RedCloth 4.1.0-universal-java

data/ext/redcloth_scan/redcloth_inline.java.rl

@@ -0,0 +1,108 @@
+/*
+ * redcloth_inline.rl
+ *
+ * Copyright (C) 2008 Jason Garber
+ */
+import java.io.IOException;
+
+import org.jruby.Ruby;
+import org.jruby.RubyArray;
+import org.jruby.RubyClass;
+import org.jruby.RubyHash;
+import org.jruby.RubyModule;
+import org.jruby.RubyNumeric;
+import org.jruby.RubyObject;
+import org.jruby.RubyString;
+import org.jruby.RubySymbol;
+import org.jruby.anno.JRubyMethod;
+import org.jruby.runtime.Block;
+import org.jruby.runtime.CallbackFactory;
+import org.jruby.runtime.builtin.IRubyObject;
+import org.jruby.exceptions.RaiseException;
+import org.jruby.runtime.load.BasicLibraryService;
+
+import org.jruby.util.ByteList;
+
+public class RedclothInline extends RedclothScanService.Base {
+
+%%{
+
+  machine redcloth_inline;
+  include redcloth_common "redcloth_common.java.rl";
+  include redcloth_inline "redcloth_inline.rl";
+
+}%%
+
+%% write data nofinal;
+
+  public IRubyObject red_pass_code(IRubyObject self, IRubyObject regs, IRubyObject ref, String meth) {
+    IRubyObject txt = ((RubyHash)regs).aref(ref);
+    if(!txt.isNil()) {
+      IRubyObject txt2 = RubyString.newEmptyString(runtime);
+      strCatEscapedForPreformatted(self, txt2, ((RubyString)txt).getByteList().bytes, ((RubyString)txt).getByteList().begin, ((RubyString)txt).getByteList().begin + ((RubyString)txt).getByteList().realSize);
+      ((RubyHash)regs).aset(ref, txt2);
+    }
+    return self.callMethod(runtime.getCurrentContext(), meth, regs);
+  }
+
+  public IRubyObject red_parse_attr(IRubyObject self, IRubyObject regs, IRubyObject ref) {
+    IRubyObject txt = ((RubyHash)regs).aref(ref);
+    IRubyObject new_regs = RedclothAttributes.attributes(self, txt);
+    return regs.callMethod(runtime.getCurrentContext(), "update", new_regs);
+  }
+
+  public IRubyObject red_parse_link_attr(IRubyObject self, IRubyObject regs, IRubyObject ref) {
+    IRubyObject txt = ((RubyHash)regs).aref(ref);
+    IRubyObject new_regs = RedclothAttributes.link_attributes(self, txt);
+    return regs.callMethod(runtime.getCurrentContext(), "update", new_regs);
+  }
+
+  public void PASS_CODE(IRubyObject H, String A, String T, int O) {
+    ((RubyString)H).append(red_pass_code(self, regs, runtime.newSymbol(A), T));
+  }
+
+  public void PARSE_ATTR(String A) {
+    red_parse_attr(self, regs, runtime.newSymbol(A));
+  }
+
+  public void PARSE_LINK_ATTR(String A) {
+    red_parse_link_attr(self, regs, runtime.newSymbol(A));
+  }
+
+  private int opts;
+  private IRubyObject buf;
+
+  public RedclothInline(IRubyObject self, byte[] data, int p, int pe, IRubyObject refs) {
+    this.runtime = self.getRuntime();
+    this.self = self;
+    
+    // This is GROSS but necessary for EOF matching
+    this.data = new byte[pe+1];
+    System.arraycopy(data, p, this.data, 0, pe);
+    this.data[pe] = 0;
+
+    this.p = 0;
+    this.pe = pe+1;
+    this.eof = this.pe;
+    this.orig_p = 0;
+    this.orig_pe = this.pe;
+    this.refs = refs;
+    this.block = RubyString.newEmptyString(runtime);
+    this.regs = runtime.getNil();
+    this.opts = 0;
+    this.buf = runtime.getNil();
+  }
+
+
+  public IRubyObject inline() {
+    %% write init;
+    %% write exec;
+
+    return block;
+  }
+
+  public static IRubyObject inline2(IRubyObject self, IRubyObject str, IRubyObject refs) {
+    ByteList bl = str.convertToString().getByteList();
+    return new RedclothInline(self, bl.bytes, bl.begin, bl.realSize, refs).inline();
+  }
+}

data/ext/redcloth_scan/redcloth_inline.rl

@@ -0,0 +1,157 @@
+/*
+ * redcloth_inline.rl
+ *
+ * Copyright (C) 2008 Jason Garber
+ */
+%%{
+
+  machine redcloth_inline;
+
+  # links
+  mtext_noquotes = mtext -- '"' ;
+  quoted_mtext = '"' mtext_noquotes '"' ;
+  mtext_including_quotes = (mtext_noquotes ' "' mtext_noquotes '" ' mtext_noquotes?)+ ;
+  link_says = ( C_noactions "."* " "* ((quoted_mtext | mtext_including_quotes | mtext_noquotes) -- '":') ) >A %{ STORE("link_text"); } ;
+  link_says_noquotes_noactions = ( C_noquotes_noactions "."* " "* ((mtext_noquotes) -- '":') ) ;
+  link = ( '"' link_says '":' %A uri %{ STORE_URL("href"); } ) >X ;
+  link_noquotes_noactions = ( '"' link_says_noquotes_noactions '":' uri ) ;
+  bracketed_link = ( '["' link_says '":' %A uri %{ STORE("href"); } :> "]" ) >X ;
+
+  # images
+  image_src = ( uri ) >A %{ STORE("src"); } ;
+  image_is = ( A2 C ". "? image_src :> title? ) ;
+  image_link = ( ":" uri >A %{ STORE_URL("href"); } ) ;
+  image = ( "["? "!" image_is "!" %A image_link? "]"? ) >X ;
+
+  # footnotes
+  footno = "[" >X %A digit+ %T "]" ;
+
+  # markup
+  end_markup_phrase = (" " | PUNCT | EOF | LF) @{ fhold; };
+  code = "["? "@" >X mtext >A %T :> "@" "]"? ;
+  code_tag_start = "<code" [^>]* ">" ;
+  code_tag_end = "</code>" ;
+  script_tag = ( "<script" [^>]* ">" (default+ -- "</script>") "</script>" LF? ) >X >A %T ;
+  notextile = "<notextile>" >X (default+ -- "</notextile>") >A %T "</notextile>";
+  strong = "["? "*" >X mtext >A %T :> "*" "]"? ;
+  b = "["? "**" >X mtext >A %T :> "**" "]"? ;
+  em = "["? "_" >X mtext >A %T :> "_" "]"? ;
+  i = "["? "__" >X mtext >A %T :> "__" "]"? ;
+  del = "[-" >X C ( mtext ) >A %T :>> "-]" ;
+  emdash_parenthetical_phrase_with_spaces = " -- " mtext " -- " ;
+  del_phrase = (( " " >A %{ STORE("beginning_space"); } "-") >X C ( mtext ) >A %T :>> ( "-" end_markup_phrase )) - emdash_parenthetical_phrase_with_spaces ;
+  ins = "["? "+" >X mtext >A %T :> "+" "]"? ;
+  sup = "[^" >X mtext >A %T :> "^]" ;
+  sup_phrase = ( "^" when starts_phrase) >X ( mtext ) >A %T :>> ( "^" end_markup_phrase ) ;
+  sub = "[~" >X mtext >A %T :> "~]" ;
+  sub_phrase = ( "~" when starts_phrase) >X ( mtext ) >A %T :>> ( "~" end_markup_phrase ) ;
+  span = "[%" >X mtext >A %T :> "%]" ;
+  span_phrase = (("%" when starts_phrase) >X ( mtext ) >A %T :>> ( "%" end_markup_phrase )) ;
+  cite = "["? "??" >X mtext >A %T :> "??" "]"? ;
+  ignore = "["? "==" >X %A mtext %T :> "==" "]"? ;
+  snip = "["? "```" >X %A mtext %T :> "```" "]"? ;
+  
+  # quotes
+  quote1 = "'" >X %A mtext %T :> "'" ;
+  non_quote_chars_or_link = (chars -- '"') | link_noquotes_noactions ;
+  mtext_inside_quotes = ( non_quote_chars_or_link (mspace non_quote_chars_or_link)* ) ;
+  html_tag_up_to_attribute_quote = "<" Name space+ NameAttr space* "=" space* ;
+  quote2 = ('"' >X %A ( mtext_inside_quotes - (mtext_inside_quotes html_tag_up_to_attribute_quote ) ) %T :> '"' ) ;
+  multi_paragraph_quote = (('"' when starts_line) >X  %A ( chars -- '"' ) %T );
+  
+  # html
+  start_tag = ( "<" Name space+ AttrSet* (AttrEnd)? ">" | "<" Name ">" ) >X >A %T ;
+  empty_tag = ( "<" Name space+ AttrSet* (AttrEnd)? "/>" | "<" Name "/>" ) >X >A %T ;
+  end_tag = ( "</" Name space* ">" ) >X >A %T ;
+  html_comment = ("<!--" (default+) :>> "-->") >X >A %T;
+
+  # glyphs
+  ellipsis = ( " "? >A %T "..." ) >X ;
+  emdash = "--" ;
+  arrow = "->" ;
+  endash = " - " ;
+  acronym = ( [A-Z] >A [A-Z0-9]{2,} %T "(" default+ >A %{ STORE("title"); } :> ")" ) >X ;
+  caps_noactions = upper{3,} ;
+  caps = ( caps_noactions >A %*T ) >X ;
+  dim_digit = [0-9.]+ ;
+  prime = ("'" | '"')?;
+  dim_noactions = dim_digit prime (("x" | " x ") dim_digit prime) %T (("x" | " x ") dim_digit prime)? ;
+  dim = dim_noactions >X >A %T ;
+  tm = [Tt] [Mm] ;
+  trademark = " "? ( "[" tm "]" | "(" tm ")" ) ;
+  reg = [Rr] ;
+  registered = " "? ( "[" reg "]" | "(" reg ")" ) ;
+  cee = [Cc] ;
+  copyright = ( "[" cee "]" | "(" cee ")" ) ;
+  entity = ( "&" %A ( '#' digit+ | ( alpha ( alpha | digit )+ ) ) %T ';' ) >X ;
+  
+  # info
+  redcloth_version = "[RedCloth::VERSION]" ;
+
+  other_phrase = phrase -- dim_noactions;
+
+  code_tag := |*
+    code_tag_end { CAT(block); fgoto main; };
+    default => esc_pre;
+  *|;
+
+  main := |*
+    
+    image { INLINE(block, "image"); };
+    
+    link { PARSE_LINK_ATTR("link_text"); PASS(block, "name", "link"); };
+    bracketed_link { PARSE_LINK_ATTR("link_text"); PASS(block, "name", "link"); };
+    
+    code { PARSE_ATTR("text"); PASS_CODE(block, "text", "code", opts); };
+    code_tag_start { CAT(block); fgoto code_tag; };
+    notextile { INLINE(block, "notextile"); };
+    strong { PARSE_ATTR("text"); PASS(block, "text", "strong"); };
+    b { PARSE_ATTR("text"); PASS(block, "text", "b"); };
+    em { PARSE_ATTR("text"); PASS(block, "text", "em"); };
+    i { PARSE_ATTR("text"); PASS(block, "text", "i"); };
+    del { PASS(block, "text", "del"); };
+    del_phrase { PASS(block, "text", "del_phrase"); };
+    ins { PARSE_ATTR("text"); PASS(block, "text", "ins"); };
+    sup { PARSE_ATTR("text"); PASS(block, "text", "sup"); };
+    sup_phrase { PARSE_ATTR("text"); PASS(block, "text", "sup_phrase"); };
+    sub { PARSE_ATTR("text"); PASS(block, "text", "sub"); };
+    sub_phrase { PARSE_ATTR("text"); PASS(block, "text", "sub_phrase"); };
+    span { PARSE_ATTR("text"); PASS(block, "text", "span"); };
+    span_phrase { PARSE_ATTR("text"); PASS(block, "text", "span_phrase"); };
+    cite { PARSE_ATTR("text"); PASS(block, "text", "cite"); };
+    ignore => ignore;
+    snip { PASS(block, "text", "snip"); };
+    quote1 { PASS(block, "text", "quote1"); };
+    quote2 { PASS(block, "text", "quote2"); };
+    multi_paragraph_quote { PASS(block, "text", "multi_paragraph_quote"); };
+    
+    ellipsis { INLINE(block, "ellipsis"); };
+    emdash { INLINE(block, "emdash"); };
+    endash { INLINE(block, "endash"); };
+    arrow { INLINE(block, "arrow"); };
+    caps { INLINE(block, "caps"); };
+    acronym { INLINE(block, "acronym"); };
+    dim { INLINE(block, "dim"); };
+    trademark { INLINE(block, "trademark"); };
+    registered { INLINE(block, "registered"); };
+    copyright { INLINE(block, "copyright"); };
+    footno { PASS(block, "text", "footno"); };
+    entity { INLINE(block, "entity"); };
+    
+    script_tag { INLINE(block, "inline_html"); };
+    start_tag { INLINE(block, "inline_html"); };
+    end_tag { INLINE(block, "inline_html"); };
+    empty_tag { INLINE(block, "inline_html"); };
+    html_comment { INLINE(block, "inline_html"); };
+    
+    redcloth_version { INLINE(block, "inline_redcloth_version"); };
+    
+    other_phrase => esc;
+    PUNCT => esc;
+    space => esc;
+    
+    EOF;
+    
+  *|;
+
+}%%;

data/ext/redcloth_scan/redcloth_scan.c.rl

@@ -0,0 +1,227 @@
+/*
+ * redcloth_scan.c.rl
+ *
+ * Copyright (C) 2008 Jason Garber
+ */
+#define redcloth_scan_c
+
+#include <ruby.h>
+#include "redcloth.h"
+
+VALUE mRedCloth, super_ParseError, super_RedCloth, super_HTML, super_LATEX;
+int SYM_escape_preformatted, SYM_escape_attributes;
+
+%%{
+
+  machine redcloth_scan;
+  include redcloth_common "redcloth_common.c.rl";
+
+  action extend { extend = rb_hash_aref(regs, ID2SYM(rb_intern("type"))); }
+
+  include redcloth_scan "redcloth_scan.rl";
+
+}%%
+
+%% write data nofinal;
+
+VALUE
+redcloth_transform(self, p, pe, refs)
+  VALUE self;
+  char *p, *pe;
+  VALUE refs;
+{
+  char *orig_p = p, *orig_pe = pe;
+  int cs, act, nest;
+  char *ts = NULL, *te = NULL, *reg = NULL, *bck = NULL, *eof = NULL;
+  VALUE html = rb_str_new2("");
+  VALUE table = rb_str_new2("");
+  VALUE block = rb_str_new2("");
+  VALUE regs; CLEAR_REGS()
+  
+  
+  VALUE list_layout = Qnil;
+  char *list_type = NULL;
+  VALUE list_index = rb_ary_new();
+  int list_continue = 0;
+  VALUE plain_block; SET_PLAIN_BLOCK("p");
+  VALUE extend = Qnil;
+  char listm[10] = "";
+  VALUE refs_found = rb_hash_new();
+  
+  %% write init;
+
+  %% write exec;
+
+  if (RSTRING_LEN(block) > 0)
+  {
+    ADD_BLOCK();
+  }
+
+  if ( NIL_P(refs) && rb_funcall(refs_found, rb_intern("empty?"), 0) == Qfalse ) {
+    return redcloth_transform(self, orig_p, orig_pe, refs_found);
+  } else {
+    rb_funcall(self, rb_intern("after_transform"), 1, html);
+    return html;
+  }
+}
+
+VALUE
+redcloth_transform2(self, str)
+  VALUE self, str;
+{
+  StringValue(str);
+  rb_funcall(self, rb_intern("before_transform"), 1, str);
+  return redcloth_transform(self, RSTRING_PTR(str), RSTRING_PTR(str) + RSTRING_LEN(str) + 1, Qnil);
+}
+
+/*
+ * Converts special characters into HTML entities.
+ */
+static VALUE
+redcloth_html_esc(int argc, VALUE* argv, VALUE self) //(self, str, level)
+{
+  VALUE str, level;
+  
+  rb_scan_args(argc, argv, "11", &str, &level);
+  
+  VALUE new_str = rb_str_new2("");
+  if (str == Qnil)
+    return new_str;
+    
+  StringValue(str);
+  
+  if (RSTRING_LEN(str) == 0)
+    return new_str;
+  
+  char *ts = RSTRING_PTR(str), *te = RSTRING_PTR(str) + RSTRING_LEN(str);
+  char *t = ts, *t2 = ts, *ch = NULL;
+  if (te <= ts) return;
+
+  while (t2 < te) {
+    ch = NULL;
+    
+    // normal + pre
+    switch (*t2)
+    {
+      case '&':  ch = "amp";    break;
+      case '>':  ch = "gt";     break;
+      case '<':  ch = "lt";     break;
+    }
+    
+    // normal (non-pre)
+    if (level != SYM_escape_preformatted) {
+      switch (*t2)
+      {
+        case '\n': ch = "br";     break;
+        case '"' : ch = "quot";   break;
+        case '\'': 
+          ch = (level == SYM_escape_attributes) ? "apos" : "squot";
+          break;
+      }
+    }
+    
+    if (ch != NULL)
+    {
+      if (t2 > t)
+        rb_str_cat(new_str, t, t2-t);
+      rb_str_concat(new_str, rb_funcall(self, rb_intern(ch), 1, rb_hash_new()));
+      t = t2 + 1;
+    }
+
+    t2++;
+  }
+  if (t2 > t)
+    rb_str_cat(new_str, t, t2-t);
+  
+  return new_str;
+}
+
+/*
+ * Converts special characters into LaTeX entities.
+ */
+static VALUE
+redcloth_latex_esc(VALUE self, VALUE str)
+{  
+  VALUE new_str = rb_str_new2("");
+  
+  if (str == Qnil)
+    return new_str;
+    
+  StringValue(str);
+  
+  if (RSTRING_LEN(str) == 0)
+    return new_str;
+  
+  char *ts = RSTRING_PTR(str), *te = RSTRING_PTR(str) + RSTRING_LEN(str);
+  char *t = ts, *t2 = ts, *ch = NULL;
+  if (te <= ts) return;
+
+  while (t2 < te) {
+    ch = NULL;
+    
+    switch (*t2) 
+    { 
+      case '{':  ch = "#123";   break;
+      case '}':  ch = "#125";   break;
+      case '\\': ch = "#92";    break;
+      case '#':  ch = "#35";    break;
+      case '$':  ch = "#36";    break;
+      case '%':  ch = "#37";    break;
+      case '&':  ch = "amp";    break;
+      case '_':  ch = "#95";    break;
+      case '^':  ch = "circ";   break;
+      case '~':  ch = "tilde";  break;
+      case '<':  ch = "lt";     break;
+      case '>':  ch = "gt";     break;
+      case '\n': ch = "#10";    break;
+    }
+
+    if (ch != NULL)
+    {
+      if (t2 > t)
+        rb_str_cat(new_str, t, t2-t);
+      VALUE opts = rb_hash_new();
+      rb_hash_aset(opts, ID2SYM(rb_intern("text")), rb_str_new2(ch));
+      rb_str_concat(new_str, rb_funcall(self, rb_intern("entity"), 1, opts));
+      t = t2 + 1;
+    }
+
+    t2++;
+  }
+  if (t2 > t)
+    rb_str_cat(new_str, t, t2-t);
+  
+  return new_str;
+}
+
+static VALUE
+redcloth_to(self, formatter)
+  VALUE self, formatter;
+{
+  char *pe, *p;
+  int len = 0;
+  
+  rb_funcall(self, rb_intern("delete!"), 1, rb_str_new2("\r"));
+  VALUE working_copy = rb_obj_clone(self);
+  rb_extend_object(working_copy, formatter);
+  if (rb_funcall(working_copy, rb_intern("lite_mode"), 0) == Qtrue) {
+    return redcloth_inline2(working_copy, self, rb_hash_new());
+  } else {
+    return redcloth_transform2(working_copy, self);
+  }
+}
+
+void Init_redcloth_scan()
+{
+  mRedCloth = rb_define_module("RedCloth");
+  /* A Textile document that can be converted to other formats. See
+   the README for Textile syntax. */
+  super_RedCloth = rb_define_class_under(mRedCloth, "TextileDoc", rb_cString);
+  rb_define_method(super_RedCloth, "to", redcloth_to, 1);
+  super_ParseError = rb_define_class_under(super_RedCloth, "ParseError", rb_eException);
+  /* Escaping */
+  rb_define_method(super_RedCloth, "html_esc", redcloth_html_esc, -1);
+  rb_define_method(super_RedCloth, "latex_esc", redcloth_latex_esc, 1);
+  SYM_escape_preformatted   = ID2SYM(rb_intern("html_escape_preformatted"));
+  SYM_escape_attributes     = ID2SYM(rb_intern("html_escape_attributes"));
+}