Add is_notice flag to the --classify option, issue aboutcode-org#3822

Signed-off-by: Aayush Kumar <[email protected]>
aayushkdev · Feb 11, 2025 · d705e5b · d705e5b
1 parent e795bc6
commit d705e5b
Show file tree

Hide file tree

Showing 19 changed files with 64 additions and 2 deletions.
diff --git a/src/summarycode/classify.py b/src/summarycode/classify.py
@@ -91,6 +91,9 @@ def get_relative_path(root_path, path):
     'readme',
 )
 
+NOTICE_STARTS_ENDS = (
+    "notice",
+)
 
 def check_resource_name_start_and_end(resource, STARTS_ENDS):
     """
@@ -111,6 +114,7 @@ def set_classification_flags(resource,
     _LEGAL=LEGAL_STARTS_ENDS,
     _MANIF=MANIFEST_ENDS,
     _README=README_STARTS_ENDS,
+    _NOTICE=NOTICE_STARTS_ENDS,
 ):
     """
     Set classification flags on the `resource` Resource
@@ -119,8 +123,9 @@ def set_classification_flags(resource,
 
     resource.is_legal = is_legal = check_resource_name_start_and_end(resource, _LEGAL)
     resource.is_readme = is_readme = check_resource_name_start_and_end(resource, _README)
+    resource.is_notice = is_notice = check_resource_name_start_and_end(resource, _NOTICE)
     # FIXME: this will never be picked up as this is NOT available in a pre-scan plugin
     has_package_data = bool(getattr(resource, 'package_data', False))
     resource.is_manifest = is_manifest = path.endswith(_MANIF) or has_package_data
-    resource.is_key_file = (resource.is_top_level and (is_readme or is_legal or is_manifest))
+    resource.is_key_file = (resource.is_top_level and (is_readme or is_legal or is_manifest or is_notice))
     return resource
diff --git a/src/summarycode/classify_plugin.py b/src/summarycode/classify_plugin.py
@@ -75,7 +75,10 @@ class FileClassifier(PostScanPlugin):
 
         ('is_key_file',
          Boolean(help='True if this file is "top-level" file and either a '
-                      'legal, readme or manifest file.')),
+                      'legal, notice, readme or manifest file.')),
+
+        ('is_notice',
+        Boolean(help='True if this file is likely a notice file')),
 
 #         ('is_doc',
 #          Boolean(help='True if this file is likely a documentation file.')),

diff --git a/tests/formattedcode/data/common/manifests-expected.yaml b/tests/formattedcode/data/common/manifests-expected.yaml
@@ -1934,6 +1934,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -1974,6 +1975,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -2092,6 +2094,7 @@ files:
     is_readme: no
     is_top_level: no
     is_key_file: no
+    is_notice: no
     detected_license_expression: cddl-1.0
     detected_license_expression_spdx: CDDL-1.0
     license_detections:
@@ -2173,6 +2176,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -2390,6 +2394,7 @@ files:
     is_readme: no
     is_top_level: no
     is_key_file: no
+    is_notice: no
     detected_license_expression: apache-2.0
     detected_license_expression_spdx: Apache-2.0
     license_detections:
@@ -2481,6 +2486,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -2631,6 +2637,7 @@ files:
     is_readme: no
     is_top_level: no
     is_key_file: no
+    is_notice: no
     detected_license_expression: mit
     detected_license_expression_spdx: MIT
     license_detections:
@@ -2702,6 +2709,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -2935,6 +2943,7 @@ files:
     is_readme: no
     is_top_level: no
     is_key_file: no
+    is_notice: no
     detected_license_expression: lgpl-3.0
     detected_license_expression_spdx: LGPL-3.0-only
     license_detections:

diff --git a/tests/formattedcode/data/yaml/package-and-licenses-expected.yaml b/tests/formattedcode/data/yaml/package-and-licenses-expected.yaml
@@ -1013,6 +1013,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: no
+    is_notice: no
     detected_license_expression:
     detected_license_expression_spdx:
     license_detections: []
@@ -1053,6 +1054,7 @@ files:
     is_readme: yes
     is_top_level: yes
     is_key_file: yes
+    is_notice: no
     detected_license_expression: apache-2.0 AND (apache-2.0 OR mit)
     detected_license_expression_spdx: Apache-2.0 AND (Apache-2.0 OR MIT)
     license_detections:
@@ -1129,6 +1131,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: yes
+    is_notice: no
     detected_license_expression: apache-2.0
     detected_license_expression_spdx: Apache-2.0
     license_detections:
@@ -1338,6 +1341,7 @@ files:
     is_readme: no
     is_top_level: yes
     is_key_file: yes
+    is_notice: no
     detected_license_expression: mit
     detected_license_expression_spdx: MIT
     license_detections:
@@ -1477,6 +1481,7 @@ files:
     is_readme: no
     is_top_level: no
     is_key_file: no
+    is_notice: no
     detected_license_expression: apache-2.0
     detected_license_expression_spdx: Apache-2.0
     license_detections:

diff --git a/tests/packagedcode/data/license_detection/multi-flavor/jquery-form-3.51.0.expected.json b/tests/packagedcode/data/license_detection/multi-flavor/jquery-form-3.51.0.expected.json
@@ -705,6 +705,7 @@
       "is_readme": true,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "detected_license_expression": "mit OR gpl-2.0",
       "detected_license_expression_spdx": "MIT OR GPL-2.0-only",
       "license_detections": [
@@ -832,6 +833,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "detected_license_expression": null,
       "detected_license_expression_spdx": null,
       "license_detections": [],
@@ -969,6 +971,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "detected_license_expression": "(mit OR gpl-2.0-plus) AND gpl-2.0",
       "detected_license_expression_spdx": "(MIT OR GPL-2.0-or-later) AND GPL-2.0-only",
       "license_detections": [
@@ -1242,6 +1245,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "detected_license_expression": "mit OR gpl-2.0",
       "detected_license_expression_spdx": "MIT OR GPL-2.0-only",
       "license_detections": [

diff --git a/tests/summarycode/data/classify/cli.expected.json b/tests/summarycode/data/classify/cli.expected.json
@@ -24,6 +24,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 8,
       "dirs_count": 1,
       "size_count": 0,
@@ -53,6 +54,7 @@
       "is_readme": true,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -82,6 +84,7 @@
       "is_readme": true,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -111,6 +114,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -140,6 +144,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -169,6 +174,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 0,
       "size_count": 0,
@@ -198,6 +204,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -227,6 +234,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -256,6 +264,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -285,6 +294,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,

diff --git a/tests/summarycode/data/classify/notice/ATTRIBUTION_NOTICE.txt b/tests/summarycode/data/classify/notice/ATTRIBUTION_NOTICE.txt
diff --git a/tests/summarycode/data/classify/notice/LICENSE-NOTICE.txt b/tests/summarycode/data/classify/notice/LICENSE-NOTICE.txt
diff --git a/tests/summarycode/data/classify/notice/NOTICE b/tests/summarycode/data/classify/notice/NOTICE
diff --git a/tests/summarycode/data/classify/notice/NOTICE-APACHE.txt b/tests/summarycode/data/classify/notice/NOTICE-APACHE.txt
diff --git a/tests/summarycode/data/classify/notice/NOTICE-LEGAL.TXT b/tests/summarycode/data/classify/notice/NOTICE-LEGAL.TXT
diff --git a/tests/summarycode/data/classify/notice/NOTICE-LEGAL.txt b/tests/summarycode/data/classify/notice/NOTICE-LEGAL.txt
diff --git a/tests/summarycode/data/classify/notice/NOTICE-README.txt b/tests/summarycode/data/classify/notice/NOTICE-README.txt
diff --git a/tests/summarycode/data/classify/notice/NOTICE.md b/tests/summarycode/data/classify/notice/NOTICE.md
diff --git a/tests/summarycode/data/classify/notice/NOTICE.txt b/tests/summarycode/data/classify/notice/NOTICE.txt
diff --git a/tests/summarycode/data/classify/notice/PATENTS-NOTICE.txt b/tests/summarycode/data/classify/notice/PATENTS-NOTICE.txt
diff --git a/tests/summarycode/data/classify/notice/THIRD_PARTY_NOTICE.txt b/tests/summarycode/data/classify/notice/THIRD_PARTY_NOTICE.txt
diff --git a/tests/summarycode/data/classify/with_package_data.expected.json b/tests/summarycode/data/classify/with_package_data.expected.json
@@ -255,6 +255,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 7,
       "dirs_count": 7,
       "size_count": 31701,
@@ -288,6 +289,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 5,
       "dirs_count": 3,
       "size_count": 27441,
@@ -321,6 +323,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -354,6 +357,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -437,6 +441,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -470,6 +475,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 2,
       "size_count": 15229,
@@ -503,6 +509,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 1,
       "size_count": 15229,
@@ -536,6 +543,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 0,
       "size_count": 15229,
@@ -614,6 +622,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -832,6 +841,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": true,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -865,6 +875,7 @@
       "is_readme": false,
       "is_top_level": true,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 2,
       "size_count": 4260,
@@ -898,6 +909,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 1,
       "size_count": 4260,
@@ -931,6 +943,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 2,
       "dirs_count": 0,
       "size_count": 4260,
@@ -964,6 +977,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,
@@ -997,6 +1011,7 @@
       "is_readme": false,
       "is_top_level": false,
       "is_key_file": false,
+      "is_notice": false,
       "files_count": 0,
       "dirs_count": 0,
       "size_count": 0,