pmacs3/mode/c.py

import os, re
from subprocess import Popen, PIPE, STDOUT
import color, context, default, method, method.shell, mode, tab
from lex import Grammar, PatternRule, RegionRule, OverridePatternRule
from mode.python import StringGrammar

# this might not be complete...
# see http://gcc.gnu.org/onlinedocs/gcc-2.95.3/cpp_3.html#SEC44
class MacroGrammar(Grammar):
    rules = [
        PatternRule('name', r'(?:(?<=#define )) *[a-zA-Z_][a-zA-Z0-9_]*'),
        PatternRule(r"unop", r"\+=|-=|\*=|/=|//=|%=|&=|\^=|>>=|<<=|\*\*=|\|="),
        PatternRule(r'binop', r"\+|<>|<<|<=|<|-|>>|>=|>|\*\*|&|\*|\||/|\^|==|//|~|!=|%"),
        PatternRule(r"delimiter", r"->|\.|\(|\)|\[|\]|{|}|@|,|:|`|;|=|\?"),
        PatternRule(r"integer", r"-?(?:0(?![x0-9])|[1-9][0-9]*|0[0-7]+|0[xX][0-9a-fA-F]+)[lL]?"),
        PatternRule(r"float", r"-?(?:[0-9]+\.[0-9]*|\.[0-9]+|(?:[0-9]|[0-9]+\.[0-9]*|\.[0-9]+)[eE][\+-]?[0-9]+)"),
        RegionRule(r'string', '"', StringGrammar, '"'),
        PatternRule(r'char', r"'.'|'\\.'|'\\[0-7]{3}'"),
        PatternRule(r"continued", r"\\\n$"),
    ]

class CGrammar(Grammar):
    rules = [
        PatternRule(r'include', r'#include(?!=[a-zA-Z0-9_])'),
        PatternRule(r'header', r'<[-A-Za-z/0-9_\.]+>|"[-A-Za-z/0-9_\.]+"'),
        RegionRule(r'macrocomment', r'#if +(?:0|NULL|FALSE)', Grammar, r'#endif'),
        RegionRule(r'macro', r'#(?:assert|cpu|define|elif|else|endif|error|ident|ifdef|ifndef|if|import|include_next|line|machine|pragma|pragma_once|system|unassert|undef|warning)(?!=[a-zA-Z0-9_])', MacroGrammar, r'\n$'),

        OverridePatternRule(r'comment', r'/\* *@@:(?P<token>[.a-zA-Z0-9_]+):(?P<mode>[.a-zA-Z0-9_]+) *\*/$'),
        OverridePatternRule(r'comment', r'// *@@:(?P<token>[.a-zA-Z0-9_]+):(?P<mode>[.a-zA-Z0-9_]+) *$'),
        RegionRule(r'comment', r'/\*', Grammar, r'\*/'),
        PatternRule(r'comment', r'//.*$'),

        PatternRule(r'keyword', r"(?:auto|break|case|char|const|continue|default|double|do|else|enum|extern|float|for|goto|if|int|long|register|return|short|signed|sizeof|static|struct|switch|typedef|union|unsigned|void|volatile|while)(?![a-zA-Z_])"),
        PatternRule(r'label', r'[a-zA-Z_][a-zA-Z0-9_]*(?=:)'),
        PatternRule(r'structname', r'(?<=struct ) *[a-zA-Z_][a-zA-Z0-9_]*'),
        PatternRule(r'enumname', r'(?<=enum ) *[a-zA-Z_][a-zA-Z0-9_]*'),
        PatternRule(r'function', r'[a-zA-Z_][a-zA-Z0-9_]*(?= *\()'),

        PatternRule(r'builtin', r"(?:NULL|TRUE|FALSE)"),
        PatternRule(r'c_type', r'[a-zA-Z_][a-zA-Z0-9_]*(?= +[a-zA-Z_])'),
        PatternRule(r'identifier', r"[a-zA-Z_][a-zA-Z0-9_]*"),
        PatternRule(r"unop", r"\+=|-=|\*=|/=|//=|%=|&=\|\^=|>>=|<<=|\*\*="),
        PatternRule(r'binop', r"\+|<>|<<|<=|<|-|>>|>=|>|\*\*|&|\*|\||/|\^|==|//|~|!=|%"),

        # this is sketchy as hell
        PatternRule(r"delimiter", r"->|\.|\(|\)|\[|\]|{|}|@|,|:|`|;|=|\?"),

        PatternRule(r"integer", r"(?:0(?![x0-9])|[1-9][0-9]*|0[0-7]+|0[xX][0-9a-fA-F]+)[lL]?"),
        PatternRule(r"float", r"[0-9]+\.[0-9]*|\.[0-9]+|(?:[0-9]|[0-9]+\.[0-9]*|\.[0-9]+)[eE][\+-]?[0-9]+"),
        RegionRule(r'string', '"', StringGrammar, '"'),
        PatternRule(r'char', r"'.'|'\\.'|'\\[0-7]{3}'"),
        PatternRule(r'spaces', r' +'),
        PatternRule(r'eol', r"\n$"),
    ]

class CTabber(tab.StackTabber):
    wst  = ('spaces', 'eol', 'comment', 'comment.start', 'comment.null', 'comment.end')
    def token_is_whitespace(self, y, i):
        token = self.get_token(y, i)
        return token.fqname() in self.wst

    def is_base(self, y):
        if y == 0:
            return True
        highlighter = self.mode.window.buffer.highlights[self.mode.name()]
        if not highlighter.tokens[y]:
            return False

        # this assumes that people aren't gonna use these macros inside of
        # blocks, which is probably ok.
        t0 = highlighter.tokens[y][0]
        if t0.name == 'macro.start' and t0.string in ('#define', '#include'):
            return True
        
        # detecting function declarations is annoying; this assumes that people
        # won't put a variable type and name on different lines, but that they
        # might do that for function return type and name.
        #
        # unfortunately, valid function return types might include any of the
        # four types of tokens below
        decl = False
        for t in highlighter.tokens[y]:
            if t.name in ('keyword', 'identifier', 'structname', 'enumname'):
                decl = True
                continue
            if decl and t.name == 'function':
                break
            else:
                decl = False
                break
        if decl:
            return True

        return False
        
    def _handle_open_token(self, currlvl, y, i):
        self._opt_pop('cont')
        token = self.get_token(y, i)
        if token.string == '{':
            self._pop_while('cond', 'cont', 'case')
            if self.is_leftmost_token(y, i):
                currlvl = self.get_curr_level()
        tab.StackTabber._handle_open_token(self, currlvl, y, i)
        return currlvl
    def _handle_close_token(self, currlvl, y, i):
        w = self.mode.tabwidth
        self._opt_pop('cont')
        token = self.get_token(y, i)
        if token.string == '}':
            self._opt_pop('case')
        currlvl = tab.StackTabber._handle_close_token(self, currlvl, y, i)
        if self.is_rightmost_token(y, i):
            if token.string == '}':
                self._pop_while('cond', 'cont', 'case')
            elif self._has_markers() and self._peek_name() == 'cond':
                pass
            else:
                if token.fqname() != 'macro.delimiter':
                    self._opt_append('cont', currlvl + w)
        return currlvl
    def _handle_other_token(self, currlvl, y, i):
        w = self.mode.tabwidth
        token  = self.get_token(y, i)
        fqname = token.fqname()
        if fqname == 'delimiter' and token.string == ';':
            self._pop_while('cond', 'cont')

        elif fqname == 'keyword':
            if token.string in ('do', 'else', 'for', 'if', 'while'):
                self._append('cond', currlvl + w)
            elif token.string == 'break':
                self._opt_pop('case', 'while', 'for')
            elif token.string == 'continue':
                self._opt_pop('while', 'for')
            elif token.string == 'case':
                self._opt_pop('case')
                currlvl = self.get_curr_level()
                self._opt_append('case', currlvl + w)

        elif fqname == 'string.start':
            self._opt_append('string', None)
        elif fqname == 'string.end':
            self._opt_pop('string')
            if self.is_rightmost_token(y, i):
                self._opt_append('cont', currlvl + w)

        # TODO: this could be a lot better
        elif fqname == 'macro':
            currlvl = 0
        elif fqname.startswith('macro.start'):
            self._opt_append('macro', None)
            currlvl = 0
        elif fqname.startswith('macro.end'):
            self._opt_pop('macro', None)                               

        elif fqname.startswith('macroblock.start'):
            self._opt_append('macroblock', None)                               
            currlvl = 0
        elif fqname.startswith('macroblock.end'):
            self._opt_pop('macroblock', None)                               

        if self.is_rightmost_token(y, i):
            if self._has_markers() and self._peek_name() == 'cond':
                pass
            elif(not fqname.startswith('string') and
                 not fqname.startswith('comment') and
                 not fqname.startswith('macro') and
                 not fqname == 'delimiter' and
                 not fqname == 'header' and
                 #not fqname == 'null' and
                 not fqname == 'spaces' and
                 not fqname == 'eol' and
                 token.string not in ('}', ';', '(', '{', '[', ',')):
                self._opt_append('cont', currlvl + w)
        return currlvl

class CCheckSyntax(method.shell.Exec):
    '''Build this C program (using the mode's make cmd)'''
    show_success = False
    args = []
    def _execute(self, w, **vargs):
        if w.application.config['c.syntax-rel-dir']:
            d = os.path.dirname(w.buffer.path)
            self._doit(w, w.buffer.path, w.application.config['c.syntax-cmd'],
                       cmdname='c-check-syntax', cmddir=d)
        else:
            self._doit(w, w.buffer.path, w.application.config['c.syntax-cmd'],
                       cmdname='c-check-syntax')

class CMake(method.shell.Exec):
    '''Build this C program (using the mode's make cmd)'''
    show_success = False
    args = []
    def _execute(self, w, **vargs):
        if w.application.config['c.make-rel-dir']:
            d = os.path.dirname(w.buffer.path)            
            self._doit(w, w.buffer.path, w.application.config['c.make-cmd'],
                       cmdname='c-make', cmddir=d)
        else:
            self._doit(w, w.buffer.path, w.application.config['c.make-cmd'],
                       cmdname='c-make')

class CContext(context.Context):
    def _regen_stack(self, y):
        if y > 0 and self.namelines[y - 1][1]:
            return list(self.namelines[y - 1][1])
        else:
            return []

    def _build_name_map(self, y1, y2, last, curr, stack):
        blen       = len(self.mode.window.buffer.lines)
        highlights = self.mode.window.get_highlighter()
        i          = y1
        starting   = curr and not bool(stack)
        while i < y2:
            if not starting and not stack:
                curr = None

            g = highlights.tokens[i]
            if not stack and len(g) > 2 and g[0].name in ('c_type', 'keyword'):
                g2 = [x for x in g if x.name != 'spaces']
                j = 0
                while j < len(g2) - 1 and g2[j].name in ('keyword', 'c_type'):
                    j += 1
                if j < len(g2) and g2[j].name == 'function':
                    curr     = g2[j].string
                    starting = True

            if curr is not None and curr not in self.names:
                self.names[curr] = i

            if i == y2 - 1 and curr != self.namelines[i][0] and y2 < blen:
                y2 += 1

            m = self.mode
            for t in g:
                if t.name == 'delimiter' and t.string == '{':
                    stack.append(t.string)
                    starting = False
                elif t.name == 'delimiter' and t.string == '}':
                    if not stack:
                        # we are totally hosed. start over. ugh.
                        #self.build_name_map()
                        raise Exception, "we're totally hosed"
                        return
                    stack.pop(-1)
            if curr:
                self.namelines[i] = (curr, tuple(stack))
            i += 1

class C(mode.Fundamental):
    modename    = 'C'
    extensions  = ['.c', '.h', '.cpp']
    tabbercls   = CTabber
    grammar     = CGrammar
    opentokens  = ('delimiter',)
    opentags    = {'(': ')', '[': ']', '{': '}'}
    closetokens = ('delimiter',)
    closetags   = {')': '(', ']': '[', '}': '{'}
    colors      = {
        'macrocomment.start':   ('red', 'default', 'bold'),
        'macrocomment.null':    ('red', 'default', 'bold'),
        'macrocomment.end':     ('red', 'default', 'bold'),
        'macro':                ('blue', 'default', 'bold'),
        'macro.start':          ('blue', 'default', 'bold'),
        'macro.name':           ('yellow', 'default', 'bold'),
        'macro.null':           ('magenta', 'default', 'bold'),
        'macro.continued':      ('red', 'default', 'bold'),
        'macro.delimiter':      ('default', 'default', 'bold'),
        'macro.integer':        ('green', 'default', 'bold'),
        'macro.float':          ('green', 'default', 'bold'),
        'macro.char':           ('green', 'default', 'bold'),
        'macro.string.start':   ('green', 'default', 'bold'),
        'macro.string.escaped': ('magenta', 'default', 'bold'),
        'macro.string.octal':   ('magenta', 'default', 'bold'),
        'macro.string.null':    ('green', 'default', 'bold'),
        'macro.string.end':     ('green', 'default', 'bold'),
        'macro.end':            ('magenta', 'default', 'bold'),
        'include':              ('blue', 'default', 'bold'),
        'header':               ('green', 'default', 'bold'),
        'structname':           ('yellow', 'default', 'bold'),
        'enumname':             ('yellow', 'default', 'bold'),
        'c_type':               ('green', 'default', 'bold'),
    }
    config = {
        'c.syntax-cmd':     "gcc -x c -fsyntax-only %(path)s",
        'c.syntax-rel-dir': False,
        'c.make-cmd':       "make",
        'c.make-rel-dir':   True,
    }
    actions = [CCheckSyntax, CMake]

    format = "%(flag)s  %(bname)-18s  (%(mname)s)  %(indent)s  %(cursor)s/%(mark)s  %(perc)s  [%(func)s]"

    def get_status_names(self):
        names = mode.Fundamental.get_status_names(self)
        c = self.window.logical_cursor()
        names['func'] = self.get_line_function(c.y)
        return names

    def __init__(self, w):
        mode.Fundamental.__init__(self, w)
        self.add_bindings('close-paren', (')',))
        self.add_bindings('close-brace', ('}',))
        self.add_bindings('close-bracket', (']',))
        self.add_bindings('c-check-syntax', ('C-c s',))
        self.add_bindings('c-make', ('C-c C-c',))
        self.context   = CContext(self)
        self.functions = None
        self.funclines = None

    def get_functions(self):
        return self.context.get_names()
    def get_function_names(self):
        return self.context.get_name_list()
    def get_line_function(self, y):
        return self.context.get_line_name(y)

install = C.install