Added LaTeX renderer.

* New exported function in API: `cmark_render_latex`.
* Added src/latex.hs.
* Updated README and man page.
* Closes #31.

8 files changed, 635 insertions, 8 deletions

diff --git a/Makefile b/Makefile
index f11431c..ddcd105 100644
--- a/Makefile
+++ b/Makefile
@@ -132,7 +132,7 @@ $(ALLTESTS): $(SPEC)
 
 leakcheck: $(ALLTESTS)
 	rc=0; \
-	for format in html man xml commonmark; do \
+	for format in html man xml latex commonmark; do \
 	  for opts in "" "--smart" "--normalize"; do \
 	     echo "cmark -t $$format $$opts" ; \
 	     cat $< | valgrind -q --leak-check=full --dsymutil=yes --error-exitcode=1 $(PROG) -t $$format $$opts >/dev/null || rc=1; \
diff --git a/README.md b/README.md
index b9be9b1..c0ca22d 100644
--- a/README.md
+++ b/README.md
@@ -11,7 +11,7 @@ rationalized version of Markdown syntax with a [spec][the spec].
 
 It provides a shared library (`libcmark`) with functions for parsing
 CommonMark documents to an abstract syntax tree (AST), manipulating
-the AST, and rendering the document to HTML, groff man,
+the AST, and rendering the document to HTML, groff man, LaTeX,
 CommonMark, or an XML representation of the AST.  It also provides a
 command-line program (`cmark`) for parsing and rendering CommonMark
 documents.
@@ -43,7 +43,7 @@ Advantages of this library:
 - **Flexible.** CommonMark input is parsed to an AST which can be
   manipulated programatically prior to rendering.
 
-- **Multiple renderers.**  Output in HTML, groff man, CommonMark,
+- **Multiple renderers.**  Output in HTML, groff man, LaTeX, CommonMark,
   and a custom XML format is supported. And it is easy to write new
   renderers to support other formats.
 
diff --git a/man/man1/cmark.1 b/man/man1/cmark.1
index 8c1c2c7..64fa697 100644
--- a/man/man1/cmark.1
+++ b/man/man1/cmark.1
@@ -10,7 +10,7 @@ file*
 .SH "DESCRIPTION"
 \fBcmark\fR
 converts Markdown formatted plain text to either HTML, groff man,
-CommonMark XML, or CommonMark, using the conventions
+CommonMark XML, LaTeX, or CommonMark, using the conventions
 described in the CommonMark spec.  It reads input from \fIstdin\fR
 or the specified files (concatenating their contents) and writes
 output to \fIstdout\fR.
@@ -18,7 +18,7 @@ output to \fIstdout\fR.
 .TP 12n
 .B \-\-to, \-t \f[I]FORMAT\f[]
 Specify output format (\f[C]html\f[], \f[C]man\f[], \f[C]xml\f[],
-\f[C]commonmark\f[]).
+\f[C]latex\f[], \f[C]commonmark\f[]).
 .TP 12n
 .B \-\-width \f[I]WIDTH\f[]
 Specify a column width to which to wrap the output. For no wrapping, use
diff --git a/man/man3/cmark.3 b/man/man3/cmark.3
index 6edac7b..6521126 100644
--- a/man/man3/cmark.3
+++ b/man/man3/cmark.3
@@ -1,4 +1,4 @@
-.TH cmark 3 "June 25, 2015" "LOCAL" "Library Functions Manual"
+.TH cmark 3 "July 03, 2015" "LOCAL" "Library Functions Manual"
 .SH
 NAME
 .PP
@@ -480,6 +480,12 @@ Render a \f[I]node\f[] tree as a groff man page, without the header.
 Render a \f[I]node\f[] tree as a commonmark document.
 
 .PP
+\fIchar *\f[] \fBcmark_render_latex\f[](\fIcmark_node *root\f[], \fIint options\f[], \fIint width\f[])
+
+.PP
+Render a \f[I]node\f[] tree as a LaTeX document.
+
+.PP
 .nf
 \fC
 .RS 0n
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 37a46d3..0324fb4 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -32,6 +32,7 @@ set(LIBRARY_SOURCES
   xml.c
   html.c
   commonmark.c
+  latex.c
   houdini_href_e.c
   houdini_html_e.c
   houdini_html_u.c
diff --git a/src/cmark.h b/src/cmark.h
index 192290c..6618301 100644
--- a/src/cmark.h
+++ b/src/cmark.h
@@ -486,6 +486,11 @@ char *cmark_render_man(cmark_node *root, int options);
 CMARK_EXPORT
 char *cmark_render_commonmark(cmark_node *root, int options, int width);
 
+/** Render a 'node' tree as a LaTeX document.
+ */
+CMARK_EXPORT
+char *cmark_render_latex(cmark_node *root, int options, int width);
+
 /** Default writer options.
  */
 #define CMARK_OPT_DEFAULT 0
diff --git a/src/latex.c b/src/latex.c
new file mode 100644
index 0000000..5eea3a0
--- /dev/null
+++ b/src/latex.c
@@ -0,0 +1,609 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+#include <ctype.h>
+
+#include "config.h"
+#include "cmark.h"
+#include "node.h"
+#include "buffer.h"
+#include "utf8.h"
+#include "scanners.h"
+
+// Functions to convert cmark_nodes to commonmark strings.
+
+struct render_state {
+	int options;
+	cmark_strbuf* buffer;
+	cmark_strbuf* prefix;
+	int column;
+	int width;
+	int need_cr;
+	int enumlevel;
+	bufsize_t last_breakable;
+	bool begin_line;
+	bool no_wrap;
+	bool in_tight_list_item;
+	bool silence;
+};
+
+static inline void cr(struct render_state *state)
+{
+	if (state->need_cr < 1) {
+		state->need_cr = 1;
+	}
+}
+
+static inline void blankline(struct render_state *state)
+{
+	if (state->need_cr < 2) {
+		state->need_cr = 2;
+	}
+}
+
+typedef enum  {
+	LITERAL,
+	NORMAL,
+	URL
+} escaping;
+
+static inline void out(struct render_state *state,
+                       cmark_chunk str,
+                       bool wrap,
+                       escaping escape)
+{
+	unsigned char* source = str.data;
+	int length = str.len;
+	unsigned char nextc;
+	int32_t c;
+	int i = 0;
+	int len;
+	cmark_chunk remainder = cmark_chunk_literal("");
+	int k = state->buffer->size - 1;
+
+	if (state->silence)
+		return;
+
+	wrap = wrap && !state->no_wrap;
+
+	if (state->in_tight_list_item && state->need_cr > 1) {
+		state->need_cr = 1;
+	}
+	while (state->need_cr) {
+		if (k < 0 || state->buffer->ptr[k] == '\n') {
+			k -= 1;
+		} else {
+			cmark_strbuf_putc(state->buffer, '\n');
+			if (state->need_cr > 1) {
+				cmark_strbuf_put(state->buffer, state->prefix->ptr,
+				                 state->prefix->size);
+			}
+		}
+		state->column = 0;
+		state->begin_line = true;
+		state->need_cr -= 1;
+	}
+
+	while (i < length) {
+		if (state->begin_line) {
+			cmark_strbuf_put(state->buffer, state->prefix->ptr,
+			                 state->prefix->size);
+			// note: this assumes prefix is ascii:
+			state->column = state->prefix->size;
+		}
+
+		len = utf8proc_iterate(source + i, length - i, &c);
+		if (len == -1) { // error condition
+			return;  // return without rendering rest of string
+		}
+		nextc = source[i + len];
+		if (c == 32 && wrap) {
+			if (!state->begin_line) {
+				cmark_strbuf_putc(state->buffer, ' ');
+				state->column += 1;
+				state->begin_line = false;
+				state->last_breakable = state->buffer->size -
+				                        1;
+				// skip following spaces
+				while (source[i + 1] == ' ') {
+					i++;
+				}
+			}
+
+		} else if (c == 10) {
+			cmark_strbuf_putc(state->buffer, '\n');
+			state->column = 0;
+			state->begin_line = true;
+			state->last_breakable = 0;
+		} else if (escape == LITERAL) {
+				utf8proc_encode_char(c, state->buffer);
+				state->column += 2;
+		} else {
+			switch(c) {
+			case 123: // '{'
+			case 125: // '}'
+			case 35: // '#'
+			case 37: // '%'
+			case 38: // '&'
+				cmark_strbuf_putc(state->buffer, '\\');
+				utf8proc_encode_char(c, state->buffer);
+				state->column += 2;
+				break;
+			case 36: // '$'
+			case 95: // '_'
+				if (escape == NORMAL) {
+					cmark_strbuf_putc(state->buffer, '\\');
+				}
+				utf8proc_encode_char(c, state->buffer);
+				break;
+			case 45 : // '-'
+				if (nextc == 45) { // prevent ligature
+					cmark_strbuf_putc(state->buffer, '\\');
+				}
+				utf8proc_encode_char(c, state->buffer);
+				break;
+			case 126: // '~'
+				if (escape == NORMAL) {
+					cmark_strbuf_puts(state->buffer,
+					  "\\textasciitilde{}");
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 94: // '^'
+				cmark_strbuf_puts(state->buffer,
+						  "\\^{}");
+				break;
+			case 92: // '\\'
+				if (escape == URL) {
+					// / acts as path sep even on windows:
+					cmark_strbuf_puts(state->buffer, "/");
+				} else {
+					cmark_strbuf_puts(state->buffer,
+							  "\\textbackslash{}");
+				}
+				break;
+			case 124: // '|'
+				cmark_strbuf_puts(state->buffer,
+						  "\\textbar{}");
+				break;
+			case 60: // '<'
+				cmark_strbuf_puts(state->buffer,
+						  "\\textless{}");
+				break;
+			case 62: // '>'
+				cmark_strbuf_puts(state->buffer,
+						  "\\textgreater{}");
+				break;
+			case 91: // '['
+			case 93: // ']'
+				cmark_strbuf_putc(state->buffer, '{');
+				utf8proc_encode_char(c, state->buffer);
+				cmark_strbuf_putc(state->buffer, '}');
+				break;
+			case 39: // '\''
+				cmark_strbuf_puts(state->buffer,
+						  "\\textquotesingle{}");
+				break;
+			case 160: // nbsp
+				cmark_strbuf_putc(state->buffer, '~');
+				break;
+			case 8230: // hellip
+				cmark_strbuf_puts(state->buffer, "\\ldots{}");
+				break;
+			case 8216: // lsquo
+				if (escape == NORMAL) {
+					cmark_strbuf_putc(state->buffer, '`');
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 8217: // rsquo
+				if (escape == NORMAL) {
+					cmark_strbuf_putc(state->buffer, '\'');
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 8220: // ldquo
+				if (escape == NORMAL) {
+					cmark_strbuf_puts(state->buffer, "``");
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 8221: // rdquo
+				if (escape == NORMAL) {
+					cmark_strbuf_puts(state->buffer, "''");
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 8212: // emdash
+				if (escape == NORMAL) {
+					cmark_strbuf_puts(state->buffer, "---");
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			case 8211: // endash
+				if (escape == NORMAL) {
+					cmark_strbuf_puts(state->buffer, "--");
+				} else {
+					utf8proc_encode_char(c, state->buffer);
+				}
+				break;
+			default:
+				utf8proc_encode_char(c, state->buffer);
+				state->column += 1;
+				state->begin_line = false;
+			}
+		}
+
+		// If adding the character went beyond width, look for an
+		// earlier place where the line could be broken:
+		if (state->width > 0 &&
+		    state->column > state->width &&
+		    !state->begin_line &&
+		    state->last_breakable > 0) {
+
+			// copy from last_breakable to remainder
+			cmark_chunk_set_cstr(&remainder, (char *) state->buffer->ptr + state->last_breakable + 1);
+			// truncate at last_breakable
+			cmark_strbuf_truncate(state->buffer, state->last_breakable);
+			// add newline, prefix, and remainder
+			cmark_strbuf_putc(state->buffer, '\n');
+			cmark_strbuf_put(state->buffer, state->prefix->ptr,
+			                 state->prefix->size);
+			cmark_strbuf_put(state->buffer, remainder.data, remainder.len);
+			state->column = state->prefix->size + remainder.len;
+			cmark_chunk_free(&remainder);
+			state->last_breakable = 0;
+			state->begin_line = false;
+		}
+
+		i += len;
+	}
+}
+
+static void lit(struct render_state *state, char *s, bool wrap)
+{
+	cmark_chunk str = cmark_chunk_literal(s);
+	out(state, str, wrap, LITERAL);
+}
+
+typedef enum  {
+	NO_LINK,
+	URL_AUTOLINK,
+	EMAIL_AUTOLINK,
+	NORMAL_LINK
+} link_type;
+
+static link_type
+get_link_type(cmark_node *node)
+{
+	cmark_chunk *title;
+	cmark_chunk *url;
+	cmark_node *link_text;
+	char *realurl;
+	int realurllen;
+	bool isemail = false;
+
+	if (node->type != CMARK_NODE_LINK) {
+		return NO_LINK;
+	}
+
+	url = &node->as.link.url;
+	if (url->len == 0 || scan_scheme(url, 0) == 0) {
+		return NO_LINK;
+	}
+
+	title = &node->as.link.title;
+	// if it has a title, we can't treat it as an autolink:
+	if (title->len > 0) {
+		return NORMAL_LINK;
+	}
+
+	link_text = node->first_child;
+	cmark_consolidate_text_nodes(link_text);
+        realurl = (char*)url->data;
+	realurllen = url->len;
+	if (strncmp(realurl, "mailto:", 7) == 0) {
+		realurl += 7;
+		realurllen -= 7;
+		isemail = true;
+	}
+	if (realurllen == link_text->as.literal.len &&
+	        strncmp(realurl,
+	                (char*)link_text->as.literal.data,
+	                link_text->as.literal.len) == 0) {
+		if (isemail) {
+			return EMAIL_AUTOLINK;
+		} else {
+			return URL_AUTOLINK;
+		}
+	} else {
+		return NORMAL_LINK;
+	}
+}
+
+// if node is a block node, returns node.
+// otherwise returns first block-level node that is an ancestor of node.
+static cmark_node*
+get_containing_block(cmark_node *node)
+{
+	while (node &&
+	       (node->type < CMARK_NODE_FIRST_BLOCK ||
+	        node->type > CMARK_NODE_LAST_BLOCK)) {
+		node = node->parent;
+	}
+	return node;
+}
+
+static int
+S_render_node(cmark_node *node, cmark_event_type ev_type,
+              struct render_state *state)
+{
+	cmark_node *tmp;
+	cmark_chunk *code;
+	int list_number;
+	char list_number_string[20];
+	bool entering = (ev_type == CMARK_EVENT_ENTER);
+	cmark_list_type list_type;
+	cmark_chunk list_name;
+	cmark_chunk url;
+	const char* roman_numerals[] = { "", "i", "ii", "iii", "iv", "v",
+					"vi", "vii", "viii", "ix", "x" };
+
+	// Don't adjust tight list status til we've started the list.
+	// Otherwise we loose the blank line between a paragraph and
+	// a following list.
+	if (!(node->type == CMARK_NODE_ITEM && node->prev == NULL &&
+	      entering)) {
+		tmp = get_containing_block(node);
+		state->in_tight_list_item =
+		    (tmp->type == CMARK_NODE_ITEM &&
+		     cmark_node_get_list_tight(tmp->parent)) ||
+		    (tmp &&
+		     tmp->parent &&
+		     tmp->parent->type == CMARK_NODE_ITEM &&
+		     cmark_node_get_list_tight(tmp->parent->parent));
+	}
+
+	switch (node->type) {
+	case CMARK_NODE_DOCUMENT:
+		if (!entering) {
+			cmark_strbuf_putc(state->buffer, '\n');
+		}
+		break;
+
+	case CMARK_NODE_BLOCK_QUOTE:
+		if (entering) {
+			lit(state, "\\begin{quote}", false);
+			cr(state);
+		} else {
+			lit(state, "\\end{quote}", false);
+			blankline(state);
+		}
+		break;
+
+	case CMARK_NODE_LIST:
+		list_type = cmark_node_get_list_type(node);
+		list_name = cmark_chunk_literal(
+			list_type == CMARK_ORDERED_LIST ?
+			"enumerate" : "itemize");
+		if (entering) {
+			if (list_type == CMARK_ORDERED_LIST) {
+				state->enumlevel++;
+			}
+			lit(state, "\\begin{", false);
+			out(state, list_name, false, false);
+			lit(state, "}", false);
+			cr(state);
+			list_number = cmark_node_get_list_start(node);
+			if (list_number > 1) {
+				snprintf(list_number_string, 19,
+					 "%d", list_number);
+				lit(state, "\\setcounter{enum", false);
+				lit(state, (char *)roman_numerals[state->enumlevel],
+				    false);
+				lit(state, "}{", false);
+				out(state,
+				    cmark_chunk_literal(list_number_string),
+				    false, NORMAL);
+				lit(state, "}", false);
+				cr(state);
+			}
+		} else {
+			if (list_type == CMARK_ORDERED_LIST) {
+				state->enumlevel--;
+			}
+			lit(state, "\\end{", false);
+			out(state, list_name, false, false);
+			lit(state, "}", false);
+			blankline(state);
+		}
+		cmark_chunk_free(&list_name);
+		break;
+
+	case CMARK_NODE_ITEM:
+		if (entering) {
+			lit(state, "\\item ", false);
+		} else {
+			cr(state);
+		}
+		break;
+
+	case CMARK_NODE_HEADER:
+		if (entering) {
+			switch (cmark_node_get_header_level(node)) {
+			case 1:
+				lit(state, "\\section", false);
+				break;
+			case 2:
+				lit(state, "\\subsection", false);
+				break;
+			case 3:
+				lit(state, "\\subsubsection", false);
+				break;
+			case 4:
+				lit(state, "\\paragraph", false);
+				break;
+			case 5:
+				lit(state, "\\subparagraph", false);
+				break;
+			}
+			lit(state, "{", false);
+		} else {
+			lit(state, "}", false);
+			blankline(state);
+		}
+		break;
+
+	case CMARK_NODE_CODE_BLOCK:
+		cr(state);
+		lit(state, "\\begin{verbatim}", false);
+		cr(state);
+		code = &node->as.code.literal;
+		out(state, node->as.code.literal, false, LITERAL);
+		cr(state);
+		lit(state, "\\end{verbatim}", false);
+		blankline(state);
+		break;
+
+	case CMARK_NODE_HTML:
+		break;
+
+	case CMARK_NODE_HRULE:
+		blankline(state);
+		lit(state, "\\begin{center}\\rule{0.5\\linewidth}{\\linethickness}\\end{center}", false);
+		blankline(state);
+		break;
+
+	case CMARK_NODE_PARAGRAPH:
+		if (!entering) {
+			blankline(state);
+		}
+		break;
+
+	case CMARK_NODE_TEXT:
+		out(state, node->as.literal, true, NORMAL);
+		break;
+
+	case CMARK_NODE_LINEBREAK:
+		lit(state, "\\\\", false);
+		cr(state);
+		break;
+
+	case CMARK_NODE_SOFTBREAK:
+		if (state->width == 0) {
+			cr(state);
+		} else {
+			lit(state, " ", true);
+		}
+		break;
+
+	case CMARK_NODE_CODE:
+		lit(state, "\\texttt{", false);
+		out(state, node->as.literal, false, NORMAL);
+		lit(state, "}", false);
+		break;
+
+	case CMARK_NODE_INLINE_HTML:
+		break;
+
+	case CMARK_NODE_STRONG:
+		if (entering) {
+			lit(state, "\\strong{", false);
+		} else {
+			lit(state, "}", false);
+		}
+		break;
+
+	case CMARK_NODE_EMPH:
+		if (entering) {
+			lit(state, "\\emph{", false);
+		} else {
+			lit(state, "}", false);
+		}
+		break;
+
+	case CMARK_NODE_LINK:
+		if (entering) {
+			url = cmark_chunk_literal(cmark_node_get_url(node));
+			switch(get_link_type(node)) {
+			case URL_AUTOLINK:
+				lit(state, "\\url{", false);
+				out(state, url, false, URL);
+				break;
+			case EMAIL_AUTOLINK:
+				lit(state, "\\href{", false);
+				out(state, url, false, URL);
+				lit(state, "}\\nolinkurl{", false);
+				break;
+			case NORMAL_LINK:
+				lit(state, "\\href{", false);
+				out(state, url, false, URL);
+				lit(state, "}{", false);
+				break;
+			case NO_LINK:
+				lit(state, "{", false);  // error?
+			}
+		} else {
+			lit(state, "}", false);
+		}
+
+		break;
+
+	case CMARK_NODE_IMAGE:
+		if (entering) {
+			url = cmark_chunk_literal(cmark_node_get_url(node));
+			lit(state, "\\protect\\includegraphics{", false);
+			out(state, url, false, URL);
+			lit(state, "}", false);
+			state->silence = true; // don't print the alt text
+		} else {
+			state->silence = false;
+		}
+		break;
+
+	default:
+		assert(false);
+		break;
+	}
+
+	return 1;
+}
+
+char *cmark_render_latex(cmark_node *root, int options, int width)
+{
+	char *result;
+	cmark_strbuf commonmark = GH_BUF_INIT;
+	cmark_strbuf prefix = GH_BUF_INIT;
+	if (CMARK_OPT_HARDBREAKS & options) {
+		width = 0;
+	}
+	struct render_state state = {
+		options, &commonmark, &prefix, 0, width,
+		0, 0, 0, true, false, false, false
+	};
+	cmark_node *cur;
+	cmark_event_type ev_type;
+	cmark_iter *iter = cmark_iter_new(root);
+
+	while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
+		cur = cmark_iter_get_node(iter);
+		if (!S_render_node(cur, ev_type, &state)) {
+			// a false value causes us to skip processing
+			// the node's contents.  this is used for
+			// autolinks.
+			cmark_iter_reset(iter, cur, CMARK_EVENT_EXIT);
+		}
+	}
+	result = (char *)cmark_strbuf_detach(&commonmark);
+
+	cmark_strbuf_free(&prefix);
+	cmark_iter_free(iter);
+	return result;
+}
diff --git a/src/main.c b/src/main.c
index 5dc97b2..e20b0dc 100644
--- a/src/main.c
+++ b/src/main.c
@@ -17,14 +17,15 @@ typedef enum {
 	FORMAT_HTML,
 	FORMAT_XML,
 	FORMAT_MAN,
-	FORMAT_COMMONMARK
+	FORMAT_COMMONMARK,
+	FORMAT_LATEX
 } writer_format;
 
 void print_usage()
 {
 	printf("Usage:   cmark [FILE*]\n");
 	printf("Options:\n");
-	printf("  --to, -t FORMAT  Specify output format (html, xml, man, commonmark)\n");
+	printf("  --to, -t FORMAT  Specify output format (html, xml, man, commonmark, latex)\n");
 	printf("  --width WIDTH    Specify wrap width (default 0 = nowrap)\n");
 	printf("  --sourcepos      Include source position attribute\n");
 	printf("  --hardbreaks     Treat newlines as hard line breaks\n");
@@ -52,6 +53,9 @@ static void print_document(cmark_node *document, writer_format writer,
 	case FORMAT_COMMONMARK:
 		result = cmark_render_commonmark(document, options, width);
 		break;
+	case FORMAT_LATEX:
+		result = cmark_render_latex(document, options, width);
+		break;
 	default:
 		fprintf(stderr, "Unknown format %d\n", writer);
 		exit(1);
@@ -125,6 +129,8 @@ int main(int argc, char *argv[])
 					writer = FORMAT_XML;
 				} else if (strcmp(argv[i], "commonmark") == 0) {
 					writer = FORMAT_COMMONMARK;
+				} else if (strcmp(argv[i], "latex") == 0) {
+					writer = FORMAT_LATEX;
 				} else {
 					fprintf(stderr,
 					        "Unknown format %s\n", argv[i]);