new: drive the license menu, parsing and sniffing from data/licenses.yml

src/new/detect.rs

@@ -23,6 +23,7 @@ use std::path::Path;
 
 use regex::Regex;
 
+use super::licenses;
 use super::options::TemplateId;
 
 /// Outcome of the detection.
@@ -109,23 +110,14 @@ fn has_shebang(path: &Path) -> bool {
     content.starts_with(b"#!")
 }
 
-/// License files looked at by [`sniff_license`], in preference order.
-const LICENSE_FILES: [&str; 5] = [
-    "LICENSE",
-    "LICENSE.md",
-    "LICENSE.txt",
-    "COPYING",
-    "COPYING.txt",
-];
-
 /// Sniff the license of the project in `dir` from its `LICENSE`/`COPYING`
 /// file: an `SPDX-License-Identifier:` line wins, otherwise the text is
-/// matched against a short list of recognizable licenses (MIT, BSD-2/3,
-/// Apache-2.0, GPL-2/3, LGPL-2.1/3, ISC). `None` when no license file
-/// exists or nothing recognizable is found.
+/// matched against the marker sets of the bundled license table
+/// (`data/licenses.yml`: MIT, BSD-2/3, Apache-2.0, GPL-2/3, LGPL-2.1/3,
+/// ISC). `None` when no license file exists or nothing recognizable is
+/// found.
 pub fn sniff_license(dir: &Path) -> Option<String> {
-    let content = LICENSE_FILES
-        .iter()
+    let content = licenses::detect_files()
         .find_map(|name| std::fs::read_to_string(dir.join(name)).ok())
         // Case variants and suffixes (LICENSE-MIT, LICENCE, cpYING…): the
         // first top-level file whose name looks like a license notice.
@@ -166,43 +158,10 @@ pub fn sniff_license(dir: &Path) -> Option<String> {
         return Some(id);
     }
 
+    // The recognizable-license markers live in the bundled table; the
+    // LICENSE_TEXTS test below is their behavioral lock.
     let text = content.to_ascii_lowercase();
-    if text.contains("apache license") && text.contains("version 2") {
-        return Some("Apache-2.0".to_string());
-    }
-    if text.contains("lesser general public license") {
-        return if text.contains("version 3") && !text.contains("version 2.1") {
-            Some("LGPL-3.0+".to_string())
-        } else {
-            Some("LGPL-2.1+".to_string())
-        };
-    }
-    if text.contains("general public license") {
-        return if text.contains("version 3") {
-            Some("GPL-3.0+".to_string())
-        } else {
-            Some("GPL-2.0+".to_string())
-        };
-    }
-    if text.contains("mit license") || text.contains("permission is hereby granted, free of charge")
-    {
-        return Some("MIT".to_string());
-    }
-    if text.contains("isc license")
-        || text.contains("permission to use, copy, modify, and/or distribute this software")
-    {
-        return Some("ISC".to_string());
-    }
-    if text.contains("redistribution and use in source and binary forms") {
-        // The third clause (name endorsement) is what sets BSD-3 apart
-        // from BSD-2.
-        return if text.contains("endorse or promote") {
-            Some("BSD-3-Clause".to_string())
-        } else {
-            Some("BSD-2-Clause".to_string())
-        };
-    }
-    None
+    licenses::detect_from_text(&text).map(str::to_string)
 }
 
 #[cfg(test)]
@@ -302,8 +261,10 @@ mod tests {
         assert_eq!(detect(dir.path()), Detection::Empty);
     }
 
-    /// Distinctive (shortened) excerpts of the recognizable license texts.
-    const LICENSE_TEXTS: [(&str, &str); 9] = [
+    /// Distinctive (shortened) excerpts of the recognizable license texts:
+    /// the behavioral lock of the marker sets in `data/licenses.yml` — a
+    /// bad marker edit fails here, not on real packages.
+    const LICENSE_TEXTS: [(&str, &str); 11] = [
         (
             "MIT",
             "MIT License\n\nPermission is hereby granted, free of charge, to any person",
@@ -337,6 +298,19 @@ mod tests {
             "ISC",
             "ISC License\nPermission to use, copy, modify, and/or distribute this software",
         ),
+        // Dual-licensed preamble: the GPL reference outranks the MIT
+        // boilerplate, like the old hardcoded cascade decided.
+        (
+            "GPL-2.0+",
+            "MIT License\n\nAlternatively, under the terms of the GNU General Public License,\
+             \nversion 2 of the License.",
+        ),
+        // LGPL text naming both versions: the 2.1 wording wins.
+        (
+            "LGPL-2.1+",
+            "GNU LESSER GENERAL PUBLIC LICENSE\nVersion 2.1, February 1999\n\
+             This is version 2.1; version 3 is available separately.",
+        ),
     ];
 
     #[test]