rpeg-multimarkdown 0.1.1 → 0.2

data/ext/markdown_peg.h

@@ -1,8 +1,9 @@
 /* markdown_peg.h */
-#include "markdown_lib.h"
-#include <glib.h>
+#ifndef MARKDOWN_PEG_H
+#define MARKDOWN_PEG_H
 
-extern char *strdup(const char *string);
+#include "markdown_lib.h"
+#include "glib.h"
 
 /* Information (label, URL and title) for a link. */
 struct Link {
@@ -133,10 +134,10 @@ void free_element_list(element * elt);
 void free_element(element *elt);
 void print_element_list(GString *out, element *elt, int format, int exts);
 
-
 element * parse_metadata_only(char *string, int extensions);
 char * extract_metadata_value(char *text, int extensions, char *key);
 
 char * metavalue_for_key(char *key, element *list);
 
 element * parse_markdown_for_opml(char *string, int extensions);
+#endif

data/ext/odf.c

@@ -14,6 +14,8 @@
 
  ***********************************************************************/
 
+#include "odf.h"
+
 
 void print_odf_header(GString *out){
     
@@ -99,7 +101,7 @@ void print_odf_header(GString *out){
     "             style:class=\"html\">\n" \
     "   <style:paragraph-properties fo:margin-left=\"0.3937in\" fo:margin-right=\"0.3937in\" fo:margin-top=\"0in\"\n" \
     "                               fo:margin-bottom=\"0.1965in\"\n" \
-    "                               fo:text-align=\"justify\"" \ 
+    "                               fo:text-align=\"justify\"" \
     "                               style:justify-single-word=\"false\"" \
     "                               fo:text-indent=\"0in\"\n" \
     "                               style:auto-text-indent=\"false\"/>\n" \
@@ -130,6 +132,10 @@ void print_odf_header(GString *out){
     "                               style:join-border=\"false\"/>\n" \
     "   <style:text-properties fo:font-size=\"6pt\" style:font-size-asian=\"6pt\" style:font-size-complex=\"6pt\"/>\n" \
     "</style:style>\n" \
+	"<style:style style:name=\"Footnote_20_anchor\" style:display-name=\"Footnote anchor\"" \
+	"              style:family=\"text\">" \
+	"    <style:text-properties style:text-position=\"super 58%\"/>" \
+	" </style:style>" \
     "</office:styles>\n");
 
     /* Automatic style information */

data/ext/odf.h

@@ -1,3 +1,11 @@
+#ifndef ODF_H
+#define ODF_H
+
 #include <stdlib.h>
 #include <stdio.h>
 #include <glib.h>
+
+void print_odf_header(GString *out);
+void print_odf_footer(GString *out);
+#endif
+

data/ext/parsing_functions.c

@@ -1,7 +1,25 @@
 /* parsing_functions.c - Functions for parsing markdown and
  * freeing element lists. */
 
-int yyparse(void);
+/* These yy_* functions come from markdown_parser.c which is
+ * generated from markdown_parser.leg
+ * */
+typedef int (*yyrule)();
+
+extern int yyparse();
+extern int yyparsefrom(yyrule);
+extern int yy_References();
+extern int yy_Notes();
+extern int yy_Doc();
+
+extern int yy_AutoLabels();
+extern int yy_DocWithMetaData();
+extern int yy_MetaDataOnly();
+extern int yy_DocForOPML();
+
+#include "utility_functions.h"
+#include "parsing_functions.h"
+#include "markdown_peg.h"
 
 static void free_element_contents(element elt);
 
@@ -41,6 +59,11 @@ static void free_element_contents(element elt) {
       case GLOSSARY:
       case GLOSSARYTERM:
       case NOTELABEL:
+      case CELLSPAN:
+      case EMDASH:
+      case ENDASH:
+      case GLOSSARYSORTKEY:
+      case MATHSPAN:
         free(elt.contents.str);
         elt.contents.str = NULL;
         break;
@@ -129,6 +152,12 @@ element * parse_markdown(char *string, int extensions, element *reference_list,
     yyparsefrom(yy_Doc);
 
     charbuf = oldcharbuf;          /* restore charbuf to original value */
+
+/*    if (parse_aborted) {
+        free_element_list(parse_result);
+        return NULL;
+    }*/
+
     return parse_result;
 
 }
@@ -144,9 +173,20 @@ element * parse_markdown_with_metadata(char *string, int extensions, element *re
     oldcharbuf = charbuf;
     charbuf = string;
 
-    yyparsefrom(yy_DocWithMetaData);
+	start_time = clock();
 
+    yyparsefrom(yy_DocWithMetaData);
     charbuf = oldcharbuf;          /* restore charbuf to original value */
+
+    /* reset start_time for subsequent passes */
+    start_time = 0;
+    
+    if (parse_aborted) {
+        parse_aborted = 0;
+        free_element_list(parse_result);
+        return NULL;
+    }
+
     return parse_result;
 
 }

data/ext/parsing_functions.h

@@ -0,0 +1,17 @@
+#ifndef PARSING_FUNCTIONS_H
+#define PARSING_FUNCTIONS_H
+/* parsing_functions.c - Functions for parsing markdown and
+ * freeing element lists. */
+
+#include "markdown_peg.h"
+
+/* free_element_list - free list of elements recursively */
+void free_element_list(element * elt);
+/* free_element - free element and contents */
+void free_element(element *elt);
+
+element * parse_references(char *string, int extensions);
+element * parse_notes(char *string, int extensions, element *reference_list);
+element * parse_markdown(char *string, int extensions, element *reference_list, element *note_list, element *label_list);
+
+#endif

data/ext/utility_functions.c

@@ -1,12 +1,12 @@
 /* utility_functions.c - List manipulation functions, element
  * constructors, and macro definitions for leg markdown parser. */
 
-extern int strcasecmp(const char *string1, const char *string2);
+#include "utility_functions.h"
+#include "markdown_peg.h"
 
-static char *label_from_string(char *str, bool obfuscate) ;
-static void localize_typography(GString *out, int character, int language, int output);
+#include <string.h>
+#include <assert.h>
 
-static void print_raw_element_list(GString *out, element *list);
 
 /**********************************************************************
 
@@ -15,14 +15,14 @@ static void print_raw_element_list(GString *out, element *list);
  ***********************************************************************/
 
 /* cons - cons an element onto a list, returning pointer to new head */
-static element * cons(element *new, element *list) {
+element * cons(element *new, element *list) {
     assert(new != NULL);
     new->next = list;
     return new;
 }
 
 /* reverse - reverse a list, returning pointer to new list */
-static element *reverse(element *list) {
+element *reverse(element *list) {
     element *new = NULL;
     element *next = NULL;
     while (list != NULL) {
@@ -34,7 +34,7 @@ static element *reverse(element *list) {
 }
 
 /* append_list - add element to end of list */
-static void append_list(element *new, element *list) {
+void append_list(element *new, element *list) {
     assert(new != NULL);
     element *step = list;
     
@@ -48,7 +48,7 @@ static void append_list(element *new, element *list) {
 
 /* concat_string_list - concatenates string contents of list of STR elements.
  * Frees STR elements as they are added to the concatenation. */
-static GString *concat_string_list(element *list) {
+GString *concat_string_list(element *list) {
     GString *result;
     element *next;
     result = g_string_new("");
@@ -69,13 +69,16 @@ static GString *concat_string_list(element *list) {
 
  ***********************************************************************/
 
-static char *charbuf = "";     /* Buffer of characters to be parsed. */
-static element *references = NULL;    /* List of link references found. */
-static element *notes = NULL;         /* List of footnotes found. */
-static element *parse_result;  /* Results of parse. */
+
+char *charbuf = "";     /* Buffer of characters to be parsed. */
+element *references = NULL;    /* List of link references found. */
+element *notes = NULL;         /* List of footnotes found. */
+element *parse_result;  /* Results of parse. */
 int syntax_extensions;  /* Syntax extensions selected. */
 
-static element *labels = NULL;      /* List of labels found in document. */
+element *labels = NULL;      /* List of labels found in document. */
+clock_t start_time = 0;                 /* Used for ensuring we're not stuck in a loop */
+bool parse_aborted = 0;      /* flag indicating we ran out of time */
 
 /**********************************************************************
 
@@ -86,7 +89,7 @@ static element *labels = NULL;      /* List of labels found in document. */
  ***********************************************************************/
 
 /* mk_element - generic constructor for element */
-static element * mk_element(int key) {
+element * mk_element(int key) {
     element *result = malloc(sizeof(element));
     result->key = key;
     result->children = NULL;
@@ -96,7 +99,7 @@ static element * mk_element(int key) {
 }
 
 /* mk_str - constructor for STR element */
-static element * mk_str(char *string) {
+element * mk_str(char *string) {
     element *result;
     assert(string != NULL);
     result = mk_element(STR);
@@ -106,7 +109,7 @@ static element * mk_str(char *string) {
 
 /* mk_str_from_list - makes STR element by concatenating a
  * reversed list of strings, adding optional extra newline */
-static element * mk_str_from_list(element *list, bool extra_newline) {
+element * mk_str_from_list(element *list, bool extra_newline) {
     element *result;
     GString *c = concat_string_list(reverse(list));
     if (extra_newline)
@@ -120,7 +123,7 @@ static element * mk_str_from_list(element *list, bool extra_newline) {
 /* mk_list - makes new list with key 'key' and children the reverse of 'lst'.
  * This is designed to be used with cons to build lists in a parser action.
  * The reversing is necessary because cons adds to the head of a list. */
-static element * mk_list(int key, element *lst) {
+element * mk_list(int key, element *lst) {
     element *result;
     result = mk_element(key);
     result->children = reverse(lst);
@@ -128,7 +131,7 @@ static element * mk_list(int key, element *lst) {
 }
 
 /* mk_link - constructor for LINK element */
-static element * mk_link(element *label, char *url, char *title, element *attr, char *id) {
+element * mk_link(element *label, char *url, char *title, element *attr, char *id) {
     element *result;
     result = mk_element(LINK);
     result->contents.link = malloc(sizeof(link));
@@ -141,12 +144,12 @@ static element * mk_link(element *label, char *url, char *title, element *attr,
 }
 
 /* extension = returns true if extension is selected */
-static bool extension(int ext) {
+bool extension(int ext) {
     return (syntax_extensions & ext);
 }
 
 /* match_inlines - returns true if inline lists match (case-insensitive...) */
-static bool match_inlines(element *l1, element *l2) {
+bool match_inlines(element *l1, element *l2) {
     while (l1 != NULL && l2 != NULL) {
         if (l1->key != l2->key)
             return false;
@@ -190,7 +193,7 @@ static bool match_inlines(element *l1, element *l2) {
 
 /* find_reference - return true if link found in references matching label.
  * 'link' is modified with the matching url and title. */
-static bool find_reference(link *result, element *label) {
+bool find_reference(link *result, element *label) {
     element *cur = references;  /* pointer to walk up list of references */
     link *curitem;
     while (cur != NULL) {
@@ -208,7 +211,7 @@ static bool find_reference(link *result, element *label) {
 /* find_note - return true if note found in notes matching label.
 if found, 'result' is set to point to matched note. */
 
-static bool find_note(element **result, char *label) {
+bool find_note(element **result, char *label) {
    element *cur = notes;  /* pointer to walk up list of notes */
    while (cur != NULL) {
        if (strcmp(label, cur->contents.str) == 0) {
@@ -222,36 +225,10 @@ static bool find_note(element **result, char *label) {
 }
 
 
-
-/**********************************************************************
-
-  Definitions for leg parser generator.
-  YY_INPUT is the function the parser calls to get new input.
-  We take all new input from (static) charbuf.
-
- ***********************************************************************/
-
-# define YYSTYPE element *
-#ifdef __DEBUG__
-# define YY_DEBUG 1
-#endif
-
-#define YY_INPUT(buf, result, max_size)              \
-{                                                    \
-    int yyc;                                         \
-    if (charbuf && *charbuf != '\0') {               \
-        yyc= *charbuf++;                             \
-    } else {                                         \
-        yyc= EOF;                                    \
-    }                                                \
-    result= (EOF == yyc) ? 0 : (*(buf)= yyc, 1);     \
-}
-
-
 /* peg-multimarkdown additions */
 
 /* print_raw_element - print an element as original text */
-static void print_raw_element(GString *out, element *elt) {
+void print_raw_element(GString *out, element *elt) {
     if (elt->key == LINK) {
         print_raw_element_list(out,elt->contents.link->label);
     } else {
@@ -264,7 +241,7 @@ static void print_raw_element(GString *out, element *elt) {
 }
 
 /* print_raw_element_list - print a list of elements as original text */
-static void print_raw_element_list(GString *out, element *list) {
+void print_raw_element_list(GString *out, element *list) {
     while (list != NULL) {
         print_raw_element(out, list);
         list = list->next;
@@ -274,7 +251,7 @@ static void print_raw_element_list(GString *out, element *list) {
 /* label_from_element_list */
 /* Returns a null-terminated string, which must be freed after use. */
 
-static char *label_from_element_list(element *list, bool obfuscate) {
+char *label_from_element_list(element *list, bool obfuscate) {
     char *label;
     char *label2;
     GString *raw = g_string_new("");
@@ -290,7 +267,7 @@ static char *label_from_element_list(element *list, bool obfuscate) {
     HTML id */
 /* Returns a null-terminated string, which must be freed after use. */
 
-static char *label_from_string(char *str, bool obfuscate) {
+char *label_from_string(char *str, bool obfuscate) {
     bool valid = FALSE;
     GString *out = g_string_new("");
     char *label;
@@ -321,7 +298,7 @@ static char *label_from_string(char *str, bool obfuscate) {
 
 /* find_label - return true if header, table, etc is found matching label.
  * 'link' is modified with the matching url and title. */
-static bool find_label(link *result, element *label) {
+bool find_label(link *result, element *label) {
     char *lab;
     element *cur = labels;  /* pointer to walk up list of references */
     GString *text = g_string_new("");
@@ -347,7 +324,7 @@ static bool find_label(link *result, element *label) {
 /* localize_typography - return the proper string, based on language chosen */
 /* Default action is English */
 
-static void localize_typography(GString *out, int character, int lang, int output) {
+void localize_typography(GString *out, int character, int lang, int output) {
 
     switch (output) {
         case HTMLOUT:
@@ -525,11 +502,38 @@ static void localize_typography(GString *out, int character, int lang, int outpu
 }
 
 /* Trim spaces at end of string */
-static void trim_trailing_whitespace(char *str) {    
+void trim_trailing_whitespace(char *str) {    
     while ( ( str[strlen(str)-1] == ' ' ) ||
         ( str[strlen(str)-1] == '\n' ) || 
         ( str[strlen(str)-1] == '\r' ) || 
         ( str[strlen(str)-1] == '\t' ) ) {
         str[strlen(str)-1] = '\0';
     }
-}
+}
+
+/* Don't let us get caught in "infinite" loop */
+bool check_timeout() {
+    /* Once we abort, keep aborting */
+    if (parse_aborted)
+        return 0;
+    
+    /* We're not timing this run */
+    if (start_time == 0)
+        return 1;
+
+    clock_t end = clock();
+    double elapsed = ((double) (end - start_time)) / CLOCKS_PER_SEC;
+    
+	/* fprintf(stderr,"%2.2f elapsed; (%4.2f CLOCKS_PER_SEC)\n",elapsed,CLOCKS_PER_SEC); */
+	/* fprintf(stderr,"%2.2f elapsed\n",elapsed); */
+	
+	
+    /* If > 3 clock seconds, then abort */
+    float max = 3;
+    if (elapsed > max) {
+        parse_aborted = 1;
+        return 0;
+    }
+    return 1;
+}
+

data/ext/utility_functions.h

@@ -0,0 +1,93 @@
+#ifndef UTILITY_FUNCTIONS_H
+#define UTILITY_FUNCTIONS_H
+
+#include <stdbool.h>
+#include <glib.h>
+
+#include "markdown_peg.h"
+
+#include <time.h>
+
+/* utility_functions.h - List manipulation functions, element
+ * constructors, and macro definitions for leg markdown parser. */
+
+
+/* cons - cons an element onto a list, returning pointer to new head */
+element * cons(element *new, element *list);
+
+/* reverse - reverse a list, returning pointer to new list */
+element *reverse(element *list);
+/* concat_string_list - concatenates string contents of list of STR elements.
+ * Frees STR elements as they are added to the concatenation. */
+GString *concat_string_list(element *list);
+/**********************************************************************
+
+  Global variables used in parsing
+
+ ***********************************************************************/
+
+extern char *charbuf;          /* Buffer of characters to be parsed. */
+extern element *references;    /* List of link references found. */
+extern element *notes;         /* List of footnotes found. */
+extern element *parse_result;  /* Results of parse. */
+extern int syntax_extensions;  /* Syntax extensions selected. */
+
+
+extern element *labels;	       /* List of labels found in document. */
+extern clock_t start_time;     /* Used for ensuring we're not stuck in a loop */
+extern bool parse_aborted;     /* flag indicating we ran out of time */
+
+/**********************************************************************
+
+  Auxiliary functions for parsing actions.
+  These make it easier to build up data structures (including lists)
+  in the parsing actions.
+
+ ***********************************************************************/
+
+/* mk_element - generic constructor for element */
+element * mk_element(int key);
+
+/* mk_str - constructor for STR element */
+element * mk_str(char *string);
+
+/* mk_str_from_list - makes STR element by concatenating a
+ * reversed list of strings, adding optional extra newline */
+element * mk_str_from_list(element *list, bool extra_newline);
+
+/* mk_list - makes new list with key 'key' and children the reverse of 'lst'.
+ * This is designed to be used with cons to build lists in a parser action.
+ * The reversing is necessary because cons adds to the head of a list. */
+element * mk_list(int key, element *lst);
+
+/* mk_link - constructor for LINK element */
+element * mk_link(element *label, char *url, char *title, element *attr, char *id);
+/* extension = returns true if extension is selected */
+bool extension(int ext);
+
+/* match_inlines - returns true if inline lists match (case-insensitive...) */
+bool match_inlines(element *l1, element *l2);
+
+/* find_reference - return true if link found in references matching label.
+ * 'link' is modified with the matching url and title. */
+bool find_reference(link *result, element *label);
+
+/* find_note - return true if note found in notes matching label.
+if found, 'result' is set to point to matched note. */
+
+bool find_note(element **result, char *label);
+
+
+char *label_from_string(char *str, bool obfuscate);
+void localize_typography(GString *out, int character, int language, int output);
+void print_raw_element_list(GString *out, element *list);   
+void append_list(element *new, element *list);
+bool find_label(link *result, element *label);
+bool check_timeout();
+void trim_trailing_whitespace(char *str);
+char *label_from_element_list(element *list, bool obfuscate);
+void print_raw_element_list(GString *out, element *list);
+void print_raw_element(GString *out, element *elt);
+
+#endif
+