compiletest: Stricter parsing of //@ normalize-* headers

Author: Zalathar

src/tools/compiletest/src/header.rs

@@ -14,6 +14,7 @@ use crate::common::{Config, Debugger, FailMode, Mode, PassMode};
 use crate::header::cfg::parse_cfg_name_directive;
 use crate::header::cfg::MatchOutcome;
 use crate::header::needs::CachedNeedsConditions;
+use crate::util::static_regex;
 use crate::{extract_cdb_version, extract_gdb_version};
 
 mod cfg;
@@ -1186,11 +1187,11 @@ impl Config {
         }
     }
 
-    fn parse_custom_normalization(&self, mut line: &str, prefix: &str) -> Option<(String, String)> {
+    fn parse_custom_normalization(&self, line: &str, prefix: &str) -> Option<(String, String)> {
         if parse_cfg_name_directive(self, line, prefix).outcome == MatchOutcome::Match {
-            let from = parse_normalization_string(&mut line)?;
-            let to = parse_normalization_string(&mut line)?;
-            Some((from, to))
+            let (regex, replacement) = parse_normalize_rule(line)
+                .unwrap_or_else(|| panic!("couldn't parse custom normalization rule: `{line}`"));
+            Some((regex, replacement))
         } else {
             None
         }
@@ -1311,24 +1312,29 @@ fn expand_variables(mut value: String, config: &Config) -> String {
     value
 }
 
-/// Finds the next quoted string `"..."` in `line`, and extract the content from it. Move the `line`
-/// variable after the end of the quoted string.
-///
-/// # Examples
-///
-/// ```
-/// let mut s = "normalize-stderr-32bit: \"something (32 bits)\" -> \"something ($WORD bits)\".";
-/// let first = parse_normalization_string(&mut s);
-/// assert_eq!(first, Some("something (32 bits)".to_owned()));
-/// assert_eq!(s, " -> \"something ($WORD bits)\".");
+/// Parses the regex and replacement values of a `//@ normalize-*` header,
+/// in the format:
+/// ```text
+/// normalize-*: "REGEX" -> "REPLACEMENT"
 /// ```
-fn parse_normalization_string(line: &mut &str) -> Option<String> {
-    // FIXME support escapes in strings.
-    let begin = line.find('"')? + 1;
-    let end = line[begin..].find('"')? + begin;
-    let result = line[begin..end].to_owned();
-    *line = &line[end + 1..];
-    Some(result)
+fn parse_normalize_rule(header: &str) -> Option<(String, String)> {
+    // FIXME(#126370): A colon after the header name should be mandatory, but
+    // currently is not, and there are many tests that lack the colon.
+    // FIXME: Support escaped double-quotes in strings.
+    let captures = static_regex!(
+        r#"(?x) # (verbose mode regex)
+        ^
+        [^:\s]+:?\s*            # (header name followed by optional colon)
+        "(?<regex>[^"]*)"       # "REGEX"
+        \s+->\s+                # ->
+        "(?<replacement>[^"]*)" # "REPLACEMENT"
+        $
+        "#
+    )
+    .captures(header)?;
+    let regex = captures["regex"].to_owned();
+    let replacement = captures["replacement"].to_owned();
+    Some((regex, replacement))
 }
 
 pub fn extract_llvm_version(version: &str) -> Option<u32> {

src/tools/compiletest/src/header/tests.rs

@@ -3,7 +3,7 @@ use std::path::Path;
 use std::str::FromStr;
 
 use crate::common::{Config, Debugger, Mode};
-use crate::header::{parse_normalization_string, EarlyProps, HeadersCache};
+use crate::header::{parse_normalize_rule, EarlyProps, HeadersCache};
 
 use super::iter_header;
 
@@ -32,35 +32,41 @@ fn make_test_description<R: Read>(
 }
 
 #[test]
-fn test_parse_normalization_string() {
-    let mut s = "normalize-stderr-32bit: \"something (32 bits)\" -> \"something ($WORD bits)\".";
-    let first = parse_normalization_string(&mut s);
-    assert_eq!(first, Some("something (32 bits)".to_owned()));
-    assert_eq!(s, " -> \"something ($WORD bits)\".");
-
-    // Nothing to normalize (No quotes)
-    let mut s = "normalize-stderr-32bit: something (32 bits) -> something ($WORD bits).";
-    let first = parse_normalization_string(&mut s);
-    assert_eq!(first, None);
-    assert_eq!(s, r#"normalize-stderr-32bit: something (32 bits) -> something ($WORD bits)."#);
-
-    // Nothing to normalize (Only a single quote)
-    let mut s = "normalize-stderr-32bit: \"something (32 bits) -> something ($WORD bits).";
-    let first = parse_normalization_string(&mut s);
-    assert_eq!(first, None);
-    assert_eq!(s, "normalize-stderr-32bit: \"something (32 bits) -> something ($WORD bits).");
-
-    // Nothing to normalize (Three quotes)
-    let mut s = "normalize-stderr-32bit: \"something (32 bits)\" -> \"something ($WORD bits).";
-    let first = parse_normalization_string(&mut s);
-    assert_eq!(first, Some("something (32 bits)".to_owned()));
-    assert_eq!(s, " -> \"something ($WORD bits).");
-
-    // Nothing to normalize (No quotes, 16-bit)
-    let mut s = "normalize-stderr-16bit: something (16 bits) -> something ($WORD bits).";
-    let first = parse_normalization_string(&mut s);
-    assert_eq!(first, None);
-    assert_eq!(s, r#"normalize-stderr-16bit: something (16 bits) -> something ($WORD bits)."#);
+fn test_parse_normalize_rule() {
+    let good_data = &[
+        (
+            r#"normalize-stderr-32bit: "something (32 bits)" -> "something ($WORD bits)""#,
+            "something (32 bits)",
+            "something ($WORD bits)",
+        ),
+        // FIXME(#126370): A colon after the header name should be mandatory,
+        // but currently is not, and there are many tests that lack the colon.
+        (
+            r#"normalize-stderr-32bit "something (32 bits)" -> "something ($WORD bits)""#,
+            "something (32 bits)",
+            "something ($WORD bits)",
+        ),
+    ];
+
+    for &(input, expected_regex, expected_replacement) in good_data {
+        let parsed = parse_normalize_rule(input);
+        let parsed =
+            parsed.as_ref().map(|(regex, replacement)| (regex.as_str(), replacement.as_str()));
+        assert_eq!(parsed, Some((expected_regex, expected_replacement)));
+    }
+
+    let bad_data = &[
+        r#"normalize-stderr-16bit: something (16 bits) -> something ($WORD bits)"#,
+        r#"normalize-stderr-32bit: something (32 bits) -> something ($WORD bits)"#,
+        r#"normalize-stderr-32bit: "something (32 bits) -> something ($WORD bits)"#,
+        r#"normalize-stderr-32bit: "something (32 bits)" -> "something ($WORD bits)"#,
+        r#"normalize-stderr-32bit: "something (32 bits)" -> "something ($WORD bits)"."#,
+    ];
+
+    for &input in bad_data {
+        let parsed = parse_normalize_rule(input);
+        assert_eq!(parsed, None);
+    }
 }
 
 #[derive(Default)]