From c7c7ff84e485ffdc595c8468716e4555e2378d68 Mon Sep 17 00:00:00 2001
From: Frank Sauerburger <frank@sauerburger.com>
Date: Thu, 3 Aug 2017 17:36:27 +0200
Subject: [PATCH] Implement Markdown parsing
Implement all parsing methods for Markdown syntax, such that all test
requirements are satisfied.
The commit provides a functioning CI and therefore closes #1.
---
doxec/__init__.py | 64 ++++++++++++++++++++++++++++++++++-------
doxec/tests/markdown.py | 15 ++++------
2 files changed, 59 insertions(+), 20 deletions(-)
diff --git a/doxec/__init__.py b/doxec/__init__.py
index 02a28ce..64ca33f 100644
--- a/doxec/__init__.py
+++ b/doxec/__init__.py
@@ -1,5 +1,6 @@
import abc
+import re
class Operation(metaclass=abc.ABCMeta):
"""
@@ -116,7 +117,16 @@ class Markdown(DoxecSyntax):
"""
See DoxecSyntax.
"""
- pass
+ while len(lines) > 0:
+ cmd = Markdown.parse_command(lines[0])
+ del lines[0]
+ if cmd is not None:
+ code = Markdown.parse_code(lines)
+ if code is not None:
+ return cmd[0], cmd[1], code
+
+ return None
+
@staticmethod
def parse_command(line):
@@ -131,10 +141,17 @@ class Markdown(DoxecSyntax):
>>> Markdown.parse_command("<!-- write file.txt -->")
('write', 'file.txt')
- >>> Markdown.parse_command("<!-- invalid line ")
- None
+ >>> Markdown.parse_command("<!-- invalid line ") is None
+ True
"""
- pass
+ match = re.match(r"<!--\s+(\S.*\S)\s+-->\s*$", line)
+ if match is None:
+ return None
+ token = re.split("\s+", match.group(1), maxsplit=1)
+ if len(token) == 2:
+ return tuple(token)
+ else:
+ return token[0], None
@staticmethod
def parse_code(lines):
@@ -146,14 +163,39 @@ class Markdown(DoxecSyntax):
None if an error.
occurs.
>>> lines = []
- >>> lines += "```bash"
- >>> lines += "$ whoami"
- >>> lines += "$ ls"
- >>> lines += "```"
+ >>> lines.append("```bash")
+ >>> lines.append("$ whoami")
+ >>> lines.append("$ ls")
+ >>> lines.append("```")
+ >>> lines.append("lalala")
>>> Markdown.parse_code(lines)
- ["$ whoami", "$ ls"]
- """
- pass
+ ['$ whoami', '$ ls']
+ >>> lines
+ ['lalala']
+ """
+ if len(lines) == 0:
+ return None
+
+ head = lines[0]
+ match = re.match(r"```.*$", head)
+ if match is None:
+ return None
+
+ del lines[0] # delete start
+
+ buf = []
+ while len(lines) > 0:
+ # found end?
+ if lines[0] == "```":
+ del lines[0]
+ return buf
+
+ # eat lines and add them to the buffer
+ buf.append(lines[0])
+ del lines[0]
+
+ # there was no end
+ return None
class Document:
diff --git a/doxec/tests/markdown.py b/doxec/tests/markdown.py
index f987f93..771f845 100644
--- a/doxec/tests/markdown.py
+++ b/doxec/tests/markdown.py
@@ -63,9 +63,6 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
retval = Markdown.parse_command("<!--WRITE hello_world.c -->")
self.assertIsNone(retval)
- retval = Markdown.parse_command("<!-- WRITE hello_world.c -->")
- self.assertIsNone(retval)
-
retval = Markdown.parse_command(" <!-- WRITE hello_world.c -->")
self.assertIsNone(retval)
@@ -83,9 +80,6 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
retval = Markdown.parse_command("<-- WRITE hello_world.c -->")
self.assertIsNone(retval)
- retval = Markdown.parse_command("<!-- WRITE hello_world.c -->")
- self.assertIsNone(retval)
-
retval = Markdown.parse_command("!-- WRITE hello_world.c -->")
self.assertIsNone(retval)
@@ -208,7 +202,7 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
retval = Markdown.parse_code(block)
self.assertIsNone(retval)
- self.assertEqual(block, [])
+ self.assertEqual(block, ['touch /tmp', 'touch /home', '```'])
def test_parse_code_missing_end(self):
"""
@@ -228,7 +222,7 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
def test_parse_code_leading_lines(self):
"""
Run parse_command on input with leading lines and check that parsing
- failed.
+ failed. The parse method should not remove any lines in this case.
"""
block = []
block.append("This is a leading line.")
@@ -236,10 +230,13 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
block.append("touch /tmp")
block.append("touch /home")
block.append("```")
+
retval = Markdown.parse_code(block)
self.assertIsNone(retval)
+ self.assertEqual(len(block), 5)
+
def test_parse_valid(self):
"""
Check that a simple example runs and returns args, command and
@@ -294,7 +291,7 @@ class MarkdownSyntaxTestCase(unittest.TestCase):
retval = Markdown.parse(doc)
self.assertIsNone(retval)
- self.assertEqual(doc, ["This caused a seg fault?"])
+ self.assertEqual(doc, []) # lines are eaten in the second round
def test_parse_empty_input(self):
"""
--
GitLab