Improved shed linting and type default.

jmchilton · jmchilton · commit 36ac6d8e6eaa · 2015-04-22T13:00:41.000-04:00
No .shed.yml fields are required anymore, but when present more are validated more carefully.

 - Validate owner string matches validation inforced by Tool Shed.
 - Validate name string matches validation inforced by Tool Shed.
 - Validate type (if specified) is valid.
 - Validate name matches type conventions.

Use new validation to more confidently select a default type based on name during creation.
diff --git a/planemo/shed.py b/planemo/shed.py
@@ -128,7 +128,7 @@ def create_repository(ctx, tsi, path, **kwds):
 
     description = repo_config.get("description", None)
     long_description = repo_config.get("long_description", None)
-    type = repo_config.get("type", "unrestricted")
+    type = repo_config.get("type", None)
     remote_repository_url = repo_config.get("remote_repository_url", None)
     homepage_url = repo_config.get("homepage_url", None)
     categories = repo_config.get("categories", [])
@@ -137,6 +137,13 @@ def create_repository(ctx, tsi, path, **kwds):
     if name is None:
         name = os.path.basename(os.path.abspath(path))
 
+    if type is None and name.startswith("package_"):
+        type = "tool_dependency_definition"
+    elif type is None and name.startswith("suite_"):
+        type = "repository_suite_definition"
+    elif type is None:
+        type = "unrestricted"
+
     # description is required, as is name.
     if description is None:
         message = "description required for automatic creation of repositories"
diff --git a/planemo/shed_lint.py b/planemo/shed_lint.py
@@ -1,4 +1,5 @@
 import os
+import re
 import yaml
 from galaxy.tools.lint import LintContext
 from planemo.lint import lint_xsd
@@ -17,6 +18,16 @@
 TOOL_DEPENDENCIES_XSD = os.path.join(XSDS_PATH, "tool_dependencies.xsd")
 REPO_DEPENDENCIES_XSD = os.path.join(XSDS_PATH, "repository_dependencies.xsd")
 
+# TODO: sync this with tool shed impl someday
+VALID_REPOSITORYNAME_RE = re.compile("^[a-z0-9\_]+$")
+VALID_PUBLICNAME_RE = re.compile("^[a-z0-9\-]+$")
+
+VALID_REPOSITORY_TYPES = [
+    "unrestricted",
+    "tool_dependency_definition",
+    "repository_suite_definition",
+]
+
 
 def lint_repository(ctx, path, **kwds):
     info("Linting repository %s" % path)
@@ -76,12 +87,60 @@ def lint_shed_yaml(path, lint_ctx):
         shed_contents = yaml.load(open(shed_yaml, "r"))
     except Exception as e:
         lint_ctx.warn("Failed to parse .shed.yml file [%s]" % str(e))
+    lint_ctx.info(".shed.yml found and appears to be valid YAML.")
+    _lint_shed_contents(lint_ctx, path, shed_contents)
+
+
+def _lint_shed_contents(lint_ctx, path, shed_contents):
+    def _lint_if_present(key, func, *args):
+        value = shed_contents.get(key, None)
+        if value is not None:
+            msg = func(value, *args)
+            if msg:
+                lint_ctx.warn(msg)
+
+    _lint_if_present("owner", _validate_repo_owner)
+    _lint_if_present("name", _validate_repo_name)
+    effective_name = shed_contents.get("name", None) or os.path.basename(path)
+    _lint_if_present("type", _validate_repo_type, effective_name)
+
+
+def _validate_repo_type(repo_type, name):
+    if repo_type not in VALID_REPOSITORY_TYPES:
+        return "Invalid repository type specified [%s]" % repo_type
+
+    is_dep = repo_type == "tool_dependency_definition"
+    is_suite = repo_type == "repository_suite_definition"
+    if is_dep and not name.startswith("package_"):
+        return ("Tool dependency definition repositories should have names "
+                "starting with package_")
+    if is_suite and not name.startswith("suite_"):
+        return ("Repository suite definition repositories should have names "
+                "starting with suite_")
+    if name.startswith("package_") or name.startswith("suite_"):
+        if repo_type == "unrestricted":
+            return ("Repository name indicated specialized repository type "
+                    "but repository is listed as unrestricted.")
+
+
+def _validate_repo_name(name):
+    msg = None
+    if len(name) < 2:
+        msg = "Repository names must be at least 2 characters in length."
+    if len(name) > 80:
+        msg = "Repository names cannot be more than 80 characters in length."
+    if not(VALID_REPOSITORYNAME_RE.match(name)):
+        msg = ("Repository names must contain only lower-case letters, "
+               "numbers and underscore.")
+    return msg
 
-    warned = False
-    for required_key in ["owner", "name"]:
-        if required_key not in shed_contents:
-            lint_ctx.warn(".shed.yml did not contain key [%s]" % required_key)
-            warned = True
 
-    if not warned:
-        lint_ctx.info(".shed.yml found and appears to be valid YAML.")
+def _validate_repo_owner(owner):
+    msg = None
+    if len(owner) < 3:
+        msg = "Owner must be at least 3 characters in length"
+    if len(owner) > 255:
+        msg = "Owner cannot be more than 255 characters in length"
+    if not(VALID_PUBLICNAME_RE.match(owner)):
+        msg = "Owner must contain only lower-case letters, numbers and '-'"
+    return msg
