fix: rebuild index when mode adds capabilities

Signed-off-by: Anton Standrik <astandrik@yandex-team.ru>

Author: astandrik

src/pipeline/pipeline.c

@@ -33,6 +33,7 @@ enum { CBM_DIR_PERMS = 0755, PL_RING = 4, PL_RING_MASK = 3, PL_SEQ_PASSES = 6, P
 #include "foundation/compat_thread.h"
 #include "foundation/profile.h"
 #include "foundation/mem.h"
+#include "yyjson/yyjson.h"
 
 #include <stdint.h>
 #include <stdio.h>
@@ -151,6 +152,66 @@ static const char *itoa_buf(int val) {
     return bufs[i];
 }
 
+static const char *index_mode_name(cbm_index_mode_t mode) {
+    switch (mode) {
+    case CBM_MODE_FULL:
+        return "full";
+    case CBM_MODE_MODERATE:
+        return "moderate";
+    case CBM_MODE_FAST:
+        return "fast";
+    default:
+        return "unknown";
+    }
+}
+
+static bool parse_index_mode(const char *properties_json, cbm_index_mode_t *out_mode) {
+    if (!properties_json || !out_mode) {
+        return false;
+    }
+
+    yyjson_doc *doc = yyjson_read(properties_json, strlen(properties_json), 0);
+    if (!doc) {
+        return false;
+    }
+
+    yyjson_val *root = yyjson_doc_get_root(doc);
+    yyjson_val *mode_value = yyjson_is_obj(root) ? yyjson_obj_get(root, "index_mode") : NULL;
+    const char *mode_name =
+        mode_value && yyjson_is_str(mode_value) ? yyjson_get_str(mode_value) : NULL;
+    bool parsed = true;
+    if (mode_name && strcmp(mode_name, "full") == 0) {
+        *out_mode = CBM_MODE_FULL;
+    } else if (mode_name && strcmp(mode_name, "moderate") == 0) {
+        *out_mode = CBM_MODE_MODERATE;
+    } else if (mode_name && strcmp(mode_name, "fast") == 0) {
+        *out_mode = CBM_MODE_FAST;
+    } else {
+        parsed = false;
+    }
+
+    yyjson_doc_free(doc);
+    return parsed;
+}
+
+static bool index_mode_covers(cbm_index_mode_t stored_mode, cbm_index_mode_t requested_mode) {
+    return stored_mode >= CBM_MODE_FULL && stored_mode <= CBM_MODE_FAST &&
+           requested_mode >= CBM_MODE_FULL && requested_mode <= CBM_MODE_FAST &&
+           stored_mode <= requested_mode;
+}
+
+static bool read_stored_index_mode(cbm_store_t *store, const char *project,
+                                   cbm_index_mode_t *out_mode) {
+    cbm_node_t project_node = {0};
+    if (cbm_store_find_node_by_qn(store, project, project, &project_node) != CBM_STORE_OK) {
+        return false;
+    }
+
+    bool parsed = parse_index_mode(project_node.properties_json, out_mode);
+    cbm_node_free_fields(&project_node);
+    return parsed;
+}
+
 /* Log current + peak RSS at a pipeline phase boundary (memory profiling). */
 static void log_phase_mem(const char *phase) {
     enum { PL_BYTES_PER_MB = 1024 * 1024 };
@@ -446,7 +507,11 @@ static int pass_structure(cbm_pipeline_t *p, const cbm_file_info_t *files, int f
     cbm_log_info("pass.start", "pass", "structure", "files", itoa_buf(file_count));
 
     /* Project node */
-    cbm_gbuf_upsert_node(p->gbuf, "Project", p->project_name, p->project_name, NULL, 0, 0, "{}");
+    char project_props[CBM_SZ_64];
+    snprintf(project_props, sizeof(project_props), "{\"index_mode\":\"%s\"}",
+             index_mode_name(p->mode));
+    cbm_gbuf_upsert_node(p->gbuf, "Project", p->project_name, p->project_name, NULL, 0, 0,
+                         project_props);
     const char *branch_qn = p->branch_qn ? p->branch_qn : p->project_name;
     const char *branch_name = p->git_ctx.branch ? p->git_ctx.branch : "working-tree";
     char branch_props[CBM_SZ_2K];
@@ -998,17 +1063,26 @@ static int try_incremental_or_delete_db(cbm_pipeline_t *p, cbm_file_info_t *file
     if (check_store && cbm_store_check_integrity(check_store)) {
         cbm_file_hash_t *hashes = NULL;
         int hash_count = 0;
+        cbm_index_mode_t stored_mode = CBM_MODE_FAST;
+        bool stored_mode_known = read_stored_index_mode(check_store, p->project_name, &stored_mode);
         cbm_store_get_file_hashes(check_store, p->project_name, &hashes, &hash_count);
         cbm_store_free_file_hashes(hashes, hash_count);
         cbm_store_close(check_store);
-        if (hash_count > 0 && file_count <= hash_count + (hash_count / PAIR_LEN)) {
+        bool mode_covered = stored_mode_known && index_mode_covers(stored_mode, p->mode);
+        if (hash_count > 0 && mode_covered && file_count <= hash_count + (hash_count / PAIR_LEN)) {
             cbm_log_info("pipeline.route", "path", "incremental", "stored_hashes",
-                         itoa_buf(hash_count));
-            int rc = cbm_pipeline_run_incremental(p, db_path, files, file_count);
+                         itoa_buf(hash_count), "stored_mode", index_mode_name(stored_mode),
+                         "requested_mode", index_mode_name(p->mode), "effective_mode",
+                         index_mode_name(stored_mode));
+            int rc = cbm_pipeline_run_incremental(p, db_path, files, file_count, stored_mode);
             free(db_path);
             return rc;
         }
-        if (hash_count > 0) {
+        if (hash_count > 0 && !mode_covered) {
+            cbm_log_info("pipeline.route", "path", "mode_upgrade_reindex", "stored_mode",
+                         stored_mode_known ? index_mode_name(stored_mode) : "unknown",
+                         "requested_mode", index_mode_name(p->mode));
+        } else if (hash_count > 0) {
             cbm_log_info("pipeline.route", "path", "mode_change_reindex", "stored_hashes",
                          itoa_buf(hash_count), "discovered", itoa_buf(file_count));
         }

src/pipeline/pipeline_incremental.c

@@ -684,11 +684,12 @@ static void dump_and_persist(cbm_gbuf_t *gbuf, const char *db_path, const char *
 /* ── Incremental pipeline entry point ────────────────────────────── */
 
 int cbm_pipeline_run_incremental(cbm_pipeline_t *p, const char *db_path, cbm_file_info_t *files,
-                                 int file_count) {
+                                 int file_count, cbm_index_mode_t effective_mode) {
     struct timespec t0;
     cbm_clock_gettime(CLOCK_MONOTONIC, &t0);
 
     const char *project = cbm_pipeline_project_name(p);
+    cbm_index_mode_t requested_mode = (cbm_index_mode_t)cbm_pipeline_get_mode(p);
 
     /* Open existing disk DB */
     cbm_store_t *store = cbm_store_open_path(db_path);
@@ -839,7 +840,7 @@ int cbm_pipeline_run_incremental(cbm_pipeline_t *p, const char *db_path, cbm_fil
         .registry = registry,
         .cancelled = cbm_pipeline_cancelled_ptr(p),
         .pipeline = p, /* so passes can record per-file skips (Track B) */
-        .mode = cbm_pipeline_get_mode(p),
+        .mode = effective_mode,
         .path_aliases = path_aliases,
         .excluded_dirs = excluded_dirs,
         .excluded_count = excluded_count,
@@ -854,7 +855,12 @@ int cbm_pipeline_run_incremental(cbm_pipeline_t *p, const char *db_path, cbm_fil
         }
     }
 
+    /* Discovery already used requested_mode. Re-extraction must preserve the
+     * stronger capabilities recorded in Project.index_mode, including full-only
+     * C/C++ Macro nodes. Restore the process-wide gate immediately afterwards. */
+    cbm_set_macro_extraction(effective_mode == CBM_MODE_FULL);
     run_extract_resolve(&ctx, changed_files, ci);
+    cbm_set_macro_extraction(requested_mode == CBM_MODE_FULL);
     cbm_pipeline_pass_k8s(&ctx, changed_files, ci);
     run_postpasses(&ctx, changed_files, ci, project);
 

src/pipeline/pipeline_internal.h

@@ -607,9 +607,11 @@ int cbm_scan_project_env_urls_excluded(const char *root_path, cbm_env_binding_t
 
 /* Run incremental re-index on an existing disk DB.
  * Classifies files by mtime+size, deletes changed nodes, re-parses changed
- * files, merges into disk DB. Returns 0 on success. */
+ * files with effective_mode capabilities, then merges into disk DB. The
+ * pipeline's requested mode still owns discovery/exclusions. Returns 0 on
+ * success. */
 int cbm_pipeline_run_incremental(cbm_pipeline_t *p, const char *db_path, cbm_file_info_t *files,
-                                 int file_count);
+                                 int file_count, cbm_index_mode_t effective_mode);
 
 /* Pipeline accessors for incremental use */
 const char *cbm_pipeline_repo_path(const cbm_pipeline_t *p);