fix(discover): keep JVM packages named like build dirs (part of #2661)

src/discover/discover.c

@@ -346,6 +346,56 @@ static bool resolve_global_excludes_path(char *out, size_t out_sz) {
     return false;
 }
 
+/* JVM source-set language roots: src/<set>/<lang>/ in Maven and Gradle layouts. */
+static const char *JVM_SOURCE_LANG_DIRS[] = {"java", "kotlin", "scala", "groovy", NULL};
+
+static bool component_equals(const char *start, size_t len, const char *word) {
+    return strlen(word) == len && strncmp(start, word, len) == 0;
+}
+
+/* True when rel_path names a directory strictly below a JVM source root
+ * ("src/main/java/com/acme/target", "app/src/test/kotlin/x/tmp"). There every
+ * directory is a package, so a built-in skip name such as "target", "tmp" or
+ * "vendor" is an ordinary package name, not build output (#2661). The root
+ * itself and directories beside it ("src/main/target") are not packages. */
+bool cbm_discover_path_in_jvm_package(const char *rel_path) {
+    if (!rel_path) {
+        return false;
+    }
+    enum { JVM_MAX_COMPONENTS = 256 };
+    const char *starts[JVM_MAX_COMPONENTS];
+    size_t lens[JVM_MAX_COMPONENTS];
+    size_t count = 0;
+    const char *p = rel_path;
+    while (*p && count < JVM_MAX_COMPONENTS) {
+        while (*p == '/' || *p == '\\') {
+            p++;
+        }
+        if (!*p) {
+            break;
+        }
+        const char *start = p;
+        while (*p && *p != '/' && *p != '\\') {
+            p++;
+        }
+        starts[count] = start;
+        lens[count] = (size_t)(p - start);
+        count++;
+    }
+    /* The language root sits at i + 2; the directory must be below it. */
+    for (size_t i = 0; i + 3 < count; i++) {
+        if (!component_equals(starts[i], lens[i], "src")) {
+            continue;
+        }
+        for (int l = 0; JVM_SOURCE_LANG_DIRS[l]; l++) {
+            if (component_equals(starts[i + 2], lens[i + 2], JVM_SOURCE_LANG_DIRS[l])) {
+                return true;
+            }
+        }
+    }
+    return false;
+}
+
 /* True when rel_path equals a suffix rule or ends with "/" + the rule. */
 bool cbm_discover_path_has_skip_suffix(const char *rel_path) {
     if (!rel_path) {
@@ -803,8 +853,10 @@ static dir_verdict_t directory_verdict(const char *entry_name, const char *rel_p
                                        const gitignore_link_t *ignore_chain,
                                        const cbm_gitignore_t *global_gi,
                                        const cbm_gitignore_t *cbmignore) {
-    if (cbm_should_skip_dir(entry_name, opts ? opts->mode : CBM_MODE_FULL) ||
-        cbm_discover_path_has_skip_suffix(rel_path)) {
+    bool builtin_skip = cbm_should_skip_dir(entry_name, opts ? opts->mode : CBM_MODE_FULL) &&
+                        (cbm_discover_is_safety_core_directory(entry_name) ||
+                         !cbm_discover_path_in_jvm_package(rel_path));
+    if (builtin_skip || cbm_discover_path_has_skip_suffix(rel_path)) {
         /* #500: a .cbmignore negation (e.g. "!obj/") whose rule is the last
          * match for this dir un-skips a built-in skip-list dir — except the
          * non-negatable safety core. Fall through so .gitignore/global/local

src/discover/discover_policy_internal.h

@@ -5,5 +5,6 @@
 /* Pure legacy policy predicates; callers provide live terminated strings. */
 bool cbm_discover_path_has_skip_suffix(const char *relative_path);
 bool cbm_discover_is_safety_core_directory(const char *name);
+bool cbm_discover_path_in_jvm_package(const char *relative_path);
 
 #endif

src/discover/inventory_filter.c

@@ -102,7 +102,9 @@ static bool cif_directory_reason(cif_context *c, size_t parent, cbm_inventory_pa
     if (!cif_poll(c)) {
         return false;
     }
-    bool builtin = cbm_should_skip_dir(name, CBM_MODE_FULL);
+    bool builtin = cbm_should_skip_dir(name, CBM_MODE_FULL) &&
+                   (cbm_discover_is_safety_core_directory(name) ||
+                    !cbm_discover_path_in_jvm_package((const char *)path.data));
     bool suffix = !builtin && cbm_discover_path_has_skip_suffix((const char *)path.data);
     bool safety =
         (builtin || suffix) && c->owner->cbm_program && cbm_discover_is_safety_core_directory(name);

tests/test_discover.c

@@ -6,6 +6,7 @@
 #include "test_framework.h"
 #include "test_helpers.h"
 #include "discover/discover.h"
+#include "discover/discover_policy_internal.h"
 #include "foundation/platform.h"
 
 typedef struct {
@@ -1328,6 +1329,73 @@ TEST(discover_cbmignore_negates_always_skip_dir) {
     PASS();
 }
 
+/* ── JVM packages named like build output (issue #2661) ─────────── */
+
+/* Below src/<set>/{java,kotlin,scala,groovy}/ every directory is a package, so
+ * "target", "tmp", "vendor" there are package names, not build output. */
+TEST(jvm_package_path_predicate_issue2661) {
+    ASSERT_TRUE(cbm_discover_path_in_jvm_package("src/main/java/com/acme/target"));
+    ASSERT_TRUE(cbm_discover_path_in_jvm_package("app/src/test/kotlin/x/tmp"));
+    ASSERT_TRUE(cbm_discover_path_in_jvm_package("src/main/scala/vendor"));
+    /* The language root itself and its siblings are not packages. */
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package("src/main/java"));
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package("src/main/target"));
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package("target"));
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package("module/target/classes"));
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package("src/java/main/tmp"));
+    ASSERT_FALSE(cbm_discover_path_in_jvm_package(NULL));
+    PASS();
+}
+
+TEST(discover_keeps_jvm_packages_named_like_build_dirs_issue2661) {
+    char *base = th_mktempdir("cbm_disc_jvm_pkg");
+    ASSERT(base != NULL);
+
+    th_write_file(TH_PATH(base, "src/main/java/com/acme/target/Foo.java"),
+                  "package com.acme.target;\nclass Foo {}\n");
+    th_write_file(TH_PATH(base, "src/test/java/com/acme/tmp/T.java"),
+                  "package com.acme.tmp;\nclass T {}\n");
+    th_write_file(TH_PATH(base, "src/main/kotlin/com/acme/vendor/V.kt"),
+                  "package com.acme.vendor\nclass V\n");
+    /* Real build output stays skipped. */
+    th_write_file(TH_PATH(base, "target/classes/Gen.java"), "class Gen {}\n");
+    th_write_file(TH_PATH(base, "src/main/target/Out.java"), "class Out {}\n");
+    /* The safety core is never un-skipped, even below a package root. */
+    th_write_file(TH_PATH(base, "src/main/java/com/.git/H.java"), "class H {}\n");
+
+    cbm_discover_opts_t opts = {0};
+    cbm_file_info_t *files = NULL;
+    int count = 0;
+    int rc = cbm_discover(base, &opts, &files, &count);
+    ASSERT_EQ(rc, 0);
+    ASSERT_TRUE(discover_has_rel_path(files, count, "src/main/java/com/acme/target/Foo.java"));
+    ASSERT_TRUE(discover_has_rel_path(files, count, "src/test/java/com/acme/tmp/T.java"));
+    ASSERT_TRUE(discover_has_rel_path(files, count, "src/main/kotlin/com/acme/vendor/V.kt"));
+    ASSERT_FALSE(discover_has_rel_path(files, count, "target/classes/Gen.java"));
+    ASSERT_FALSE(discover_has_rel_path(files, count, "src/main/target/Out.java"));
+    ASSERT_FALSE(discover_has_rel_path(files, count, "src/main/java/com/.git/H.java"));
+    ASSERT_EQ(count, 3);
+    cbm_discover_free(files, count);
+
+    /* Moderate/fast mode: a package named like a fast-mode skip dir is kept
+     * below a JVM source root, while the same name at the top stays skipped. */
+    th_write_file(TH_PATH(base, "src/main/java/com/acme/integration/I.java"),
+                  "package com.acme.integration;\nclass I {}\n");
+    th_write_file(TH_PATH(base, "integration/Top.java"), "class Top {}\n");
+
+    cbm_discover_opts_t mod_opts = {.mode = CBM_MODE_MODERATE};
+    files = NULL;
+    count = 0;
+    rc = cbm_discover(base, &mod_opts, &files, &count);
+    ASSERT_EQ(rc, 0);
+    ASSERT_TRUE(discover_has_rel_path(files, count, "src/main/java/com/acme/integration/I.java"));
+    ASSERT_FALSE(discover_has_rel_path(files, count, "integration/Top.java"));
+
+    cbm_discover_free(files, count);
+    th_cleanup(base);
+    PASS();
+}
+
 /* ── Laravel compiled Blade view cache (issue #1735) ─────────────── */
 
 /* storage/framework/views/ holds Laravel's compiled Blade templates: generated
@@ -2381,6 +2449,8 @@ SUITE(discover) {
     RUN_TEST(skip_venv);
     RUN_TEST(skip_dist);
     RUN_TEST(skip_target);
+    RUN_TEST(jvm_package_path_predicate_issue2661);
+    RUN_TEST(discover_keeps_jvm_packages_named_like_build_dirs_issue2661);
     RUN_TEST(skip_vendor);
     RUN_TEST(skip_vendored);
     RUN_TEST(skip_terraform);

tests/test_doc_mentions_jvm.c

@@ -578,8 +578,10 @@ TEST(doc_mentions_jvm_java_families_and_reasons) {
     jt_write(&r, "src/main/java/b/Helper.java", JT_HELPER);
     jt_write(&r, "src/main/java/a/Shapes.java", JT_SHAPES);
     jt_write(&r, "src/main/java/a/Widget.java", JT_WIDGET);
-    /* a real package in a directory the indexer skips by name */
-    jt_write(&r, "src/main/java/a/env/Config.java", "package a.env;\npublic class Config {}\n");
+    /* a real package in a layout discovery still skips by name: a plain
+     * src/<pkg>/ tree (the open part of #2661); below src/<set>/<lang>/ the
+     * same package is kept since #2724 */
+    jt_write(&r, "src/a/env/Config.java", "package a.env;\npublic class Config {}\n");
     jt_write(&r, "src/test/java/a/GadgetTests.java",
              "package a;\n/** Tests {@link Gadget}. */\npublic class GadgetTests {}\n");
     ASSERT_EQ(dm_index(r.repo, r.db, NULL), 0);
@@ -643,8 +645,8 @@ TEST(doc_mentions_jvm_java_families_and_reasons) {
     ASSERT_EQ(jt_edge(db, "a.Widget", "a.GadgetTests"), 0);
     /* ... while test code binds product code */
     ASSERT_EQ(jt_edge(db, "a.GadgetTests", "a.Gadget"), 1);
-    /* not_indexed: a package of the repository in a directory discovery
-     * skips, named by an import and by its qualified name */
+    /* not_indexed: a package of the repository in a layout discovery still
+     * skips by name, named by an import and by its qualified name */
     ASSERT_STR_EQ(jt_reason(db, w, "Config"), "not_indexed");
     ASSERT_STR_EQ(jt_reason(db, w, "a.env.Other"), "not_indexed");
     /* graph_gap: a package (no node), a record component, an interface
@@ -1117,8 +1119,10 @@ TEST(doc_mentions_jvm_kotlin_families_and_reasons) {
     jt_write(&r, "src/main/kotlin/far/Helper.kt",
              "package far\n\nclass Helper {\n    fun run() {}\n}\n");
     jt_write(&r, "src/main/kotlin/far/sub/Deep.kt", "package far.sub\n\nclass Deep\n");
-    /* a real package in a directory the indexer skips by name */
-    jt_write(&r, "src/main/kotlin/k/env/Config.kt", "package k.env\n\nclass Config\n");
+    /* a real package in a layout discovery still skips by name: a plain
+     * src/<pkg>/ tree (the open part of #2661); below src/<set>/<lang>/ the
+     * same package is kept since #2724 */
+    jt_write(&r, "src/k/env/Config.kt", "package k.env\n\nclass Config\n");
     jt_write(&r, "src/test/kotlin/k/GadgetTests.kt", "package k\n\nclass GadgetTests\n");
     jt_write(&r, "src/test/kotlin/k/samples/Samples.kt",
              "package k.samples\n\nfun widgetSample() {}\n");
@@ -1170,7 +1174,7 @@ TEST(doc_mentions_jvm_kotlin_families_and_reasons) {
     ASSERT_STR_EQ(jt_reason(db, w, "far.sub"), "graph_gap");
     /* test_only_target */
     ASSERT_STR_EQ(jt_reason(db, w, "GadgetTests"), "test_only_target");
-    /* not_indexed: a package of the repository in a directory discovery skips */
+    /* not_indexed: a package of the repository in a layout discovery still skips by name */
     ASSERT_STR_EQ(jt_reason(db, w, "k.env.Config"), "not_indexed");
     /* unparseable: a path with an empty name in it */
     ASSERT_STR_EQ(jt_reason(db, w, "Gadget..spin"), "unparseable");