rbtr-lang-bash 2026.7.0.dev0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1 @@
1
+ """Bash language plugin package."""
@@ -0,0 +1,34 @@
1
+ ; Top-level comments (Bash: single `comment` type).
2
+ (comment) @comment
3
+
4
+ (function_definition
5
+ name: (word) @_fn_name) @function
6
+
7
+ (command
8
+ name: (command_name
9
+ (word) @_cmd)
10
+ .
11
+ (word) @_import_module
12
+ (#eq? @_cmd "source")) @import
13
+
14
+ (command
15
+ name: (command_name
16
+ (word) @_cmd)
17
+ .
18
+ (word) @_import_module
19
+ (#eq? @_cmd ".")) @import
20
+
21
+ (program
22
+ (variable_assignment
23
+ name: (variable_name) @_var_name) @variable)
24
+
25
+ (program
26
+ (declaration_command
27
+ (variable_assignment
28
+ name: (variable_name) @_var_name)) @variable)
29
+
30
+ (command
31
+ name: (command_name (word) @_cmd)
32
+ argument: (concatenation (word) @_var_name)
33
+ (#eq? @_cmd "alias")
34
+ (#match? @_var_name "=")) @variable
@@ -0,0 +1,64 @@
1
+ """Bash language plugin.
2
+
3
+ Provides function extraction only — Bash has no import system,
4
+ class system, or module structure.
5
+
6
+ Extracted chunks::
7
+
8
+ deploy() { echo deploying; } → function "deploy", scope ""
9
+ function setup { ... } → function "setup", scope ""
10
+ alias ll="ls -l" → variable "ll", scope ""
11
+
12
+ No classes or methods are extracted.
13
+
14
+ An `alias` name parses as one `word` fused with its `=` (`ll=`), which no
15
+ query can split. The `name_extractor` strips the trailing `=`; no other
16
+ bash name ends in one.
17
+ """
18
+
19
+ from __future__ import annotations
20
+
21
+ from typing import TYPE_CHECKING
22
+
23
+ from rbtr.languages.registration import (
24
+ LanguageRegistration,
25
+ NameResolver,
26
+ QueryExtraction,
27
+ load_query,
28
+ )
29
+
30
+ if TYPE_CHECKING:
31
+ from tree_sitter import Node
32
+
33
+
34
+ # ── Query ────────────────────────────────────────────────────────────
35
+
36
+
37
+ # ── Plugin ───────────────────────────────────────────────────────────
38
+
39
+
40
+ bash = LanguageRegistration(
41
+ id="bash",
42
+ extensions=frozenset({".sh", ".bash", ".zsh"}),
43
+ filenames=frozenset(
44
+ {
45
+ "Bashrc",
46
+ ".bashrc",
47
+ ".bash_profile",
48
+ ".zshrc",
49
+ }
50
+ ),
51
+ grammar_module="tree_sitter_bash",
52
+ extraction=QueryExtraction(
53
+ query=load_query(__package__, "bash"),
54
+ ),
55
+ extraction_serial=4,
56
+ )
57
+
58
+
59
+ @bash.name_extractor
60
+ def _strip_alias_eq(
61
+ resolver: NameResolver, capture_name: str, node: Node, captures: dict[str, list[Node]]
62
+ ) -> str:
63
+ """Default name, with the `=` the grammar fuses onto an alias removed."""
64
+ return resolver(capture_name, node, captures).rstrip("=")
File without changes
File without changes
@@ -0,0 +1,4 @@
1
+ [
2
+ "bash.sh::source ./lib/colours.sh -> lib/colours.sh::RED [imports]",
3
+ "bash.sh::source ./lib/colours.sh -> lib/colours.sh::RESET [imports]"
4
+ ]
@@ -0,0 +1,254 @@
1
+ [
2
+ {
3
+ "id": "d39ad69c069b6285",
4
+ "blob_sha": "sha1",
5
+ "file_path": "bash.sh",
6
+ "kind": "comment",
7
+ "name": "<anonymous>",
8
+ "scope": "",
9
+ "language": "bash",
10
+ "content": "#!/usr/bin/env bash\n# Greeter — print greetings for named recipients.\n#\n# Bash has no classes or methods; the plugin extracts functions,\n# top-level variable assignments (including export/declare/readonly),\n# aliases, and source/. imports.",
11
+ "line_start": 1,
12
+ "line_end": 6,
13
+ "metadata": {
14
+ "module": "",
15
+ "names": "",
16
+ "dots": "",
17
+ "language_hint": ""
18
+ }
19
+ },
20
+ {
21
+ "id": "ca906d45e5c8e1da",
22
+ "blob_sha": "sha1",
23
+ "file_path": "bash.sh",
24
+ "kind": "import",
25
+ "name": "source ./lib/colours.sh",
26
+ "scope": "",
27
+ "language": "bash",
28
+ "content": "source ./lib/colours.sh",
29
+ "line_start": 8,
30
+ "line_end": 8,
31
+ "metadata": {
32
+ "module": "./lib/colours.sh",
33
+ "names": "",
34
+ "dots": "",
35
+ "language_hint": ""
36
+ }
37
+ },
38
+ {
39
+ "id": "19b95c899d99e1f5",
40
+ "blob_sha": "sha1",
41
+ "file_path": "bash.sh",
42
+ "kind": "import",
43
+ "name": ". /etc/greeter.conf",
44
+ "scope": "",
45
+ "language": "bash",
46
+ "content": ". /etc/greeter.conf",
47
+ "line_start": 9,
48
+ "line_end": 9,
49
+ "metadata": {
50
+ "module": "/etc/greeter.conf",
51
+ "names": "",
52
+ "dots": "",
53
+ "language_hint": ""
54
+ }
55
+ },
56
+ {
57
+ "id": "dbf8d8210a1c9118",
58
+ "blob_sha": "sha1",
59
+ "file_path": "bash.sh",
60
+ "kind": "variable",
61
+ "name": "DEFAULT_GREETING",
62
+ "scope": "",
63
+ "language": "bash",
64
+ "content": "DEFAULT_GREETING=\"Hello\"",
65
+ "line_start": 11,
66
+ "line_end": 11,
67
+ "metadata": {
68
+ "module": "",
69
+ "names": "",
70
+ "dots": "",
71
+ "language_hint": ""
72
+ }
73
+ },
74
+ {
75
+ "id": "610d295f73434926",
76
+ "blob_sha": "sha1",
77
+ "file_path": "bash.sh",
78
+ "kind": "comment",
79
+ "name": "<anonymous>",
80
+ "scope": "",
81
+ "language": "bash",
82
+ "content": "# trailing comment: its own chunk",
83
+ "line_start": 11,
84
+ "line_end": 11,
85
+ "metadata": {
86
+ "module": "",
87
+ "names": "",
88
+ "dots": "",
89
+ "language_hint": ""
90
+ }
91
+ },
92
+ {
93
+ "id": "aa482bbea0e16ce3",
94
+ "blob_sha": "sha1",
95
+ "file_path": "bash.sh",
96
+ "kind": "variable",
97
+ "name": "LOCALE",
98
+ "scope": "",
99
+ "language": "bash",
100
+ "content": "# Standalone note, separated by blank lines from any definition.\n# Second line of the same block.\nLOCALE=\"${LANG:-en}\"",
101
+ "line_start": 13,
102
+ "line_end": 15,
103
+ "metadata": {
104
+ "module": "",
105
+ "names": "",
106
+ "dots": "",
107
+ "language_hint": ""
108
+ }
109
+ },
110
+ {
111
+ "id": "fd2d219510e272c5",
112
+ "blob_sha": "sha1",
113
+ "file_path": "bash.sh",
114
+ "kind": "variable",
115
+ "name": "API_URL",
116
+ "scope": "",
117
+ "language": "bash",
118
+ "content": "export API_URL=\"https://example.com\"",
119
+ "line_start": 16,
120
+ "line_end": 16,
121
+ "metadata": {
122
+ "module": "",
123
+ "names": "",
124
+ "dots": "",
125
+ "language_hint": ""
126
+ }
127
+ },
128
+ {
129
+ "id": "28c8e045bbc3c2fc",
130
+ "blob_sha": "sha1",
131
+ "file_path": "bash.sh",
132
+ "kind": "variable",
133
+ "name": "MAX_RETRIES",
134
+ "scope": "",
135
+ "language": "bash",
136
+ "content": "readonly MAX_RETRIES=3",
137
+ "line_start": 17,
138
+ "line_end": 17,
139
+ "metadata": {
140
+ "module": "",
141
+ "names": "",
142
+ "dots": "",
143
+ "language_hint": ""
144
+ }
145
+ },
146
+ {
147
+ "id": "46c076160d937a23",
148
+ "blob_sha": "sha1",
149
+ "file_path": "bash.sh",
150
+ "kind": "variable",
151
+ "name": "COUNTER",
152
+ "scope": "",
153
+ "language": "bash",
154
+ "content": "declare -i COUNTER=0",
155
+ "line_start": 18,
156
+ "line_end": 18,
157
+ "metadata": {
158
+ "module": "",
159
+ "names": "",
160
+ "dots": "",
161
+ "language_hint": ""
162
+ }
163
+ },
164
+ {
165
+ "id": "9be9ed61fbcec8cf",
166
+ "blob_sha": "sha1",
167
+ "file_path": "bash.sh",
168
+ "kind": "variable",
169
+ "name": "greet",
170
+ "scope": "",
171
+ "language": "bash",
172
+ "content": "alias greet=\"format_greeting\"",
173
+ "line_start": 19,
174
+ "line_end": 19,
175
+ "metadata": {
176
+ "module": "",
177
+ "names": "",
178
+ "dots": "",
179
+ "language_hint": ""
180
+ }
181
+ },
182
+ {
183
+ "id": "1c05f30beda1010c",
184
+ "blob_sha": "sha1",
185
+ "file_path": "bash.sh",
186
+ "kind": "function",
187
+ "name": "format_greeting",
188
+ "scope": "",
189
+ "language": "bash",
190
+ "content": "# Format a greeting for a single recipient.\nformat_greeting() {\n local name=\"$1\"\n echo \"${DEFAULT_GREETING}, ${name} (${LOCALE})\"\n}",
191
+ "line_start": 21,
192
+ "line_end": 25,
193
+ "metadata": {
194
+ "module": "",
195
+ "names": "",
196
+ "dots": "",
197
+ "language_hint": ""
198
+ }
199
+ },
200
+ {
201
+ "id": "f6bed6e22b836142",
202
+ "blob_sha": "sha1",
203
+ "file_path": "bash.sh",
204
+ "kind": "function",
205
+ "name": "greet_all",
206
+ "scope": "",
207
+ "language": "bash",
208
+ "content": "# Greet every argument in turn.\ngreet_all() {\n for name in \"$@\"; do\n format_greeting \"$name\"\n done\n}",
209
+ "line_start": 27,
210
+ "line_end": 32,
211
+ "metadata": {
212
+ "module": "",
213
+ "names": "",
214
+ "dots": "",
215
+ "language_hint": ""
216
+ }
217
+ },
218
+ {
219
+ "id": "c01f22efbc31cfad",
220
+ "blob_sha": "sha1",
221
+ "file_path": "lib/colours.sh",
222
+ "kind": "variable",
223
+ "name": "RED",
224
+ "scope": "",
225
+ "language": "bash",
226
+ "content": "#!/usr/bin/env bash\n# Terminal colour codes.\nRED='\\033[0;31m'",
227
+ "line_start": 1,
228
+ "line_end": 3,
229
+ "metadata": {
230
+ "module": "",
231
+ "names": "",
232
+ "dots": "",
233
+ "language_hint": ""
234
+ }
235
+ },
236
+ {
237
+ "id": "472aaed6fa440473",
238
+ "blob_sha": "sha1",
239
+ "file_path": "lib/colours.sh",
240
+ "kind": "variable",
241
+ "name": "RESET",
242
+ "scope": "",
243
+ "language": "bash",
244
+ "content": "RESET='\\033[0m'",
245
+ "line_start": 4,
246
+ "line_end": 4,
247
+ "metadata": {
248
+ "module": "",
249
+ "names": "",
250
+ "dots": "",
251
+ "language_hint": ""
252
+ }
253
+ }
254
+ ]
@@ -0,0 +1,112 @@
1
+ """Bash docstring-extraction test cases.
2
+
3
+ tree-sitter-bash uses a single `comment` node for any `#` line. Function
4
+ docs are a `#` comment run directly above the definition; a shebang
5
+ (`#!/bin/bash`) attaches to the first function only when no blank line
6
+ separates them.
7
+
8
+ Each `@case` returns `(lang, source, symbol_name, snippet)`; see
9
+ `test_docstrings.py` for the assertion direction per tag.
10
+ """
11
+
12
+ from __future__ import annotations
13
+
14
+ from pytest_cases import case
15
+
16
+ type DocstringCase = tuple[str, str, str, str]
17
+
18
+
19
+ @case(tags=["documented", "canonical", "exterior_doc"])
20
+ def case_bash_hash_doc_on_function() -> DocstringCase:
21
+ """Canonical `#` comment above a shell function."""
22
+ src = """\
23
+ # Greet the user.
24
+ greet() {
25
+ echo hello
26
+ }
27
+ """
28
+ return "bash", src, "greet", "Greet the user"
29
+
30
+
31
+ @case(tags=["documented", "canonical", "exterior_doc"])
32
+ def case_bash_multi_line_hash_doc() -> DocstringCase:
33
+ """Multi-line `#` comment run."""
34
+ src = """\
35
+ # Greet the user.
36
+ #
37
+ # Reads the name from $1.
38
+ greet() {
39
+ echo hi $1
40
+ }
41
+ """
42
+ return "bash", src, "greet", "Reads the name from"
43
+
44
+
45
+ @case(tags=["documented", "edge_case", "exterior_doc"])
46
+ def case_bash_shebang_attached_to_first_function() -> DocstringCase:
47
+ """When a function follows the shebang with no blank line,
48
+ the shebang attaches — an honest consequence of the
49
+ flexible attachment policy. Recording this as a case so
50
+ any future tightening does not regress silently.
51
+ """
52
+ src = """\
53
+ #!/bin/bash
54
+ # Entry point.
55
+ main() {
56
+ echo hi
57
+ }
58
+ """
59
+ return "bash", src, "main", "#!/bin/bash"
60
+
61
+
62
+ @case(tags=["documented", "unconventional", "exterior_doc"])
63
+ def case_bash_comment_with_script_style_heading() -> DocstringCase:
64
+ """Heading-like comment above a function attaches."""
65
+ src = """\
66
+ # ==== helpers ====
67
+ # Trim whitespace from $1.
68
+ trim() {
69
+ echo "$1"
70
+ }
71
+ """
72
+ return "bash", src, "trim", "Trim whitespace"
73
+
74
+
75
+ @case(tags=["undocumented", "no_docs"])
76
+ def case_bash_fn_without_doc() -> DocstringCase:
77
+ """Undocumented shell function."""
78
+ src = """\
79
+ bare() {
80
+ echo hi
81
+ }
82
+ """
83
+ return "bash", src, "bare", "PHANTOM_DOC_TEXT_SHOULD_NEVER_APPEAR"
84
+
85
+
86
+ @case(tags=["undocumented", "boundary_not_attached"])
87
+ def case_bash_doc_detached_by_blank_line() -> DocstringCase:
88
+ """Blank line breaks attachment."""
89
+ src = """\
90
+ # Orphan.
91
+
92
+ later() {
93
+ echo hi
94
+ }
95
+ """
96
+ return "bash", src, "later", "Orphan"
97
+
98
+
99
+ @case(tags=["undocumented", "invalid"])
100
+ def case_bash_shebang_separated_by_blank_line() -> DocstringCase:
101
+ """When a blank line separates the shebang from the first
102
+ function, the shebang stays detached — the same
103
+ blank-line rule applies uniformly to all comment kinds.
104
+ """
105
+ src = """\
106
+ #!/bin/bash
107
+
108
+ main() {
109
+ echo hi
110
+ }
111
+ """
112
+ return "bash", src, "main", "#!/bin/bash"
@@ -0,0 +1,138 @@
1
+ """Bash extraction test cases.
2
+
3
+ Each `@case` returns test data consumed by `test_extraction.py` via
4
+ `pytest-cases`.
5
+ """
6
+
7
+ from __future__ import annotations
8
+
9
+ from pytest_cases import case
10
+
11
+ type SymbolCase = tuple[str, str, list[tuple[str, str, str]]]
12
+ type MixedCase = tuple[str, str, set[str], list[tuple[str, str]]]
13
+
14
+
15
+ @case(tags=["symbol"])
16
+ def case_bash_function_keyword() -> SymbolCase:
17
+ """function deploy { ... }."""
18
+ src = """\
19
+ function deploy {
20
+ echo deploying
21
+ }
22
+ """
23
+ return "bash", src, [("function", "deploy", "")]
24
+
25
+
26
+ @case(tags=["symbol"])
27
+ def case_bash_function_keyword_parens() -> SymbolCase:
28
+ """function deploy() { ... }."""
29
+ src = """\
30
+ function deploy() {
31
+ echo deploying
32
+ }
33
+ """
34
+ return "bash", src, [("function", "deploy", "")]
35
+
36
+
37
+ @case(tags=["symbol"])
38
+ def case_bash_function_posix() -> SymbolCase:
39
+ """deploy() { ... } — POSIX syntax."""
40
+ src = """\
41
+ deploy() {
42
+ echo deploying
43
+ }
44
+ """
45
+ return "bash", src, [("function", "deploy", "")]
46
+
47
+
48
+ @case(tags=["symbol"])
49
+ def case_bash_multiple_functions() -> SymbolCase:
50
+ """Multiple shell functions."""
51
+ src = """\
52
+ function setup {
53
+ echo setup
54
+ }
55
+
56
+ function teardown {
57
+ echo teardown
58
+ }
59
+
60
+ run() {
61
+ echo run
62
+ }
63
+ """
64
+ return (
65
+ "bash",
66
+ src,
67
+ [
68
+ ("function", "setup", ""),
69
+ ("function", "teardown", ""),
70
+ ("function", "run", ""),
71
+ ],
72
+ )
73
+
74
+
75
+ @case(tags=["symbol"])
76
+ def case_bash_alias() -> SymbolCase:
77
+ """An alias is a variable, named without the `=` the grammar fuses on."""
78
+ src = """\
79
+ alias ll="ls -l"
80
+ """
81
+ return "bash", src, [("variable", "ll", "")]
82
+
83
+
84
+ @case(tags=["symbol"])
85
+ def case_bash_function_local_vars() -> SymbolCase:
86
+ """Function with local variables."""
87
+ src = """\
88
+ setup() {
89
+ local dir="/tmp"
90
+ mkdir -p "$dir"
91
+ }
92
+ """
93
+ return "bash", src, [("function", "setup", "")]
94
+
95
+
96
+ @case(tags=["symbol"])
97
+ def case_bash_function_conditionals() -> SymbolCase:
98
+ """Function with conditionals."""
99
+ src = """\
100
+ check() {
101
+ if [ -f /tmp/x ]; then
102
+ echo yes
103
+ fi
104
+ }
105
+ """
106
+ return "bash", src, [("function", "check", "")]
107
+
108
+
109
+ # ── Mixed ───────────────────────────────────────────────────────────
110
+
111
+
112
+ @case(tags=["symbol"])
113
+ def case_bash_assignment() -> SymbolCase:
114
+ """Top-level variable assignment."""
115
+ return "bash", "MAX=100\n", [("variable", "MAX", "")]
116
+
117
+
118
+ @case(tags=["mixed"])
119
+ def case_bash_full_script() -> MixedCase:
120
+ """Realistic shell script with doc comments on every
121
+ function. Expected-kinds tuple pins symbol extraction;
122
+ content invariants are in `test_docstrings.py`.
123
+ """
124
+ src = """\
125
+ #!/bin/bash
126
+
127
+ # Deploy the current build to the given environment.
128
+ deploy() {
129
+ local env="$1"
130
+ echo "deploying to $env"
131
+ }
132
+
133
+ # Roll back the last deploy.
134
+ rollback() {
135
+ echo "rolling back"
136
+ }
137
+ """
138
+ return "bash", src, {"function"}, []
@@ -0,0 +1,34 @@
1
+ #!/usr/bin/env bash
2
+ # Greeter — print greetings for named recipients.
3
+ #
4
+ # Bash has no classes or methods; the plugin extracts functions,
5
+ # top-level variable assignments (including export/declare/readonly),
6
+ # aliases, and source/. imports.
7
+
8
+ source ./lib/colours.sh
9
+ . /etc/greeter.conf
10
+
11
+ DEFAULT_GREETING="Hello" # trailing comment: its own chunk
12
+
13
+ # Standalone note, separated by blank lines from any definition.
14
+ # Second line of the same block.
15
+ LOCALE="${LANG:-en}"
16
+ export API_URL="https://example.com"
17
+ readonly MAX_RETRIES=3
18
+ declare -i COUNTER=0
19
+ alias greet="format_greeting"
20
+
21
+ # Format a greeting for a single recipient.
22
+ format_greeting() {
23
+ local name="$1"
24
+ echo "${DEFAULT_GREETING}, ${name} (${LOCALE})"
25
+ }
26
+
27
+ # Greet every argument in turn.
28
+ greet_all() {
29
+ for name in "$@"; do
30
+ format_greeting "$name"
31
+ done
32
+ }
33
+
34
+ greet_all "$@"
@@ -0,0 +1,4 @@
1
+ #!/usr/bin/env bash
2
+ # Terminal colour codes.
3
+ RED='\033[0;31m'
4
+ RESET='\033[0m'
@@ -0,0 +1,50 @@
1
+ """Bash docstring-extraction tests.
2
+
3
+ Bash docs are exterior (a leading `#` comment run); there are no
4
+ interior-doc cases, so only the documented / undocumented / exterior-doc
5
+ checks apply.
6
+ """
7
+
8
+ from __future__ import annotations
9
+
10
+ from pytest_cases import parametrize_with_cases
11
+
12
+ from rbtr.git import FileEntry
13
+ from rbtr.languages.extract import extract_file
14
+
15
+
16
+ @parametrize_with_cases(
17
+ "lang, source, name, snippet", cases=".cases_docstrings", has_tag="documented"
18
+ )
19
+ def test_documented_chunk_includes_doc_text(
20
+ lang: str, source: str, name: str, snippet: str
21
+ ) -> None:
22
+ """By default the chunk content carries the symbol's docs."""
23
+ chunks = extract_file(FileEntry("input", "sha1", source.encode()), lang)
24
+ chunk = next(c for c in chunks if c.name == name)
25
+ assert snippet in chunk.content, (
26
+ f"expected {snippet!r} in {lang}.{name} content; got:\n{chunk.content!r}"
27
+ )
28
+
29
+
30
+ @parametrize_with_cases(
31
+ "lang, source, name, snippet", cases=".cases_docstrings", has_tag="undocumented"
32
+ )
33
+ def test_no_phantom_documentation(lang: str, source: str, name: str, snippet: str) -> None:
34
+ """Symbols without documentation do not gain any in content."""
35
+ chunks = extract_file(FileEntry("input", "sha1", source.encode()), lang)
36
+ chunk = next(c for c in chunks if c.name == name)
37
+ assert snippet not in chunk.content, (
38
+ f"unexpected {snippet!r} in {lang}.{name} content; got:\n{chunk.content!r}"
39
+ )
40
+
41
+
42
+ @parametrize_with_cases(
43
+ "lang, source, name, snippet", cases=".cases_docstrings", has_tag="exterior_doc"
44
+ )
45
+ def test_leading_doc_folds_into_symbol(lang: str, source: str, name: str, snippet: str) -> None:
46
+ """A leading comment block folds into its symbol's chunk content."""
47
+ chunk = next(
48
+ c for c in extract_file(FileEntry("input", "sha1", source.encode()), lang) if c.name == name
49
+ )
50
+ assert snippet in chunk.content
@@ -0,0 +1,54 @@
1
+ """Bash extraction tests.
2
+
3
+ Construct/mixed cases (`cases_extraction.py`) drive the shared checks;
4
+ the function at the end pins bash's source/. import behaviour.
5
+ """
6
+
7
+ from __future__ import annotations
8
+
9
+ from pytest_cases import parametrize_with_cases
10
+
11
+ from rbtr.git import FileEntry
12
+ from rbtr.index.models import ChunkKind
13
+ from rbtr.languages.extract import extract_file
14
+
15
+
16
+ @parametrize_with_cases("lang, source, expected", cases=".cases_extraction", has_tag="symbol")
17
+ def test_extracts_expected_symbols(lang: str, source: str, expected: list) -> None:
18
+ """Each expected (kind, name, scope) tuple appears in the output."""
19
+ chunks = extract_file(FileEntry("input", "sha1", source.encode()), lang)
20
+ symbols = [(c.kind, c.name, c.scope) for c in chunks]
21
+ for exp in expected:
22
+ assert exp in symbols, f"expected {exp} not found in {symbols}"
23
+
24
+
25
+ @parametrize_with_cases(
26
+ "lang, source, expected_kinds, expected_methods", cases=".cases_extraction", has_tag="mixed"
27
+ )
28
+ def test_extracts_all_expected_kinds(
29
+ lang: str,
30
+ source: str,
31
+ expected_kinds: set[str],
32
+ expected_methods: list[tuple[str, str]],
33
+ ) -> None:
34
+ """Realistic source produces all expected chunk kinds and method scoping."""
35
+ chunks = extract_file(FileEntry("input", "sha1", source.encode()), lang)
36
+ kinds = {c.kind for c in chunks}
37
+ for kind in expected_kinds:
38
+ assert kind in kinds, f"expected kind {kind!r} not in {kinds}"
39
+ methods = [(c.name, c.scope) for c in chunks if c.kind == ChunkKind.METHOD]
40
+ for name, scope in expected_methods:
41
+ assert (name, scope) in methods, f"expected method ({name}, {scope}) not in {methods}"
42
+
43
+
44
+ def test_bash_source_and_dot_extracted_as_imports() -> None:
45
+ """source/. commands are captured as imports."""
46
+ src = """\
47
+ source ./env.sh
48
+ . /etc/profile
49
+ """
50
+ chunks = extract_file(FileEntry("input", "sha1", src.encode()), "bash")
51
+ imports = [c for c in chunks if c.kind == ChunkKind.IMPORT]
52
+ assert len(imports) == 2
53
+ modules = {c.metadata.module for c in imports}
54
+ assert modules == {"./env.sh", "/etc/profile"}
@@ -0,0 +1,82 @@
1
+ """Bash sample extraction: the `samples/bash/` project through the real pipeline.
2
+
3
+ The snapshots are the golden record of what bash extraction produces;
4
+ regenerate with `pytest --snapshot-update` after an intended change.
5
+ Engine-wide invariants (determinism, line numbers, syntax-error recovery)
6
+ are covered once in core, not re-run per language.
7
+ """
8
+
9
+ from __future__ import annotations
10
+
11
+ from pathlib import Path
12
+ from typing import TYPE_CHECKING
13
+
14
+ import pytest
15
+ from tree_sitter import Parser
16
+
17
+ from rbtr.git import FileEntry
18
+ from rbtr.index.models import Chunk, ChunkKind, Edge
19
+ from rbtr.languages.edges import build_resolution_map, infer_import_edges
20
+ from rbtr.languages.extract import extract_file
21
+ from rbtr.languages.manager import get_manager
22
+ from rbtr.testing import render_edges
23
+
24
+ if TYPE_CHECKING:
25
+ from syrupy.assertion import SnapshotAssertion
26
+
27
+
28
+ @pytest.fixture
29
+ def project() -> list[tuple[str, str]]:
30
+ """The `(relative path, text)` files of the `samples/bash/` project."""
31
+ root = Path(__file__).parent / "samples" / "bash"
32
+ return [
33
+ (str(p.relative_to(root)), p.read_text()) for p in sorted(root.rglob("*")) if p.is_file()
34
+ ]
35
+
36
+
37
+ @pytest.fixture
38
+ def chunks(project: list[tuple[str, str]]) -> list[Chunk]:
39
+ """Chunks from every project file, each via the real `extract_file`."""
40
+ manager = get_manager()
41
+ out: list[Chunk] = []
42
+ for path, text in project:
43
+ lang = manager.detect_language(path) or "bash"
44
+ out.extend(extract_file(FileEntry(path, "sha1", text.encode()), lang))
45
+ return out
46
+
47
+
48
+ @pytest.fixture
49
+ def edges(project: list[tuple[str, str]], chunks: list[Chunk]) -> list[Edge]:
50
+ """Import edges inferred across the project's files."""
51
+ manager = get_manager()
52
+ return infer_import_edges(chunks, {p for p, _ in project}, build_resolution_map(manager))
53
+
54
+
55
+ def test_emits_expected_kinds(chunks: list[Chunk]) -> None:
56
+ """The sample exercises bash's function, variable, and import chunks."""
57
+ kinds = {c.kind for c in chunks}
58
+ assert {
59
+ ChunkKind.FUNCTION,
60
+ ChunkKind.VARIABLE,
61
+ ChunkKind.IMPORT,
62
+ ChunkKind.COMMENT,
63
+ } <= kinds
64
+
65
+
66
+ def test_parses_cleanly(project: list[tuple[str, str]]) -> None:
67
+ """Every project file is valid source — no tree-sitter ERROR/MISSING nodes."""
68
+ manager = get_manager()
69
+ for path, text in project:
70
+ grammar = manager.grammar(manager.detect_language(path) or "bash")
71
+ assert grammar is not None
72
+ assert not Parser(grammar).parse(text.encode()).root_node.has_error, path
73
+
74
+
75
+ def test_extraction_matches_snapshot(chunks: list[Chunk], snapshot_json: SnapshotAssertion) -> None:
76
+ assert chunks == snapshot_json
77
+
78
+
79
+ def test_edges_match_snapshot(
80
+ chunks: list[Chunk], edges: list[Edge], snapshot_json: SnapshotAssertion
81
+ ) -> None:
82
+ assert render_edges(edges, chunks) == snapshot_json
@@ -0,0 +1,8 @@
1
+ Metadata-Version: 2.4
2
+ Name: rbtr-lang-bash
3
+ Version: 2026.7.0.dev0
4
+ Summary: rbtr — Bash language plugin
5
+ License-Expression: MIT
6
+ Requires-Dist: rbtr==2026.7.0.dev0
7
+ Requires-Dist: tree-sitter-bash
8
+ Requires-Python: >=3.13
@@ -0,0 +1,18 @@
1
+ rbtr_lang_bash/__init__.py,sha256=gIm_kg0Z44nZ487E3GfD3rnqm01SbLASuvJz9IMeURQ,36
2
+ rbtr_lang_bash/bash.scm,sha256=6z_T45eiyqNWD7VS42YV5zYeFGdT7TjdTuK40vjxp2Y,693
3
+ rbtr_lang_bash/plugin.py,sha256=ERZQS7oc8yeZxA9E-3lSY3bOh4RpjkjdUn4lnI7NeZM,1883
4
+ rbtr_lang_bash/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
5
+ rbtr_lang_bash/tests/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
6
+ rbtr_lang_bash/tests/__snapshots__/test_samples/test_edges_match_snapshot.json,sha256=T2W5YHLPuKqYLJZBcLQRHRhK2yfSMn-JCkg_U8H3JSU,147
7
+ rbtr_lang_bash/tests/__snapshots__/test_samples/test_extraction_matches_snapshot.json,sha256=MVRWRbVx4xr3GG8WANzvamnrMP2eAikAWdAbpdPVYI4,5869
8
+ rbtr_lang_bash/tests/cases_docstrings.py,sha256=LGUR8vywW6-VZ_ggGJ_sU3cwWQFhLN1zTXwjYbYbS-E,2783
9
+ rbtr_lang_bash/tests/cases_extraction.py,sha256=-GFzlKNMemgTCcZ6GVal05iXd1p_ppW74q9uG-pAivs,2977
10
+ rbtr_lang_bash/tests/samples/bash/bash.sh,sha256=nNGz3bDP5J8WztL2CUOzUEZZmSofbm4NldwxRzatCSc,848
11
+ rbtr_lang_bash/tests/samples/bash/lib/colours.sh,sha256=IA4qng78MNbspyC9-_hv1OB0uLFDqn0h4z63JhWZI7Y,78
12
+ rbtr_lang_bash/tests/test_docstrings.py,sha256=h-KpUyUEa5mYldkaihXNMHrPcpvnfdMLbMqDIEEkKxo,1885
13
+ rbtr_lang_bash/tests/test_extraction.py,sha256=IM6nsVCwWLr90abB2pX8gIlfumo-e2U1MNVQRtTWZwQ,2089
14
+ rbtr_lang_bash/tests/test_samples.py,sha256=NopyYvZ-ii6HleZFbqXadw0LV406DOStB3uemYvRQCQ,2855
15
+ rbtr_lang_bash-2026.7.0.dev0.dist-info/WHEEL,sha256=CoDSoyhtC_eO_tlxRYzsTraPv1fPJRXFx91k6ISeAvA,81
16
+ rbtr_lang_bash-2026.7.0.dev0.dist-info/entry_points.txt,sha256=wBTXG4ULNpbpy6Ab2s3Xn3gkLpbxle5HHhP_57BwHx0,52
17
+ rbtr_lang_bash-2026.7.0.dev0.dist-info/METADATA,sha256=yQGc4z4irpb2YCfLiYjW6IzLsgbnI_r_mNOMMBNjaR0,220
18
+ rbtr_lang_bash-2026.7.0.dev0.dist-info/RECORD,,
@@ -0,0 +1,4 @@
1
+ Wheel-Version: 1.0
2
+ Generator: uv 0.11.28
3
+ Root-Is-Purelib: true
4
+ Tag: py3-none-any
@@ -0,0 +1,3 @@
1
+ [rbtr.languages]
2
+ bash = rbtr_lang_bash.plugin:bash
3
+