Merge pull request #4352 from rlaverde/markdown-support

PR: Add Markdown syntax highlighter.

spyder/config/utils.py

@@ -45,6 +45,7 @@
     (_("Enaml files"), ('.enaml',)),
     (_("Configuration files"), ('.properties', '.session', '.ini', '.inf',
                                 '.reg', '.cfg', '.desktop')),
+    (_("Markdown files"), ('.md', )),
 ]
 
 # Filter for all files

spyder/utils/sourcecode.py

@@ -33,6 +33,7 @@
                  'Cpp': ('c', 'cc', 'cpp', 'cxx', 'h', 'hh', 'hpp', 'hxx'),
                  'OpenCL': ('cl',),
                  'Yaml':('yaml','yml'),
+                 "Markdown": ('md', ),
                  }
 
 PYTHON_LIKE_LANGUAGES = ('Python', 'Cython', 'Enaml')

spyder/utils/syntaxhighlighters.py

@@ -910,6 +910,130 @@ class HtmlSH(BaseWebSH):
     PROG = re.compile(make_html_patterns(), re.S)
 
 
+# =============================================================================
+# Markdown highlighter
+# =============================================================================
+
+def make_md_patterns():
+    h1 = '^#[^#]+'
+    h2 = '^##[^#]+'
+    h3 = '^###[^#]+'
+    h4 = '^####[^#]+'
+    h5 = '^#####[^#]+'
+    h6 = '^######[^#]+'
+
+    titles = any('title', [h1, h2, h3, h4, h5, h6])
+
+    html_tags = any("builtin", [r"<", r"[\?/]?>", r"(?<=<).*?(?=[ >])"])
+    html_symbols = '&[^; ].+;'
+    html_comment = '<!--.+-->'
+
+    strikethrough = any('strikethrough', [r'(~~)(.*?)~~'])
+    strong = any('strong', [r'(\*\*)(.*?)\*\*'])
+
+    italic = r'(__)(.*?)__'
+    emphasis = r'(//)(.*?)//'
+    italic = any('italic', [italic, emphasis])
+
+    # links - (links) after [] or links after []:
+    link_html = (r'(?<=(\]\())[^\(\)]*(?=\))|'
+                 '(<https?://[^>]+>)|'
+                 '(<[^ >]+@[^ >]+>)')
+    # link/image references - [] or ![]
+    link = r'!?\[[^\[\]]*\]'
+    links = any('link', [link_html, link])
+
+    # blockquotes and lists -  > or - or * or 0.
+    blockquotes = (r'(^>+.*)'
+                   r'|(^(?:    |\t)*[0-9]+\. )'
+                   r'|(^(?:    |\t)*- )'
+                   r'|(^(?:    |\t)*\* )')
+    # code
+    code = any('code', ['^`{3,}.*$'])
+    inline_code = any('inline_code', ['`[^`]*`'])
+
+    # math - $$
+    math = any('number', [r'^(?:\${2}).*$', html_symbols])
+
+    comment = any('comment', [blockquotes, html_comment])
+
+    return '|'.join([titles, comment, html_tags, math, links, italic, strong,
+                     strikethrough, code, inline_code])
+
+
+class MarkdownSH(BaseSH):
+    """Markdown Syntax Highlighter"""
+    # Syntax highlighting rules:
+    PROG = re.compile(make_md_patterns(), re.S)
+    NORMAL = 0
+    CODE = 1
+
+    def highlightBlock(self, text):
+        text = to_text_string(text)
+        previous_state = self.previousBlockState()
+
+        if previous_state == self.CODE:
+            self.setFormat(0, len(text), self.formats["code"])
+        else:
+            previous_state = self.NORMAL
+            self.setFormat(0, len(text), self.formats["normal"])
+
+        self.setCurrentBlockState(previous_state)
+
+        match = self.PROG.search(text)
+        while match:
+            for key, value in list(match.groupdict().items()):
+                start, end = match.span(key)
+
+                if value:
+                    previous_state = self.previousBlockState()
+
+                    if previous_state == self.CODE:
+                        if key == "code":
+                            # Change to normal
+                            self.setFormat(0, len(text),
+                                           self.formats["normal"])
+                            self.setCurrentBlockState(self.NORMAL)
+                        else:
+                            continue
+                    else:
+                        if key == "code":
+                            # Change to code
+                            self.setFormat(0, len(text), self.formats["code"])
+                            self.setCurrentBlockState(self.CODE)
+                            continue
+
+                    self.setFormat(start, end - start, self.formats[key])
+
+            match = self.PROG.search(text, match.end())
+
+        self.highlight_spaces(text)
+
+    def setup_formats(self, font=None):
+        super(MarkdownSH, self).setup_formats(font)
+
+        font = QTextCharFormat(self.formats['normal'])
+        font.setFontItalic(True)
+        self.formats['italic'] = font
+
+        self.formats['strong'] = self.formats['definition']
+
+        font = QTextCharFormat(self.formats['normal'])
+        font.setFontStrikeOut(True)
+        self.formats['strikethrough'] = font
+
+        font = QTextCharFormat(self.formats['string'])
+        font.setUnderlineStyle(True)
+        self.formats['link'] = font
+
+        self.formats['code'] = self.formats['string']
+        self.formats['inline_code'] = self.formats['string']
+
+        font = QTextCharFormat(self.formats['keyword'])
+        font.setFontWeight(QFont.Bold)
+        self.formats['title'] = font
+
+
 #==============================================================================
 # Pygments based omni-parser
 #==============================================================================

spyder/utils/tests/test_syntaxhighlighters.py

@@ -9,7 +9,7 @@
 from qtpy.QtWidgets import QApplication
 from qtpy.QtGui import QTextDocument
 
-from spyder.utils.syntaxhighlighters import HtmlSH, PythonSH
+from spyder.utils.syntaxhighlighters import HtmlSH, PythonSH, MarkdownSH
 
 def compare_formats(actualFormats, expectedFormats, sh):
     assert len(actualFormats) == len(expectedFormats)
@@ -46,6 +46,30 @@ def test_HtmlSH_unclosed_commend():
     res = [(0, 3, 'normal')]
     compare_formats(doc.firstBlock().layout().additionalFormats(), res, sh)
 
+
+def test_Markdown_basic():
+    txt = "Some __random__ **text** with ~~different~~ [styles](link_url)"
+
+    doc = QTextDocument(txt)
+    sh = MarkdownSH(doc, color_scheme='Spyder')
+    sh.rehighlightBlock(doc.firstBlock())
+
+    res = [(0, 5, 'normal'),  # |Some|
+           (5, 10, 'italic'),  # |__random__|
+           (15, 1, 'normal'),  # | |
+           (16, 8, 'strong'),  # |**text**|
+           (24, 6, 'normal'),  # |with|
+           (30, 13, 'italic'),  # |~~diferents~~|
+           (43, 1, 'normal'),  # | |
+           (44, 8, 'string'),  # |[styles]|
+           (52, 1, 'normal'),  # |(|
+           (53, 8, 'string'),  # |(link_url)|
+           (61, 1, 'normal'),  # ||
+           ]
+
+    compare_formats(doc.firstBlock().layout().additionalFormats(), res, sh)
+
+
 @pytest.mark.parametrize('line', ['# --- First variant',
                                   '#------ 2nd variant',
                                   '### 3rd variant'])

spyder/widgets/sourcecode/codeeditor.py

@@ -334,6 +334,7 @@ class CodeEditor(TextEditBaseWidget):
                  'Cpp': (sh.CppSH, '//', None),
                  'OpenCL': (sh.OpenCLSH, '//', None),
                  'Enaml': (sh.EnamlSH, '#', PythonCFM),
+                 'Markdown': (sh.MarkdownSH, '#', None),
                 }
 
     TAB_ALWAYS_INDENTS = ('py', 'pyw', 'python', 'c', 'cpp', 'cl', 'h')