Correctly handle a few more mis-identified locales

indexer/src/man.rs

@@ -32,10 +32,15 @@ pub fn parse_path(path: &str) -> Option<(&str, &str, &str)> {
     }
 
     let cap = match RE.captures(path) { Some(x) => x, None => return None };
-    let locale = cap.get(1).map(|e| e.as_str()).unwrap_or("");
+    let mut locale = cap.get(1).map(|e| e.as_str()).unwrap_or("");
     let name = cap.get(2).unwrap().as_str();
     let section = cap.get(3).unwrap().as_str();
 
+    // Some weird directories that happen to match the locale
+    if locale.contains("openmpi") || locale.contains("mpich") || locale.contains("mvapich") {
+        locale = "";
+    }
+
     // Not everything matching the regex is necessarily a man page, exclude some special cases.
     match (name, section, locale) {
         // Files that totally aren't man pages
@@ -47,9 +52,10 @@ pub fn parse_path(path: &str) -> Option<(&str, &str, &str)> {
         (_,            "bz2",  _) |
         (_,            "xz",   _) |
         (_,            "html", _) => None,
-        // Some weird directories that happen to match the locale
+        // Some more weird directories that happen to match the locale
         (n, s, "5man") |
         (n, s, "c")    |
+        (n, s, "man")  |
         (n, s, "man1") |
         (n, s, "man2") |
         (n, s, "man3") |
@@ -59,7 +65,8 @@ pub fn parse_path(path: &str) -> Option<(&str, &str, &str)> {
         (n, s, "man7") |
         (n, s, "man8") |
         (n, s, "Man-Part1") |
-        (n, s, "Man-Part2") => Some((n, s, "")),
+        (n, s, "Man-Part2") |
+        (n, s, "overrides") => Some((n, s, "")),
         // Nothing special!
         x => Some(x)
     }

sql/update-2021-12-16.sql

@@ -0,0 +1,8 @@
+-- 'overrides' is not a valid locale.  Rather, that was a directory that CentOS
+-- used for updated man pages without overwriting the pages already installed
+-- on the system.
+-- 'man' is just a badly nested directory.
+-- The others are openmpi using non-standard directories for some reason.
+WITH b(id) AS (SELECT id FROM locales WHERE locale IN('man', 'overrides') OR locale ~ '(openmpi|mpich|mvapich)'),
+     f     AS (UPDATE files SET locale = 0 WHERE locale IN(SELECT id FROM b))
+DELETE FROM locales WHERE id IN(SELECT id FROM b);