Improve how we detect license in Windows PEs

pombredanne · pombredanne · commit e2e82a5bc280 · 2022-10-02T16:18:10.000+02:00
If there is a license tag, use it for dteection, otherwise use all
declared licensing information.

Add new rules to improve license detection accuracy.

Signed-off-by: Philippe Ombredanne &lt;pombredanne@nexb.com&gt;
diff --git a/src/licensedcode/data/rules/cc0-1.0_176.yml b/src/licensedcode/data/rules/cc0-1.0_176.yml
@@ -1,3 +1,4 @@
 license_expression: cc0-1.0
 is_license_reference: yes
-relevance: 100
+relevance: 50
+
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_570.RULE b/src/licensedcode/data/rules/gpl-1.0-plus_570.RULE
@@ -0,0 +1,3 @@
+License
+This program is free software; you can redistribute it and/or modify it under 
+the terms of the GNU General Public License
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_570.yml b/src/licensedcode/data/rules/gpl-1.0-plus_570.yml
@@ -0,0 +1,2 @@
+license_expression: gpl-1.0-plus
+is_license_notice: yes
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_571.RULE b/src/licensedcode/data/rules/gpl-1.0-plus_571.RULE
@@ -0,0 +1,2 @@
+you can redistribute it and/or modify it under 
+the terms of the GNU General Public License;see www.gnu.org/copyleft/gpl.html.
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_571.yml b/src/licensedcode/data/rules/gpl-1.0-plus_571.yml
@@ -0,0 +1,4 @@
+license_expression: gpl-1.0-plus
+is_license_notice: yes
+ignorable_urls:
+    - http://www.gnu.org/copyleft/gpl.html
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_572.RULE b/src/licensedcode/data/rules/gpl-1.0-plus_572.RULE
@@ -0,0 +1,2 @@
+you can redistribute it and/or modify it under 
+the terms of the GNU General Public License
diff --git a/src/licensedcode/data/rules/gpl-1.0-plus_572.yml b/src/licensedcode/data/rules/gpl-1.0-plus_572.yml
@@ -0,0 +1,3 @@
+license_expression: gpl-1.0-plus
+is_license_notice: yes
+relevance: 100
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_454.RULE b/src/licensedcode/data/rules/lgpl-2.1-plus_454.RULE
@@ -0,0 +1,3 @@
+License
+This program is free software; you can redistribute it and/or modify it under 
+the terms of the GNU Lesser General Public License;see www.gnu.org/copyleft/lesser.html.
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_454.yml b/src/licensedcode/data/rules/lgpl-2.1-plus_454.yml
@@ -0,0 +1,4 @@
+license_expression: lgpl-2.1-plus
+is_license_notice: yes
+ignorable_urls:
+    - http://www.gnu.org/copyleft/lesser.html
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_455.RULE b/src/licensedcode/data/rules/lgpl-2.1-plus_455.RULE
@@ -0,0 +1,2 @@
+This program is free software; you can redistribute it and/or modify it under 
+the terms of the GNU Lesser General Public License;see www.gnu.org/copyleft/lesser.html.
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_455.yml b/src/licensedcode/data/rules/lgpl-2.1-plus_455.yml
@@ -0,0 +1,4 @@
+license_expression: lgpl-2.1-plus
+is_license_notice: yes
+ignorable_urls:
+    - http://www.gnu.org/copyleft/lesser.html
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_456.RULE b/src/licensedcode/data/rules/lgpl-2.1-plus_456.RULE
@@ -0,0 +1,2 @@
+This program is free software; you can redistribute it and/or modify it under 
+the terms of the GNU Lesser General Public License
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_456.yml b/src/licensedcode/data/rules/lgpl-2.1-plus_456.yml
@@ -0,0 +1,2 @@
+license_expression: lgpl-2.1-plus
+is_license_notice: yes
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_457.RULE b/src/licensedcode/data/rules/lgpl-2.1-plus_457.RULE
@@ -0,0 +1,2 @@
+you can redistribute it and/or modify it under 
+the terms of the GNU Lesser General Public License
diff --git a/src/licensedcode/data/rules/lgpl-2.1-plus_457.yml b/src/licensedcode/data/rules/lgpl-2.1-plus_457.yml
@@ -0,0 +1,2 @@
+license_expression: lgpl-2.1-plus
+is_license_notice: yes
diff --git a/src/licensedcode/data/rules/lgpl-2.1_153.RULE b/src/licensedcode/data/rules/lgpl-2.1_153.RULE
@@ -1,3 +1,3 @@
 This program is free software; you can redistribute it and/or modify
-it under the terms of the GNU Lesser General Public License         
-version 2.1 as published by the Free Software Foundation;           
+it under the terms of the {{GNU Lesser General Public License         
+version 2.1 as published by the Free Software Foundation; }}           
diff --git a/src/licensedcode/data/rules/proprietary-license_788.RULE b/src/licensedcode/data/rules/proprietary-license_788.RULE
diff --git a/src/licensedcode/data/rules/proprietary-license_788.yml b/src/licensedcode/data/rules/proprietary-license_788.yml
@@ -0,0 +1,2 @@
+license_expression: proprietary-license
+is_license_text: yes
diff --git a/src/packagedcode/win_pe.py b/src/packagedcode/win_pe.py
@@ -342,6 +342,10 @@ def parse(cls, location):
         )
 
         if not package_data.license_expression and package_data.declared_license:
-            package_data.license_expression = models.compute_normalized_license(package_data.declared_license)
+            if License:
+                lexp = models.compute_normalized_license(License)
+            else:
+                lexp = models.compute_normalized_license(package_data.declared_license)
+            package_data.license_expression = lexp
 
         yield package_data
diff --git a/tests/packagedcode/data/win_pe/libiconv2.dll.package-expected.json b/tests/packagedcode/data/win_pe/libiconv2.dll.package-expected.json
@@ -30,7 +30,7 @@
     "code_view_url": null,
     "vcs_url": null,
     "copyright": "\u00a9 2004 Free Software Foundation <www.fsf.org>",
-    "license_expression": "(lgpl-2.1 AND lgpl-2.1-plus) AND unknown",
+    "license_expression": "lgpl-2.1-plus",
     "declared_license": {
       "LegalCopyright": "\u00a9 2004 Free Software Foundation <www.fsf.org>",
       "LegalTrademarks": "GNU\u00ae, LibIconv\u00ae, libiconv2\u00ae",
diff --git a/tests/packagedcode/data/win_pe/libintl3.dll.package-expected.json b/tests/packagedcode/data/win_pe/libintl3.dll.package-expected.json
@@ -30,7 +30,7 @@
     "code_view_url": null,
     "vcs_url": null,
     "copyright": "\u00a9 2005 Free Software Foundation <www.fsf.org>",
-    "license_expression": "gpl-2.0-plus AND unknown",
+    "license_expression": "gpl-1.0-plus",
     "declared_license": {
       "LegalCopyright": "\u00a9 2005 Free Software Foundation <www.fsf.org>",
       "LegalTrademarks": "GNU\u00ae, GetText\u00ae, libintl3\u00ae",
diff --git a/tests/packagedcode/data/win_pe/tre4.dll.package-expected.json b/tests/packagedcode/data/win_pe/tre4.dll.package-expected.json
@@ -30,7 +30,7 @@
     "code_view_url": null,
     "vcs_url": null,
     "copyright": "\u00a9 2008 Ville Laurikari <vl@iki.fi>",
-    "license_expression": "(gpl-1.0-plus AND gpl-1.0-plus) AND unknown",
+    "license_expression": "gpl-1.0-plus",
     "declared_license": {
       "LegalCopyright": "\u00a9 2008 Ville Laurikari <vl@iki.fi>",
       "LegalTrademarks": "GnuWin32\u00ae, Tre\u00ae, tre4\u00ae",

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+License`
	`2`	`+This program is free software; you can redistribute it and/or modify it under`
	`3`	`+the terms of the GNU General Public License`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+license_expression: gpl-1.0-plus`
	`2`	`+is_license_notice: yes`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+you can redistribute it and/or modify it under`
	`2`	`+the terms of the GNU General Public License;see www.gnu.org/copyleft/gpl.html.`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,2 @@`
	`1`	`+This program is free software; you can redistribute it and/or modify it under`
	`2`	`+the terms of the GNU Lesser General Public License;see www.gnu.org/copyleft/lesser.html.`