Use a branchless binary search for extension<->MIME-Types

This uses the excellent branchless binary search by Orlon Peters[1],
based on the Malte Skarupke's version[2] of Leonard E Shar's version of
a binary search.  It's a fascinating implementation that ends up
becoming just a very tight loop using CMOV instructions, and finishing
with another CMOV instruction.

It's so clean that I removed the fast path using STRING_SWITCH as
that's not necessary anymore.  A nice side effect of this change is
that the string "application/octet-stream" doesn't appear in the binary
anymore (except in the debug version due to assertions), as it's now
part of the compressed blob that mimegen generates.

A good segue to this commit would be porting the other usage of
bsearch() to this implementation.

[1] https://orlp.net/blog/bitwise-binary-search/
[2] https://probablydance.com/2023/04/27/beautiful-branchless-binary-search/

Author: lpereira

src/bin/tools/mimegen.c

@@ -161,29 +161,6 @@ static char *compress_output(const struct output *output, size_t *outlen)
     return compressed;
 }
 
-static bool is_builtin_ext(const char *ext)
-{
-    /* STRING_SWITCH_L() is not used here to not bring in lwan.h */
-    /* FIXME: maybe use an X-macro to keep in sync with lwan-tables.c? */
-    if (strcaseequal_neutral(ext, "css"))
-        return true;
-    if (strcaseequal_neutral(ext, "gif"))
-        return true;
-    if (strcaseequal_neutral(ext, "htm"))
-        return true;
-    if (strcaseequal_neutral(ext, "html"))
-        return true;
-    if (strcaseequal_neutral(ext, "jpg"))
-        return true;
-    if (strcaseequal_neutral(ext, "js"))
-        return true;
-    if (strcaseequal_neutral(ext, "png"))
-        return true;
-    if (strcaseequal_neutral(ext, "txt"))
-        return true;
-    return false;
-}
-
 int main(int argc, char *argv[])
 {
     /* 32k is sufficient for the provided mime.types, but we can reallocate
@@ -258,11 +235,6 @@ int main(int argc, char *argv[])
                 ext[8] = '\0';
             }
 
-            /* Lwan has a fast-path for some common extensions, so don't bundle them
-             * in this table if not really needed. */
-            if (is_builtin_ext(ext))
-                continue;
-
             k = strdup(ext);
             v = strdup(mime_type);
 
@@ -286,6 +258,22 @@ int main(int argc, char *argv[])
         }
     }
 
+    {
+        char *k = strdup("bin");
+        char *v = strdup("application/octet-stream");
+        if (!k || !v) {
+            fprintf(stderr, "Could not allocate memory\n");
+            fclose(fp);
+            return 1;
+        }
+        int r = hash_add_unique(ext_mime, k, v);
+        if (r != 0 && r != -EEXIST) {
+            fprintf(stderr, "Could not add fallback mime entry\n");
+            fclose(fp);
+            return 1;
+        }
+    }
+
     /* Get sorted list of extensions. */
     exts = calloc(hash_get_count(ext_mime), sizeof(char *));
     if (!exts) {
@@ -305,6 +293,7 @@ int main(int argc, char *argv[])
         fclose(fp);
         return 1;
     }
+    ssize_t bin_index = -1;
     for (i = 0; i < hash_get_count(ext_mime); i++) {
         uint64_t ext_lower = 0;
 
@@ -322,6 +311,9 @@ int main(int argc, char *argv[])
             fclose(fp);
             return 1;
         }
+
+        if (bin_index < 0 && streq(exts[i], "bin"))
+            bin_index = (ssize_t)i;
     }
     for (i = 0; i < hash_get_count(ext_mime); i++) {
         if (output_append(&output, hash_find(ext_mime, exts[i])) < 0) {
@@ -331,6 +323,12 @@ int main(int argc, char *argv[])
         }
     }
 
+    if (bin_index < 0) {
+        fprintf(stderr, "Could not find fallback item after sorting!\n");
+        fclose(fp);
+        return 1;
+    }
+
     /* Compress blob. */
     compressed = compress_output(&output, &compressed_size);
     if (!compressed) {
@@ -349,10 +347,15 @@ int main(int argc, char *argv[])
 #else
     printf("/* Compressed with zlib (deflate) */\n");
 #endif
+
+    unsigned int entries_floor = 1u << (31 - __builtin_clz(hash_get_count(ext_mime)));
+
     printf("#pragma once\n");
     printf("#define MIME_UNCOMPRESSED_LEN %zu\n", output.used);
     printf("#define MIME_COMPRESSED_LEN %lu\n", compressed_size);
     printf("#define MIME_ENTRIES %d\n", hash_get_count(ext_mime));
+    printf("#define MIME_ENTRIES_FLOOR %d\n", entries_floor);
+    printf("#define MIME_ENTRY_FALLBACK %ld\n", bin_index);
     printf("static const unsigned char mime_entries_compressed[] = {\n");
     for (i = 1; compressed_size; compressed_size--, i++)
         printf("0x%02x,%c", compressed[i - 1] & 0xff, " \n"[i % 13 == 0]);

src/lib/lwan-tables.c

@@ -36,6 +36,7 @@
 
 static unsigned char uncompressed_mime_entries[MIME_UNCOMPRESSED_LEN];
 static char *mime_types[MIME_ENTRIES];
+static uint64_t *mime_extensions;
 static bool mime_entries_initialized = false;
 
 void lwan_tables_shutdown(void)
@@ -86,6 +87,7 @@ void lwan_tables_init(void)
         mime_types[i] = (char *)ptr;
         ptr += strlen((const char *)ptr) + 1;
     }
+    mime_extensions = (uint64_t *)uncompressed_mime_entries;
 
     mime_entries_initialized = true;
 
@@ -120,34 +122,25 @@ LWAN_SELF_TEST(status_codes)
 #undef ASSERT_STATUS
 }
 
-static int compare_mime_entry(const void *a, const void *b)
+static ALWAYS_INLINE const char *bsearch_mime_type(uint64_t ext)
 {
-    const uint64_t exta = string_as_uint64((const char *)a);
-    const uint64_t extb = string_as_uint64((const char *)b);
-
-    return (exta > extb) - (exta < extb);
+    /* Based on https://orlp.net/blog/bitwise-binary-search/ */
+    int64_t b = ext > mime_extensions[MIME_ENTRIES / 2]
+                    ? MIME_ENTRIES - MIME_ENTRIES_FLOOR
+                    : -1;
+    for (uint64_t bit = MIME_ENTRIES_FLOOR >> 1; bit != 0; bit >>= 1) {
+        if (ext > mime_extensions[b + (int64_t)bit])
+            b += (int64_t)bit;
+    }
+    return mime_types[mime_extensions[b + 1] == ext ? b + 1
+                                                    : MIME_ENTRY_FALLBACK];
 }
 
-const char *
-lwan_determine_mime_type_for_file_name(const char *file_name)
+const char *lwan_determine_mime_type_for_file_name(const char *file_name)
 {
     char *last_dot = strrchr(file_name, '.');
-    if (UNLIKELY(!last_dot))
-        goto fallback;
-
-    STRING_SWITCH_L(last_dot) {
-    case STR4_INT_L('.','c','s','s'): return "text/css";
-    case STR4_INT_L('.','g','i','f'): return "image/gif";
-    case STR4_INT_L('.','h','t','m'): return "text/html";
-    case STR4_INT_L('.','j','p','g'): return "image/jpeg";
-    case STR4_INT_L('.','j','s',' '): return "text/javascript";
-    case STR4_INT_L('.','p','n','g'): return "image/png";
-    case STR4_INT_L('.','t','x','t'): return "text/plain";
-    }
-
-    if (LIKELY(*last_dot)) {
+    if (LIKELY(last_dot && *last_dot)) {
         uint64_t key = 0;
-        const unsigned char *extension;
 
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wstringop-truncation"
@@ -157,17 +150,11 @@ lwan_determine_mime_type_for_file_name(const char *file_name)
          * 8 bytes per extension. */
         strncpy((char *)&key, last_dot + 1, 8);
 #pragma GCC diagnostic pop
-        key &= ~0x2020202020202020ull;
-        key = htobe64(key);
 
-        extension = bsearch(&key, uncompressed_mime_entries, MIME_ENTRIES, 8,
-                            compare_mime_entry);
-        if (LIKELY(extension))
-            return mime_types[(extension - uncompressed_mime_entries) / 8];
+        return bsearch_mime_type(htobe64(key & ~0x2020202020202020ull));
     }
 
-fallback:
-    return "application/octet-stream";
+    return mime_types[MIME_ENTRY_FALLBACK];
 }
 
 #include "lookup-http-status.h" /* genrated by statuslookupgen */