speed up linting

tools/lint/lib/checks/frontmatter.py

@@ -25,7 +25,9 @@ def run(self, name, meta, source):
         for parsing_event in meta.parsing_events:
             if not isinstance(parsing_event, yaml.ScalarEvent):
                 continue
-            if parsing_event.style is not None:
+            # The pure-Python parser reports plain scalars with a style of
+            # `None`; the libyaml parser uses the empty string.
+            if parsing_event.style:
                 continue
             if parsing_event.start_mark.line != parsing_event.end_mark.line:
                 return 'YAML multiline scalar values in flow notation are disallowed (use "|" or ">")'

tools/lint/lib/frontmatter.py

@@ -1,23 +1,18 @@
 import re
 import yaml
 
+try:
+    # The libyaml-backed parser is roughly an order of magnitude faster than
+    # the pure-Python one, but it is not available in every environment.
+    from yaml import CSafeLoader as Loader
+except ImportError:
+    from yaml import SafeLoader as Loader
+
 class Result(dict):
     def __init__(self, meta, events):
         self.parsing_events = events
         super(Result, self).__init__(**meta)
 
-class MyLoader(yaml.SafeLoader):
-    events = None
-
-    def __init__(self, *args, **kwargs):
-        MyLoader.events = []
-        super(MyLoader, self).__init__(*args, **kwargs)
-
-    def get_event(self):
-        event = super(MyLoader, self).get_event()
-        MyLoader.events.append(event)
-        return event
-
 def parse(src):
     '''Parse the YAML-formatted metadata found in a given string of source
     code. Tolerate missing or invalid metadata; those conditions are handled by
@@ -27,8 +22,13 @@ def parse(src):
     if not match:
         return None
 
+    # NB: Call strip() to match parseTestRecord.
+    attrs = match.group(1).strip()
+
     try:
-        # NB: Call strip() to match parseTestRecord.
-        return Result(yaml.load(match.group(1).strip(), MyLoader), MyLoader.events)
+        # The event stream takes a second pass, since the
+        # libyaml parser cannot be instrumented from Python.
+        events = list(yaml.parse(attrs, Loader=Loader))
+        return Result(yaml.load(attrs, Loader), events)
     except (yaml.scanner.ScannerError, yaml.parser.ParserError):
         return None

tools/lint/lint.py

@@ -76,6 +76,10 @@ def checks(features):
 
 def lint(file_names, features):
     errors = dict()
+    # The checks are stateless with respect to the files they inspect, so they
+    # are created once and reused for every file (some read data files during
+    # construction, which is expensive to repeat).
+    all_checks = checks(features)
 
     for file_name in file_names:
         if not file_name.endswith((".js", ".json")):
@@ -84,7 +88,7 @@ def lint(file_names, features):
         with open(file_name, 'r') as f:
             content = f.read()
         meta = lib.frontmatter.parse(content)
-        for check in checks(features):
+        for check in all_checks:
             error = check.run(file_name, meta, content)
 
             if error is not None:

tools/lint/requirements.txt

@@ -1 +1 @@
-PyYAML==5.1.2
+PyYAML==6.0.3