common replace rule, new coloring library

files/mvregex

@@ -9,6 +9,7 @@ function helpexit() {
     echo 'Modes:'
     echo '    -n to match non-ascii and non-printable characters, and replace to [arg1]'
     echo '    -f to replace match [arg1] with format [arg2], ex: -f "[0-9]\+" "%04d"'
+    echo '    -p to replace problematic characters [^\w()[]-.] with [arg1]'
     
     exit
 }
@@ -39,6 +40,10 @@ do  [[ "${!i}" = "-h" ]] && helpexit
             MODE=printf
             continue
         }
+    [[ "${!i}" = "-p" ]] && {
+            SRC='[^]\[0-9a-zA-Z_.()-]'
+            continue
+        }
     [[ -z "$SRC" ]] && {
             SRC="${!i}"
             continue

reporting/bc.py

@@ -0,0 +1,60 @@
+# Python library for ansi colorization
+
+class ansi:
+    K="\033[1;30m"
+    R="\033[1;31m"
+    G="\033[1;32m"
+    B="\033[1;34m"
+    Y="\033[1;33m"
+    M="\033[1;35m"
+    C="\033[1;36m"
+    W="\033[1;37m"
+    
+    k="\033[0;30m"
+    r="\033[0;31m"
+    g="\033[0;32m"
+    b="\033[0;34m"
+    y="\033[0;33m"
+    m="\033[0;35m"
+    c="\033[0;36m"
+    w="\033[0;37m"
+    
+    bk="\033[40m"
+    br="\033[41m"
+    bg="\033[42m"
+    by="\033[43m"
+    bb="\033[44m"
+    bm="\033[45m"
+    bc="\033[46m"
+    bw="\033[47m"
+
+    S = '\033[1m'#strong
+    U = '\033[4m'#underline
+    Z = '\033[0m'#zero colors
+    CLRLIN = '\033[2J'
+    CLREND = '\033[K'
+    CLRSCR = CLRLIN+"\033[0;0H"
+    
+    color_keys="K,R,G,B,Y,M,C,W,k,r,g,b,y,m,c,w,S,U,Z,bk,br,bg,by,bb,bm,bc,bw,CLRLIN,CLREND,CLRSCR".split(",")
+    color_list=[K,R,G,B,Y,M,C,W,k,r,g,b,y,m,c,w,S,U,Z,bk,br,bg,by,bb,bm,bc,bw,CLRLIN,CLREND,CLRSCR]
+        
+    def pos(self,y,x):
+        return "\033["+str(y)+";"+str(x)+"H"
+
+    def posprint(self, y,x,s):
+        sys.stdout.write( self.pos(y,x) + str(s) )
+
+    def clear(self):
+        sys.stdout.write( self.CLR+self.pos(0,0) )
+    def clear_to_end(self):
+        sys.stdout.write( self.CLREND )
+
+    def color_string(self,s):
+        for i,c in enumerate(self.color_keys):
+            s=s.replace("${"+c+"}",self.color_list[i])
+        return s
+    def nocolor_string(self,s):
+        for i,c in enumerate(self.color_keys):
+            s=s.replace("${"+c+"}","")
+        return s
+

reporting/md-color.py

@@ -0,0 +1,97 @@
+#!/usr/bin/env python
+
+import sys,os,re
+from argparse import ArgumentParser
+sys.path.append(os.path.dirname(os.path.realpath(__file__)))
+import bc
+
+__author__ = "Ville Rantanen <ville.q.rantanen@gmail.com>"
+__version__ = "0.1"
+
+
+def setup_options():
+    ''' Create command line options '''
+    usage='''
+Color notation renderer in ANSI codes
+Special syntaxes:
+    * Colors: insert string ${C}, where C is one of KRGBYMCWkrgbymcwSUZ
+
+'''
+ 
+    parser=ArgumentParser(description=usage,
+                          epilog=__author__)
+
+    parser.add_argument("-v","--version",action="version",version=__version__)
+    parser.add_argument("--no-color","-n",action="store_false",dest="color",default=True,
+                     help="Disable color.")
+    parser.add_argument("-z",action="store_true",dest="zero",default=False,
+                     help="Reset coloring at the end of each line.")
+    parser.add_argument("filename",type=str, 
+                     help="File to show, - for stdin")
+    opts=parser.parse_args()
+    return opts
+
+block_match={
+    'block_code': (re.compile(r'^( {4}[^*])(.*)$'),'{}\\1\\2','${c}'),
+    'multiline_code' : (re.compile(r'^ *(`{3,}|~{3,}) *(\S*)?'),'${c}\\1\\2','${c}'),  # ```lang
+    'block_quote': (re.compile(r'^ *>'),False),
+
+    'hrule': (re.compile(r'^ {0,3}[-*_]([-*_]){2,}$'),False),
+    'heading' : (re.compile(r'^ *(#{1,6}) *([^\n]+?) *#* *(?:\n+|$)'),'${W}\\1 \\2','${W}'),
+    'lheading' : (re.compile(r'^(=+|-+)$'),'${W}\\1','${W}'),
+    'list_bullet':(re.compile(r'^( *)([*+-]|\d+\.)( +)'),'\\1${y}\\2${Z}\\3'),
+    'text':       (re.compile(r'^[^\n]+'),False),
+    'empty':      (re.compile(r'^$'),False)
+}
+blocks=['block_code', 'block_quote', 'multiline_code','hrule', 'heading','lheading','list_bullet', 'text', 'empty']
+
+inline_match={
+    'code': (re.compile(r'(`+)([^`]+)(`+)'),'${c}\\1\\2\\3')  # `code`
+}
+inlines=['code']
+
+
+bc=bc.ansi()
+opts=setup_options()
+if opts.filename=="-":
+    f=sys.stdin
+else:
+    f=open(opts.filename,'r')
+block='text'
+new_block='text'
+multiline_block=False
+for row in f:
+    if not row:
+        continue
+    row=row.decode('utf-8').rstrip("\n\r ")
+    for match in blocks:
+        if block_match[match][0].match(row):
+            new_block=match
+            if match.startswith('multiline'):
+                if multiline_block:
+                    multiline_block=False
+                else:
+                    multiline_block=match
+            break
+    if multiline_block: 
+        new_block=multiline_block
+        
+    if new_block!=block:
+        block=new_block
+        if not multiline_block:
+            sys.stdout.write(bc.Z)
+    #print(multiline_block)
+    #print(block)
+    if block_match[block][1]:
+        row=block_match[block][0].sub(block_match[block][1],row)
+    #print(row)
+    if opts.color:
+        colored=bc.color_string(row)
+    else:
+        colored=bc.nocolor_string(row)
+    sys.stdout.write(colored.encode('utf-8'))
+    if opts.zero:
+        sys.stdout.write(bc.Z)
+    sys.stdout.write("\n")
+
+