Add rule_url and update scancode URLs

* add a new attribute `rule_url` to LicenseMatch results.
* deleted License attributes 'scancode_data_url' and 'scancode_text_url' and
  replaced them with a single License attribute 'scancode_url'
* regen test expectations accordingly

Signed-off-by: Ayan Sinha Mahapatra <ayansmahapatra@gmail.com>

Author: AyanSinhaMahapatra

src/licensedcode/match.py

@@ -777,9 +777,10 @@ def get_mapping(
                 else:
                     matched_text = self.matched_text(whole_lines=False, highlight=False)
 
-        SCANCODE_BASE_URL = 'https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses'
-        SCANCODE_LICENSE_TEXT_URL = SCANCODE_BASE_URL + '/{}.LICENSE'
-        SCANCODE_LICENSE_DATA_URL = SCANCODE_BASE_URL + '/{}.yml'
+        SCANCODE_DATA_BASE_URL = 'https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data'
+        SCANCODE_LICENSE_URL = SCANCODE_DATA_BASE_URL + '/licenses/{}.LICENSE'
+        SCANCODE_LICENSE_RULE_URL = SCANCODE_DATA_BASE_URL + '/licenses/{}'
+        SCANCODE_RULE_URL = SCANCODE_DATA_BASE_URL + '/rules/{}'
 
         result = {}
 
@@ -794,6 +795,13 @@ def get_mapping(
         # LicenseDB Level Information (Rule that was matched)
         result['license_expression'] = self.rule.license_expression
         result['rule_identifier'] = self.rule.identifier
+        if self.matcher == "1-spdx-id":
+            result['rule_url'] = None
+        elif self.rule.is_from_license:
+            result['rule_url'] = SCANCODE_LICENSE_RULE_URL.format(self.rule.identifier)
+        else:
+            result['rule_url'] = SCANCODE_RULE_URL.format(self.rule.identifier)
+
         result['referenced_filenames'] = self.rule.referenced_filenames
         result['is_license_text'] = self.rule.is_license_text
         result['is_license_notice'] = self.rule.is_license_notice
@@ -823,22 +831,21 @@ def get_mapping(
             detected_license['homepage_url'] = lic.homepage_url
             detected_license['text_url'] = lic.text_urls[0] if lic.text_urls else ''
             detected_license['reference_url'] = license_url_template.format(lic.key)
-            detected_license['scancode_text_url'] = SCANCODE_LICENSE_TEXT_URL.format(lic.key)
-            detected_license['scancode_data_url'] = SCANCODE_LICENSE_DATA_URL.format(lic.key)
+            detected_license['scancode_url'] = SCANCODE_LICENSE_URL.format(lic.key)
 
             spdx_key = lic.spdx_license_key
             detected_license['spdx_license_key'] = spdx_key
 
             if spdx_key:
                 is_license_ref = spdx_key.lower().startswith('licenseref-')
                 if is_license_ref:
-                    spdx_url = SCANCODE_LICENSE_TEXT_URL.format(lic.key)
+                    spdx_url = SCANCODE_LICENSE_URL.format(lic.key)
                 else:
                     # TODO: Is this replacing spdx_key???
                     spdx_key = lic.spdx_license_key.rstrip('+')
                     spdx_url = spdx_license_url.format(spdx_key)
             else:
-                spdx_url = ''
+                spdx_url = None
             detected_license['spdx_url'] = spdx_url
 
         return result

tests/cluecode/data/plugin_filter_clues/filtered-expected.json

@@ -37,6 +37,7 @@
               "matcher": "3-seq",
               "license_expression": "apache-1.1",
               "rule_identifier": "apache-1.1_63.RULE",
+              "rule_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/rules/apache-1.1_63.RULE",
               "referenced_filenames": [],
               "is_license_text": true,
               "is_license_notice": false,
@@ -57,8 +58,7 @@
                   "homepage_url": "http://www.apache.org/licenses/",
                   "text_url": "http://apache.org/licenses/LICENSE-1.1",
                   "reference_url": "https://scancode-licensedb.aboutcode.org/apache-1.1",
-                  "scancode_text_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/apache-1.1.LICENSE",
-                  "scancode_data_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/apache-1.1.yml",
+                  "scancode_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/apache-1.1.LICENSE",
                   "spdx_license_key": "Apache-1.1",
                   "spdx_url": "https://spdx.org/licenses/Apache-1.1"
                 }

tests/cluecode/data/plugin_filter_clues/filtered-expected2.json

@@ -37,6 +37,7 @@
               "matcher": "2-aho",
               "license_expression": "pygres-2.2",
               "rule_identifier": "pygres-2.2_2.RULE",
+              "rule_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/rules/pygres-2.2_2.RULE",
               "referenced_filenames": [],
               "is_license_text": true,
               "is_license_notice": false,
@@ -57,8 +58,7 @@
                   "homepage_url": null,
                   "text_url": "http://shell.vex.net/viewvc.cgi/pygresql/trunk/module/pgmodule.c?view=markup&pathrev=431",
                   "reference_url": "https://scancode-licensedb.aboutcode.org/pygres-2.2",
-                  "scancode_text_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pygres-2.2.LICENSE",
-                  "scancode_data_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pygres-2.2.yml",
+                  "scancode_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pygres-2.2.LICENSE",
                   "spdx_license_key": "LicenseRef-scancode-pygres-2.2",
                   "spdx_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pygres-2.2.LICENSE"
                 }

tests/cluecode/data/plugin_filter_clues/filtered-expected3.json

@@ -37,6 +37,7 @@
               "matcher": "1-hash",
               "license_expression": "pcre",
               "rule_identifier": "pcre.LICENSE",
+              "rule_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pcre.LICENSE",
               "referenced_filenames": [],
               "is_license_text": true,
               "is_license_notice": false,
@@ -57,8 +58,7 @@
                   "homepage_url": "http://www.pcre.org/licence.txt",
                   "text_url": "http://www.pcre.org/licence.txt",
                   "reference_url": "https://scancode-licensedb.aboutcode.org/pcre",
-                  "scancode_text_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pcre.LICENSE",
-                  "scancode_data_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pcre.yml",
+                  "scancode_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pcre.LICENSE",
                   "spdx_license_key": "LicenseRef-scancode-pcre",
                   "spdx_url": "https://github.com/nexB/scancode-toolkit/tree/develop/src/licensedcode/data/licenses/pcre.LICENSE"
                 }