Better cleanup package manifest fields

platformio/package/manifest/parser.py

@@ -334,7 +334,7 @@ class LibraryJsonManifestParser(BaseManifestParser):
         if "authors" in data:
             data["authors"] = self._parse_authors(data["authors"])
         if "platforms" in data:
-            data["platforms"] = self._parse_platforms(data["platforms"]) or None
+            data["platforms"] = self._fix_platforms(data["platforms"]) or None
         if "export" in data:
             data["export"] = self._parse_export(data["export"])
         if "dependencies" in data:
@@ -367,16 +367,11 @@ class LibraryJsonManifestParser(BaseManifestParser):
         return [self.cleanup_author(author) for author in raw]
 
     @staticmethod
-    def _parse_platforms(raw):
+    def _fix_platforms(items):
-        assert isinstance(raw, list)
+        assert isinstance(items, list)
-        result = []
+        if "espressif" in items:
-        # renamed platforms
+            items[items.index("espressif")] = "espressif8266"
-        for item in raw:
+        return items
-            if item == "espressif":
-                item = "espressif8266"
-            if item not in result:
-                result.append(item)
-        return result
 
     @staticmethod
     def _parse_export(raw):
@@ -485,7 +480,7 @@ class LibraryPropertiesManifestParser(BaseManifestParser):
                 repository=repository or None,
                 description=self._parse_description(data),
                 platforms=self._parse_platforms(data) or ["*"],
-                keywords=self._parse_keywords(data),
+                keywords=self._parse_keywords(data) or None,
                 export=self._parse_export(),
             )
         )
@@ -527,18 +522,17 @@ class LibraryPropertiesManifestParser(BaseManifestParser):
             lines[0] += "."
         return " ".join(lines)
 
-    @staticmethod
+    def _parse_keywords(self, properties):
-    def _parse_keywords(properties):
+        return self.str_to_list(
-        result = []
+            re.split(
-        for item in re.split(r"[\s/]+", properties.get("category", "uncategorized")):
+                r"[\s/]+",
-            item = item.strip()
+                properties.get("category", ""),
-            if not item:
+            ),
-                continue
+            lowercase=True,
-            result.append(item.lower())
+            unique=True,
-        return result
+        )
 
-    @staticmethod
+    def _parse_platforms(self, properties):
-    def _parse_platforms(properties):
         result = []
         platforms_map = {
             "avr": "atmelavr",
@@ -557,9 +551,9 @@ class LibraryPropertiesManifestParser(BaseManifestParser):
                 continue
             if arch == "*":
                 return ["*"]
-            if arch in platforms_map and platforms_map[arch] not in result:
+            if arch in platforms_map:
                 result.append(platforms_map[arch])
-        return result
+        return self.str_to_list(result, lowercase=True, unique=True)
 
     def _parse_authors(self, properties):
         if "author" not in properties:
@@ -655,19 +649,17 @@ class PlatformJsonManifestParser(BaseManifestParser):
     def parse(self, contents):
         data = json.loads(contents)
         if "keywords" in data:
-            data["keywords"] = self.str_to_list(data["keywords"], sep=",", unique=True)
+            data["keywords"] = self.str_to_list(
-        if "frameworks" in data:
+                data["keywords"], sep=",", lowercase=True, unique=True
-            data["frameworks"] = self._parse_frameworks(data["frameworks"])
+            )
+        if "frameworks" in data and not isinstance(data["frameworks"], dict):
+            data["frameworks"] = self.str_to_list(
+                data["frameworks"].keys(), lowercase=True, unique=True
+            )
         if "packages" in data:
             data["dependencies"] = self._parse_dependencies(data["packages"])
         return data
 
-    @staticmethod
-    def _parse_frameworks(raw):
-        if not isinstance(raw, dict):
-            return None
-        return [name.lower() for name in raw.keys()]
-
     @staticmethod
     def _parse_dependencies(raw):
         result = []
@@ -694,16 +686,13 @@ class PackageJsonManifestParser(BaseManifestParser):
         data = self._parse_repository(data)
         return data
 
-    @staticmethod
+    def _parse_system(self, data):
-    def _parse_system(data):
         if "system" not in data:
             return data
         if data["system"] in ("*", ["*"], "all"):
             del data["system"]
             return data
-        if not isinstance(data["system"], list):
+        data["system"] = self.str_to_list(data["system"], lowercase=True, unique=True)
-            data["system"] = [data["system"]]
-        data["system"] = [s.strip().lower() for s in data["system"]]
         return data
 
     @staticmethod

tests/package/test_manager.py

@@ -202,7 +202,7 @@ def test_install_from_registry(isolated_pio_core, tmpdir_factory):
     lm.install("AsyncMqttClient-esphome @ 0.8.4", silent=True)
     assert len(lm.get_installed()) == 3
     pkg = lm.get_package("AsyncTCP-esphome")
-    assert pkg.metadata.spec.owner == "ottowinter"
+    assert pkg.metadata.spec.owner == "esphome"
     assert not lm.get_package("non-existing-package")
     # mbed library
     assert lm.install("wolfSSL", silent=True)

tests/package/test_manifest.py

@@ -206,6 +206,7 @@ version=1.2.3
 author=SomeAuthor <info AT author.com>, Maintainer Author (nickname) <www.example.com>
 maintainer=Maintainer Author (nickname) <www.example.com>
 sentence=This is Arduino library
+category=Signal Input/Output
 customField=Custom Value
 depends=First Library (=2.0.0), Second Library (>=1.2.0), Third
 ignore_empty_field=
@@ -229,7 +230,8 @@ includes=Arduino.h, Arduino Space.hpp
                 {"name": "SomeAuthor", "email": "info@author.com"},
                 {"name": "Maintainer Author", "maintainer": True},
             ],
-            "keywords": ["uncategorized"],
+            "category": "Signal Input/Output",
+            "keywords": ["signal", "input", "output"],
             "headers": ["Arduino.h", "Arduino Space.hpp"],
             "customField": "Custom Value",
             "depends": "First Library (=2.0.0), Second Library (>=1.2.0), Third",
@@ -297,6 +299,7 @@ maintainer=Rocket Scream Electronics
     assert data["authors"] == [
         {"name": "Rocket Scream Electronics", "maintainer": True}
     ]
+    assert "keywords" not in data
 
 
 def test_library_json_schema():
@@ -559,7 +562,7 @@ def test_platform_json_schema():
   "name": "atmelavr",
   "title": "Atmel AVR",
   "description": "Atmel AVR 8- and 32-bit MCUs deliver a unique combination of performance, power efficiency and design flexibility. Optimized to speed time to market-and easily adapt to new ones-they are based on the industrys most code-efficient architecture for C and assembly programming.",
-  "keywords": "arduino, atmel, avr",
+  "keywords": "arduino, atmel, avr, MCU",
   "homepage": "http://www.atmel.com/products/microcontrollers/avr/default.aspx",
   "license": "Apache-2.0",
   "engines": {
@@ -619,7 +622,7 @@ def test_platform_json_schema():
                 "on the industrys most code-efficient architecture for C and "
                 "assembly programming."
             ),
-            "keywords": ["arduino", "atmel", "avr"],
+            "keywords": ["arduino", "atmel", "avr", "mcu"],
             "homepage": "http://www.atmel.com/products/microcontrollers/avr/default.aspx",
             "license": "Apache-2.0",
             "repository": {
@@ -648,7 +651,7 @@ def test_package_json_schema():
     "description": "SCons software construction tool",
     "keywords": "SCons, build",
     "homepage": "http://www.scons.org",
-    "system": ["linux_armv6l", "linux_armv7l", "linux_armv8l"],
+    "system": ["linux_armv6l", "linux_armv7l", "linux_armv8l", "LINUX_ARMV7L"],
     "version": "3.30101.0"
 }
 """