markly 0.1.0

data/ext/markly/inlines.h

@@ -0,0 +1,29 @@
+#ifndef CMARK_INLINES_H
+#define CMARK_INLINES_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "references.h"
+
+cmark_chunk cmark_clean_url(cmark_mem *mem, cmark_chunk *url);
+cmark_chunk cmark_clean_title(cmark_mem *mem, cmark_chunk *title);
+
+CMARK_GFM_EXPORT
+void cmark_parse_inlines(cmark_parser *parser,
+                         cmark_node *parent,
+                         cmark_map *refmap,
+                         int options);
+
+bufsize_t cmark_parse_reference_inline(cmark_mem *mem, cmark_chunk *input,
+                                       cmark_map *refmap);
+
+void cmark_inlines_add_special_character(unsigned char c, bool emphasis);
+void cmark_inlines_remove_special_character(unsigned char c, bool emphasis);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

data/ext/markly/iterator.c

@@ -0,0 +1,159 @@
+#include <assert.h>
+#include <stdlib.h>
+
+#include "config.h"
+#include "node.h"
+#include "cmark-gfm.h"
+#include "iterator.h"
+
+cmark_iter *cmark_iter_new(cmark_node *root) {
+  if (root == NULL) {
+    return NULL;
+  }
+  cmark_mem *mem = root->content.mem;
+  cmark_iter *iter = (cmark_iter *)mem->calloc(1, sizeof(cmark_iter));
+  iter->mem = mem;
+  iter->root = root;
+  iter->cur.ev_type = CMARK_EVENT_NONE;
+  iter->cur.node = NULL;
+  iter->next.ev_type = CMARK_EVENT_ENTER;
+  iter->next.node = root;
+  return iter;
+}
+
+void cmark_iter_free(cmark_iter *iter) { iter->mem->free(iter); }
+
+static bool S_is_leaf(cmark_node *node) {
+  switch (node->type) {
+  case CMARK_NODE_HTML_BLOCK:
+  case CMARK_NODE_THEMATIC_BREAK:
+  case CMARK_NODE_CODE_BLOCK:
+  case CMARK_NODE_TEXT:
+  case CMARK_NODE_SOFTBREAK:
+  case CMARK_NODE_LINEBREAK:
+  case CMARK_NODE_CODE:
+  case CMARK_NODE_HTML_INLINE:
+    return 1;
+  }
+  return 0;
+}
+
+cmark_event_type cmark_iter_next(cmark_iter *iter) {
+  cmark_event_type ev_type = iter->next.ev_type;
+  cmark_node *node = iter->next.node;
+
+  iter->cur.ev_type = ev_type;
+  iter->cur.node = node;
+
+  if (ev_type == CMARK_EVENT_DONE) {
+    return ev_type;
+  }
+
+  /* roll forward to next item, setting both fields */
+  if (ev_type == CMARK_EVENT_ENTER && !S_is_leaf(node)) {
+    if (node->first_child == NULL) {
+      /* stay on this node but exit */
+      iter->next.ev_type = CMARK_EVENT_EXIT;
+    } else {
+      iter->next.ev_type = CMARK_EVENT_ENTER;
+      iter->next.node = node->first_child;
+    }
+  } else if (node == iter->root) {
+    /* don't move past root */
+    iter->next.ev_type = CMARK_EVENT_DONE;
+    iter->next.node = NULL;
+  } else if (node->next) {
+    iter->next.ev_type = CMARK_EVENT_ENTER;
+    iter->next.node = node->next;
+  } else if (node->parent) {
+    iter->next.ev_type = CMARK_EVENT_EXIT;
+    iter->next.node = node->parent;
+  } else {
+    assert(false);
+    iter->next.ev_type = CMARK_EVENT_DONE;
+    iter->next.node = NULL;
+  }
+
+  return ev_type;
+}
+
+void cmark_iter_reset(cmark_iter *iter, cmark_node *current,
+                      cmark_event_type event_type) {
+  iter->next.ev_type = event_type;
+  iter->next.node = current;
+  cmark_iter_next(iter);
+}
+
+cmark_node *cmark_iter_get_node(cmark_iter *iter) { return iter->cur.node; }
+
+cmark_event_type cmark_iter_get_event_type(cmark_iter *iter) {
+  return iter->cur.ev_type;
+}
+
+cmark_node *cmark_iter_get_root(cmark_iter *iter) { return iter->root; }
+
+void cmark_consolidate_text_nodes(cmark_node *root) {
+  if (root == NULL) {
+    return;
+  }
+  cmark_iter *iter = cmark_iter_new(root);
+  cmark_strbuf buf = CMARK_BUF_INIT(iter->mem);
+  cmark_event_type ev_type;
+  cmark_node *cur, *tmp, *next;
+
+  while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
+    cur = cmark_iter_get_node(iter);
+    if (ev_type == CMARK_EVENT_ENTER && cur->type == CMARK_NODE_TEXT &&
+        cur->next && cur->next->type == CMARK_NODE_TEXT) {
+      cmark_strbuf_clear(&buf);
+      cmark_strbuf_put(&buf, cur->as.literal.data, cur->as.literal.len);
+      tmp = cur->next;
+      while (tmp && tmp->type == CMARK_NODE_TEXT) {
+        cmark_iter_next(iter); // advance pointer
+        cmark_strbuf_put(&buf, tmp->as.literal.data, tmp->as.literal.len);
+        cur->end_column = tmp->end_column;
+        next = tmp->next;
+        cmark_node_free(tmp);
+        tmp = next;
+      }
+      cmark_chunk_free(iter->mem, &cur->as.literal);
+      cur->as.literal = cmark_chunk_buf_detach(&buf);
+    }
+  }
+
+  cmark_strbuf_free(&buf);
+  cmark_iter_free(iter);
+}
+
+void cmark_node_own(cmark_node *root) {
+  if (root == NULL) {
+    return;
+  }
+  cmark_iter *iter = cmark_iter_new(root);
+  cmark_event_type ev_type;
+  cmark_node *cur;
+
+  while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
+    cur = cmark_iter_get_node(iter);
+    if (ev_type == CMARK_EVENT_ENTER) {
+      switch (cur->type) {
+      case CMARK_NODE_TEXT:
+      case CMARK_NODE_HTML_INLINE:
+      case CMARK_NODE_CODE:
+      case CMARK_NODE_HTML_BLOCK:
+        cmark_chunk_to_cstr(iter->mem, &cur->as.literal);
+        break;
+      case CMARK_NODE_LINK:
+        cmark_chunk_to_cstr(iter->mem, &cur->as.link.url);
+        cmark_chunk_to_cstr(iter->mem, &cur->as.link.title);
+        break;
+      case CMARK_NODE_CUSTOM_INLINE:
+        cmark_chunk_to_cstr(iter->mem, &cur->as.custom.on_enter);
+        cmark_chunk_to_cstr(iter->mem, &cur->as.custom.on_exit);
+        break;
+      }
+    }
+  }
+
+  cmark_iter_free(iter);
+}

data/ext/markly/iterator.h

@@ -0,0 +1,26 @@
+#ifndef CMARK_ITERATOR_H
+#define CMARK_ITERATOR_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "cmark-gfm.h"
+
+typedef struct {
+  cmark_event_type ev_type;
+  cmark_node *node;
+} cmark_iter_state;
+
+struct cmark_iter {
+  cmark_mem *mem;
+  cmark_node *root;
+  cmark_iter_state cur;
+  cmark_iter_state next;
+};
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

data/ext/markly/latex.c

@@ -0,0 +1,466 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+
+#include "config.h"
+#include "cmark-gfm.h"
+#include "node.h"
+#include "buffer.h"
+#include "utf8.h"
+#include "scanners.h"
+#include "render.h"
+#include "syntax_extension.h"
+
+#define OUT(s, wrap, escaping) renderer->out(renderer, node, s, wrap, escaping)
+#define LIT(s) renderer->out(renderer, node, s, false, LITERAL)
+#define CR() renderer->cr(renderer)
+#define BLANKLINE() renderer->blankline(renderer)
+#define LIST_NUMBER_STRING_SIZE 20
+
+static CMARK_INLINE void outc(cmark_renderer *renderer, cmark_node *node,
+                              cmark_escaping escape,
+                              int32_t c, unsigned char nextc) {
+  if (escape == LITERAL) {
+    cmark_render_code_point(renderer, c);
+    return;
+  }
+
+  switch (c) {
+  case 123: // '{'
+  case 125: // '}'
+  case 35:  // '#'
+  case 37:  // '%'
+  case 38:  // '&'
+    cmark_render_ascii(renderer, "\\");
+    cmark_render_code_point(renderer, c);
+    break;
+  case 36: // '$'
+  case 95: // '_'
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "\\");
+    }
+    cmark_render_code_point(renderer, c);
+    break;
+  case 45:             // '-'
+    if (nextc == 45) { // prevent ligature
+      cmark_render_ascii(renderer, "-{}");
+    } else {
+      cmark_render_ascii(renderer, "-");
+    }
+    break;
+  case 126: // '~'
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "\\textasciitilde{}");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 94: // '^'
+    cmark_render_ascii(renderer, "\\^{}");
+    break;
+  case 92: // '\\'
+    if (escape == URL) {
+      // / acts as path sep even on windows:
+      cmark_render_ascii(renderer, "/");
+    } else {
+      cmark_render_ascii(renderer, "\\textbackslash{}");
+    }
+    break;
+  case 124: // '|'
+    cmark_render_ascii(renderer, "\\textbar{}");
+    break;
+  case 60: // '<'
+    cmark_render_ascii(renderer, "\\textless{}");
+    break;
+  case 62: // '>'
+    cmark_render_ascii(renderer, "\\textgreater{}");
+    break;
+  case 91: // '['
+  case 93: // ']'
+    cmark_render_ascii(renderer, "{");
+    cmark_render_code_point(renderer, c);
+    cmark_render_ascii(renderer, "}");
+    break;
+  case 34: // '"'
+    cmark_render_ascii(renderer, "\\textquotedbl{}");
+    // requires \usepackage[T1]{fontenc}
+    break;
+  case 39: // '\''
+    cmark_render_ascii(renderer, "\\textquotesingle{}");
+    // requires \usepackage{textcomp}
+    break;
+  case 160: // nbsp
+    cmark_render_ascii(renderer, "~");
+    break;
+  case 8230: // hellip
+    cmark_render_ascii(renderer, "\\ldots{}");
+    break;
+  case 8216: // lsquo
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "`");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 8217: // rsquo
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "\'");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 8220: // ldquo
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "``");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 8221: // rdquo
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "''");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 8212: // emdash
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "---");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  case 8211: // endash
+    if (escape == NORMAL) {
+      cmark_render_ascii(renderer, "--");
+    } else {
+      cmark_render_code_point(renderer, c);
+    }
+    break;
+  default:
+    cmark_render_code_point(renderer, c);
+  }
+}
+
+typedef enum {
+  NO_LINK,
+  URL_AUTOLINK,
+  EMAIL_AUTOLINK,
+  NORMAL_LINK,
+  INTERNAL_LINK
+} link_type;
+
+static link_type get_link_type(cmark_node *node) {
+  size_t title_len, url_len;
+  cmark_node *link_text;
+  char *realurl;
+  int realurllen;
+  bool isemail = false;
+
+  if (node->type != CMARK_NODE_LINK) {
+    return NO_LINK;
+  }
+
+  const char *url = cmark_node_get_url(node);
+  cmark_chunk url_chunk = cmark_chunk_literal(url);
+
+  if (url && *url == '#') {
+    return INTERNAL_LINK;
+  }
+
+  url_len = strlen(url);
+  if (url_len == 0 || scan_scheme(&url_chunk, 0) == 0) {
+    return NO_LINK;
+  }
+
+  const char *title = cmark_node_get_title(node);
+  title_len = strlen(title);
+  // if it has a title, we can't treat it as an autolink:
+  if (title_len == 0) {
+
+    link_text = node->first_child;
+    cmark_consolidate_text_nodes(link_text);
+
+    if (!link_text)
+      return NO_LINK;
+
+    realurl = (char *)url;
+    realurllen = (int)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;
+      }
+    }
+  }
+
+  return NORMAL_LINK;
+}
+
+static int S_get_enumlevel(cmark_node *node) {
+  int enumlevel = 0;
+  cmark_node *tmp = node;
+  while (tmp) {
+    if (tmp->type == CMARK_NODE_LIST &&
+        cmark_node_get_list_type(node) == CMARK_ORDERED_LIST) {
+      enumlevel++;
+    }
+    tmp = tmp->parent;
+  }
+  return enumlevel;
+}
+
+static int S_render_node(cmark_renderer *renderer, cmark_node *node,
+                         cmark_event_type ev_type, int options) {
+  int list_number;
+  int enumlevel;
+  char list_number_string[LIST_NUMBER_STRING_SIZE];
+  bool entering = (ev_type == CMARK_EVENT_ENTER);
+  cmark_list_type list_type;
+  bool allow_wrap = renderer->width > 0 && !(CMARK_OPT_NOBREAKS & options);
+
+  if (node->extension && node->extension->latex_render_func) {
+    node->extension->latex_render_func(node->extension, renderer, node, ev_type, options);
+    return 1;
+  }
+
+  switch (node->type) {
+  case CMARK_NODE_DOCUMENT:
+    break;
+
+  case CMARK_NODE_BLOCK_QUOTE:
+    if (entering) {
+      LIT("\\begin{quote}");
+      CR();
+    } else {
+      LIT("\\end{quote}");
+      BLANKLINE();
+    }
+    break;
+
+  case CMARK_NODE_LIST:
+    list_type = cmark_node_get_list_type(node);
+    if (entering) {
+      LIT("\\begin{");
+      LIT(list_type == CMARK_ORDERED_LIST ? "enumerate" : "itemize");
+      LIT("}");
+      CR();
+      list_number = cmark_node_get_list_start(node);
+      if (list_number > 1) {
+        enumlevel = S_get_enumlevel(node);
+        // latex normally supports only five levels
+        if (enumlevel >= 1 && enumlevel <= 5) {
+          snprintf(list_number_string, LIST_NUMBER_STRING_SIZE, "%d",
+                   list_number);
+          LIT("\\setcounter{enum");
+          switch (enumlevel) {
+          case 1: LIT("i"); break;
+          case 2: LIT("ii"); break;
+          case 3: LIT("iii"); break;
+          case 4: LIT("iv"); break;
+          case 5: LIT("v"); break;
+          default: LIT("i"); break;
+	  }
+          LIT("}{");
+          OUT(list_number_string, false, NORMAL);
+          LIT("}");
+        }
+        CR();
+      }
+    } else {
+      LIT("\\end{");
+      LIT(list_type == CMARK_ORDERED_LIST ? "enumerate" : "itemize");
+      LIT("}");
+      BLANKLINE();
+    }
+    break;
+
+  case CMARK_NODE_ITEM:
+    if (entering) {
+      LIT("\\item ");
+    } else {
+      CR();
+    }
+    break;
+
+  case CMARK_NODE_HEADING:
+    if (entering) {
+      switch (cmark_node_get_heading_level(node)) {
+      case 1:
+        LIT("\\section");
+        break;
+      case 2:
+        LIT("\\subsection");
+        break;
+      case 3:
+        LIT("\\subsubsection");
+        break;
+      case 4:
+        LIT("\\paragraph");
+        break;
+      case 5:
+        LIT("\\subparagraph");
+        break;
+      }
+      LIT("{");
+    } else {
+      LIT("}");
+      BLANKLINE();
+    }
+    break;
+
+  case CMARK_NODE_CODE_BLOCK:
+    CR();
+    LIT("\\begin{verbatim}");
+    CR();
+    OUT(cmark_node_get_literal(node), false, LITERAL);
+    CR();
+    LIT("\\end{verbatim}");
+    BLANKLINE();
+    break;
+
+  case CMARK_NODE_HTML_BLOCK:
+    break;
+
+  case CMARK_NODE_CUSTOM_BLOCK:
+    CR();
+    OUT(entering ? cmark_node_get_on_enter(node) : cmark_node_get_on_exit(node),
+        false, LITERAL);
+    CR();
+    break;
+
+  case CMARK_NODE_THEMATIC_BREAK:
+    BLANKLINE();
+    LIT("\\begin{center}\\rule{0.5\\linewidth}{\\linethickness}\\end{center}");
+    BLANKLINE();
+    break;
+
+  case CMARK_NODE_PARAGRAPH:
+    if (!entering) {
+      BLANKLINE();
+    }
+    break;
+
+  case CMARK_NODE_TEXT:
+    OUT(cmark_node_get_literal(node), allow_wrap, NORMAL);
+    break;
+
+  case CMARK_NODE_LINEBREAK:
+    LIT("\\\\");
+    CR();
+    break;
+
+  case CMARK_NODE_SOFTBREAK:
+    if (options & CMARK_OPT_HARDBREAKS) {
+      LIT("\\\\");
+      CR();
+    } else if (renderer->width == 0 && !(CMARK_OPT_NOBREAKS & options)) {
+      CR();
+    } else {
+      OUT(" ", allow_wrap, NORMAL);
+    }
+    break;
+
+  case CMARK_NODE_CODE:
+    LIT("\\texttt{");
+    OUT(cmark_node_get_literal(node), false, NORMAL);
+    LIT("}");
+    break;
+
+  case CMARK_NODE_HTML_INLINE:
+    break;
+
+  case CMARK_NODE_CUSTOM_INLINE:
+    OUT(entering ? cmark_node_get_on_enter(node) : cmark_node_get_on_exit(node),
+        false, LITERAL);
+    break;
+
+  case CMARK_NODE_STRONG:
+    if (entering) {
+      LIT("\\textbf{");
+    } else {
+      LIT("}");
+    }
+    break;
+
+  case CMARK_NODE_EMPH:
+    if (entering) {
+      LIT("\\emph{");
+    } else {
+      LIT("}");
+    }
+    break;
+
+  case CMARK_NODE_LINK:
+    if (entering) {
+      const char *url = cmark_node_get_url(node);
+      // requires \usepackage{hyperref}
+      switch (get_link_type(node)) {
+      case URL_AUTOLINK:
+        LIT("\\url{");
+        OUT(url, false, URL);
+        LIT("}");
+        return 0; // Don't process further nodes to avoid double-rendering artefacts
+      case EMAIL_AUTOLINK:
+        LIT("\\href{");
+        OUT(url, false, URL);
+        LIT("}\\nolinkurl{");
+        break;
+      case NORMAL_LINK:
+        LIT("\\href{");
+        OUT(url, false, URL);
+        LIT("}{");
+        break;
+      case INTERNAL_LINK:
+        LIT("\\protect\\hyperlink{");
+        OUT(url + 1, false, URL);
+        LIT("}{");
+        break;
+      case NO_LINK:
+        LIT("{"); // error?
+      }
+    } else {
+      LIT("}");
+    }
+
+    break;
+
+  case CMARK_NODE_IMAGE:
+    if (entering) {
+      LIT("\\protect\\includegraphics{");
+      // requires \include{graphicx}
+      OUT(cmark_node_get_url(node), false, URL);
+      LIT("}");
+      return 0;
+    }
+    break;
+
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
+  case CMARK_NODE_FOOTNOTE_REFERENCE:
+    // TODO
+    break;
+
+  default:
+    assert(false);
+    break;
+  }
+
+  return 1;
+}
+
+char *cmark_render_latex(cmark_node *root, int options, int width) {
+  return cmark_render_latex_with_mem(root, options, width, cmark_node_mem(root));
+}
+
+char *cmark_render_latex_with_mem(cmark_node *root, int options, int width, cmark_mem *mem) {
+  return cmark_render(mem, root, options, width, outc, S_render_node);
+}