Add indexer arugment '--micromamba-download-url'

* Add is_file_compressed() function to utils
* Call micromamba_install() from tests

7 files changed, 104 insertions, 8 deletions

diff --git a/src/cli/stasis_indexer/args.c b/src/cli/stasis_indexer/args.c
index 0d0e9b9..ccc69e6 100644
--- a/src/cli/stasis_indexer/args.c
+++ b/src/cli/stasis_indexer/args.c
@@ -7,6 +7,7 @@ struct option long_options[] = {
     {"verbose", no_argument, 0, 'v'},
     {"unbuffered", no_argument, 0, 'U'},
     {"web", no_argument, 0, 'w'},
+    {"micromamba-download-url", required_argument, 0, OPT_MICROMAMBA_DOWNLOAD_URL},
     {0, 0, 0, 0},
 };
 
@@ -16,6 +17,7 @@ const char *long_options_help[] = {
     "Increase output verbosity",
     "Disable line buffering",
     "Generate HTML indexes (requires pandoc)",
+    "Set micromamba download URL",
     NULL,
 };
 
diff --git a/src/cli/stasis_indexer/helpers.c b/src/cli/stasis_indexer/helpers.c
index bdb538d..b89622b 100644
--- a/src/cli/stasis_indexer/helpers.c
+++ b/src/cli/stasis_indexer/helpers.c
@@ -157,6 +157,7 @@ int micromamba_configure(const struct Delivery *ctx, struct MicromambaInfo *m) {
     m->conda_prefix = globals.conda_install_prefix;
     m->micromamba_prefix = micromamba_prefix;
     m->download_dir = ctx->storage.tmpdir;
+    m->download_url = globals.micromamba_download_url;
 
     const size_t pathvar_len = strlen(getenv("PATH")) + strlen(m->micromamba_prefix) + strlen(m->conda_prefix) + 3 + 4 + 1;
     // ^^^^^^^^^^^^^^^^^^
diff --git a/src/cli/stasis_indexer/include/args.h b/src/cli/stasis_indexer/include/args.h
index 543aa4b..080863c 100644
--- a/src/cli/stasis_indexer/include/args.h
+++ b/src/cli/stasis_indexer/include/args.h
@@ -3,6 +3,8 @@
 
 #include <getopt.h>
 
+#define OPT_MICROMAMBA_DOWNLOAD_URL 1000
+
 extern struct option long_options[];
 void usage(char *name);
 
diff --git a/src/cli/stasis_indexer/stasis_indexer_main.c b/src/cli/stasis_indexer/stasis_indexer_main.c
index c13d175..5d86f29 100644
--- a/src/cli/stasis_indexer/stasis_indexer_main.c
+++ b/src/cli/stasis_indexer/stasis_indexer_main.c
@@ -197,6 +197,13 @@ int main(const int argc, char *argv[]) {
             case 'w':
                 do_html = 1;
                 break;
+            case OPT_MICROMAMBA_DOWNLOAD_URL:
+                globals.micromamba_download_url = strdup(optarg);
+                if (!globals.micromamba_download_url) {
+                    SYSERROR("unable to allocate memory for micromamba_download_url");
+                    exit(1);
+                }
+                break;
             case '?':
             default:
                 exit(1);
@@ -326,6 +333,13 @@ int main(const int argc, char *argv[]) {
         SYSWARN("Unable to open stasis configuration file: %s", cfg_path);
     }
 
+    if (globals.micromamba_download_url && isempty(globals.micromamba_download_url)) {
+        // safeguard against supplying a zero-length URL
+        // this covers the case where the user supplied it as an argument and/or in the config file
+        SYSERROR("micromamba download URL cannot be empty");
+        exit(1);
+    }
+
     indexer_init_dirs(&ctx, workdir);
 
     msg(STASIS_MSG_L1, "%s delivery root %s\n",
diff --git a/src/lib/core/conda.c b/src/lib/core/conda.c
index 444cfaa..5c7779f 100644
--- a/src/lib/core/conda.c
+++ b/src/lib/core/conda.c
@@ -36,9 +36,10 @@ int micromamba_install(const struct MicromambaInfo *info) {
     // https://github.com/mamba-org/micromamba-releases/releases/download/${version}/micromamba-${arch}
 
     // Micromamba hosts binaries on github and on their own website. Prefer github.
+    // The "latest" binary from micromamba's site is compressed with bzip2 (06/2026)
     const char *url_fmts[] = {
         info->download_url,
-        "https://github.com/mamba-org/micromamba-releases/releases/latest/download/micromamba-%s-%s",
+        "https://github.com/mamba-org/micromamba-releases/releases/latest/download/micromamba-%s-%s.tar.bz2",
         "https://micro.mamba.pm/api/micromamba/%s-%s/latest",
     };
     const size_t url_fmts_max = sizeof(url_fmts) / sizeof(url_fmts[0]);
@@ -84,14 +85,26 @@ int micromamba_install(const struct MicromambaInfo *info) {
     snprintf(mmbin, sizeof(mmbin), "%s/micromamba", info->micromamba_prefix);
 
     if (access(mmbin, F_OK)) {
-        char untarcmd[PATH_MAX * 2];
         mkdirs(info->micromamba_prefix, 0755);
-        snprintf(untarcmd, sizeof(untarcmd),
-            "tar -xvf %s -C %s --strip-components=1 bin/micromamba 1>/dev/null",
-            installer_path, info->micromamba_prefix);
-        int untarcmd_status = system(untarcmd);
-        if (untarcmd_status) {
-            return -1;
+
+        if (is_file_compressed(installer_path)) {
+            char untarcmd[PATH_MAX * 2] = {0};
+            snprintf(untarcmd, sizeof(untarcmd),
+                "tar -xvf %s -C %s --strip-components=1 bin/micromamba 1>/dev/null",
+                installer_path, info->micromamba_prefix);
+            int untarcmd_status = system(untarcmd);
+            if (untarcmd_status) {
+                return -1;
+            }
+        } else {
+            if (copy2(installer_path, mmbin, CT_PERM)) {
+                SYSERROR("unable to copy %s to %s", installer_path, mmbin);
+                return -1;
+            }
+            if (chmod(mmbin, 0755)) {
+                SYSERROR("unable to set permissions: %s (%s)", mmbin, strerror(errno));
+                return -1;
+            }
         }
     }
     return 0;
diff --git a/src/lib/core/include/utils.h b/src/lib/core/include/utils.h
index 98b8ae8..3f0fe9f 100644
--- a/src/lib/core/include/utils.h
+++ b/src/lib/core/include/utils.h
@@ -492,4 +492,22 @@ int get_random_bytes(char *result, size_t maxlen);
 int non_format_len(const char *s);
 
 char *center_text(const char *s, size_t maxwidth);
+
+/**
+ * Check magic bytes against known compression formats
+ *
+ * ```c
+ * const char *filename = "/path/to/file.zip";
+ * if (is_file_compressed(filename)) {
+ *     // file is compressed
+ * } else {
+ *     // file is not compressed
+ * }
+ * ```
+ *
+ * @param filename path to maybe-compressed file
+ * @return 0 if not compressed
+ * @return 1 if compressed
+ */
+int is_file_compressed(const char *filename);
 #endif //STASIS_UTILS_H
diff --git a/src/lib/core/utils.c b/src/lib/core/utils.c
index b4a520d..152c5c5 100644
--- a/src/lib/core/utils.c
+++ b/src/lib/core/utils.c
@@ -1273,3 +1273,49 @@ char *center_text(const char *s, const size_t maxwidth) {
     return result;
 }
 
+int is_file_compressed(const char *filename) {
+    FILE *fp = fopen(filename, "rb");
+    if (!fp) {
+        SYSERROR("Unable to open for reading: %s", filename);
+        return -1;
+    }
+
+    // Container for magic bytes
+    struct Magic {
+        const unsigned char *byte;
+        const size_t size;
+    };
+
+    // Array of magic bytes for different compression types
+    const struct Magic magic[] = {
+        {(unsigned char *) "BZh", 3},                       // bzip2
+        {(unsigned char *) "\x1f\x8b", 2},                  // gzip
+        {(unsigned char *) "\xfd\x37\x7a\x58\x5a\x00", 6},  // xz
+        {(unsigned char *) "PK\03\04", 3},                  // zip
+        {(unsigned char *) "PK\05\06", 3},                  // zip (empty)
+        {(unsigned char *) "PK\07\08", 3},                  // zip (spanned)
+        {(unsigned char *) "\xfd\x2f\xb5\x28", 4}           // zstd
+    };
+    unsigned char buf[8] = {0}; // unsigned long
+    size_t bytes_read = 0;
+    bytes_read = fread(buf, 1, sizeof(buf), fp);
+    if (bytes_read < sizeof(buf)) {
+        SYSWARN("consumed fewer than %zu bytes (%zu) from %s", sizeof(buf), bytes_read, filename);
+        fclose(fp);
+        // well, the file isn't compressed.
+        return 0;
+    }
+    fclose(fp);
+
+    // Compare known magic bytes to consumed data
+    for (size_t i = 0; i < sizeof(magic) / sizeof(magic[0]); i++) {
+        if (memcmp(buf, magic[i].byte, magic[i].size) == 0) {
+            // match
+            return 1;
+        }
+    }
+
+    // no match
+    return 0;
+}
+