From 2273ce77b819895022d62dec8a8446f9e0345260 Mon Sep 17 00:00:00 2001
From: Jake Stanger <mail@jstanger.dev>
Date: Tue, 31 Oct 2023 22:30:23 +0000
Subject: [PATCH] feat: trim whitespace from multi-line strings

Resolves #24
---
 assets/inputs/string_multiline.corn       |  6 ++++
 assets/outputs/json/string.json           |  2 +-
 assets/outputs/json/string_multiline.json |  3 ++
 assets/outputs/toml/string.toml           |  6 ++--
 assets/outputs/toml/string_multiline.toml |  5 +++
 assets/outputs/yaml/string.yml            |  3 +-
 assets/outputs/yaml/string_multiline.yml  |  4 +++
 libcorn/src/parser.rs                     | 40 +++++++++++++++++++++++
 libcorn/src/wasm.rs                       |  1 -
 9 files changed, 64 insertions(+), 6 deletions(-)
 create mode 100644 assets/inputs/string_multiline.corn
 create mode 100644 assets/outputs/json/string_multiline.json
 create mode 100644 assets/outputs/toml/string_multiline.toml
 create mode 100644 assets/outputs/yaml/string_multiline.yml

diff --git a/assets/inputs/string_multiline.corn b/assets/inputs/string_multiline.corn
new file mode 100644
index 0000000..c1d25a0
--- /dev/null
+++ b/assets/inputs/string_multiline.corn
@@ -0,0 +1,6 @@
+{
+    foo = "
+    hello
+    world
+    "
+}
\ No newline at end of file
diff --git a/assets/outputs/json/string.json b/assets/outputs/json/string.json
index 61c3a14..53e6ede 100644
--- a/assets/outputs/json/string.json
+++ b/assets/outputs/json/string.json
@@ -1,6 +1,6 @@
 {
   "foo": "bar",
-  "bar": "\"\\\n\r\t",
+  "bar": "\"\\\n",
   "baz": "a",
   "qux": ""
 }
diff --git a/assets/outputs/json/string_multiline.json b/assets/outputs/json/string_multiline.json
new file mode 100644
index 0000000..888d65d
--- /dev/null
+++ b/assets/outputs/json/string_multiline.json
@@ -0,0 +1,3 @@
+{
+  "foo": "hello\nworld\n"
+}
diff --git a/assets/outputs/toml/string.toml b/assets/outputs/toml/string.toml
index 5d05e07..f4cc605 100644
--- a/assets/outputs/toml/string.toml
+++ b/assets/outputs/toml/string.toml
@@ -1,7 +1,7 @@
 foo = "bar"
-bar = """
-\"\\
-\r\t"""
+bar = '''
+"\
+'''
 baz = "a"
 qux = ""
 
diff --git a/assets/outputs/toml/string_multiline.toml b/assets/outputs/toml/string_multiline.toml
new file mode 100644
index 0000000..d2a49da
--- /dev/null
+++ b/assets/outputs/toml/string_multiline.toml
@@ -0,0 +1,5 @@
+foo = """
+hello
+world
+"""
+
diff --git a/assets/outputs/yaml/string.yml b/assets/outputs/yaml/string.yml
index 6a78af1..f1649d4 100644
--- a/assets/outputs/yaml/string.yml
+++ b/assets/outputs/yaml/string.yml
@@ -1,5 +1,6 @@
 foo: bar
-bar: "\"\\\n\r\t"
+bar: |
+  "\
 baz: a
 qux: ''
 
diff --git a/assets/outputs/yaml/string_multiline.yml b/assets/outputs/yaml/string_multiline.yml
new file mode 100644
index 0000000..58f9fac
--- /dev/null
+++ b/assets/outputs/yaml/string_multiline.yml
@@ -0,0 +1,4 @@
+foo: |
+  hello
+  world
+
diff --git a/libcorn/src/parser.rs b/libcorn/src/parser.rs
index af80470..1964ab8 100644
--- a/libcorn/src/parser.rs
+++ b/libcorn/src/parser.rs
@@ -124,6 +124,12 @@ impl<'a> CornParser<'a> {
             };
         }
 
+        let full_string = if full_string.contains('\n') {
+            trim_multiline_string(&full_string)
+        } else {
+            full_string
+        };
+
         Ok(Cow::Owned(full_string))
     }
 
@@ -315,6 +321,40 @@ impl<'a> CornParser<'a> {
     }
 }
 
+/// Takes a multiline string and trims the maximum amount of
+/// whitespace at the start of each line
+/// while preserving formatting.
+///
+/// Based on code from `indoc` crate:
+/// <https://github.com/dtolnay/indoc/blob/60b5fa29ba4f98b479713621a1f4ec96155caaba/src/unindent.rs#L15-L51>
+
+fn trim_multiline_string(string: &str) -> String {
+    let ignore_first_line = string.starts_with('\n') || string.starts_with("\r\n");
+
+    let spaces = string
+        .lines()
+        .skip(1)
+        .map(|line| line.chars().take_while(char::is_ascii_whitespace).count())
+        .min()
+        .unwrap_or_default();
+
+    let mut result = String::with_capacity(string.len());
+    for (i, line) in string.lines().enumerate() {
+        if i > 1 || (i == 1 && !ignore_first_line) {
+            result.push('\n');
+        }
+        if i == 0 {
+            // Do not un-indent anything on same line as opening quote
+            result.push_str(line);
+        } else if line.len() > spaces {
+            // Whitespace-only lines may have fewer than the number of spaces
+            // being removed
+            result.push_str(&line[spaces..]);
+        }
+    }
+    result
+}
+
 /// Parses the input string into a `Config`
 /// containing the resolved inputs
 /// and a map of values representing the top-level object.
diff --git a/libcorn/src/wasm.rs b/libcorn/src/wasm.rs
index e0943e4..b688bbf 100644
--- a/libcorn/src/wasm.rs
+++ b/libcorn/src/wasm.rs
@@ -10,7 +10,6 @@ use wasm_bindgen_test::*;
 #[wasm_bindgen]
 pub fn parse(corn: &str) -> Result<JsValue, JsValue> {
     console_error_panic_hook::set_once();
-
     let res = crate::parse(corn);
 
     match res {