From 884f187c49a190984f6c7ca3f3d4123cb1e04ef4 Mon Sep 17 00:00:00 2001
From: Simon Sapin <simon@apollographql.com>
Date: Mon, 6 Nov 2023 12:37:36 +0100
Subject: [PATCH] Serialize multi-line strings as block strings

Fixes https://github.com/apollographql/apollo-rs/issues/664
---
 crates/apollo-compiler/src/ast/serialize.rs   | 99 +++++++++++++++++++
 .../serializer/ok/0015_supergraph.graphql     |  6 +-
 2 files changed, 104 insertions(+), 1 deletion(-)

diff --git a/crates/apollo-compiler/src/ast/serialize.rs b/crates/apollo-compiler/src/ast/serialize.rs
index fab07b720..40c185a18 100644
--- a/crates/apollo-compiler/src/ast/serialize.rs
+++ b/crates/apollo-compiler/src/ast/serialize.rs
@@ -98,6 +98,19 @@ impl<'config, 'fmt, 'fmt2> State<'config, 'fmt, 'fmt2> {
         Ok(())
     }
 
+    /// Panics if newlines are disabled
+    fn require_new_line(&mut self) -> fmt::Result {
+        let prefix = self
+            .config
+            .ident_prefix
+            .expect("require_new_line called with newlines disabled");
+        self.write("\n")?;
+        for _ in 0..self.indent_level {
+            self.write(prefix)?;
+        }
+        Ok(())
+    }
+
     pub(crate) fn newlines_enabled(&self) -> bool {
         self.config.ident_prefix.is_some()
     }
@@ -799,6 +812,9 @@ fn curly_brackets_space_separated<T>(
 }
 
 fn serialize_string_value(state: &mut State, mut str: &str) -> fmt::Result {
+    if state.newlines_enabled() && prefer_block_string(str) && can_be_block_string(str) {
+        return serialize_block_string(state, str);
+    }
     // TODO: use block string when it would be possible AND would look nicer
     // The parsed value of a block string cannot have:
     // * non-\n control characters (not representable)
@@ -829,6 +845,89 @@ fn serialize_string_value(state: &mut State, mut str: &str) -> fmt::Result {
     state.write("\"")
 }
 
+fn serialize_block_string(state: &mut State, str: &str) -> fmt::Result {
+    const TRIPLE_QUOTE: &str = "\"\"\"";
+    const ESCAPED_TRIPLE_QUOTE: &str = "\\\"\"\"";
+    const _: () = assert!(TRIPLE_QUOTE.len() == 3);
+    const _: () = assert!(ESCAPED_TRIPLE_QUOTE.len() == 4);
+
+    state.write(TRIPLE_QUOTE)?;
+    // `can_be_block_string` excludes \r, so the only remaining line terminator is \n
+    for mut line in str.split('\n') {
+        if line.is_empty() {
+            // Skip indentation which would be trailing whitespace
+            state.write("\n")?;
+            continue;
+        }
+        state.require_new_line()?;
+        loop {
+            if let Some((before, after)) = line.split_once(TRIPLE_QUOTE) {
+                state.write(before)?;
+                state.write(ESCAPED_TRIPLE_QUOTE)?;
+                line = after;
+            } else {
+                state.write(line)?;
+                break;
+            }
+        }
+    }
+    state.require_new_line()?;
+    state.write(TRIPLE_QUOTE)
+}
+
+fn prefer_block_string(value: &str) -> bool {
+    // If a value is multi-line, it likely looks nicer as a block string
+    value.contains('\n')
+}
+
+/// Is it possible to create a serialization that, when fed through
+/// [BlockStringValue](https://spec.graphql.org/October2021/#BlockStringValue()),
+/// returns exactly `value`?
+fn can_be_block_string(value: &str) -> bool {
+    // `BlockStringValue` splits its inputs at any `LineTerminator` (\n, \r\n, or \r)
+    // and eventually joins lines but always with \n. So its output can never contain \r
+    if value.contains('\r') {
+        return false;
+    }
+
+    /// <https://spec.graphql.org/October2021/#WhiteSpace>
+    fn trim_start_graphql_whitespace(value: &str) -> &str {
+        value.trim_start_matches([' ', '\t'])
+    }
+
+    // With the above, \n is the only remaining LineTerminator
+    let mut lines = value.split('\n');
+    if lines
+        .next()
+        .is_some_and(|first| trim_start_graphql_whitespace(first).is_empty())
+        || lines
+            .next_back()
+            .is_some_and(|last| trim_start_graphql_whitespace(last).is_empty())
+    {
+        // Leading or trailing whitespace-only line would be trimmed by `BlockStringValue`
+        return false;
+    }
+
+    let common_indent = {
+        let mut lines = value.split('\n');
+        // Skip the first line, the one following """
+        lines.next();
+
+        let each_line_indent_utf8_len = lines.filter_map(|line| {
+            let after_indent = trim_start_graphql_whitespace(line);
+            if !after_indent.is_empty() {
+                Some(line.len() - after_indent.len())
+            } else {
+                None // skip whitespace-only lines
+            }
+        });
+        each_line_indent_utf8_len.min().unwrap_or(0)
+    };
+    // If there is common indent `BlockStringValue` would remove it
+    // and incorrectly round-trip to a different value.
+    common_indent == 0
+}
+
 fn serialize_description(state: &mut State, description: &Option<NodeStr>) -> fmt::Result {
     if let Some(description) = description {
         serialize_string_value(state, description)?;
diff --git a/crates/apollo-compiler/test_data/serializer/ok/0015_supergraph.graphql b/crates/apollo-compiler/test_data/serializer/ok/0015_supergraph.graphql
index 48caa2de9..e1cb5427e 100644
--- a/crates/apollo-compiler/test_data/serializer/ok/0015_supergraph.graphql
+++ b/crates/apollo-compiler/test_data/serializer/ok/0015_supergraph.graphql
@@ -33,7 +33,11 @@ type Book implements Product @join__owner(graph: BOOKS) @join__type(graph: BOOKS
   year: Int @join__field(graph: BOOKS)
   similarBooks: [Book]! @join__field(graph: BOOKS)
   metadata: [MetadataOrError] @join__field(graph: BOOKS)
-  "\"\"\"To be in stock or not to be in stock\"\"\"\n\nthat is the question"
+  """
+  \"""To be in stock or not to be in stock\"""
+
+  that is the question
+  """
   inStock: Boolean @join__field(graph: INVENTORY)
   isCheckedOut: Boolean @join__field(graph: INVENTORY)
   upc: String! @join__field(graph: PRODUCT)