31 exclusion list (#40)

* feat: added exclusion list to the available configuration

Closes #31

README.md

@@ -4,15 +4,15 @@
 
 The following directories contain the projects:
 
-| Directory Name | Project description |
-| --------------:|:------------------- |
-| antidotum (latin antidote) | `anitodutm` contains the R bindings (ie. an **R package**) for the Rust formatting library |
-| aqua (latin water) | `aqua` houses a Rust library that implements a basic R **tokenizer** |
-| balnea (latin bath) | `balnea` contains a Rust library that acts as **an entrypoint** to general formatting workflows |
-| scopa (latin broom) | `scopa` conains **a Visual Studio Code extension** that acts as a code formatter for R language |
-| spongia (lating sponge) | `spongia` contains a Rust library that implements an imperfect **R parser** |
-| tergo (latin to clean) | `tergo` contains **a command line interface (CLI)** that uses the `balnea` library to format R code |
-| unguentum (latin perfume) | `unguentum` houses a Rust library that implements all the **formatting tools** |
+|             Directory Name | Project description                                                                                                                                                                   |
+| -------------------------: | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
+| antidotum (latin antidote) | [`anitodutm`](./antidotum/tergo) contains the R bindings (ie. an **R package**) for the Rust formatting library                                                                       |
+|         aqua (latin water) | [`aqua`](./aqua/) houses a Rust library that implements a basic R **tokenizer**                                                                                                       |
+|        balnea (latin bath) | [`balnea`](./balnea/) contains a Rust library that acts as **an entrypoint** to general formatting workflows                                                                          |
+|        scopa (latin broom) | [`scopa`](./scopa) contains [**a Visual Studio Code extension**](https://marketplace.visualstudio.com/items?itemName=konradpagacz.tergo) that acts as a code formatter for R language |
+|    spongia (lating sponge) | [`spongia`](./spongia) contains a Rust library that implements an imperfect **R parser**                                                                                              |
+|     tergo (latin to clean) | [`tergo`](./tergo) contains **a command line interface (CLI)** that uses the `balnea` library to format R code                                                                        |
+|  unguentum (latin perfume) | [`unguentum`](./unguentum) houses a Rust library that implements all the **formatting tools**                                                                                         |
 
 ## Authors
 

antidotum/tergo/.Rbuildignore

@@ -7,3 +7,4 @@
 ^src/rust/repo_vendor/
 ^data-raw/
 ^cran-comments\.md$
+^tergo\.toml

antidotum/tergo/R/styling.R

@@ -115,16 +115,33 @@ style_pkg <- function(path = ".",
   # Define ANSI Color Codes and Unicode Symbols
   green_tick <- "\u001B[32m\u2714\u001B[0m"
   red_cross <- "\u001B[31m\u274C\u001B[0m"
+  # Define ANSI Color Codes and Unicode Symbols for a yellow dot
+  yellow_dot <- "\u001B[33m\u2022\u001B[0m"
 
+  ignored_paths <- vapply(
+    config$exclusion_list,
+    function(ignored_path) {
+      if (startsWith(ignored_path, "./")) {
+        ignored_path <- substr(ignored_path, 3, nchar(ignored_path))
+      }
+      file.path(path, ignored_path)
+    },
+    FUN.VALUE = character(1),
+    USE.NAMES = FALSE
+  )
   success_count <- 0
-
+  skipped_count <- 0
   for (file in files) {
     tryCatch(
       {
-        style_file(file, config)
-        success_count <- success_count + 1
-        # Print File Path and Green Tick
-        if (verbose) cat(sprintf("%s %s\n", basename(file), green_tick))
+        succes <- style_file_internal(file, config, ignored_paths)
+        if (succes) {
+          success_count <- success_count + 1
+          if (verbose) cat(sprintf("%s %s\n", file, green_tick))
+        } else {
+          skipped_count <- skipped_count + 1
+          if (verbose) cat(sprintf("%s %s\n", file, yellow_dot))
+        }
       },
       error = function(err) {
         # Print File Path, Red Cross, and Error Message
@@ -138,7 +155,8 @@ style_pkg <- function(path = ".",
     cat("\nSummary:\n")
     cat(sprintf("  %s Files processed : %d\n", summary_bullet, length(files)))
     cat(sprintf("  %s Successful      : %d\n", green_tick, success_count))
-    cat(sprintf("  %s Failed          : %d\n", red_cross, length(files) - success_count))
+    cat(sprintf("  %s Skipped         : %d\n", yellow_dot, skipped_count))
+    cat(sprintf("  %s Failed          : %d\n", red_cross, length(files) - success_count - skipped_count))
   }
 
   invisible(NULL)
@@ -150,8 +168,9 @@ style_pkg <- function(path = ".",
 #' To see possible configuration options, see [get_default_config()].
 #'
 #' @inheritParams style
-#' @param file (`character`) the file to format
-#' @return No return value, called for side effects.
+#' @param file (`character`) path to the file to format.
+#' @return (`logical`) whether the file was formatted successfully
+#' or skipped. `TRUE` - formatted successfully, `FALSE` - skipped.
 #'
 #' @export
 #' @examples
@@ -162,6 +181,12 @@ style_pkg <- function(path = ".",
 #' style_file(file = tmp, configuration = list())
 #' unlink(tmp)
 style_file <- function(file, configuration = list()) {
+  ignored_paths <- configuration$exclusion_list
+  if (!is.null(ignored_paths)) {
+    if (any(Map(function(ignored_path) startsWith(file, ignored_path), ignored_paths))) {
+      return(FALSE)
+    }
+  }
   if (!file.exists(file)) {
     stop("File " + file + " does not exist")
   }
@@ -174,7 +199,49 @@ style_file <- function(file, configuration = list()) {
     stop("Failed to style the file. Error: ", truncate_error(formatted[[2]]))
   }
   write(x = formatted[[2]], file = file)
-  invisible(NULL)
+  TRUE
+}
+
+#' Check whether a path is in ignored paths
+#' @return (`logical`) whether the path is in the ignored paths.
+#' @keywords internal
+is_in_ignored_paths <- function(path, ignored_paths) {
+  if (!is.null(ignored_paths)) {
+    if (
+      any(
+        vapply(
+          ignored_paths,
+          FUN = function(ignored_path) startsWith(path, ignored_path),
+          FUN.VALUE = logical(1),
+          USE.NAMES = FALSE
+        )
+      )
+    ) {
+      return(TRUE)
+    }
+  }
+  FALSE
+}
+
+#' Style a file internal
+#' @keywords internal
+style_file_internal <- function(file, configuration, ignored_paths) {
+  if (is_in_ignored_paths(file, ignored_paths)) {
+    return(FALSE)
+  }
+  if (!file.exists(file)) {
+    stop("File " + file + " does not exist")
+  }
+  size <- file.info(file)$size
+  code <- readChar(con = file, nchars = size)
+  formatted <- format_code(code, configuration)
+  if (formatted[[1]] == "success") {
+    formatted[[2]]
+  } else {
+    stop("Failed to style the file.")
+  }
+  write(x = formatted[[2]], file = file)
+  TRUE
 }
 
 #' Style text

antidotum/tergo/src/rust/Cargo.toml

@@ -8,7 +8,7 @@ crate-type = ["staticlib"]
 name = "tergo"
 
 [dependencies]
-tergo-lib = "0.2.8"
+tergo-lib = "0.2.9"
 toml = "0.8.19"
 extendr-api = "*"
 

antidotum/tergo/src/rust/src/lib.rs

@@ -115,7 +115,7 @@ fn format_code(source_code: &str, configuration: extendr_api::List) -> extendr_a
                         "Unknown function line breaks in the configuration value. Allowed: single, double, hanging."
                     )
                 }
-            },
+            }
             None => default_config.function_line_breaks,
         },
         match config_to_bool(
@@ -126,6 +126,15 @@ fn format_code(source_code: &str, configuration: extendr_api::List) -> extendr_a
             Ok(value) => value,
             Err(error) => return error,
         },
+        match configuration.get("exclusion_list") {
+            Some(list) => match list.as_string_vector() {
+                Some(arr) => arr,
+                None => {
+                    return list!(ERROR, "exclusion_list must be an array of strings.");
+                }
+            },
+            None => default_config.exclusion_list.0,
+        }
     );
 
     match tergo_lib::tergo_format(source_code, Some(&config)) {
@@ -147,44 +156,29 @@ fn format_code(source_code: &str, configuration: extendr_api::List) -> extendr_a
 /// @keywords internal
 #[extendr]
 fn get_config(path: &str) -> extendr_api::List {
-    match std::fs::read_to_string(path) {
+    let config = match std::fs::read_to_string(path) {
         Ok(config_file) => {
-            let config: Config = toml::from_str(&config_file).unwrap_or_else(|_| Config::default());
-            list!(
-                indent = config.indent.0,
-                line_length = config.line_length.0,
-                embracing_op_no_nl = config.embracing_op_no_nl.0,
-                allow_nl_after_assignment = config.allow_nl_after_assignment.0,
-                space_before_complex_rhs_in_formula = config.space_before_complex_rhs_in_formula.0,
-                strip_suffix_whitespace_in_function_defs =
-                    config.strip_suffix_whitespace_in_function_defs.0,
-                function_line_breaks = match config.function_line_breaks {
-                    FunctionLineBreaks::Hanging => "hanging",
-                    FunctionLineBreaks::Double => "double",
-                    FunctionLineBreaks::Single => "single",
-                },
-                insert_newline_in_quote_call = config.insert_newline_in_quote_call.0
-            )
-        }
-        Err(_) => {
-            let config = Config::default();
-            list!(
-                indent = config.indent.0,
-                line_length = config.line_length.0,
-                embracing_op_no_nl = config.embracing_op_no_nl.0,
-                allow_nl_after_assignment = config.allow_nl_after_assignment.0,
-                space_before_complex_rhs_in_formula = config.space_before_complex_rhs_in_formula.0,
-                strip_suffix_whitespace_in_function_defs =
-                    config.strip_suffix_whitespace_in_function_defs.0,
-                function_line_breaks = match config.function_line_breaks {
-                    FunctionLineBreaks::Hanging => "hanging",
-                    FunctionLineBreaks::Double => "double",
-                    FunctionLineBreaks::Single => "single",
-                },
-                insert_newline_in_quote_call = config.insert_newline_in_quote_call.0
-            )
+            toml::from_str::<Config>(&config_file).unwrap_or_else(|_| Config::default())
         }
-    }
+        Err(_) => Config::default(),
+    };
+
+    list!(
+        indent = config.indent.0,
+        line_length = config.line_length.0,
+        embracing_op_no_nl = config.embracing_op_no_nl.0,
+        allow_nl_after_assignment = config.allow_nl_after_assignment.0,
+        space_before_complex_rhs_in_formula = config.space_before_complex_rhs_in_formula.0,
+        strip_suffix_whitespace_in_function_defs =
+            config.strip_suffix_whitespace_in_function_defs.0,
+        function_line_breaks = match config.function_line_breaks {
+            FunctionLineBreaks::Hanging => "hanging",
+            FunctionLineBreaks::Double => "double",
+            FunctionLineBreaks::Single => "single",
+        },
+        insert_newline_in_quote_call = config.insert_newline_in_quote_call.0,
+        exclusion_list = config.exclusion_list.0
+    )
 }
 
 /// Get the default configuration
@@ -240,7 +234,8 @@ fn get_default_config() -> extendr_api::List {
             FunctionLineBreaks::Double => "double",
             FunctionLineBreaks::Single => "single",
         },
-        insert_newline_in_quote_call = config.insert_newline_in_quote_call.0
+        insert_newline_in_quote_call = config.insert_newline_in_quote_call.0,
+        exclusion_list = config.exclusion_list.0
     )
 }
 

antidotum/tergo/tergo.toml

@@ -0,0 +1 @@
+exclusion_list = ["./R/extendr-wrappers.R", "./src", "tergo.Rcheck"]

antidotum/tergo/tests/testthat/test-style_text.R

@@ -21,4 +21,3 @@ testthat::test_that("style_text validates the configuration", {
     )
   )
 })
-

antidotum/tergo/vignettes/styling_with_tergo.Rmd

@@ -84,6 +84,7 @@ tergo::style(config = list(
 | `strip_suffix_whitespace_in_function_defs` | boolean | TRUE | Trim trailing whitespace in functions |
 | `function_line_breaks`             | string  | "hanging" | `"single"`/`"double"`/`"hanging"` function argument formatting ([ref](https://style.tidyverse.org/functions.html#multi-line-function-definitions)) |
 | `insert_newline_in_quote_call`     | boolean | TRUE   | Add newlines in long `quote()` calls |
+| `exclusion_list` | array of strings | [] | Files/directories to exclude from formatting |
 
 ## Why Choose `tergo`?
 - **Blazing Fast**: Formats large codebases 100-1000x faster than alternatives

balnea/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "tergo-lib"
-version = "0.2.8"
+version = "0.2.9"
 edition = "2021"
 description = "A tool to format R code"
 license = "MIT"
@@ -14,7 +14,7 @@ path = "src/lib.rs"
 [dependencies]
 tokenizer = { package = "tergo-tokenizer", path = "../aqua", version = "0.2.2" }
 parser = { package = "tergo-parser", path = "../spongia", version = "0.2.3" }
-formatter = { package = "tergo-formatter", path = "../unguentum", version = "0.2.6" }
+formatter = { package = "tergo-formatter", path = "../unguentum", version = "0.2.8" }
 log = "0.4.21"
 env_logger = "0.11.3"
 serde = { version = "1.0.210", features = ["derive"] }