Fix SPDX and other output plugins

Modify and update spdx and other output plugins to use LicenseDetection output data format for licenses. Signed-off-by: Ayan Sinha Mahapatra <[email protected]>
aboutcode-org · Jun 1, 2022 · e68f0de · e68f0de
1 parent 5826429
commit e68f0de
Show file tree

Hide file tree

Showing 9 changed files with 2,620 additions and 1,343 deletions.
diff --git a/src/formattedcode/output_spdx.py b/src/formattedcode/output_spdx.py
@@ -29,6 +29,7 @@
 from commoncode.fileutils import parent_directory
 from commoncode.text import python_safe_name
 from formattedcode import FileOptionType
+from licensedcode.detection import get_matches_from_detections
 from plugincode.output import output_impl
 from plugincode.output import OutputPlugin
 import scancode_config
@@ -276,42 +277,45 @@ def write_spdx(
             chk_sum=Algorithm('SHA1', file_data.get('sha1') or '')
         )
 
-        file_licenses = file_data.get('licenses')
-        if file_licenses:
+        file_license_detections = file_data.get('licenses')
+        license_matches = get_matches_from_detections(file_license_detections)
+        if license_matches:
             all_files_have_no_license = False
-            for file_license in file_licenses:
-                license_key = file_license.get('key')
-
-                spdx_id = file_license.get('spdx_license_key')
-                if not spdx_id:
-                    spdx_id = f'LicenseRef-scancode-{license_key}'
-                is_license_ref = spdx_id.lower().startswith('licenseref-')
-
-                if not is_license_ref:
-                    spdx_license = License.from_identifier(spdx_id)
-                else:
-                    spdx_license = ExtractedLicense(spdx_id)
-                    spdx_license.name = file_license.get('short_name')
-                    # FIXME: replace this with the licensedb URL
-                    comment = (
-                        f'See details at https://github.com/nexB/scancode-toolkit'
-                        f'/blob/develop/src/licensedcode/data/licenses/{license_key}.yml\n'
-                    )
-                    spdx_license.comment = comment
-                    text = file_license.get('matched_text')
-                    # always set some text, even if we did not extract the
-                    # matched text
-                    if not text:
-                        text = comment
-                    spdx_license.text = text
-                    doc.add_extr_lic(spdx_license)
-
-                # Add licenses in the order they appear in the file. Maintaining
-                # the order might be useful for provenance purposes.
-                file_entry.add_lics(spdx_license)
-                package.add_lics_from_file(spdx_license)
-
-        elif file_licenses is None:
+            for match in license_matches:
+                file_licenses = match["licenses"]
+                for file_license in file_licenses:
+                    license_key = file_license.get('key')
+
+                    spdx_id = file_license.get('spdx_license_key')
+                    if not spdx_id:
+                        spdx_id = f'LicenseRef-scancode-{license_key}'
+                    is_license_ref = spdx_id.lower().startswith('licenseref-')
+
+                    if not is_license_ref:
+                        spdx_license = License.from_identifier(spdx_id)
+                    else:
+                        spdx_license = ExtractedLicense(spdx_id)
+                        spdx_license.name = file_license.get('short_name')
+                        # FIXME: replace this with the licensedb URL
+                        comment = (
+                            f'See details at https://github.com/nexB/scancode-toolkit'
+                            f'/blob/develop/src/licensedcode/data/licenses/{license_key}.yml\n'
+                        )
+                        spdx_license.comment = comment
+                        text = match.get('matched_text')
+                        # always set some text, even if we did not extract the
+                        # matched text
+                        if not text:
+                            text = comment
+                        spdx_license.text = text
+                        doc.add_extr_lic(spdx_license)
+
+                    # Add licenses in the order they appear in the file. Maintaining
+                    # the order might be useful for provenance purposes.
+                    file_entry.add_lics(spdx_license)
+                    package.add_lics_from_file(spdx_license)
+
+        elif license_matches is None:
             all_files_have_no_license = False
             file_entry.add_lics(NoAssert())
 

diff --git a/src/licensedcode/detection.py b/src/licensedcode/detection.py
@@ -753,6 +753,9 @@ def get_matches_from_detections(license_detections):
     `license_detections` list of LicenseDetection dicts.
     """
     license_matches = []
+    if not license_detections:
+        return license_matches
+
     for detection in license_detections:
         license_matches.extend(detection["matches"])