Added lint command

Signed-off-by: Ole Herman Schumacher Elgesem <ole.elgesem@northern.tech>

Author: olehermanse

README.md

@@ -35,12 +35,41 @@ cfengine help
 cfengine run
 ```
 
+Not implemented yet - TODOs:
+
+- The command could automatically detect that you have CFEngine installed on a
+  remote hub, and run it there instead (using cf-remote)
+- Handle when `cf-agent` is not installed, help users install.
+- Prompt / help users do what they meant (i.e. build and deploy and run).
+
 ### Automatically format source code
 
 ```bash
 cfengine format
 ```
 
+Not implemented yet - TODOs:
+
+- Automatically break up and indent method calls, function calls, and nested function calls.
+- Smarter placement of comments based on context.
+- The command should be able to take a filename as an argument, and also operate using stdin and stdout.
+  (Receive file content on stdin, file type using command line arg, output formatted file to stdout).
+
+### Check for errors in source code
+
+```bash
+cfengine lint
+```
+
+Note that since we use a different parser than `cf-agent` / `cf-promises`, they are not 100% in sync.
+`cf-agent` could point out something as a syntax error, while `cfengine lint` does not and vice versa.
+We aim to make the tree-sitter parser (used in this tool) more strict in general, so that when `cfengine lint` is happy with your policy, `cf-agent` will also accept it.
+(But the opposite is not a goal, that `cfengine lint` must accept any policy `cf-agent` would find acceptable).
+
+Not implemented yet - TODOs:
+
+- The command should be able to take a filename as an argument, and also take file content from stdin
+
 ## Supported platforms and versions
 
 This tool will only support a limited number of platforms, it is not int.

src/cfengine_cli/commands.py

@@ -1,8 +1,20 @@
+import os
+from cfengine_cli.lint import lint_cfbs_json, lint_json, lint_policy_file
 from cfengine_cli.shell import user_command
 from cfengine_cli.paths import bin
 from cfengine_cli.version import cfengine_cli_version_string
 from cfengine_cli.format import format_policy_file, format_json_file
-from cfbs.utils import find
+from cfbs.utils import find, user_error
+
+
+def require_cfagent():
+    if not os.path.exists(bin("cf-agent")):
+        user_error(f"cf-agent not found at {bin('cf-agent')}")
+
+
+def require_cfhub():
+    if not os.path.exists(bin("cf-hub")):
+        user_error(f"cf-hub not found at {bin('cf-hub')}")
 
 
 def format() -> int:
@@ -17,14 +29,37 @@ def format() -> int:
     return 0
 
 
+def lint() -> int:
+    errors = 0
+    for filename in find(".", extension=".json"):
+        if filename.startswith("./."):
+            continue
+        if filename.endswith("/cfbs.json"):
+            lint_cfbs_json(filename)
+            continue
+        errors += lint_json(filename)
+
+    for filename in find(".", extension=".cf"):
+        if filename.startswith("./."):
+            continue
+        errors += lint_policy_file(filename)
+
+    if errors == 0:
+        return 0
+    return 1
+
+
 def report() -> int:
+    require_cfhub()
+    require_cfagent()
     user_command(f"{bin('cf-agent')} -KIf update.cf && {bin('cf-agent')} -KI")
     user_command(f"{bin('cf-hub')} --query rebase -H 127.0.0.1")
     user_command(f"{bin('cf-hub')} --query delta -H 127.0.0.1")
     return 0
 
 
 def run() -> int:
+    require_cfagent()
     user_command(f"{bin('cf-agent')} -KIf update.cf && {bin('cf-agent')} -KI")
     return 0
 

src/cfengine_cli/lint.py

@@ -0,0 +1,127 @@
+"""
+Linting of CFEngine related files.
+
+Currently implemented for:
+- *.cf (policy files)
+- cfbs.json (CFEngine Build project files)
+- *.json (basic JSON syntax checking)
+
+Usage:
+$ cfengine lint
+
+TODOS:
+- It would be nice if we refactored validate_config in cfbs
+  so it would take a simple dictionary (JSON) instead of a
+  special CFBSConfig object.
+"""
+
+import sys
+import os
+import json
+import tree_sitter_cfengine as tscfengine
+from tree_sitter import Language, Parser, Node, Tree
+from cfbs.validate import validate_config
+from cfbs.cfbs_config import CFBSConfig
+
+
+def lint_cfbs_json(filename):
+    assert os.path.isfile(filename)
+    assert filename.endswith("cfbs.json")
+
+    config = CFBSConfig.get_instance(filename=filename, non_interactive=True)
+    r = validate_config(config)
+
+    if r == 0:
+        print(f"PASS: {filename}")
+        return
+    print(f"FAIL: {filename}")
+
+
+def lint_json(filename):
+    assert os.path.isfile(filename)
+    with open(filename, "r") as f:
+        data = f.read()
+
+    try:
+        data = json.loads(data)
+    except:
+        print(f"FAIL: {filename} (invalid JSON)")
+        return 1
+    print(f"PASS: {filename}")
+    return 0
+
+
+def _highlight_range(node, lines):
+    line = node.range.start_point[0] + 1
+    column = node.range.start_point[1]
+
+    length = len(lines[line - 1]) - column
+    if node.range.start_point[0] == node.range.end_point[0]:
+        # Starts and ends on same line:
+        length = node.range.end_point[1] - node.range.start_point[1]
+    assert length >= 1
+    print("")
+    if line >= 2:
+        print(lines[line - 2])
+    print(lines[line - 1])
+    marker = "^"
+    if length > 2:
+        marker += "-" * (length - 2)
+    if length > 1:
+        marker += "^"
+    print(" " * column + marker)
+
+
+def _text(node):
+    return node.text.decode()
+
+
+def _walk(filename, lines, node) -> int:
+    line = node.range.start_point[0] + 1
+    column = node.range.start_point[1]
+    errors = 0
+    # Checking for syntax errors (already detected by parser / grammar).
+    # These are represented in the syntax tree as special ERROR nodes.
+    if node.type == "ERROR":
+        _highlight_range(node, lines)
+        print(f"Error: Syntax error at {filename}:{line}:{column}")
+        errors += 1
+
+    if node.type == "attribute_name":
+        if _text(node) == "ifvarclass":
+            _highlight_range(node, lines)
+            print(
+                f"Error: Use 'if' instead of 'ifvarclass' (deprecated) at {filename}:{line}:{column}"
+            )
+            errors += 1
+
+    for node in node.children:
+        errors += _walk(filename, lines, node)
+
+    return errors
+
+
+def lint_policy_file(filename):
+    assert os.path.isfile(filename)
+    assert filename.endswith(".cf")
+    PY_LANGUAGE = Language(tscfengine.language())
+    parser = Parser(PY_LANGUAGE)
+
+    with open(filename, "rb") as f:
+        original_data = f.read()
+    tree = parser.parse(original_data)
+    lines = original_data.decode().split("\n")
+
+    root_node = tree.root_node
+    assert root_node.type == "source_file"
+    errors = 0
+    if not root_node.children:
+        print(f"Error: Empty policy file '{filename}'")
+        errors += 1
+    errors += _walk(filename, lines, root_node)
+    if errors == 0:
+        print(f"PASS: {filename}")
+        return 0
+
+    print(f"FAIL: {filename} ({errors} error{'s' if errors > 0 else ''})")
+    return errors

src/cfengine_cli/main.py

@@ -38,6 +38,11 @@ def _get_arg_parser():
 
     subp.add_parser("format", help="Autoformat .json and .cf files")
 
+    subp.add_parser(
+        "lint",
+        help="Look for syntax errors and other simple mistakes",
+    )
+
     subp.add_parser(
         "report",
         help="Run the agent and hub commands necessary to get new reporting data",
@@ -70,6 +75,8 @@ def run_command_with_args(command, _) -> int:
         return commands.version()
     if command == "format":
         return commands.format()
+    if command == "lint":
+        return commands.lint()
     if command == "report":
         return commands.report()
     if command == "run":