0
@@ -2,7 +2,7 @@ ALL : markdown
0
 
0
 PROGRAM=markdown
0
 MYGETOPTDIR=my_getopt-1.5
0
-OBJS=$(MYGETOPTDIR)/my_getopt.o markdown_peg.o
0
+OBJS=$(MYGETOPTDIR)/my_getopt.o markdown_parser.o markdown_output.o
0
 PEGDIR=peg-0.1.4
0
 LEG=$(PEGDIR)/leg
0
 
0
@@ -15,7 +15,7 @@ $(LEG):
0
 markdown : markdown.c $(OBJS)
0
   $(CC) -Wall -O3 -ansi -o $@ $(OBJS) $<
0
 
0
-markdown_peg.c : markdown_peg.leg $(LEG)
0
+markdown_parser.c : markdown_parser.leg $(LEG)
0
   $(LEG) -o $@ $<
0
 
0
 .PHONY: clean test

0
@@ -27,570 +27,6 @@ static int extensions;
0
 
0
 /**********************************************************************
0
 
0
-  Utility functions for printing
0
-
0
- ***********************************************************************/
0
-
0
-static int padded = 2;      /* Number of newlines after last output.
0
-                               Starts at 2 so no newlines are needed at start.
0
-                               */
0
-
0
-/* pad - add newlines if needed */
0
-void pad(int num) {
0
-    while (num-- > padded)
0
-        printf("\n");
0
-    padded = num;
0
-}
0
-
0
-/**********************************************************************
0
-
0
-  Functions for printing Elements as HTML
0
-
0
- ***********************************************************************/
0
-
0
-/* print_html_string - print string, escaping for HTML  
0
- * If obfuscate selected, convert characters to hex or decimal entities at random */
0
-void print_html_string(char *str, bool obfuscate) {
0
-    while (*str != '\0') {
0
-        switch (*str) {
0
-        case '&':
0
-            printf("&");
0
-            break;
0
-        case '<':
0
-            printf("&lt;");
0
-            break;
0
-        case '>':
0
-            printf("&gt;");
0
-            break;
0
-        case '"':
0
-            printf("&quot;");
0
-            break;
0
-        default:
0
-            if (obfuscate) {
0
-                if (rand() % 2 == 0)
0
-                    printf("&#%d;", *str);
0
-                else
0
-                    printf("&#x%x;", *str);
0
-            }
0
-            else
0
-                putchar(*str);
0
-            break;
0
-        }
0
-    str++;
0
-    }
0
-}
0
-
0
-void print_html_element(element elt, bool obfuscate);
0
-
0
-/* print_html_element_list - print a list of elements as HTML */
0
-void print_html_element_list(item *list, bool obfuscate) {
0
-    while (list != NULL) {
0
-        print_html_element((*list).val, obfuscate);
0
-        list = (*list).next;
0
-    }
0
-}
0
-
0
-/* print_html_element - print an element as HTML */
0
-void print_html_element(element elt, bool obfuscate) {
0
-    int lev;
0
-    char *contents;
0
-    switch (elt.key) {
0
-    case SPACE:
0
-        printf("%s", elt.contents.str);
0
-        break;
0
-    case LINEBREAK:
0
-        printf("<br/>");
0
-        break;
0
-    case STR:
0
-        print_html_string(elt.contents.str, obfuscate);
0
-        break;
0
-    case ELLIPSIS:
0
-        printf("&hellip;");
0
-        break;
0
-    case CODE:
0
-        printf("<code>");
0
-        print_html_string(elt.contents.str, obfuscate);
0
-        printf("</code>");
0
-        break;
0
-    case HTML:
0
-        printf(elt.contents.str);
0
-        break;
0
-    case LINK:
0
-        if (strstr(elt.contents.link.url, "mailto:") == elt.contents.link.url)
0
-            obfuscate = true;  /* obfuscate mailto: links */
0
-        printf("<a href=\"");
0
-        print_html_string(elt.contents.link.url, obfuscate);
0
-        printf("\"");
0
-        if (strlen(elt.contents.link.title) > 0) {
0
-            printf(" title=\"");
0
-            print_html_string(elt.contents.link.title, obfuscate);
0
-            printf("\"");
0
-        }
0
-        printf(">");     
0
-        print_html_element_list(elt.contents.link.label, obfuscate);
0
-        printf("</a>");
0
-        break;
0
-    case IMAGE:
0
-        printf("<img src=\"");
0
-        print_html_string(elt.contents.link.url, obfuscate);
0
-        printf("\" alt=\"");
0
-        print_html_element_list(elt.contents.link.label, obfuscate);
0
-        printf("\"");
0
-        if (strlen(elt.contents.link.title) > 0) {
0
-            printf(" title=\"");
0
-            print_html_string(elt.contents.link.title, obfuscate);
0
-            printf("\"");
0
-        }
0
-        printf(" />");
0
-        break;
0
-    case EMPH:
0
-        printf("<em>");
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        printf("</em>");
0
-        break;
0
-    case STRONG:
0
-        printf("<strong>");
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        printf("</strong>");
0
-        break;
0
-    case LIST:
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        break;
0
-    case H1: case H2: case H3: case H4: case H5: case H6:
0
-        lev = elt.key - H1 + 1;  /* assumes H1 ... H6 are in order */
0
-        pad(2);
0
-        printf("<h%1d>", lev);
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        printf("</h%1d>", lev);
0
-        padded = 0;
0
-        break;
0
-    case PLAIN:
0
-        pad(1);
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        padded = 0;
0
-        break;
0
-    case PARA:
0
-        pad(2);
0
-        printf("<p>");
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        printf("</p>");
0
-        padded = 0;
0
-        break;
0
-    case HRULE:
0
-        pad(2);
0
-        printf("<hr />");
0
-        padded = 0;
0
-        break;
0
-    case HTMLBLOCK:
0
-        pad(2);
0
-        printf(elt.contents.str);
0
-        padded = 0;
0
-        break;
0
-    case VERBATIM:
0
-        pad(2);
0
-        printf("<pre><code>");
0
-        print_html_string(elt.contents.str, obfuscate);
0
-        printf("</code></pre>");
0
-        padded = 0;
0
-        break;
0
-    case BULLETLIST:
0
-        pad(2);
0
-        printf("<ul>");
0
-        padded = 0;
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        pad(1);
0
-        printf("</ul>");
0
-        padded = 0;
0
-        break;
0
-    case ORDEREDLIST:
0
-        pad(2);
0
-        printf("<ol>");
0
-        padded = 0;
0
-        print_html_element_list(elt.contents.list, obfuscate);
0
-        pad(1);
0
-        printf("</ol>");
0
-        padded = 0;
0
-        break;
0
-    case LISTITEM:
0
-        pad(1);
0
-        printf("<li>");
0
-        padded = true;
0
-        /* \001 is used to indicate boundaries between nested lists when there
0
-         * is no blank line.  We split the string by \001 and parse
0
-         * each chunk separately. */
0
-        contents = strtok(elt.contents.str, "\001");
0
-        print_html_element(markdown(contents, extensions), obfuscate);
0
-        while ((contents = strtok(NULL, "\001")))
0
-            print_html_element(markdown(contents, extensions), obfuscate);
0
-        printf("</li>");
0
-        padded = 0;
0
-        break;
0
-    case BLOCKQUOTE:
0
-        pad(2);
0
-        printf("<blockquote>");
0
-        print_html_element(markdown(elt.contents.str, extensions), obfuscate);
0
-        printf("</blockquote>");
0
-        padded = 0;
0
-        break;
0
-    case REFERENCE:
0
-        /* Nonprinting */
0
-        break;
0
-    default: 
0
-        fprintf(stderr, "print_html_element encountered unknown element key = %d\n", elt.key); 
0
-        exit(EXIT_FAILURE);
0
-        break;
0
-    }
0
-}
0
-
0
-/**********************************************************************
0
-
0
-  Functions for printing Elements as LaTeX
0
-
0
- ***********************************************************************/
0
-
0
-/* print_latex_string - print string, escaping for LaTeX */
0
-void print_latex_string(char *str) {
0
-    while (*str != '\0') {
0
-        switch (*str) {
0
-          case '{': case '}': case '$': case '%':
0
-          case '&': case '_': case '#':
0
-            printf("\\%c", *str);
0
-            break;
0
-        case '^':
0
-            printf("\\^{}");
0
-            break;
0
-        case '\\':
0
-            printf("\\textbackslash{}");
0
-            break;
0
-        case '~':
0
-            printf("\\ensuremath{\\sim}");
0
-            break;
0
-        case '|':
0
-            printf("\\textbar{}");
0
-            break;
0
-        case '<':
0
-            printf("\\textless{}");
0
-            break;
0
-        case '>':
0
-            printf("\\textgreater{}");
0
-            break;
0
-        default:
0
-            putchar(*str);
0
-            break;
0
-        }
0
-    str++;
0
-    }
0
-}
0
-
0
-void print_latex_element(element elt);
0
-
0
-/* print_latex_element_list - print a list of elements as LaTeX */
0
-void print_latex_element_list(item *list) {
0
-    while (list != NULL) {
0
-        print_latex_element((*list).val);
0
-        list = (*list).next;
0
-    }
0
-}
0
-
0
-/* print_latex_element - print an element as LaTeX */
0
-void print_latex_element(element elt) {
0
-    int lev;
0
-    int i;
0
-    char *contents;
0
-    switch (elt.key) {
0
-    case SPACE:
0
-        printf("%s", elt.contents.str);
0
-        break;
0
-    case LINEBREAK:
0
-        printf("\\\\\n");
0
-        break;
0
-    case STR:
0
-        print_latex_string(elt.contents.str);
0
-        break;
0
-    case ELLIPSIS:
0
-        printf("\\ldots{}");
0
-        break;
0
-    case CODE:
0
-        printf("\\texttt{");
0
-        print_latex_string(elt.contents.str);
0
-        printf("}");
0
-        break;
0
-    case HTML:
0
-        /* don't print HTML */
0
-        break;
0
-    case LINK:
0
-        printf("\\href{%s}{", elt.contents.link.url);
0
-        print_latex_element_list(elt.contents.link.label);
0
-        printf("}");
0
-        break;
0
-    case IMAGE:
0
-        printf("\\includegraphics{%s}", elt.contents.link.url);
0
-        break;
0
-    case EMPH:
0
-        printf("\\emph{");
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("}");
0
-        break;
0
-    case STRONG:
0
-        printf("\\textbf{");
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("}");
0
-        break;
0
-    case LIST:
0
-        print_latex_element_list(elt.contents.list);
0
-        break;
0
-    case H1: case H2: case H3:
0
-        lev = elt.key - H1 + 1;  /* assumes H1 ... H6 are in order */
0
-        printf("\\");
0
-        for (i = elt.key; i > H1; i--)
0
-            printf("sub");
0
-        printf("section{");
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("}\n\n");
0
-        break;
0
-    case H4: case H5: case H6:
0
-        printf("\\noindent\\textbf{");
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("}\n");
0
-    case PLAIN:
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("\n");
0
-        break;
0
-    case PARA:
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("\n\n"); 
0
-        break;
0
-    case HRULE:
0
-        printf("\\begin{center}\\rule{3in}{0.4pt}\\end{center}\n");
0
-        break;
0
-    case HTMLBLOCK:
0
-        /* don't print HTML block */
0
-        break;
0
-    case VERBATIM:
0
-        printf("\\begin{verbatim}\n");
0
-        print_latex_string(elt.contents.str);
0
-        printf("\n\\end{verbatim}\n");
0
-        break;
0
-    case BULLETLIST:
0
-        printf("\\begin{itemize}\n");
0
-        padded = 0;
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("\\end{itemize}\n\n");
0
-        break;
0
-    case ORDEREDLIST:
0
-        printf("\\begin{enumerate}\n");
0
-        print_latex_element_list(elt.contents.list);
0
-        printf("\\end{enumerate}\n\n");
0
-        break;
0
-    case LISTITEM:
0
-        printf("\\item ");
0
-        /* \001 is used to indicate boundaries between nested lists when there
0
-         * is no blank line.  We split the string by \001 and parse
0
-         * each chunk separately. */
0
-        contents = strtok(elt.contents.str, "\001");
0
-        print_latex_element(markdown(contents, extensions));
0
-        while ((contents = strtok(NULL, "\001")))
0
-            print_latex_element(markdown(contents, extensions));
0
-        break;
0
-    case BLOCKQUOTE:
0
-        printf("\\begin{quote}");
0
-        print_latex_element(markdown(elt.contents.str, extensions));
0
-        printf("\\end{quote}\n\n");
0
-        break;
0
-    case REFERENCE:
0
-        /* Nonprinting */
0
-        break;
0
-    default: 
0
-        fprintf(stderr, "print_latex_element encountered unknown element key = %d\n", elt.key); 
0
-        exit(EXIT_FAILURE);
0
-        break;
0
-    }
0
-}
0
-
0
-/**********************************************************************
0
-
0
-  Functions for printing Elements as groff (mm macros)
0
-
0
- ***********************************************************************/
0
-
0
-static bool in_list_item = false; /* True if we're parsing contents of a list item. */
0
-
0
-/* print_groff_string - print string, escaping for groff */
0
-void print_groff_string(char *str) {
0
-    while (*str != '\0') {
0
-        switch (*str) {
0
-        case '\\':
0
-            printf("\\e");
0
-            break;
0
-        default:
0
-            putchar(*str);
0
-            break;
0
-        }
0
-    str++;
0
-    }
0
-}
0
-
0
-void print_groff_mm_element(element elt, int count);
0
-
0
-/* print_groff_mm_element_list - print a list of elements as groff ms */
0
-void print_groff_mm_element_list(item *list) {
0
-    int count = 1;
0
-    while (list != NULL) {
0
-        print_groff_mm_element((*list).val, count);
0
-        list = (*list).next;
0
-        count++;
0
-    }
0
-}
0
-
0
-/* print_groff_mm_element - print an element as groff ms */
0
-void print_groff_mm_element(element elt, int count) {
0
-    int lev;
0
-    char *contents;
0
-    switch (elt.key) {
0
-    case SPACE:
0
-        printf("%s", elt.contents.str);
0
-        padded = 0;
0
-        break;
0
-    case LINEBREAK:
0
-        pad(1);
0
-        printf(".br");
0
-        padded = 0;
0
-        break;
0
-    case STR:
0
-        print_groff_string(elt.contents.str);
0
-        padded = 0;
0
-        break;
0
-    case ELLIPSIS:
0
-        printf("...");
0
-        break;
0
-    case CODE:
0
-        printf("\\fC");
0
-        print_groff_string(elt.contents.str);
0
-        printf("\\fR");
0
-        padded = 0;
0
-        break;
0
-    case HTML:
0
-        /* don't print HTML */
0
-        break;
0
-    case LINK:
0
-        print_groff_mm_element_list(elt.contents.link.label);
0
-        printf(" (%s)", elt.contents.link.url);
0
-        padded = 0;
0
-        break;
0
-    case IMAGE:
0
-        printf("[IMAGE: ");
0
-        print_groff_mm_element_list(elt.contents.link.label);
0
-        printf("]");
0
-        padded = 0;
0
-        /* not supported */
0
-        break;
0
-    case EMPH:
0
-        printf("\\fI");
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        printf("\\fR");
0
-        padded = 0;
0
-        break;
0
-    case STRONG:
0
-        printf("\\fB");
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        printf("\\fR");
0
-        padded = 0;
0
-        break;
0
-    case LIST:
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        padded = 0;
0
-        break;
0
-    case H1: case H2: case H3: case H4: case H5: case H6:
0
-        lev = elt.key - H1 + 1;
0
-        pad(1);
0
-        printf(".H %d \"", lev);
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        printf("\"");
0
-        padded = 0;
0
-        break;
0
-    case PLAIN:
0
-        pad(1);
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        padded = 0;
0
-        break;
0
-    case PARA:
0
-        pad(1);
0
-        if (!in_list_item || count != 1)
0
-            printf(".P\n");
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        padded = 0;
0
-        break;
0
-    case HRULE:
0
-        pad(1);
0
-        printf("\\l'\\n(.lu*8u/10u'");
0
-        padded = 0;
0
-        break;
0
-    case HTMLBLOCK:
0
-        /* don't print HTML block */
0
-        break;
0
-    case VERBATIM:
0
-        pad(1);
0
-        printf(".VERBON 2\n");
0
-        print_groff_string(elt.contents.str);
0
-        printf(".VERBOFF");
0
-        padded = 0;
0
-        break;
0
-    case BULLETLIST:
0
-        pad(1);
0
-        printf(".BL");
0
-        padded = 0;
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        pad(1);
0
-        printf(".LE 1");
0
-        padded = 0;
0
-        break;
0
-    case ORDEREDLIST:
0
-        pad(1);
0
-        printf(".AL");
0
-        padded = 0;
0
-        print_groff_mm_element_list(elt.contents.list);
0
-        pad(1);
0
-        printf(".LE 1");
0
-        padded = 0;
0
-        break;
0
-    case LISTITEM:
0
-        pad(1);
0
-        printf(".LI\n");
0
-        in_list_item = true;
0
-        /* \001 is used to indicate boundaries between nested lists when there
0
-         * is no blank line.  We split the string by \001 and parse
0
-         * each chunk separately. */
0
-        contents = strtok(elt.contents.str, "\001");
0
-        padded = 2;
0
-        print_groff_mm_element(markdown(contents, extensions), 1);
0
-        while ((contents = strtok(NULL, "\001"))) {
0
-            padded = 2;
0
-            print_groff_mm_element(markdown(contents, extensions), 1);
0
-        }
0
-        in_list_item = false;
0
-        break;
0
-    case BLOCKQUOTE:
0
-        pad(1);
0
-        printf(".DS I\n");
0
-        padded = 2;
0
-        print_groff_mm_element(markdown(elt.contents.str, extensions), 1);
0
-        pad(1);
0
-        printf(".DE");
0
-        padded = 0;
0
-        break;
0
-    case REFERENCE:
0
-        /* Nonprinting */
0
-        break;
0
-    default: 
0
-        fprintf(stderr, "print_groff_mm_element encountered unknown element key = %d\n", elt.key); 
0
-        exit(EXIT_FAILURE);
0
-        break;
0
-    }
0
-}
0
-
0
-
0
-/**********************************************************************
0
-
0
   Main program and auxiliary functions.
0
 
0
   Reads input from files on command line, or from stdin
0
@@ -653,13 +89,6 @@ int main(int argc, char * argv[]) {
0
     char *format = 0;
0
     char *exts = 0;
0
 
0
-    /* Output formats. */
0
-    enum formats { HTML_FORMAT,
0
-                   LATEX_FORMAT,
0
-                   GROFF_MM_FORMAT
0
-                 };
0
-
0
- 
0
     int output_format = HTML_FORMAT;
0
 
0
     char *shortopts = "Vhx::o:t:";
0
@@ -799,21 +228,7 @@ int main(int argc, char * argv[]) {
0
    
0
     element parsed_input = markdown(inputbuf, extensions);
0
 
0
-    switch (output_format) {
0
-    case HTML_FORMAT:
0
-        print_html_element(parsed_input, false);
0
-        break;
0
-    case LATEX_FORMAT:
0
-        print_latex_element(parsed_input);
0
-        break;
0
-    case GROFF_MM_FORMAT:
0
-        print_groff_mm_element(parsed_input, 1);
0
-        break;
0
-    default:
0
-        fprintf(stderr, "print_element - unknown format = %d\n", output_format); 
0
-        exit(EXIT_FAILURE);
0
-        break;
0
-    }
0
+    print_element(parsed_input, output_format);
0
 
0
     printf("\n");
0