Merge lp:~adiroiban/pocket-lint/1082046-json-checker into lp:pocket-lint

[Adi Roiban]

Description
===========

The is no checker for JSON files.

JSON files are text files, but since no long lines breaker is supported in JSON, long lines are accepted for JSON.

All other plain text file checks should be applied.

Changes
=======

* A mimetype was forced for JSON since on WIndows it might not be automatically recognized.
* I have enabled the checker for ReStructuredText, since at the last merge I have only created the checker but not enabled it.
* I moved the checker_for_emtpy_last_line in the AnyTextCheckMixin together with the test.

I have a few drive by changes. Let me know if you want me to move them in another merge request.

I was not sure how to test basic plain text checks for JSON, so I just added simple test.

Let me know what needs to be changed for accepting this patch.

Thanks!

[Curtis Hovey (sinzui)]

Thank you. I will merge the branch.

review: Approve (code)

=== modified file 'pocketlint/formatcheck.py'
--- pocketlint/formatcheck.py	2012-04-25 13:09:13 +0000
+++ pocketlint/formatcheck.py	2012-12-01 20:25:22 +0000
@@ -169,6 +169,7 @@
     DOCTEST = object()
     CSS = object()
     JAVASCRIPT = object()
+    JSON = object()
     SH = object()
     XML = object()
     XSLT = object()
@@ -182,6 +183,7 @@
 
     XML_LIKE = (XML, XSLT, HTML, ZPT, ZCML, DOCBOOK)
 
+    mimetypes.add_type('application/json', '.json')
     mimetypes.add_type('application/x-zope-configuation', '.zcml')
     mimetypes.add_type('application/x-zope-page-template', '.pt')
     mimetypes.add_type('text/x-python-doctest', '.doctest')
@@ -199,6 +201,7 @@
         'text/x-log': LOG,
         'text/x-rst': RESTRUCTUREDTEXT,
         'application/javascript': JAVASCRIPT,
+        'application/json': JSON,
         'application/xml': XML,
         'application/x-sh': SH,
         'application/x-zope-configuation': ZCML,
@@ -307,6 +310,10 @@
             checker_class = XMLChecker
         elif self.language is Language.JAVASCRIPT:
             checker_class = JavascriptChecker
+        elif self.language is Language.JSON:
+            checker_class = JSONChecker
+        elif self.language is Language.RESTRUCTUREDTEXT:
+            checker_class = ReStructuredTextChecker
         elif self.language is Language.LOG:
             # Log files are not source, but they are often in source code
             # trees.
@@ -346,6 +353,19 @@
             self.message(
                 line_no, 'Line contains a tab character.', icon='info')
 
+    def check_empty_last_line(self, total_lines):
+        """Chech the files ends with an emtpy line and not with double empty
+        line.
+
+        This will avoid merge conflicts.
+        """
+        if self.text[-1] != '\n' or self.text[-2:] == '\n\n':
+            self.message(
+                total_lines,
+                'File does not ends with an empty line.',
+                icon='info',
+                )
+
 
 class AnyTextChecker(BaseChecker, AnyTextMixin):
     """Verify the text of the document."""
@@ -611,7 +631,7 @@
 
 
 class JavascriptChecker(BaseChecker, AnyTextMixin):
-    """Check python source code."""
+    """Check JavaScript source code."""
 
     HERE = os.path.dirname(__file__)
     FULLJSLINT = os.path.join(HERE, 'contrib/fulljslint.js')
@@ -652,8 +672,56 @@
             self.check_tab(line_no, line)
 
 
+class JSONChecker(BaseChecker, AnyTextMixin):
+    """Check JSON files."""
+
+    def check(self):
+        """Check JSON file using basic text checks and custom checks."""
+        if not self.text:
+            return
+
+        # Line independent checks.
+        for line_no, line in enumerate(self.text.splitlines()):
+            line_no += 1
+            self.check_trailing_whitespace(line_no, line)
+            self.check_conflicts(line_no, line)
+            self.check_tab(line_no, line)
+        last_lineno = line_no
+
+        self.check_load()
+        self.check_empty_last_line(last_lineno)
+
+    def check_length(self, line_no, line):
+        """JSON files can have long lines."""
+        return
+
+    def check_load(self):
+        """Check that JSON can be deserialized/loaded."""
+        try:
+            import json
+        except ImportError:
+            try:
+                from simplejson import json
+            except ImportError:
+                raise AssertionError('JSON module could not be loaded.')
+
+        try:
+            json.loads(self.text)
+        except ValueError, error:
+            line_number = 0
+            message = error.message
+            match = re.search(r"(.*): line (\d+)", message)
+            if match:
+                try:
+                    line_number = int(match.group(2))
+                except:
+                    # If we can not find the line number,
+                    # just fall back to default.
+                    line_number = 0
+            self.message(line_number, message, icon='error')
+
 class ReStructuredTextChecker(BaseChecker, AnyTextMixin):
-    """Check reStructuredText ource code."""
+    """Check reStructuredText source code."""
 
     # Taken from rst documentation.
     delimiter_characters = [
@@ -667,8 +735,11 @@
 
     def check(self):
         """Check the syntax of the reStructuredText code."""
+        if not self.text:
+            return
+
         self.check_lines()
-        self.check_empty_last_line()
+        self.check_empty_last_line(len(self.lines))
 
     def check_lines(self):
         """Call each line checker for each line in text."""
@@ -878,21 +949,6 @@
 
         return True
 
-    def check_empty_last_line(self):
-        """Chech the files ends with an emtpy line and not with double empty
-        line.
-
-        This will avoid merge conflicts.
-        """
-        if len(self.lines) < 2:
-            return
-        if self.text[-1] != '\n' or self.text[-2:] == '\n\n':
-            self.message(
-                len(self.lines),
-                'File does not ends with an empty line.',
-                icon='info',
-                )
-
 
 def get_option_parser():
     """Return the option parser for this program."""
 
=== added file 'pocketlint/tests/test_json.py'
--- pocketlint/tests/test_json.py	1970-01-01 00:00:00 +0000
+++ pocketlint/tests/test_json.py	2012-12-01 20:25:22 +0000
@@ -0,0 +1,123 @@
+"""
+Tests for JSON files.
+"""
+
+from pocketlint.formatcheck import JSONChecker
+from pocketlint.tests import CheckerTestCase
+
+
+class TestJSON(CheckerTestCase):
+    """Verify JSON validation."""
+
+    def test_empty_file(self):
+        """
+        No errors are be reported for empty files.
+        """
+        self.reporter.call_count = 0
+        content = ''
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual([], self.reporter.messages)
+        self.assertEqual(0, self.reporter.call_count)
+
+    def test_long_line(self):
+        """
+        No errors are be reported for long lines.
+        """
+        self.reporter.call_count = 0
+        content = '{"1": "' + 'a' * 100 + '"}\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual([], self.reporter.messages)
+        self.assertEqual(0, self.reporter.call_count)
+
+    def test_trailing_spaces(self):
+        """
+        Short test to check that trailing spaces are catched.
+        """
+        self.reporter.call_count = 0
+        content = '{} \n'
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual(
+            [(1, 'Line has trailing whitespace.')],
+            self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_conflict_markups(self):
+        """
+        Short test to check that merge conflicts are catched.
+        """
+        self.reporter.call_count = 0
+        content = '{}\n>>>>>>>\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual(
+            (2, 'File has conflicts.'),
+            self.reporter.messages[0])
+
+    def test_tabs(self):
+        """
+        Short test to check that tab characters are catched.
+        """
+        self.reporter.call_count = 0
+        content = '\t{}\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual(
+            [(1, 'Line contains a tab character.')],
+            self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_unable_to_compile(self):
+        """
+        Line number '0' is reported if JSON can not be compiled at all.
+        """
+        content = '\n\nno-json-object\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+
+        checker.check()
+
+        self.assertEqual(
+            [(0, 'No JSON object could be decoded')],
+            self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_compile_error_with_line(self):
+        """
+        When the JSON module identifies an error at a line, it will be
+        set by pocketlint.
+        """
+        content = '{\n1: "something"}\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+        checker.check()
+
+        self.assertEqual(
+            [(2, 'Expecting property name: line 2 column 1 (char 2)')],
+            self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_compile_error_on_multiple_line(self):
+        """
+        When the JSON error is reported on multiple lines, only the first
+        line is reported as the line number.
+        """
+        content = '{}\n}\n}\n'
+        checker = JSONChecker('bogus', content, self.reporter)
+        checker.check()
+
+        self.assertEqual(
+            [(2,
+             'Extra data: line 2 column 1 - line 4 column 1 (char 3 - 7)')],
+            self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
 
=== modified file 'pocketlint/tests/test_restructuredtext.py'
--- pocketlint/tests/test_restructuredtext.py	2012-04-21 17:02:14 +0000
+++ pocketlint/tests/test_restructuredtext.py	2012-12-01 20:25:22 +0000
@@ -137,7 +137,7 @@
             'the second and last line witout newline'
             )
         checker = ReStructuredTextChecker('bogus', content, self.reporter)
-        checker.check_empty_last_line()
+        checker.check_empty_last_line(2)
         expected = [(
             2, 'File does not ends with an empty line.')]
         self.assertEqual(expected, self.reporter.messages)
@@ -151,7 +151,7 @@
             '\n'
             )
         checker = ReStructuredTextChecker('bogus', content, self.reporter)
-        checker.check_empty_last_line()
+        checker.check_empty_last_line(3)
         expected = [(
             3, 'File does not ends with an empty line.')]
         self.assertEqual(expected, self.reporter.messages)
 
=== modified file 'pocketlint/tests/test_text.py'
--- pocketlint/tests/test_text.py	2012-01-29 21:24:20 +0000
+++ pocketlint/tests/test_text.py	2012-12-01 20:25:22 +0000
@@ -75,3 +75,49 @@
         self.assertEqual(
             [(1, 'Line exceeds 49 characters.')],
             self.reporter.messages)
+
+    def test_no_empty_last_line(self):
+        """
+        An error is reported if file does not end with a new lines.
+        """
+        content = (
+            'Some first line\n'
+            'the second and last line witout newline'
+            )
+        checker = AnyTextChecker('bogus', content, self.reporter)
+        checker.check_empty_last_line(2)
+        expected = [(
+            2, 'File does not ends with an empty line.')]
+        self.assertEqual(expected, self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_multiple_empty_last_lines(self):
+        """
+        An error is reported if file ends with multiple new lines.
+        """
+        content = (
+            'Some first line\n'
+            'the second and last\n'
+            '\n'
+            )
+        checker = AnyTextChecker('bogus', content, self.reporter)
+        checker.check_empty_last_line(3)
+        expected = [(
+            3, 'File does not ends with an empty line.')]
+        self.assertEqual(expected, self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)
+
+    def test_single_last_line_no_newline(self):
+        """
+        An error is reported if file contains a single line and does not end
+        with a new lines.
+        """
+        content = (
+            'the second and last line witout newline'
+            )
+        checker = AnyTextChecker('bogus', content, self.reporter)
+        checker.check_empty_last_line(2)
+        expected = [(
+            2, 'File does not ends with an empty line.')]
+        self.assertEqual(expected, self.reporter.messages)
+        self.assertEqual(1, self.reporter.call_count)