-    #[structopt(short = "W", long = "password", help = "User password")]
+
+    #[structopt(short = "W", long, help = "User password")]
     pub password: Option<String>,
+
     #[structopt(
         long = "pg_dump",
         help = "pg_dump file location",
@@ -60,14 +58,21 @@ pub struct Options {
 
     #[structopt(
         long = "accept_invalid_hostnames",
-        help = "Accept or not invalid hostnames when using SSL [default: false]"
+        help = "Accept or not invalid hostnames when using SSL"
     )]
-    pub accept_invalid_hostnames: Option<bool>,
+    pub accept_invalid_hostnames: bool,
+
     #[structopt(
         long = "accept_invalid_certs",
-        help = "Accept or not invalid certificates (e.g., self-signed) when using SSL [default: false]"
+        help = "Accept or not invalid certificates (e.g., self-signed) when using SSL"
+    )]
+    pub accept_invalid_certs: bool,
+
+    #[structopt(
+        name = "PG_DUMP_ARGS",
+        help = "The remaining arguments are passed directly to `pg_dump` calls. You should add `--` before <DBNAME> in such cases"
     )]
-    pub accept_invalid_certs: Option<bool>,
+    pub pg_dump_args: Vec<String>,
 }
 
 impl Options {
@@ -106,7 +111,7 @@ impl Options {
 
 #[cfg(test)]
 mod tests {
-    use super::Options;
+    use super::*;
 
     #[test]
     fn parse_empty_config() {
@@ -120,8 +125,9 @@ mod tests {
             username: None,
             password: None,
             pg_dump_location: "pg_dump".to_string(),
-            accept_invalid_hostnames: Some(false),
-            accept_invalid_certs: Some(false),
+            accept_invalid_hostnames: false,
+            accept_invalid_certs: false,
+            pg_dump_args: vec![],
         };
 
         let expected = "postgres://hostname/test".to_string();
@@ -140,8 +146,9 @@ mod tests {
             username: None,
             password: None,
             pg_dump_location: "pg_dump".to_string(),
-            accept_invalid_hostnames: Some(false),
-            accept_invalid_certs: Some(false),
+            accept_invalid_hostnames: false,
+            accept_invalid_certs: false,
+            pg_dump_args: vec![],
         };
 
         let cfg2 = Options {
@@ -166,4 +173,28 @@ mod tests {
         assert_eq!(cfg3.database_url().unwrap().to_string(), expected3);
         assert_eq!(cfg4.database_url().unwrap().to_string(), expected4);
     }
+
+    #[test]
+    fn parse_args() {
+        let cmd = vec![
+            "pg_datanymizer",
+            "-c",
+            "some_config.yml",
+            "-f",
+            "some_file.sql",
+            "--",
+            "postgres://user@hostname/test",
+            "--no-owner",
+            "--no-acl",
+        ];
+        let options = Options::from_iter(cmd);
+
+        assert_eq!(
+            options.database_url().unwrap().as_str(),
+            "postgres://user@hostname/test"
+        );
+        assert_eq!(options.config, "some_config.yml");
+        assert_eq!(options.file, Some("some_file.sql".to_string()));
+        assert_eq!(options.pg_dump_args, vec!["--no-owner", "--no-acl"]);
+    }
 }

datanymizer_dumper/src/postgres/dumper.rs

@@ -18,23 +18,31 @@ pub struct PgDumper {
     engine: Engine,
     dump_writer: DumpWriter,
     pg_dump_location: String,
+    pg_dump_args: Vec<String>,
     progress_bar: ProgressBar,
 }
 
 impl PgDumper {
-    pub fn new(engine: Engine, pg_dump_location: String, target: Option<String>) -> Result<Self> {
+    pub fn new(
+        engine: Engine,
+        pg_dump_location: String,
+        target: Option<String>,
+        pg_dump_args: Vec<String>,
+    ) -> Result<Self> {
         let dump_writer = DumpWriter::new(target)?;
         let pb: ProgressBar = if dump_writer.can_log_to_stdout() {
             ProgressBar::new(0)
         } else {
             ProgressBar::hidden()
         };
+
         Ok(Self {
             engine,
             dump_writer,
             pg_dump_location,
             schema_inspector: PgSchemaInspector {},
             progress_bar: pb,
+            pg_dump_args,
         })
     }
 
@@ -45,6 +53,7 @@ impl PgDumper {
         let db_url = self.engine.settings.source.get_database_url();
 
         let dump_output = Command::new(program)
+            .args(&self.pg_dump_args)
             .args(&args)
             .args(&table_args)
             .arg(&db_url)

docs/pg_datanymizer.md

@@ -5,7 +5,7 @@
 #### Usage:
 
 ```
-pg_datanymizer [OPTIONS] <DBNAME>
+pg_datanymizer [OPTIONS] <DBNAME> [PG_DUMP_ARGS]...
 ```
 
 #### FLAGS
@@ -22,8 +22,8 @@ pg_datanymizer [OPTIONS] <DBNAME>
 | `-f`, `--file` `<FILE>`                                   | Path to the dump output file, example: `/tmp/dump.sql`
 | `-c`, `--config` `<config>`                               | Path to the config file. Default: `./config.yml`
 | `--pg_dump` `<pg-dump-location>`                          | Postgres `pg_dump` utility program file location. Default: just `pg_dump`
-| `--accept_invalid_certs` `<accept-invalid-certs>`         | Accept or not invalid certificates (e.g., self-signed) when using SSL. Default: `false` 
-| `--accept_invalid_hostnames` `<accept-invalid-hostnames>` | Accept or not invalid hostnames when using SSL. Default: `false`
+| `--accept_invalid_certs`                                  | Accept invalid certificates (e.g., self-signed) when using SSL
+| `--accept_invalid_hostnames`                              | Accept invalid hostnames when using SSL
 | When `<DBNAME>` is just a database name (not a full url):
 | `-h`, `--host` `<host>`                                   | Database server host or a socket directory. Default: `localhost`
 | `-W`, `--password` `<password>`                           | User password
@@ -32,7 +32,7 @@ pg_datanymizer [OPTIONS] <DBNAME>
 
 #### ARGS
 
-| Name       | Description
-|---         |---  
-| `<DBNAME>` | Postgres database URL, e.g. `postgres://postgres:password@localhost:5432/database_name?sslmode=disable` (you can omit some parts), or just a database name, e.g. `my_db`  
-
+| Name             | Description
+|---               |---  
+| `<DBNAME>`       | Postgres database URL, e.g. `postgres://postgres:password@localhost:5432/database_name?sslmode=disable` (you can omit some parts), or just a database name, e.g. `my_db`  
+| `<PG_DUMP_ARGS>` | The remaining arguments are passed directly to `pg_dump` calls. You should add `--` before `<DBNAME>` in such cases