Merge origin/main into fix/adr-reindex-race

Signed-off-by: Martin Vogel <martin.vogel.tech@gmail.com>

internal/cbm/extract_calls.c

@@ -2249,11 +2249,15 @@ static char *extract_callee_name(CBMArena *a, TSNode node, const char *source, C
     // method is `new`.  The constructor body lives in `initialize`, so a callee
     // of "new" never resolves.  Redirect to the receiver type name so the call
     // links to the class/constructor like every other language's `new T()`.
+    // Scope-resolved receivers (`Admin::User.new`) get the same treatment —
+    // the callee carries the full constant path ("Admin::User"), whose bare
+    // leaf ("User") is what resolution joins on.
     if (lang == CBM_LANG_RUBY) {
         TSNode m = ts_node_child_by_field_name(node, TS_FIELD("method"));
         TSNode recv = ts_node_child_by_field_name(node, TS_FIELD("receiver"));
         if (!ts_node_is_null(m) && !ts_node_is_null(recv) &&
-            strcmp(ts_node_type(recv), "constant") == 0) {
+            (strcmp(ts_node_type(recv), "constant") == 0 ||
+             strcmp(ts_node_type(recv), "scope_resolution") == 0)) {
             char *mt = cbm_node_text(a, m, source);
             if (mt && strcmp(mt, "new") == 0) {
                 char *rt = cbm_node_text(a, recv, source);

src/discover/discover.h

@@ -58,6 +58,12 @@ CBMLanguage cbm_disambiguate_frm(const char *path);
  * CBM_LANG_RESCRIPT. On read failure, defaults to CBM_LANG_RESCRIPT. */
 CBMLanguage cbm_disambiguate_res(const char *path);
 
+/* Disambiguate .d files by reading first 4KB of content.
+ * Returns CBM_LANG_COUNT (not source) for a make-style dependency file written
+ * by rustc/cargo, gcc/clang -MD or CMake ("target ...: prereq ..."), otherwise
+ * CBM_LANG_DLANG. On read failure or any doubt, defaults to CBM_LANG_DLANG. */
+CBMLanguage cbm_disambiguate_d(const char *path);
+
 /* Disambiguate .inc files by reading first 4KB of content.
  * Returns CBM_LANG_OBJECTSCRIPT_ROUTINE if it looks like an ObjectScript
  * include (a "ROUTINE <Uppercase>" header), otherwise CBM_LANG_BITBAKE.

src/discover/language.c

@@ -1436,6 +1436,107 @@ CBMLanguage cbm_disambiguate_res(const char *path) {
     return lang_res_bytes(head, n);
 }
 
+/* A make rule's targets end at a ':' followed by whitespace or the end of the
+ * file, which skips drive-letter colons ("C:/", "C:\"). */
+static bool is_rule_colon(const char *p, bool whole_file) {
+    char next = p[SKIP_ONE];
+    return *p == ':' && (next == ' ' || next == '\t' || next == '\r' || next == '\n' ||
+                         (next == '\0' && whole_file));
+}
+
+/* Bytes of D whitespace at p, or 0: space, tab, vertical tab, form feed, CR
+ * and the UTF-8 line and paragraph separators U+2028/U+2029, all of which split
+ * D tokens. '\n' ends the line and is the caller's. */
+static size_t d_space_len(const char *p) {
+    if (*p == ' ' || *p == '\t' || *p == '\v' || *p == '\f' || *p == '\r') {
+        return SKIP_ONE;
+    }
+    if (strncmp(p, "\xE2\x80\xA8", SLEN("\xE2\x80\xA8")) == 0 ||
+        strncmp(p, "\xE2\x80\xA9", SLEN("\xE2\x80\xA9")) == 0) {
+        return SLEN("\xE2\x80\xA8");
+    }
+    return 0;
+}
+
+/* True if the line at p is a make rule "target...: prereq..." whose targets all
+ * look like paths (contain '/', '\' or '.'); a backslash-escaped space stays
+ * inside its target. D source fails this: "public:", "@safe:", "extern(C):"
+ * and "import a.b : c" all have a non-path word or D punctuation before the
+ * colon. An attribute ('@', as in "@1.0:") or a comment between tokens
+ * ("public/+ +/:") is D even with a '.' or '/' before the colon, so a target
+ * holding either is never dep-info. */
+static bool is_dep_rule_line(const char *p, bool whole_file) {
+    bool in_target = false;
+    bool path_like = false;
+    bool any_target = false;
+    for (; *p && *p != '\n'; p++) {
+        if (is_rule_colon(p, whole_file)) {
+            return in_target ? path_like : any_target;
+        }
+        size_t space = d_space_len(p);
+        if (space) {
+            if (in_target && !path_like) {
+                return false;
+            }
+            in_target = false;
+            path_like = false;
+            p += space - SKIP_ONE;
+            continue;
+        }
+        if (strchr("(){};=\"',@", *p)) {
+            return false;
+        }
+        if (*p == '/' && (p[SKIP_ONE] == '*' || p[SKIP_ONE] == '+' || p[SKIP_ONE] == '/')) {
+            return false;
+        }
+        in_target = true;
+        any_target = true;
+        path_like = path_like || *p == '/' || *p == '\\' || *p == '.';
+        if (*p == '\\' && p[SKIP_ONE] == ' ') {
+            p++; /* escaped space: part of this target */
+        }
+    }
+    return false;
+}
+
+/* Disambiguate .d files: shared by D source and make-style dependency files
+ * written by rustc/cargo (target/<profile>/deps/<crate>-<hash>.d), gcc/clang -MD
+ * and CMake (<object>.o.d). Those are "target: prereq ..." rules, often tens of
+ * KB on one line, which the D grammar parses slowly into nothing but a module
+ * node, so they are reported as unsupported (CBM_LANG_COUNT). Defaults to D on
+ * any doubt (preserves existing behaviour). `n` is the count of head bytes buf
+ * was made from. */
+static CBMLanguage lang_d_text(const char *buf, size_t n) {
+    /* A colon at the end of buf only ends the rule if nothing was cut off. */
+    bool whole_file = n < LANG_PROBE_HEAD && strlen(buf) == n;
+
+    const char *p = buf;
+    if (strncmp(p, "\xEF\xBB\xBF", SLEN("\xEF\xBB\xBF")) == 0) {
+        p += SLEN("\xEF\xBB\xBF");
+    }
+    while (*p && isspace((unsigned char)*p)) {
+        p++;
+    }
+    /* A shebang or a leading D comment (including the DUB single-file
+     * "/+ dub.sdl:" recipe) is D; dep-info never starts that way. */
+    if (*p == '#' ||
+        (p[0] == '/' && (p[SKIP_ONE] == '/' || p[SKIP_ONE] == '*' || p[SKIP_ONE] == '+'))) {
+        return CBM_LANG_DLANG;
+    }
+    return is_dep_rule_line(p, whole_file) ? CBM_LANG_COUNT : CBM_LANG_DLANG;
+}
+
+CBMLanguage cbm_disambiguate_d(const char *path) {
+    unsigned char head[LANG_PROBE_HEAD];
+    size_t n = 0;
+    if (!lang_read_head(path, head, sizeof(head), &n)) {
+        return CBM_LANG_DLANG;
+    }
+    char buf[LANG_PROBE_HEAD + SKIP_ONE];
+    lang_head_text(buf, LANG_PROBE_HEAD, head, n);
+    return lang_d_text(buf, n);
+}
+
 /* Disambiguate .cls files: shared by InterSystems ObjectScript UDL, Salesforce
  * Apex and Visual Basic 6 class modules (#721). ObjectScript class files begin
  * with a line of the form "Class <UppercasePackage>..."; VB6 class modules
@@ -1611,6 +1712,13 @@ static const char *lang_probe_extension(const char *filename) {
     return (dot && lang_name_in(dot, PROBED)) ? dot : NULL;
 }
 
+/* .d is shared by D and make-style dep-info (cargo, gcc -MD). Probe only while
+ * .d maps to D, so a user override to another language wins. */
+static bool lang_probe_d(CBMLanguage lang, const char *filename) {
+    const char *dot = strrchr(filename, '.');
+    return lang == CBM_LANG_DLANG && dot && strcmp(dot, ".d") == 0;
+}
+
 /* "<Export generator=" in the first line-probe bytes: an ObjectScript Studio
  * export, which is XML by name. */
 static bool lang_objectscript_export(const unsigned char *head, size_t n) {
@@ -1633,6 +1741,9 @@ size_t cbm_language_probe_bytes_with(const cbm_userconfig_t *config, const char
     if (ext) {
         return strcmp(ext, ".cfc") == 0 ? LANG_PROBE_CFC : LANG_PROBE_HEAD;
     }
+    if (lang_probe_d(lang, filename)) {
+        return LANG_PROBE_HEAD;
+    }
     return lang == CBM_LANG_XML ? LANG_PROBE_LINE : 0;
 }
 
@@ -1676,6 +1787,11 @@ CBMLanguage cbm_language_classify_with(const cbm_userconfig_t *config, const cha
             /* .res: ReScript, or a binary Godot / Windows resource (#2176) */
             lang = readable ? lang_res_bytes(head, head_len) : CBM_LANG_RESCRIPT;
         }
+    } else if (lang_probe_d(lang, filename)) {
+        /* D, or a make-style dep-info file */
+        char buf[LANG_PROBE_HEAD + SKIP_ONE];
+        lang_head_text(buf, LANG_PROBE_HEAD, head, head_len);
+        lang = readable ? lang_d_text(buf, head_len) : CBM_LANG_DLANG;
     }
     if (lang == CBM_LANG_XML && readable && lang_objectscript_export(head, head_len)) {
         return CBM_LANG_OBJECTSCRIPT_EXPORT;

src/pipeline/doc_links_md.c

@@ -10,18 +10,23 @@
  * document says (the field tests' rule for bare file names).
  *
  * Into the file: a line range picks the innermost code definition that holds
- * the WHOLE range; a member (`path::Name`, `path::Owner::name`) picks the one
- * definition of that name in the file. Otherwise the edge goes to the file,
- * and the range stays on it ("target_lines"). Every edge is exact: the
- * document wrote the path.
+ * the WHOLE range; a member (`path::Name`, `path::Owner::name`; a call form
+ * `path::name()` reads as `path::name`) picks the one definition of that name
+ * in the file. Otherwise the edge goes to the file, and the range stays on it
+ * ("target_lines"); so does a member the code file writes without a
+ * definition the graph binds (a macro, a prototype, an enum member, a class
+ * attribute). Every edge is exact: the document wrote the path.
  *
  * Unresolved:
  *   missing    the path is not in the graph, but its directory is, or a
  *              directory above it that the reference wrote is: the document
  *              names something that is not there (doc rot; a moved or
- *              renamed directory)
+ *              renamed directory); or a member whose name the code file
+ *              does not write at all
  *   ambiguous  a file name alone that is not next to the document but exists
- *              elsewhere in the repository: the document does not say which
+ *              elsewhere in the repository: the document does not say which;
+ *              or a member name the file defines more than once, with no
+ *              owner to pick one
  * Not a reference to this repository at all (LOCAL: no edge, no row): a link
  * back into the same document; a path that is neither indexed nor of a type
  * the indexer reads (an image, an archive, an extensionless page); a path
@@ -35,11 +40,14 @@
 #include "discover/discover.h" /* cbm_language_for_filename */
 #include "doclink.h"
 #include "foundation/arena.h"
+#include "foundation/compat_fs.h" /* cbm_fopen */
 #include "foundation/constants.h"
 #include "foundation/hash_table.h"
+#include "foundation/limits.h" /* cbm_max_file_bytes */
 #include "foundation/mem_core.h"
 #include "graph_buffer/graph_buffer.h"
 
+#include <ctype.h>
 #include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -90,6 +98,7 @@ typedef struct {
 typedef struct {
     const char *project;
     size_t project_len;
+    const char *repo;       /* the repository root (NULL: unknown; no file is read) */
     const char **run_paths; /* run_file -> rel_path (NULL: not a Markdown file of this run) */
     int run_count;
     /* base name -> number of File nodes with it (a file name alone that is
@@ -287,6 +296,7 @@ static void *mdr_build(const cbm_doclink_build_in_t *in) {
     cbm_arena_init(&x->arena);
     x->project = in->ctx ? in->ctx->project_name : NULL;
     x->project_len = x->project ? strlen(x->project) : 0;
+    x->repo = in->ctx ? in->ctx->repo_path : NULL;
     x->run_count = in->run_file_count;
     x->run_paths = in->run_file_count > 0
                        ? (const char **)cbm_calloc(CBM_MEM_CLASS_OTHER,
@@ -438,14 +448,31 @@ static bool mdr_code_label(const char *label) {
     return false;
 }
 
+/* The length of the member text as a name: a call form (`name()`,
+ * `Owner.name(a, b)`) ends before its first '('. */
+static size_t mdr_member_len(const char *member) {
+    size_t n = strlen(member);
+    if (n > 0 && member[n - SKIP_ONE] == ')') {
+        const char *open = memchr(member, '(', n);
+        if (open) {
+            return (size_t)(open - member);
+        }
+    }
+    return n;
+}
+
 /* The member text's last segment (after "::" or "."), and the one before it. */
 static void mdr_member_names(const char *member, const char **name, size_t *name_len,
                              const char **owner, size_t *owner_len) {
     const char *segs[PAIR_LEN] = {NULL, NULL};
     size_t lens[PAIR_LEN] = {0, 0};
     const char *s = member;
-    while (*s) {
+    const char *end = member + mdr_member_len(member);
+    while (s < end) {
         size_t n = strcspn(s, ":.");
+        if (n > (size_t)(end - s)) {
+            n = (size_t)(end - s);
+        }
         if (n > 0) {
             segs[0] = segs[SKIP_ONE];
             lens[0] = lens[SKIP_ONE];
@@ -453,7 +480,7 @@ static void mdr_member_names(const char *member, const char **name, size_t *name
             lens[SKIP_ONE] = n;
         }
         s += n;
-        while (*s == ':' || *s == '.') {
+        while (s < end && (*s == ':' || *s == '.')) {
             s++;
         }
     }
@@ -499,12 +526,24 @@ static void mdr_file_defs(const mdr_index_t *x, const char *path, int *lo, int *
     *hi = e;
 }
 
+/* How a member lookup ended: found, or why not. */
+typedef enum {
+    MDR_MEMBER_FOUND = 0,
+    MDR_MEMBER_MISSING,   /* no definition of that name in the file */
+    MDR_MEMBER_AMBIGUOUS, /* several, and the owner does not pick one */
+} mdr_member_t;
+
 /* The code definition of file `path` that the range or the member names, or
  * NULL. Read from the index's definitions by file, never from DEFINES edges:
  * an incremental run's graph holds the edges of the files it re-extracts
- * only, while every node is there. */
+ * only, while every node is there. *member_result (when not NULL) says how a
+ * member lookup ended. */
 static const cbm_gbuf_node_t *mdr_segment(const mdr_index_t *x, const char *path, uint32_t first,
-                                          uint32_t last, const char *member) {
+                                          uint32_t last, const char *member,
+                                          mdr_member_t *member_result) {
+    if (member_result) {
+        *member_result = MDR_MEMBER_FOUND;
+    }
     int lo = 0;
     int hi = 0;
     mdr_file_defs(x, path, &lo, &hi);
@@ -537,6 +576,9 @@ static const cbm_gbuf_node_t *mdr_segment(const mdr_index_t *x, const char *path
     size_t owner_len;
     mdr_member_names(member, &name, &name_len, &owner, &owner_len);
     if (!name) {
+        if (member_result) {
+            *member_result = MDR_MEMBER_MISSING;
+        }
         return NULL;
     }
     const cbm_gbuf_node_t *hit = NULL;
@@ -558,7 +600,70 @@ static const cbm_gbuf_node_t *mdr_segment(const mdr_index_t *x, const char *path
     if (owned_hits == SKIP_ONE) {
         return owned;
     }
-    return owned_hits == 0 && hits == SKIP_ONE ? hit : NULL;
+    if (owned_hits == 0 && hits == SKIP_ONE) {
+        return hit;
+    }
+    if (member_result) {
+        *member_result = hits == 0 ? MDR_MEMBER_MISSING : MDR_MEMBER_AMBIGUOUS;
+    }
+    return NULL;
+}
+
+/* A byte of an identifier: a letter, a digit, '_', and every byte of a
+ * multi-byte character (pipeline_incremental.c doc_row_ident_byte). */
+static bool mdr_ident_byte(unsigned char c) {
+    return isalnum(c) || c == '_' || c >= CBM_SZ_128;
+}
+
+/* Does text[0, n) hold name[0, len) as a whole identifier (case-sensitive)? */
+static bool mdr_writes_ident(const char *text, size_t n, const char *name, size_t len) {
+    for (size_t i = 0; len > 0 && i + len <= n;) {
+        const char *hit = memchr(text + i, name[0], n - len - i + SKIP_ONE);
+        if (!hit) {
+            return false;
+        }
+        size_t at = (size_t)(hit - text);
+        if (memcmp(hit, name, len) == 0 && (at == 0 || !mdr_ident_byte((unsigned char)hit[-1])) &&
+            (at + len == n || !mdr_ident_byte((unsigned char)hit[len]))) {
+            return true;
+        }
+        i = at + SKIP_ONE;
+    }
+    return false;
+}
+
+/* Does the code file `path` write the member's name at all? A name the graph
+ * holds no definition for (a macro, a prototype, an enum member, a class
+ * attribute) is still there when the file writes it. Read from the file,
+ * never cached (resolvers run on worker threads). True -- no claim -- when
+ * the file cannot be read whole: a member is missing only when the file was
+ * read and does not write its name. */
+static bool mdr_file_writes(const mdr_index_t *x, const char *path, const char *member) {
+    const char *name;
+    const char *owner;
+    size_t name_len;
+    size_t owner_len;
+    mdr_member_names(member, &name, &name_len, &owner, &owner_len);
+    char abs_path[CBM_PATH_MAX];
+    int w = name && x->repo ? snprintf(abs_path, sizeof(abs_path), "%s/%s", x->repo, path) : 0;
+    if (w <= 0 || (size_t)w >= sizeof(abs_path)) {
+        return true;
+    }
+    FILE *f = cbm_fopen(abs_path, "rb");
+    if (!f) {
+        return true;
+    }
+    long size = fseek(f, 0, SEEK_END) == 0 ? ftell(f) : CBM_NOT_FOUND;
+    char *text = size >= 0 && size <= cbm_max_file_bytes() && fseek(f, 0, SEEK_SET) == 0
+                     ? (char *)cbm_alloc(CBM_MEM_CLASS_OTHER, (size_t)size + SKIP_ONE)
+                     : NULL;
+    bool writes = true;
+    if (text && fread(text, SKIP_ONE, (size_t)size, f) == (size_t)size) {
+        writes = mdr_writes_ident(text, (size_t)size, name, name_len);
+    }
+    (void)fclose(f);
+    cbm_free(CBM_MEM_CLASS_OTHER, text);
+    return writes;
 }
 
 /* Is `path` of a file type the indexer reads? */
@@ -1183,9 +1288,24 @@ static void mdr_resolve_path(const mdr_index_t *x, const cbm_gbuf_t *graph, cons
                 out->kind = CBM_DOCLINK_LOCAL; /* a link into the same document */
                 return;
             }
-            const cbm_gbuf_node_t *seg = (r->first > 0 || r->member)
-                                             ? mdr_segment(x, t, r->first, r->last, r->member)
-                                             : NULL;
+            mdr_member_t member_result = MDR_MEMBER_FOUND;
+            const cbm_gbuf_node_t *seg =
+                (r->first > 0 || r->member)
+                    ? mdr_segment(x, t, r->first, r->last, r->member, &member_result)
+                    : NULL;
+            /* A named member that a code file does not define is no coarser
+             * link to the file: it is a stale or unclear reference. One the
+             * file writes without a definition the graph binds (a macro, a
+             * prototype, an enum member) is there, and keeps the file edge.
+             * A document's own headings are not definitions, so members of
+             * other files keep the file edge too. */
+            if (!seg && r->member && member_result != MDR_MEMBER_FOUND && mdr_code_file(t) &&
+                (member_result == MDR_MEMBER_AMBIGUOUS || !mdr_file_writes(x, t, r->member))) {
+                out->kind = CBM_DOCLINK_UNRESOLVED;
+                out->reason = member_result == MDR_MEMBER_AMBIGUOUS ? CBM_DOCLINK_REASON_AMBIGUOUS
+                                                                    : CBM_DOCLINK_REASON_MISSING;
+                return;
+            }
             mdr_edge(out, seg ? seg : file, r->first, r->last);
             return;
         }
@@ -1425,7 +1545,8 @@ static void mdr_resolve(const void *index, void *state, int run_file, const CBMD
 
 const cbm_gbuf_node_t *cbm_doclink_md_segment(const void *md_index, const char *path,
                                               uint32_t first, uint32_t last, const char *member) {
-    return md_index ? mdr_segment((const mdr_index_t *)md_index, path, first, last, member) : NULL;
+    return md_index ? mdr_segment((const mdr_index_t *)md_index, path, first, last, member, NULL)
+                    : NULL;
 }
 
 /* ── A region of a file's text (the field test's H8 bind_range) ─────── */

src/pipeline/pass_definitions.c

@@ -1000,6 +1000,7 @@ int cbm_pipeline_pass_definitions(cbm_pipeline_ctx_t *ctx, const cbm_file_info_t
                 }
                 cbm_free(CBM_MEM_CLASS_EXTRACT, local_cache);
             }
+            cbm_work_arena_release();
             return CBM_NOT_FOUND;
         }
 
@@ -1133,6 +1134,16 @@ int cbm_pipeline_pass_definitions(cbm_pipeline_ctx_t *ctx, const cbm_file_info_t
         total_calls += result->calls.count;
 
         if (local_cache) {
+            /* The cache lives through every later pass: hold the result as the
+             * parallel extract does (pass_parallel.c), without its tree (the
+             * cross-file LSP re-parses) and compacted. A PHP result keeps its
+             * tree: the vendor-import check reads each call and usage site's
+             * syntax from it (php_import_kind_at_site, pass_pkgmap.c), and this
+             * route keeps no source to fall back on. */
+            if (lang != CBM_LANG_PHP) {
+                cbm_free_tree(result);
+            }
+            cbm_result_compact(result);
             local_cache[i] = result;
         } else {
             /* Cache unavailable: imports for this file can still only
@@ -1146,6 +1157,7 @@ int cbm_pipeline_pass_definitions(cbm_pipeline_ctx_t *ctx, const cbm_file_info_t
             cbm_free_result(result);
         }
     }
+    cbm_work_arena_release(); /* the working arena compaction kept between files */
 
     /* Every file's nodes exist now: link methods and diagram elements to
      * their parents (without the cache each file linked its own above). */
@@ -1209,5 +1221,6 @@ configured_failure:
         }
         cbm_free(CBM_MEM_CLASS_EXTRACT, local_cache);
     }
+    cbm_work_arena_release(); /* compaction keeps it between files: release on every exit */
     return CBM_PIPELINE_ABORT_PRESERVE_DB;
 }

src/pipeline/pipeline_incremental.c

@@ -1582,7 +1582,12 @@ static int surface_added_names(const char *stored_json, const char *fresh_json,
  *     change to it is GLOBAL (it sets the global usings of a whole project);
  *   - a REMOVED name can make another file's unresolved reference resolve (an
  *     overload group shrinks) or change its reason: the files whose rows
- *     mention it re-resolve with the closure. */
+ *     mention it re-resolve with the closure;
+ *   - an unresolved `path::member` row (missing, ambiguous) depends on the
+ *     member's file as a whole, its text included (a member the file writes
+ *     without a definition is no missing member, doc_links_md.c) and the size
+ *     of a same-named group: the files whose such rows name a changed or
+ *     deleted file, by its base name, re-resolve with the closure. */
 
 /* Add s[0..n) to a heap-keyed name set. false when it cannot be recorded (too
  * long for a key, or out of memory): callers fail closed. */
@@ -1730,6 +1735,58 @@ static void doc_row_name_dependents(const cbm_doc_link_row_t *rows, int n,
     }
 }
 
+/* Is an unresolved row's reason one a member lookup gives (doc_links_md.c)? */
+static bool doc_row_member_reason(const char *reason) {
+    return reason && (strcmp(reason, "missing") == 0 || strcmp(reason, "ambiguous") == 0);
+}
+
+/* Files whose unresolved `path::member` rows name a member of one of `paths`
+ * (the changed and deleted files), and no edge leads from such a row to the
+ * file: a body edit can write or drop a missing member's name without
+ * changing a definition, and a same-named group can shrink to one without
+ * removing its name. Matched by the base name before "::", since a
+ * reference may be written relative to the document or to the root. When
+ * memory runs out every such row's file is added: re-resolving a file that did
+ * not need it costs time, keeping a stale row costs correctness. Keys are
+ * borrowed from `rows`. */
+static void doc_row_member_dependents(const cbm_doc_link_row_t *rows, int n, char *const *paths,
+                                      int npaths, CBMHashTable *out_paths) {
+    if (npaths <= 0) {
+        return;
+    }
+    CBMHashTable *bases = cbm_ht_create(CBM_SZ_64); /* NULL: fail closed, every member row */
+    for (int i = 0; bases && i < npaths; i++) {
+        const char *slash = strrchr(paths[i], '/');
+        const char *base = slash ? slash + SKIP_ONE : paths[i];
+        cbm_ht_set(bases, base, (void *)base); /* borrowed from `paths` */
+    }
+    for (int i = 0; i < n; i++) {
+        const char *raw = rows[i].raw;
+        const char *rel = rows[i].rel_path;
+        const char *sep = raw ? strstr(raw, "::") : NULL;
+        if (!sep || !rel || !rel[0] || !doc_row_member_reason(rows[i].reason) ||
+            cbm_ht_get(out_paths, rel)) {
+            continue;
+        }
+        const char *base = sep;
+        while (base > raw && base[-1] != '/') {
+            base--;
+        }
+        char key[CBM_SZ_512];
+        size_t bl = (size_t)(sep - base);
+        bool hit = !bases || bl >= sizeof(key);
+        if (!hit && bl > 0) {
+            memcpy(key, base, bl);
+            key[bl] = '\0';
+            hit = cbm_ht_get(bases, key) != NULL;
+        }
+        if (hit) {
+            cbm_ht_set(out_paths, rel, (void *)rel);
+        }
+    }
+    cbm_ht_free(bases);
+}
+
 /* Doc-link carry-forward state of the legacy partial route: the previous
  * rows, the stored scopes of the files it does not re-extract, and the set
  * of re-extracted or deleted paths whose old rows it replaces. */
@@ -3176,8 +3233,11 @@ static int closure_try_plan(cbm_pipeline_t *p, cbm_store_t *store, const char *p
         }
     }
     /* Files whose unresolved doc-link references name something a changed or
-     * deleted file no longer declares re-resolve too. */
+     * deleted file no longer declares re-resolve too, and so do the files
+     * whose unresolved `path::member` rows name a changed or deleted file. */
     doc_row_name_dependents(plan->doc_rows, plan->doc_row_count, removed_names, row_deps);
+    doc_row_member_dependents(plan->doc_rows, plan->doc_row_count, changed_paths,
+                              n_changed + n_deleted, row_deps);
     {
         row_dep_walk_t walk = {.closure = closure_set, .files = files_by_path, .added = 0};
         cbm_ht_foreach(row_deps, row_dep_visitor, &walk);

tests/test_discover.c

@@ -6,6 +6,7 @@
 #include "test_framework.h"
 #include "test_helpers.h"
 #include "discover/discover.h"
+#include "discover/userconfig.h"
 #include "discover/discover_policy_internal.h"
 #include "foundation/platform.h"
 
@@ -2444,6 +2445,49 @@ TEST(shebang_oversized_first_line_unindexed) {
     PASS();
 }
 
+/* ── .d: D source vs make/cargo dep-info ───────────────────────── */
+
+/* rustc/cargo dep-info (and gcc -MD, CMake .o.d) files share .d with D source
+ * but are Makefile rules the D grammar parses slowly into nothing useful, so
+ * they must not be indexed as D. */
+static const char CARGO_DEP_INFO[] = "/home/u/proj/t/debug/deps/foo-abc123.d: src/lib.rs src/a.rs\n"
+                                     "\n"
+                                     "src/lib.rs:\n"
+                                     "src/a.rs:\n";
+
+TEST(discover_d_dep_info_not_indexed_as_dlang) {
+    CBMLanguage dep;
+    CBMLanguage src;
+    ASSERT(shebang_probe("cbm_disc_dotd", "t/debug/deps/foo-abc123.d", CARGO_DEP_INFO, &dep));
+    ASSERT(shebang_probe("cbm_disc_dotd", "app.d", "module app;\nimport std.stdio : writeln;\n",
+                         &src));
+    ASSERT_EQ(dep, CBM_LANG_COUNT);
+    ASSERT_EQ(src, CBM_LANG_DLANG);
+    PASS();
+}
+
+/* The probe runs only while .d maps to D, so a user override to another
+ * language wins. */
+TEST(discover_d_user_override_skips_dep_info_probe) {
+    char *cfg_dir = th_mktempdir("cbm_disc_dotd_uc");
+    ASSERT(cfg_dir != NULL);
+    ASSERT_EQ(th_write_file(TH_PATH(cfg_dir, ".codebase-memory.json"),
+                            "{\"extra_extensions\":{\".d\":\"makefile\"}}"),
+              0);
+    cbm_userconfig_t *cfg = cbm_userconfig_load(cfg_dir);
+    th_cleanup(cfg_dir);
+    ASSERT_NOT_NULL(cfg);
+
+    cbm_set_user_lang_config(cfg);
+    CBMLanguage lang;
+    bool ok = shebang_probe("cbm_disc_dotd_uc", "foo-abc123.d", CARGO_DEP_INFO, &lang);
+    cbm_set_user_lang_config(NULL);
+    cbm_userconfig_free(cfg);
+    ASSERT(ok);
+    ASSERT_EQ(lang, CBM_LANG_MAKEFILE);
+    PASS();
+}
+
 /* ── Suite ─────────────────────────────────────────────────────── */
 
 SUITE(discover) {
@@ -2539,6 +2583,10 @@ SUITE(discover) {
     RUN_TEST(shebang_embedded_nul_unindexed);
     RUN_TEST(shebang_oversized_first_line_unindexed);
 
+    /* .d: D source vs make/cargo dep-info */
+    RUN_TEST(discover_d_dep_info_not_indexed_as_dlang);
+    RUN_TEST(discover_d_user_override_skips_dep_info_probe);
+
     /* Integration tests (cross-platform) */
     RUN_TEST(discover_simple);
     RUN_TEST(discover_wide_sibling_fanout_exceeds_initial_walk_stack);

tests/test_doc_links_md.c

@@ -388,6 +388,166 @@ TEST(doc_links_md_resolve) {
     PASS();
 }
 
+/* #2587: a `path::member` reference whose member the code file does not
+ * define was an exact edge to the file, so a renamed or deleted definition
+ * looked like a valid, coarser link. It is a missing reference; a member the
+ * file defines more than once, with no owner to pick one, is ambiguous. */
+TEST(doc_links_md_member_not_defined_issue2587) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "/tmp/cbm_dlmd_XXXXXX");
+    ASSERT_NOT_NULL(cbm_mkdtemp(tmp));
+    th_write_file(TH_PATH(tmp, "docs/members.md"),
+                  "# Members\n"
+                  "\n"
+                  "See `src/util/rounding.py::clamp` and "
+                  "`src/util/rounding.py::no_such_function`.\n"
+                  "Areas: `src/shapes.py::Square::area`, `src/shapes.py::area`.\n"
+                  "Notes: `docs/other.md::Intro`.\n");
+    th_write_file(TH_PATH(tmp, "docs/other.md"), "# Intro\n\nText.\n");
+    th_write_file(TH_PATH(tmp, "src/util/rounding.py"), "def clamp(x, lo, hi):\n"
+                                                        "    return max(lo, min(hi, x))\n");
+    th_write_file(TH_PATH(tmp, "src/shapes.py"), "class Circle:\n"
+                                                 "    def area(self):\n"
+                                                 "        return 3\n"
+                                                 "\n"
+                                                 "\n"
+                                                 "class Square:\n"
+                                                 "    def area(self):\n"
+                                                 "        return 4\n");
+    char db[512];
+    snprintf(db, sizeof(db), "%s/md.db", tmp);
+    ASSERT_EQ(dm_index(tmp, db, NULL), 0);
+    char props[512];
+    char reason[64];
+
+    /* a member the file defines, alone or through its owner */
+    ASSERT_EQ(md_edge(db, "Members", "rounding.clamp", props, sizeof(props)), 1);
+    ASSERT_EQ(md_edge(db, "Members", "Square.area", props, sizeof(props)), 1);
+    ASSERT_EQ(md_edge(db, "Members", "Circle.area", props, sizeof(props)), 0);
+
+    /* a member the file does not define: no edge to the file, a missing row */
+    ASSERT_EQ(dm_count(db, "SELECT COUNT(*) FROM edges e JOIN nodes s ON s.id = e.source_id "
+                           "JOIN nodes t ON t.id = e.target_id WHERE e.type = 'MENTIONS' "
+                           "AND s.qualified_name LIKE '%.Members' AND t.label = 'File' "
+                           "AND t.file_path = 'src/util/rounding.py'"),
+              0);
+    dm_row(db, "docs/members.md", "src/util/rounding.py::no_such_function", reason, sizeof(reason),
+           NULL, 0);
+    ASSERT_STR_EQ(reason, "missing");
+
+    /* defined twice and no owner written: ambiguous, not the file */
+    dm_row(db, "docs/members.md", "src/shapes.py::area", reason, sizeof(reason), NULL, 0);
+    ASSERT_STR_EQ(reason, "ambiguous");
+    ASSERT_EQ(dm_count(db, "SELECT COUNT(*) FROM edges e JOIN nodes s ON s.id = e.source_id "
+                           "JOIN nodes t ON t.id = e.target_id WHERE e.type = 'MENTIONS' "
+                           "AND s.qualified_name LIKE '%.Members' AND t.label = 'File' "
+                           "AND t.file_path = 'src/shapes.py'"),
+              0);
+
+    /* a document's headings are no definitions: the file edge stays */
+    ASSERT_EQ(dm_count(db, "SELECT COUNT(*) FROM edges e JOIN nodes s ON s.id = e.source_id "
+                           "JOIN nodes t ON t.id = e.target_id WHERE e.type = 'MENTIONS' "
+                           "AND s.qualified_name LIKE '%.Members' AND t.label = 'File' "
+                           "AND t.file_path = 'docs/other.md'"),
+              1);
+
+    th_cleanup(tmp);
+    PASS();
+}
+
+/* MENTIONS edges from the section whose qualified name ends with `section` to
+ * the File node of `file`. */
+static int md_file_edges(const char *db, const char *section, const char *file) {
+    char sql[512];
+    snprintf(sql, sizeof(sql),
+             "SELECT COUNT(*) FROM edges e JOIN nodes s ON s.id = e.source_id "
+             "JOIN nodes t ON t.id = e.target_id WHERE e.type = 'MENTIONS' "
+             "AND s.qualified_name LIKE '%%.%s' AND t.label = 'File' AND t.file_path = '%s'",
+             section, file);
+    return dm_count(db, sql);
+}
+
+/* #2587, the members a code file has although the graph defines no node for
+ * them: a call form, a macro, a prototype, a namespace, an enum member, a
+ * class attribute. Each is there -- the file writes it -- so it keeps its
+ * edge; only a name the file does not write at all is missing. */
+static const struct {
+    const char *section;
+    const char *raw;
+    const char *file;
+    bool missing;
+} MD_WRITTEN[] = {
+    {"CallForm", "src/util.py::clamp()", NULL, false},
+    {"Macro", "include/limits.h::MAX_ITEMS", "include/limits.h", false},
+    {"FunctionMacro", "include/limits.h::SQUARE", "include/limits.h", false},
+    {"Prototype", "include/limits.h::limits_init", "include/limits.h", false},
+    {"Namespace", "src/ns.ts::Shapes", "src/ns.ts", false},
+    {"EnumMember", "src/color.py::Color::RED", "src/color.py", false},
+    {"Attribute", "src/model.py::Model::name", "src/model.py", false},
+    /* not written: `blue` and LIGHT_BLUE are no BLUE */
+    {"NotWritten", "src/color.py::Color::BLUE", "src/color.py", true},
+    {"NotDefined", "src/util.py::no_such_function", "src/util.py", true},
+};
+
+TEST(doc_links_md_member_written_issue2587) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "/tmp/cbm_dlmd_XXXXXX");
+    ASSERT_NOT_NULL(cbm_mkdtemp(tmp));
+    char doc[2048] = "";
+    size_t used = 0;
+    for (size_t i = 0; i < sizeof(MD_WRITTEN) / sizeof(MD_WRITTEN[0]); i++) {
+        used += (size_t)snprintf(doc + used, sizeof(doc) - used, "# %s\n\nSee `%s`.\n\n",
+                                 MD_WRITTEN[i].section, MD_WRITTEN[i].raw);
+        ASSERT(used < sizeof(doc));
+    }
+    th_write_file(TH_PATH(tmp, "docs/written.md"), doc);
+    th_write_file(TH_PATH(tmp, "src/util.py"), "def clamp(x, lo, hi):\n"
+                                               "    return max(lo, min(hi, x))\n");
+    th_write_file(TH_PATH(tmp, "include/limits.h"), "#ifndef LIMITS_H\n"
+                                                    "#define LIMITS_H\n"
+                                                    "#define MAX_ITEMS 64\n"
+                                                    "#define SQUARE(x) ((x) * (x))\n"
+                                                    "int limits_init(void);\n"
+                                                    "#endif\n");
+    th_write_file(TH_PATH(tmp, "src/ns.ts"), "export namespace Shapes {\n"
+                                             "    export const unit = 1;\n"
+                                             "}\n");
+    th_write_file(TH_PATH(tmp, "src/color.py"), "from enum import Enum\n"
+                                                "\n"
+                                                "\n"
+                                                "class Color(Enum):\n"
+                                                "    RED = 1\n"
+                                                "    LIGHT_BLUE = 2  # blue\n");
+    th_write_file(TH_PATH(tmp, "src/model.py"), "class Model:\n"
+                                                "    name: str\n");
+    char db[512];
+    snprintf(db, sizeof(db), "%s/md.db", tmp);
+    ASSERT_EQ(dm_index(tmp, db, NULL), 0);
+    char props[512];
+    char reason[64];
+
+    /* every reference, under its own heading: an edge, or a missing row */
+    int wrong = 0;
+    for (size_t i = 0; i < sizeof(MD_WRITTEN) / sizeof(MD_WRITTEN[0]); i++) {
+        dm_row(db, "docs/written.md", MD_WRITTEN[i].raw, reason, sizeof(reason), NULL, 0);
+        int edges = MD_WRITTEN[i].file
+                        ? md_file_edges(db, MD_WRITTEN[i].section, MD_WRITTEN[i].file)
+                        : dm_mentions_from(db, MD_WRITTEN[i].section);
+        bool ok = MD_WRITTEN[i].missing ? strcmp(reason, "missing") == 0 && edges == 0
+                                        : reason[0] == '\0' && edges >= 1;
+        if (!ok) {
+            printf("  %s: row '%s', %d edge(s)\n", MD_WRITTEN[i].raw, reason, edges);
+            wrong++;
+        }
+    }
+    ASSERT_EQ(wrong, 0);
+    /* the call form binds the function itself */
+    ASSERT_EQ(md_edge(db, "CallForm", "util.clamp", props, sizeof(props)), 1);
+
+    th_cleanup(tmp);
+    PASS();
+}
+
 /* ── incremental == full ─────────────────────────────────────────── */
 
 TEST(doc_links_md_incremental) {
@@ -444,6 +604,76 @@ TEST(doc_links_md_incremental) {
     PASS();
 }
 
+/* #2587 incremental == full: whether a `path::member` row is missing depends
+ * on the member file's text, and the row leads to that file by no edge. A
+ * body edit that writes the name (no definition added) turns the row into a
+ * file edge, one that drops it turns the edge back into the row; a same-named
+ * group that shrinks to one turns an ambiguous row into an edge. The group is
+ * Ruby: a Python file whose surface changes takes the full route (the Python
+ * namespace admission), and a language with a doc-link scope (Rust, Go, ...)
+ * reports the lost method as a removed name, so neither reaches the rule for
+ * ambiguous member rows. */
+TEST(doc_links_md_member_missing_incremental) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "/tmp/cbm_dlmd_inc_XXXXXX");
+    ASSERT_NOT_NULL(cbm_mkdtemp(tmp));
+    char repo[512];
+    snprintf(repo, sizeof(repo), "%s/repo", tmp);
+    th_write_file(TH_PATH(repo, "docs/helpers.md"), "# Helpers\n\nSee `src/tools.py::helper2`.\n");
+    th_write_file(TH_PATH(repo, "docs/areas.md"), "# Areas\n\nSee `src/shapes.rb::area`.\n");
+    th_write_file(TH_PATH(repo, "src/tools.py"), "def helper():\n"
+                                                 "    return 1\n");
+    th_write_file(TH_PATH(repo, "src/shapes.rb"), "class Circle\n"
+                                                  "  def area\n"
+                                                  "    3\n"
+                                                  "  end\n"
+                                                  "end\n"
+                                                  "\n"
+                                                  "class Square\n"
+                                                  "  def area\n"
+                                                  "    4\n"
+                                                  "  end\n"
+                                                  "end\n");
+    char db[512];
+    char full_db[512];
+    snprintf(db, sizeof(db), "%s/inc.db", tmp);
+    snprintf(full_db, sizeof(full_db), "%s/full.db", tmp);
+    ASSERT_EQ(dm_index(repo, db, NULL), 0);
+    char reason[64];
+    dm_row(db, "docs/helpers.md", "src/tools.py::helper2", reason, sizeof(reason), NULL, 0);
+    ASSERT_STR_EQ(reason, "missing");
+    dm_row(db, "docs/areas.md", "src/shapes.rb::area", reason, sizeof(reason), NULL, 0);
+    ASSERT_STR_EQ(reason, "ambiguous");
+
+    /* the body writes the name, no definition added: the row becomes an edge */
+    th_write_file(TH_PATH(repo, "src/tools.py"), "def helper():\n"
+                                                 "    return 1  # helper2 follows\n");
+    ASSERT_EQ(dm_step(repo, db, full_db, "member written", CBM_INCREMENTAL_ROUTE_CLOSURE_REPAIR),
+              0);
+    ASSERT_EQ(md_file_edges(db, "Helpers", "src/tools.py"), 1);
+    /* and drops it again: missing again */
+    th_write_file(TH_PATH(repo, "src/tools.py"), "def helper():\n"
+                                                 "    return 1\n");
+    ASSERT_EQ(dm_step(repo, db, full_db, "member dropped", CBM_INCREMENTAL_ROUTE_CLOSURE_REPAIR),
+              0);
+    dm_row(db, "docs/helpers.md", "src/tools.py::helper2", reason, sizeof(reason), NULL, 0);
+    ASSERT_STR_EQ(reason, "missing");
+    /* one of two `area` methods goes, the name stays: the one left binds */
+    th_write_file(TH_PATH(repo, "src/shapes.rb"), "class Circle\n"
+                                                  "end\n"
+                                                  "\n"
+                                                  "class Square\n"
+                                                  "  def area\n"
+                                                  "    4\n"
+                                                  "  end\n"
+                                                  "end\n");
+    ASSERT_EQ(dm_step(repo, db, full_db, "group shrinks", CBM_INCREMENTAL_ROUTE_CLOSURE_REPAIR), 0);
+    char props[512];
+    ASSERT_EQ(md_edge(db, "Areas", "Square.area", props, sizeof(props)), 1);
+    th_cleanup(tmp);
+    PASS();
+}
+
 /* #2588: a reference into a directory that was moved or renamed named a
  * path whose directory is gone, so it counted as no reference to this
  * repository: no edge and no row. A directory above it that the reference
@@ -914,7 +1144,10 @@ SUITE(doc_links_md) {
     RUN_TEST(doc_links_md_repeated_headings);
     RUN_TEST(doc_links_md_classify_span);
     RUN_TEST(doc_links_md_resolve);
+    RUN_TEST(doc_links_md_member_not_defined_issue2587);
+    RUN_TEST(doc_links_md_member_written_issue2587);
     RUN_TEST(doc_links_md_incremental);
+    RUN_TEST(doc_links_md_member_missing_incremental);
     RUN_TEST(doc_links_md_moved_directory_issue2588);
     RUN_TEST(doc_links_md_adr);
     RUN_TEST(doc_links_md_adr_label_choice);

tests/test_edge_imports.c

@@ -2150,6 +2150,45 @@ TEST(ei_php_vendor_types_never_bind_by_name_issue1186) {
     PASS();
 }
 
+/* A vendor `use function` import shares its local name with a project class the
+ * file also imports. `new Thing()`, a `Thing $t` parameter and `extends Thing`
+ * name the class, so the function import must not veto their edges. The check
+ * reads each site's syntax from the parse tree, which the sequential
+ * definitions pass (a repo of 50 files or fewer) keeps for PHP results. */
+TEST(ei_php_vendor_function_import_keeps_same_named_class_sites) {
+    static const EILangFile f[] = {
+        {"composer.json", "{\"autoload\":{\"psr-4\":{\"App\\\\\":\"app/\"}}}\n"},
+        {"app/Models/Thing.php", "<?php\nnamespace App\\Models;\n\nclass Thing {\n"
+                                 "    public function __construct() {}\n"
+                                 "    public function go() { return 1; }\n}\n"},
+        {"app/Http/UsesThing.php", "<?php\nnamespace App\\Http;\n\n"
+                                   "use App\\Models\\Thing;\nuse function Vendor\\Lib\\Thing;\n\n"
+                                   "class UsesThing extends Thing {\n"
+                                   "    public function run(Thing $t) {\n"
+                                   "        $made = new Thing();\n"
+                                   "        return $made->go();\n"
+                                   "    }\n}\n"}};
+    static const char *const uses = "app/Http/UsesThing.php";
+    static const char *const thing = "app/Models/Thing.php";
+    EILangProj lp;
+    cbm_store_t *store = ei_index_files(&lp, f, 3);
+    int calls = store ? ei_php_checked_file_edges(store, lp.project, uses, "CALLS", thing) : -1;
+    int usages = store ? ei_php_checked_file_edges(store, lp.project, uses, "USAGE", thing) : -1;
+    int inherits =
+        store ? ei_php_checked_file_edges(store, lp.project, uses, "INHERITS", thing) : -1;
+    ei_cleanup(&lp, store);
+    if (calls != 2 || usages != 2 || inherits != 1) {
+        fprintf(stderr,
+                "  UsesThing.php -> Thing.php: CALLS %d (want 2), USAGE %d (want 2), "
+                "INHERITS %d (want 1)\n",
+                calls, usages, inherits);
+    }
+    ASSERT_EQ(calls, 2);    /* new Thing() and $made->go() */
+    ASSERT_EQ(usages, 2);   /* extends Thing and the Thing $t parameter */
+    ASSERT_EQ(inherits, 1); /* never vetoed: control */
+    PASS();
+}
+
 /* Build the metadata through the production writer, then challenge its
  * readback. A File node or an empty edge query alone is never completeness. */
 static bool ei_php_seed_result(cbm_gbuf_t *gb, const char *project, const char *rel,
@@ -2538,6 +2577,7 @@ SUITE(edge_imports) {
     RUN_TEST(ei_php_null_namespace_map_keeps_php_candidate_issue1186);
     RUN_TEST(ei_php_require_path_keeps_generic_resolution_issue1186);
     RUN_TEST(ei_php_vendor_types_never_bind_by_name_issue1186);
+    RUN_TEST(ei_php_vendor_function_import_keeps_same_named_class_sites);
     RUN_TEST(ei_php_binding_snapshot_requires_current_complete_evidence);
     RUN_TEST(ei_php_binding_snapshot_groups_kinds_and_invalidates_failed_rebuild);
     RUN_TEST(ei_php_psr4_missing_class_leaves_type_evidence_unknown);

tests/test_extraction.c

@@ -857,6 +857,57 @@ TEST(ruby_module) {
     PASS();
 }
 
+/* A Ruby `Const.new` call is rewritten to name the CONSTANT, not the method:
+ * the constructor body lives in `initialize`, so the call should link the class
+ * like every other language's `new T()`. The rewrite fired only for bare
+ * `constant` receivers, so a scope-resolved receiver kept the whole call text
+ * as its callee ("Admin::User.new"), which resolves to nothing and cannot join
+ * the Ruby LSP's constructor rows, whose callee is the class QN.
+ *
+ * Both spellings must land on the bare constant path; a retained ".new" suffix
+ * is the bug's signature. */
+TEST(ruby_constructor_callee_is_constant_path) {
+    CBMFileResult *r = extract("module Admin\n  class User\n  end\nend\n"
+                               "class Widget\nend\n"
+                               "class Maker\n"
+                               "  def build\n"
+                               "    Admin::User.new\n"
+                               "    Widget.new\n"
+                               "  end\n"
+                               "end\n",
+                               CBM_LANG_RUBY, "t", "maker.rb");
+    ASSERT_NOT_NULL(r);
+    ASSERT_FALSE(r->has_error);
+
+    int scoped = 0, bare = 0, kept_new = 0;
+    for (int i = 0; i < r->calls.count; i++) {
+        const char *callee = r->calls.items[i].callee_name;
+        if (strcmp(callee, "Admin::User") == 0)
+            scoped = 1;
+        if (strcmp(callee, "Widget") == 0)
+            bare = 1;
+        /* Either failure mode: the method name survived on its own, or the
+         * whole `Recv.new` expression did. */
+        size_t n = strlen(callee);
+        if (strcmp(callee, "new") == 0 || (n >= 4 && strcmp(callee + n - 4, ".new") == 0))
+            kept_new = 1;
+    }
+    if (!scoped || !bare || kept_new) {
+        printf("  ruby callees (%d):\n", r->calls.count);
+        for (int i = 0; i < r->calls.count; i++)
+            printf("    '%s'\n", r->calls.items[i].callee_name);
+    }
+    /* Scope-resolved receiver reduces to the constant path; its bare leaf is
+     * what resolution joins on. */
+    ASSERT(scoped);
+    /* Bare receiver: pre-existing behaviour, guarded against regression. */
+    ASSERT(bare);
+    ASSERT_FALSE(kept_new);
+
+    cbm_free_result(r);
+    PASS();
+}
+
 /* --- C# --- */
 TEST(csharp_class) {
     CBMFileResult *r = extract("namespace App { public class Service { public void Run() {} public "
@@ -13493,6 +13544,7 @@ SUITE(extraction) {
     RUN_TEST(php_function);
     RUN_TEST(ruby_class);
     RUN_TEST(ruby_module);
+    RUN_TEST(ruby_constructor_callee_is_constant_path);
     RUN_TEST(csharp_class);
     RUN_TEST(csharp_interface);
     RUN_TEST(csharp_extension_block_members);

tests/test_language.c

@@ -885,6 +885,73 @@ TEST(lang_cfc_default_on_read_fail) {
     PASS();
 }
 
+/* ── .d: D source vs make/cargo dep-info ───────────────────────── */
+
+/* Write content to a temp .d file and return cbm_disambiguate_d() for it. A
+ * setup failure returns -1 so neither a D nor a COUNT expectation can pass
+ * vacuously. */
+static int disambiguate_d_content(const char *content) {
+    char path[256];
+    snprintf(path, sizeof(path), "%s/test_lang_dotd.d", cbm_tmpdir());
+    if (!write_probe_file(path, content)) {
+        return -1;
+    }
+    CBMLanguage lang = cbm_disambiguate_d(path);
+    remove(path);
+    return (int)lang;
+}
+
+TEST(lang_d_dep_info_unsupported) {
+    /* cargo (absolute target), gcc -MD continuation, CMake object rule, Windows
+     * drive letter (not the rule colon), escaped space (one target, not "app"),
+     * and the GNU make manual's "foo.o foo.d : ..." (two targets, spaced colon). */
+    ASSERT_EQ(disambiguate_d_content("/home/u/t/debug/deps/foo-abc123.d: src/lib.rs src/a.rs\n"
+                                     "\nsrc/lib.rs:\n"),
+              CBM_LANG_COUNT);
+    ASSERT_EQ(disambiguate_d_content("foo.o: foo.c /usr/include/stdio.h \\\n"
+                                     " /usr/include/features.h\n"),
+              CBM_LANG_COUNT);
+    ASSERT_EQ(disambiguate_d_content("CMakeFiles/x.dir/a.cpp.o: \\\n /src/a.cpp\n"),
+              CBM_LANG_COUNT);
+    ASSERT_EQ(disambiguate_d_content("C:/b/foo.o: C:/s/foo.c \\\r\n C:/s/foo.h\r\n"),
+              CBM_LANG_COUNT);
+    ASSERT_EQ(disambiguate_d_content("out/my\\ app: main.c\n"), CBM_LANG_COUNT);
+    ASSERT_EQ(disambiguate_d_content("foo.o foo.d : foo.c defs.h\n"), CBM_LANG_COUNT);
+    PASS();
+}
+
+TEST(lang_d_source_stays_dlang) {
+    ASSERT_EQ(disambiguate_d_content("module a;\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("import std.stdio : writeln;\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("@safe:\nvoid f() {}\n"), CBM_LANG_DLANG);
+    /* A leading comment is D, even DUB's "/+ dub.sdl:" single-file recipe. */
+    ASSERT_EQ(disambiguate_d_content("/+ dub.sdl:\n    name \"hello\"\n+/\nvoid main() {}\n"),
+              CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("enum E : ubyte {\n    a,\n}\n"), CBM_LANG_DLANG);
+    /* A float UDA ("@ FloatLiteral") has a '.' before its colon, alone or glued
+     * to a keyword. */
+    ASSERT_EQ(disambiguate_d_content("@1.0:\nvoid f() {}\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("nothrow@1.0:\nvoid f() {}\n"), CBM_LANG_DLANG);
+    /* A comment between tokens puts a '/' before the colon. */
+    ASSERT_EQ(disambiguate_d_content("public/**/:\nvoid f() {}\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("public/+ +/:\nvoid f() {}\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("public//note: x\nvoid f() {}\n"), CBM_LANG_DLANG);
+    /* Every D whitespace splits tokens, not only space, tab and CR: form feed,
+     * vertical tab, U+2028 and U+2029. */
+    ASSERT_EQ(disambiguate_d_content("import\fstd.stdio : writeln;\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("import\vstd.stdio : writeln;\n"), CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("import\xE2\x80\xA8"
+                                     "std.stdio : writeln;\n"),
+              CBM_LANG_DLANG);
+    ASSERT_EQ(disambiguate_d_content("import\xE2\x80\xA9"
+                                     "std.stdio : writeln;\n"),
+              CBM_LANG_DLANG);
+    /* Default on doubt: empty or unreadable. */
+    ASSERT_EQ(disambiguate_d_content(""), CBM_LANG_DLANG);
+    ASSERT_EQ(cbm_disambiguate_d("/tmp/nonexistent_file_12345.d"), CBM_LANG_DLANG);
+    PASS();
+}
+
 /* --- New languages (auto-generated) --- */
 TEST(lang_ext_solidity) {
     ASSERT_EQ(cbm_language_for_extension(".sol"), CBM_LANG_SOLIDITY);
@@ -1398,6 +1465,7 @@ TEST(lang_probe_bytes_follow_the_name) {
     ASSERT_EQ(cbm_language_probe_bytes("foo.inc"), 4096);
     ASSERT_EQ(cbm_language_probe_bytes("Form1.frm"), 4096);
     ASSERT_EQ(cbm_language_probe_bytes("App.res"), 4096);
+    ASSERT_EQ(cbm_language_probe_bytes("app.d"), 4096);
     ASSERT_EQ(cbm_language_probe_bytes("Widget.cfc"), 16384);
     ASSERT_EQ(cbm_language_probe_bytes("pom.xml"), 255);
     /* An unknown name falls back to a shebang probe of the first line. */
@@ -1439,6 +1507,9 @@ TEST(lang_classify_matches_every_content_rule) {
         {"p.frm", "#procedure foo\nLocal F = a;\n", CBM_LANG_FORM, cbm_disambiguate_frm},
         {"q.res", "RSRC\0\0\0\1binary", CBM_LANG_COUNT, cbm_disambiguate_res},
         {"r.res", "let x = 1\n", CBM_LANG_RESCRIPT, cbm_disambiguate_res},
+        {"dep.d", "target/debug/deps/app-0a1b2c.d: src/main.rs src/lib.rs\n", CBM_LANG_COUNT,
+         cbm_disambiguate_d},
+        {"app.d", "module app;\nimport std.stdio : writeln;\n", CBM_LANG_DLANG, cbm_disambiguate_d},
         {"s.xml", "<?xml version=\"1.0\"?>\n<Export generator=\"IRIS\" version=\"26\">\n",
          CBM_LANG_OBJECTSCRIPT_EXPORT, NULL},
         {"t.xml", "<?xml version=\"1.0\"?>\n<project/>\n", CBM_LANG_XML, NULL},
@@ -1486,6 +1557,7 @@ TEST(lang_classify_unreadable_content_keeps_the_name_default) {
     ASSERT_EQ(cbm_language_classify("a.cfc", NULL, 0, false, false), CBM_LANG_CFSCRIPT);
     ASSERT_EQ(cbm_language_classify("a.frm", NULL, 0, false, false), CBM_LANG_FORM);
     ASSERT_EQ(cbm_language_classify("a.res", NULL, 0, false, false), CBM_LANG_RESCRIPT);
+    ASSERT_EQ(cbm_language_classify("a.d", NULL, 0, false, false), CBM_LANG_DLANG);
     ASSERT_EQ(cbm_language_classify("a.xml", NULL, 0, false, false), CBM_LANG_XML);
     ASSERT_EQ(cbm_language_classify("run-tests", NULL, 0, false, false), CBM_LANG_COUNT);
     ASSERT_EQ(cbm_language_classify("main.go", NULL, 0, false, false), CBM_LANG_GO);
@@ -1716,6 +1788,8 @@ SUITE(language) {
     RUN_TEST(lang_frm_form_stays_form);
     RUN_TEST(lang_res_binary_resource_unsupported);
     RUN_TEST(lang_res_rescript_stays_rescript);
+    RUN_TEST(lang_d_dep_info_unsupported);
+    RUN_TEST(lang_d_source_stays_dlang);
 
     /* Go test ports */
     /* New languages */

tests/test_parallel.c

@@ -770,6 +770,127 @@ TEST(parallel_post_extract_projection_spills_before_resolve) {
     PASS();
 }
 
+/* ── Sequential cache: held like the parallel one ─────────────────── */
+
+/* The parallel extract caches each result without its parse tree and
+ * compacted (pass_parallel.c). The sequential definitions pass cached the raw
+ * result, tree and working arena included, and that cache lives through every
+ * later pass: about 5x the memory per file, for every file of the repo. PHP
+ * keeps its tree: the vendor-import check reads its sites' syntax from it. */
+TEST(sequential_definitions_cache_drops_trees_and_compacts) {
+    if (ensure_parity_setup() != 0)
+        FAIL("setup failed");
+    cbm_discover_opts_t opts = {.mode = CBM_MODE_FULL};
+    cbm_file_info_t *files = NULL;
+    int file_count = 0;
+    ASSERT_EQ(cbm_discover(g_par_tmpdir, &opts, &files, &file_count), 0);
+    ASSERT_GT(file_count, 0);
+
+    cbm_gbuf_t *gbuf = cbm_gbuf_new("par-test", g_par_tmpdir);
+    cbm_registry_t *reg = cbm_registry_new();
+    CBMFileResult **cache = (CBMFileResult **)calloc((size_t)file_count, sizeof(CBMFileResult *));
+    ASSERT_NOT_NULL(cache);
+    atomic_int cancelled;
+    atomic_init(&cancelled, 0);
+    cbm_pipeline_ctx_t ctx = {
+        .project_name = "par-test",
+        .repo_path = g_par_tmpdir,
+        .gbuf = gbuf,
+        .registry = reg,
+        .cancelled = &cancelled,
+        .result_cache = cache,
+    };
+
+    cbm_init();
+    cbm_work_arena_release(); /* start from a thread that keeps nothing */
+    ASSERT_EQ(cbm_pipeline_pass_definitions(&ctx, files, file_count), 0);
+    ASSERT_FALSE(cbm_work_arena_keeping()); /* ...and end the pass that way */
+
+    int cached = 0;
+    int php_trees = 0;
+    for (int i = 0; i < file_count; i++) {
+        const CBMFileResult *r = cache[i];
+        if (!r) {
+            continue;
+        }
+        cached++;
+        if (files[i].language == CBM_LANG_PHP) {
+            ASSERT_NOT_NULL(r->cached_tree);
+            php_trees++;
+        } else {
+            ASSERT_NULL(r->cached_tree);
+        }
+        /* One exact block: capacity == bytes used, no dead headroom. */
+        ASSERT_EQ(r->arena.nblocks, 1);
+        ASSERT_EQ(cbm_arena_capacity(&r->arena), cbm_arena_total(&r->arena));
+    }
+    ASSERT_EQ(cached, file_count);
+    ASSERT_GT(php_trees, 0); /* app/Models.php and app/Services.php */
+
+    for (int i = 0; i < file_count; i++) {
+        cbm_free_result(cache[i]);
+    }
+    free(cache);
+    cbm_registry_free(reg);
+    cbm_gbuf_free(gbuf);
+    cbm_discover_free(files, file_count);
+    PASS();
+}
+
+/* A configured C test convention whose file cannot be extracted aborts the
+ * sequential pass (CBM_PIPELINE_ABORT_PRESERVE_DB). The Python file compacted
+ * before it handed its working arena back to this thread, so that exit has to
+ * release it like the others. */
+TEST(sequential_definitions_configured_failure_releases_work_arena) {
+    char dir[256];
+    snprintf(dir, sizeof(dir), "/tmp/cbm_seqcf_XXXXXX");
+    ASSERT_NOT_NULL(cbm_mkdtemp(dir));
+    char py_path[320];
+    char c_path[320];
+    snprintf(py_path, sizeof(py_path), "%s/a.py", dir);
+    snprintf(c_path, sizeof(c_path), "%s/z.c", dir);
+    ASSERT_EQ(th_write_file(py_path, "def first():\n    return 1\n"), 0);
+    ASSERT_EQ(th_write_file(c_path, "void second(void) {}\n"), 0);
+    cbm_file_info_t files[] = {
+        {.path = py_path, .rel_path = "a.py", .language = CBM_LANG_PYTHON},
+        {.path = c_path, .rel_path = "z.c", .language = CBM_LANG_C},
+    };
+    static const char config[] =
+        "{\"test_impact\":{\"version\":1,\"tests\":{\"conventions\":[{\"language\":\"c\","
+        "\"role\":\"case\",\"define_macro\":\"CHECK\",\"name_args\":[0]}]}}}";
+    cbm_test_declarations_t *declarations =
+        cbm_test_declarations_parse(config, sizeof(config) - 1, false);
+    ASSERT_NOT_NULL(declarations);
+
+    cbm_gbuf_t *gbuf = cbm_gbuf_new("seq-cf", dir);
+    cbm_registry_t *reg = cbm_registry_new();
+    atomic_int cancelled;
+    atomic_init(&cancelled, 0);
+    cbm_pipeline_ctx_t ctx = {
+        .project_name = "seq-cf",
+        .repo_path = dir,
+        .gbuf = gbuf,
+        .registry = reg,
+        .cancelled = &cancelled,
+        .test_declarations = declarations,
+        .test_fault_mask = CBM_PIPELINE_TEST_FAULT_EXTRACT_NULL, /* z.c extracts to NULL */
+    };
+
+    cbm_init();
+    cbm_work_arena_release(); /* start from a thread that keeps nothing */
+    int rc = cbm_pipeline_pass_definitions(&ctx, files, 2);
+    bool keeping = cbm_work_arena_keeping();
+    cbm_work_arena_release();
+
+    cbm_registry_free(reg);
+    cbm_gbuf_free(gbuf);
+    cbm_test_declarations_free(declarations);
+    rm_rf(dir);
+    ASSERT_EQ(rc, CBM_PIPELINE_ABORT_PRESERVE_DB);
+    ASSERT_FALSE(keeping);
+    PASS();
+}
+
 /* ── Empty file list ──────────────────────────────────────────────── */
 
 TEST(parallel_empty_files) {
@@ -5599,6 +5720,8 @@ SUITE(parallel) {
     RUN_TEST(parallel_total_edges);
     RUN_TEST(parallel_spill_mode_builds_the_same_graph);
     RUN_TEST(parallel_post_extract_projection_spills_before_resolve);
+    RUN_TEST(sequential_definitions_cache_drops_trees_and_compacts);
+    RUN_TEST(sequential_definitions_configured_failure_releases_work_arena);
     RUN_TEST(parallel_empty_files);
     RUN_TEST(parallel_args_json_no_overflow);
     RUN_TEST(parallel_wide_definition_keeps_fp_sp_bt_f14);

tests/test_pipeline.c

@@ -636,6 +636,71 @@ TEST(pipeline_structure_nodes) {
     PASS();
 }
 
+/* Number of `label` nodes whose file_path starts with `prefix`, or -1. */
+static int count_nodes_with_path_prefix(cbm_store_t *s, const char *project, const char *label,
+                                        const char *prefix) {
+    cbm_node_t *nodes = NULL;
+    int count = 0;
+    if (cbm_store_find_nodes_by_label(s, project, label, &nodes, &count) != CBM_STORE_OK) {
+        return -1;
+    }
+    int hits = 0;
+    for (int i = 0; i < count; i++) {
+        if (nodes[i].file_path && strncmp(nodes[i].file_path, prefix, strlen(prefix)) == 0) {
+            hits++;
+        }
+    }
+    cbm_store_free_nodes(nodes, count);
+    return hits;
+}
+
+/* Make-style dep-info .d files are dropped at discovery, so the graph holds no
+ * File or Module node for them and no Folder for a directory that held only
+ * dep-info. Real D source beside a dep-info file keeps its nodes and folder. */
+TEST(pipeline_d_dep_info_leaves_no_nodes) {
+    char tmp[256];
+    snprintf(tmp, sizeof(tmp), "/tmp/cbm_dotd_graph_XXXXXX");
+    ASSERT_NOT_NULL(cbm_mkdtemp(tmp));
+    char repo[512];
+    snprintf(repo, sizeof(repo), "%s/repo", tmp);
+    write_temp_file(repo, "cargo-target/debug/deps/app-0123abcd.d",
+                    "/r/cargo-target/debug/deps/app-0123abcd.d: src/main.rs\n\nsrc/main.rs:\n");
+    write_temp_file(repo, "src/app.d", "module app;\n\nvoid main() {}\n");
+    write_temp_file(repo, "src/app.o.d", "src/app.o: src/app.d\n");
+
+    char db_path[512];
+    snprintf(db_path, sizeof(db_path), "%s/test.db", tmp);
+    cbm_pipeline_t *p = cbm_pipeline_new(repo, db_path, CBM_MODE_FULL);
+    ASSERT_NOT_NULL(p);
+    ASSERT_EQ(cbm_pipeline_run(p), 0);
+    cbm_store_t *s = cbm_store_open_path(db_path);
+    ASSERT_NOT_NULL(s);
+    const char *project = cbm_pipeline_project_name(p);
+
+    int dep_folders = count_nodes_with_path_prefix(s, project, "Folder", "cargo-target");
+    int dep_files = count_nodes_with_path_prefix(s, project, "File", "cargo-target");
+    int dep_modules = count_nodes_with_path_prefix(s, project, "Module", "cargo-target");
+    int obj_files = count_nodes_with_path_prefix(s, project, "File", "src/app.o.d");
+    int obj_modules = count_nodes_with_path_prefix(s, project, "Module", "src/app.o.d");
+    int src_folders = count_nodes_with_path_prefix(s, project, "Folder", "src");
+    int src_files = count_nodes_with_path_prefix(s, project, "File", "src/app.d");
+    int src_modules = count_nodes_with_path_prefix(s, project, "Module", "src/app.d");
+
+    cbm_store_close(s);
+    cbm_pipeline_free(p);
+    th_rmtree(tmp);
+
+    ASSERT_EQ(dep_folders, 0);
+    ASSERT_EQ(dep_files, 0);
+    ASSERT_EQ(dep_modules, 0);
+    ASSERT_EQ(obj_files, 0);
+    ASSERT_EQ(obj_modules, 0);
+    ASSERT_EQ(src_folders, 1);
+    ASSERT_EQ(src_files, 1);
+    ASSERT_EQ(src_modules, 1);
+    PASS();
+}
+
 /* Issue #516: an ADR stored via manage_adr (project_summaries) must survive a
  * full re-index. A full re-index deletes the DB and rebuilds it from the graph
  * buffer, which writes an empty project_summaries table; the fix captures the
@@ -29290,6 +29355,7 @@ SUITE(pipeline) {
     RUN_TEST(pipeline_structure_nodes);
     RUN_TEST(pipeline_stem_folder_collision_keeps_both_nodes);
     RUN_TEST(pipeline_kustomize_entry_beside_same_stem_file);
+    RUN_TEST(pipeline_d_dep_info_leaves_no_nodes);
     RUN_TEST(pipeline_committed_counts_match_persisted);
     RUN_TEST(pipeline_adr_survives_full_reindex);
     RUN_TEST(pipeline_export_error_snapshot_on_artifact_failure);