pkg/codesearch: do indexing of struct/union/enum

Update #6469

18 files changed, 245 insertions, 26 deletions

diff --git a/pkg/codesearch/database.go b/pkg/codesearch/database.go
index 800c5759b..edeb761cc 100644
--- a/pkg/codesearch/database.go
+++ b/pkg/codesearch/database.go
@@ -44,7 +44,7 @@ type LineRange struct {
 var DatabaseFormatHash = func() string {
 	// Semantic version should be bumped when the schema does not change,
 	// but stored values changes.
-	const semanticVersion = "1"
+	const semanticVersion = "2"
 	schema, err := jsonschema.For[Database](nil)
 	if err != nil {
 		panic(err)
diff --git a/pkg/codesearch/testdata/query-def-source-header b/pkg/codesearch/testdata/query-def-source-header
index 9e221f57b..edd8d2303 100644
--- a/pkg/codesearch/testdata/query-def-source-header
+++ b/pkg/codesearch/testdata/query-def-source-header
@@ -2,7 +2,7 @@ def-source source0.c function_with_comment_in_header yes
 
 function function_with_comment_in_header is defined in source0.c:
 
-  19:	void function_with_comment_in_header()
-  20:	{
-  21:		same_name_in_several_files();
-  22:	}
+  24:	void function_with_comment_in_header()
+  25:	{
+  26:		same_name_in_several_files();
+  27:	}
diff --git a/pkg/codesearch/testdata/query-def-source-open b/pkg/codesearch/testdata/query-def-source-open
index c8c11b8ef..79d0f8977 100644
--- a/pkg/codesearch/testdata/query-def-source-open
+++ b/pkg/codesearch/testdata/query-def-source-open
@@ -2,10 +2,10 @@ def-source source0.c open yes
 
 function open is defined in source0.c:
 
-   6:	/*
-   7:	 * Comment about open.
-   8:	 */
-   9:	int open()
-  10:	{
-  11:		return 0;
-  12:	}
+  11:	/*
+  12:	 * Comment about open.
+  13:	 */
+  14:	int open()
+  15:	{
+  16:		return 0;
+  17:	}
diff --git a/pkg/codesearch/testdata/query-def-source-struct0 b/pkg/codesearch/testdata/query-def-source-struct0
new file mode 100644
index 000000000..adf011a3c
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct0
@@ -0,0 +1,8 @@
+def-source source0.c struct_in_c_file yes
+
+struct struct_in_c_file is defined in source0.c:
+
+   6:	struct struct_in_c_file {
+   7:		int X;
+   8:		struct some_struct by_value;
+   9:	};
diff --git a/pkg/codesearch/testdata/query-def-source-struct1 b/pkg/codesearch/testdata/query-def-source-struct1
new file mode 100644
index 000000000..f6b5bba0b
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct1
@@ -0,0 +1,8 @@
+def-source source0.c some_struct yes
+
+struct some_struct is defined in source0.h:
+
+  17:	struct some_struct {
+  18:		int x;
+  19:		int y;
+  20:	};
diff --git a/pkg/codesearch/testdata/query-def-source-struct2 b/pkg/codesearch/testdata/query-def-source-struct2
new file mode 100644
index 000000000..5de681a75
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct2
@@ -0,0 +1,5 @@
+def-source source0.c some_struct_t yes
+
+typedef some_struct_t is defined in source0.h:
+
+  22:	typedef struct some_struct some_struct_t;
diff --git a/pkg/codesearch/testdata/query-def-source-struct3 b/pkg/codesearch/testdata/query-def-source-struct3
new file mode 100644
index 000000000..0d4f4e29e
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct3
@@ -0,0 +1,11 @@
+def-source source0.c some_struct_with_a_comment yes
+
+struct some_struct_with_a_comment is defined in source0.h:
+
+  24:	/*
+  25:	 * This should not require an explanation.
+  26:	 */
+  27:	struct some_struct_with_a_comment {
+  28:		int x;
+  29:		struct some_struct* other_struct;
+  30:	};
diff --git a/pkg/codesearch/testdata/query-def-source-struct4 b/pkg/codesearch/testdata/query-def-source-struct4
new file mode 100644
index 000000000..bd0c03d90
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct4
@@ -0,0 +1,7 @@
+def-source source0.c typedefed_struct_t yes
+
+typedef typedefed_struct_t is defined in source0.h:
+
+  32:	typedef struct {
+  33:		int x;
+  34:	} typedefed_struct_t;
diff --git a/pkg/codesearch/testdata/query-def-source-struct5 b/pkg/codesearch/testdata/query-def-source-struct5
new file mode 100644
index 000000000..b86f38484
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct5
@@ -0,0 +1,7 @@
+def-source source0.c another_struct yes
+
+struct another_struct is defined in source0.h:
+
+  36:	typedef struct another_struct {
+  37:		int x;
+  38:	} another_struct_t;
diff --git a/pkg/codesearch/testdata/query-def-source-struct6 b/pkg/codesearch/testdata/query-def-source-struct6
new file mode 100644
index 000000000..77d9a1102
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct6
@@ -0,0 +1,7 @@
+def-source source0.c another_struct_t yes
+
+typedef another_struct_t is defined in source0.h:
+
+  36:	typedef struct another_struct {
+  37:		int x;
+  38:	} another_struct_t;
diff --git a/pkg/codesearch/testdata/query-def-source-struct7 b/pkg/codesearch/testdata/query-def-source-struct7
new file mode 100644
index 000000000..3d6212614
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct7
@@ -0,0 +1,8 @@
+def-source source0.c some_union yes
+
+union some_union is defined in source0.h:
+
+  40:	union some_union {
+  41:		int x;
+  42:		void* p;
+  43:	};
diff --git a/pkg/codesearch/testdata/query-def-source-struct8 b/pkg/codesearch/testdata/query-def-source-struct8
new file mode 100644
index 000000000..93c7c6516
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct8
@@ -0,0 +1,8 @@
+def-source source0.c some_enum yes
+
+enum some_enum is defined in source0.h:
+
+  45:	enum some_enum {
+  46:		enum_foo = 1,
+  47:		enum_bar = 2,
+  48:	};
diff --git a/pkg/codesearch/testdata/query-def-source-struct9 b/pkg/codesearch/testdata/query-def-source-struct9
new file mode 100644
index 000000000..20056963a
--- /dev/null
+++ b/pkg/codesearch/testdata/query-def-source-struct9
@@ -0,0 +1,5 @@
+def-source source0.c some_enum_t yes
+
+typedef some_enum_t is defined in source0.h:
+
+  50:	typedef enum some_enum some_enum_t;
diff --git a/pkg/codesearch/testdata/query-file-index-source b/pkg/codesearch/testdata/query-file-index-source
index c238079d0..3263deca3 100644
--- a/pkg/codesearch/testdata/query-file-index-source
+++ b/pkg/codesearch/testdata/query-file-index-source
@@ -5,3 +5,4 @@ file source0.c defines the following entities:
 function close
 function function_with_comment_in_header
 function open
+struct struct_in_c_file
diff --git a/pkg/codesearch/testdata/query-find-references-normal b/pkg/codesearch/testdata/query-find-references-normal
index 1bc1a4be1..2c19827cc 100644
--- a/pkg/codesearch/testdata/query-find-references-normal
+++ b/pkg/codesearch/testdata/query-find-references-normal
@@ -2,10 +2,10 @@ find-references source0.c same_name_in_several_files "" 3 10
 
 same_name_in_several_files has 1 references:
 
-function function_with_comment_in_header calls it at source0.c:21
-  19:	void function_with_comment_in_header()
-  20:	{
-  21:		same_name_in_several_files();
-  22:	}
+function function_with_comment_in_header calls it at source0.c:26
+  24:	void function_with_comment_in_header()
+  25:	{
+  26:		same_name_in_several_files();
+  27:	}
 
 
diff --git a/pkg/codesearch/testdata/source0.c b/pkg/codesearch/testdata/source0.c
index 384c4c119..2d312ff99 100644
--- a/pkg/codesearch/testdata/source0.c
+++ b/pkg/codesearch/testdata/source0.c
@@ -3,6 +3,11 @@
 
 #include "source0.h"
 
+struct struct_in_c_file {
+	int X;
+	struct some_struct by_value;
+};
+
 /*
  * Comment about open.
  */
diff --git a/pkg/codesearch/testdata/source0.c.json b/pkg/codesearch/testdata/source0.c.json
index cb80dc021..ba6b41751 100644
--- a/pkg/codesearch/testdata/source0.c.json
+++ b/pkg/codesearch/testdata/source0.c.json
@@ -1,13 +1,23 @@
 {
 	"definitions": [
 		{
+			"kind": "enum",
+			"name": "some_enum",
+			"body": {
+				"file": "source0.h",
+				"start_line": 45,
+				"end_line": 48
+			},
+			"comment": {}
+		},
+		{
 			"kind": "function",
 			"name": "close",
 			"type": "int ()",
 			"body": {
 				"file": "source0.c",
-				"start_line": 14,
-				"end_line": 17
+				"start_line": 19,
+				"end_line": 22
 			},
 			"comment": {}
 		},
@@ -29,8 +39,8 @@
 			"type": "void ()",
 			"body": {
 				"file": "source0.c",
-				"start_line": 19,
-				"end_line": 22
+				"start_line": 24,
+				"end_line": 27
 			},
 			"comment": {},
 			"refs": [
@@ -38,7 +48,7 @@
 					"kind": "calls",
 					"entity_kind": "function",
 					"name": "same_name_in_several_files",
-					"line": 21
+					"line": 26
 				}
 			]
 		},
@@ -48,14 +58,108 @@
 			"type": "int ()",
 			"body": {
 				"file": "source0.c",
-				"start_line": 6,
-				"end_line": 12
+				"start_line": 11,
+				"end_line": 17
 			},
 			"comment": {
 				"file": "source0.c",
-				"start_line": 6,
-				"end_line": 8
+				"start_line": 11,
+				"end_line": 13
+			}
+		},
+		{
+			"kind": "struct",
+			"name": "another_struct",
+			"body": {
+				"file": "source0.h",
+				"start_line": 36,
+				"end_line": 38
+			},
+			"comment": {}
+		},
+		{
+			"kind": "struct",
+			"name": "some_struct",
+			"body": {
+				"file": "source0.h",
+				"start_line": 17,
+				"end_line": 20
+			},
+			"comment": {}
+		},
+		{
+			"kind": "struct",
+			"name": "some_struct_with_a_comment",
+			"body": {
+				"file": "source0.h",
+				"start_line": 24,
+				"end_line": 30
+			},
+			"comment": {
+				"file": "source0.h",
+				"start_line": 24,
+				"end_line": 26
 			}
+		},
+		{
+			"kind": "struct",
+			"name": "struct_in_c_file",
+			"body": {
+				"file": "source0.c",
+				"start_line": 6,
+				"end_line": 9
+			},
+			"comment": {}
+		},
+		{
+			"kind": "typedef",
+			"name": "another_struct_t",
+			"body": {
+				"file": "source0.h",
+				"start_line": 36,
+				"end_line": 38
+			},
+			"comment": {}
+		},
+		{
+			"kind": "typedef",
+			"name": "some_enum_t",
+			"body": {
+				"file": "source0.h",
+				"start_line": 50,
+				"end_line": 50
+			},
+			"comment": {}
+		},
+		{
+			"kind": "typedef",
+			"name": "some_struct_t",
+			"body": {
+				"file": "source0.h",
+				"start_line": 22,
+				"end_line": 22
+			},
+			"comment": {}
+		},
+		{
+			"kind": "typedef",
+			"name": "typedefed_struct_t",
+			"body": {
+				"file": "source0.h",
+				"start_line": 32,
+				"end_line": 34
+			},
+			"comment": {}
+		},
+		{
+			"kind": "union",
+			"name": "some_union",
+			"body": {
+				"file": "source0.h",
+				"start_line": 40,
+				"end_line": 43
+			},
+			"comment": {}
 		}
 	]
 }
\ No newline at end of file
diff --git a/pkg/codesearch/testdata/source0.h b/pkg/codesearch/testdata/source0.h
index d1540ebce..9549feb9d 100644
--- a/pkg/codesearch/testdata/source0.h
+++ b/pkg/codesearch/testdata/source0.h
@@ -13,3 +13,38 @@ static inline int func_in_header()
 {
 	return 0;
 }
+
+struct some_struct {
+	int x;
+	int y;
+};
+
+typedef struct some_struct some_struct_t;
+
+/*
+ * This should not require an explanation.
+ */
+struct some_struct_with_a_comment {
+	int x;
+	struct some_struct* other_struct;
+};
+
+typedef struct {
+	int x;
+} typedefed_struct_t;
+
+typedef struct another_struct {
+	int x;
+} another_struct_t;
+
+union some_union {
+	int x;
+	void* p;
+};
+
+enum some_enum {
+	enum_foo = 1,
+	enum_bar = 2,
+};
+
+typedef enum some_enum some_enum_t;