feat(plip): stage files onto SD via /game/file

/game/file only wrote SPIFFS (960 KB, too small for a voice pack). A
path prefixed sd/ (or /sdcard/) now routes to the microSD: the card is
mounted on demand (audio_ensure_sd, independent of hook state) and the
parent dir is created. MAX_BODY 256->512 KB so full say() hints fit
(streamed in 2 KB chunks, RAM-safe). Enable FATFS LFN_HEAP so >8.3
filenames work (greet_0142738200.wav etc). Validated: 40-sample FR
voice pack staged to /sdcard/voice on the bench PLIP (960 MB card).

plip_voice/main/audio.c

@@ -137,6 +137,12 @@ static void ensure_sd_mounted(void)
     }
 }
 
+bool audio_ensure_sd(void)
+{
+    ensure_sd_mounted();
+    return s_sd_mounted;
+}
+
 /* ── WAV parser ──────────────────────────────────────────────────────────── */
 
 static esp_err_t parse_wav_header(const uint8_t *buf, size_t len, wav_info_t *out)

plip_voice/main/audio.h

@@ -39,6 +39,12 @@ esp_err_t audio_play_async(const char *path);
 /* Stop current playback immediately. */
 esp_err_t audio_stop(void);
 
+/* Mount the microSD card at /sdcard on demand (best-effort, idempotent) and
+ * report whether it is mounted. Unlike playback, this does NOT depend on the
+ * hook state — it lets the REST layer stage a voice pack onto the SD while the
+ * handset is down. Returns true if /sdcard is usable. */
+bool audio_ensure_sd(void);
+
 /* True while the audio worker is playing a clip (tone/WAV). The conversation
  * LISTEN loop polls this to stay half-duplex: never capture while playing
  * (avoids the earpiece→mic feedback that saturated the line). */

plip_voice/main/net.c

@@ -22,6 +22,7 @@
 #include <string.h>
 #include <stdio.h>
 #include <errno.h>
+#include <sys/stat.h>
 
 #include "audio.h"
 #include "tones.h"
@@ -53,7 +54,8 @@
 
 #define SPIFFS_LABEL  "storage"
 #define SPIFFS_BASE   "/spiffs"
-#define MAX_BODY      (256 * 1024)
+#define MAX_BODY      (512 * 1024)   /* streamed in 2 KB chunks → RAM-safe; raised
+                                        so full-length say() hint WAVs fit on SD */
 
 static volatile bool    s_connected  = false;
 static httpd_handle_t   s_httpd      = NULL;
@@ -209,14 +211,39 @@ static esp_err_t handle_file_post(httpd_req_t *req)
     if (req->content_len <= 0 || req->content_len > MAX_BODY)
         return send_json(req, "413 Payload Too Large", "{\"error\":\"too large\"}");
 
-    ensure_spiffs();
+    /* Route to the microSD when the path is sd-prefixed ("sd/…", "sdcard/…"
+     * or "/sdcard/…"); otherwise stage in SPIFFS (default). The SD is mounted
+     * on demand here so a voice pack can be staged without an off-hook play. */
+    const char *base;
+    const char *rel = fpath;
+    bool to_sd = false;
+    if      (strncmp(fpath, "sd/", 3) == 0)       { rel = fpath + 3; to_sd = true; }
+    else if (strncmp(fpath, "sdcard/", 7) == 0)   { rel = fpath + 7; to_sd = true; }
+    else if (strncmp(fpath, "/sdcard/", 8) == 0)  { rel = fpath + 8; to_sd = true; }
 
-    /* Build full SPIFFS path. */
-    char full[160];
-    if (fpath[0] == '/') {
-        snprintf(full, sizeof(full), "%s%s", SPIFFS_BASE, fpath);
+    if (to_sd) {
+        if (!audio_ensure_sd())
+            return send_json(req, "503 Service Unavailable", "{\"error\":\"no sd card\"}");
+        base = PLIP_SD_MOUNT;   /* "/sdcard" */
     } else {
-        snprintf(full, sizeof(full), "%s/%s", SPIFFS_BASE, fpath);
+        ensure_spiffs();
+        base = SPIFFS_BASE;
+    }
+    while (*rel == '/') rel++;   /* avoid a double slash when joining */
+
+    char full[192];
+    snprintf(full, sizeof(full), "%s/%s", base, rel);
+
+    /* Create the immediate parent dir (one level) so a pack can live in a
+     * subfolder like /sdcard/voice/<file>.wav — fopen won't create it. FAT
+     * supports mkdir; SPIFFS is flat so we only do this for the SD path. */
+    if (to_sd) {
+        char *slash = strrchr(full, '/');
+        if (slash && slash != full) {
+            *slash = '\0';
+            mkdir(full, 0775);   /* best-effort; ignore EEXIST */
+            *slash = '/';
+        }
     }
 
     FILE *f = fopen(full, "wb");

plip_voice/sdkconfig.defaults

@@ -52,3 +52,7 @@ CONFIG_PLIP_HOOK_ACTIVE_HIGH=y
 # Example: CONFIG_PLIP_GATEWAY_URL="http://192.168.0.175:8401"
 #          CONFIG_PLIP_GATEWAY_TOKEN="testtoken"
 # Default (Kconfig): http://192.168.0.50:8401 with empty token.
+
+# Long file names on the SD (voice sample pack uses >8.3 names)
+CONFIG_FATFS_LFN_HEAP=y
+CONFIG_FATFS_MAX_LFN=255