markly 0.7.0 → 0.8.0

data/ext/markly/markly.h

@@ -11,6 +11,6 @@
 
 #define CSTR2SYM(s) (ID2SYM(rb_intern((s))))
 
-void Init_markly();
+void Init_markly(void);
 
 #endif

data/ext/markly/node.c

@@ -5,10 +5,42 @@
 #include "node.h"
 #include "syntax_extension.h"
 
+/**
+ * Expensive safety checks are off by default, but can be enabled
+ * by calling cmark_enable_safety_checks().
+ */
+static bool enable_safety_checks = false;
+
+void cmark_enable_safety_checks(bool enable) {
+  enable_safety_checks = enable;
+}
+
 static void S_node_unlink(cmark_node *node);
 
 #define NODE_MEM(node) cmark_node_mem(node)
 
+void cmark_register_node_flag(cmark_node_internal_flags *flags) {
+  static cmark_node_internal_flags nextflag = CMARK_NODE__REGISTER_FIRST;
+
+  // flags should be a pointer to a global variable and this function
+  // should only be called once to initialize its value.
+  if (*flags) {
+    fprintf(stderr, "flag initialization error in cmark_register_node_flag\n");
+    abort();
+  }
+
+  // Check that we haven't run out of bits.
+  if (nextflag == 0) {
+    fprintf(stderr, "too many flags in cmark_register_node_flag\n");
+    abort();
+  }
+
+  *flags = nextflag;
+  nextflag <<= 1;
+}
+
+void cmark_init_standard_node_flags(void) {}
+
 bool cmark_node_can_contain_type(cmark_node *node, cmark_node_type child_type) {
   if (child_type == CMARK_NODE_DOCUMENT) {
       return false;
@@ -48,8 +80,6 @@ bool cmark_node_can_contain_type(cmark_node *node, cmark_node_type child_type) {
 }
 
 static bool S_can_contain(cmark_node *node, cmark_node *child) {
-  cmark_node *cur;
-
   if (node == NULL || child == NULL) {
     return false;
   }
@@ -57,14 +87,16 @@ static bool S_can_contain(cmark_node *node, cmark_node *child) {
     return 0;
   }
 
-  // Verify that child is not an ancestor of node or equal to node.
-  cur = node;
-  do {
-    if (cur == child) {
-      return false;
-    }
-    cur = cur->parent;
-  } while (cur != NULL);
+  if (enable_safety_checks) {
+    // Verify that child is not an ancestor of node or equal to node.
+    cmark_node *cur = node;
+    do {
+      if (cur == child) {
+        return false;
+      }
+      cur = cur->parent;
+    } while (cur != NULL);
+  }
 
   return cmark_node_can_contain_type(node, (cmark_node_type) child->type);
 }
@@ -301,6 +333,14 @@ cmark_node *cmark_node_last_child(cmark_node *node) {
   }
 }
 
+cmark_node *cmark_node_parent_footnote_def(cmark_node *node) {
+  if (node == NULL) {
+    return NULL;
+  } else {
+    return node->parent_footnote_def;
+  }
+}
+
 void *cmark_node_get_user_data(cmark_node *node) {
   if (node == NULL) {
     return NULL;
@@ -337,6 +377,7 @@ const char *cmark_node_get_literal(cmark_node *node) {
   case CMARK_NODE_HTML_INLINE:
   case CMARK_NODE_CODE:
   case CMARK_NODE_FOOTNOTE_REFERENCE:
+  case CMARK_NODE_FOOTNOTE_DEFINITION:
     return cmark_chunk_to_cstr(NODE_MEM(node), &node->as.literal);
 
   case CMARK_NODE_CODE_BLOCK:
@@ -524,6 +565,31 @@ int cmark_node_set_list_tight(cmark_node *node, int tight) {
   }
 }
 
+int cmark_node_get_item_index(cmark_node *node) {
+  if (node == NULL) {
+    return 0;
+  }
+
+  if (node->type == CMARK_NODE_ITEM) {
+    return node->as.list.start;
+  } else {
+    return 0;
+  }
+}
+
+int cmark_node_set_item_index(cmark_node *node, int idx) {
+  if (node == NULL || idx < 0) {
+    return 0;
+  }
+
+  if (node->type == CMARK_NODE_ITEM) {
+    node->as.list.start = idx;
+    return 1;
+  } else {
+    return 0;
+  }
+}
+
 const char *cmark_node_get_fence_info(cmark_node *node) {
   if (node == NULL) {
     return NULL;

data/ext/markly/node.h

@@ -52,8 +52,14 @@ enum cmark_node__internal_flags {
   CMARK_NODE__OPEN = (1 << 0),
   CMARK_NODE__LAST_LINE_BLANK = (1 << 1),
   CMARK_NODE__LAST_LINE_CHECKED = (1 << 2),
+
+  // Extensions can register custom flags by calling `cmark_register_node_flag`.
+  // This is the starting value for the custom flags.
+  CMARK_NODE__REGISTER_FIRST = (1 << 3),
 };
 
+typedef uint16_t cmark_node_internal_flags;
+
 struct cmark_node {
   cmark_strbuf content;
 
@@ -72,10 +78,18 @@ struct cmark_node {
   int end_column;
   int internal_offset;
   uint16_t type;
-  uint16_t flags;
+  cmark_node_internal_flags flags;
 
   cmark_syntax_extension *extension;
 
+  /**
+   * Used during cmark_render() to cache the most recent non-NULL
+   * extension, if you go up the parent chain like this:
+   *
+   * node->parent->...parent->extension
+   */
+  cmark_syntax_extension *ancestor_extension;
+
   union {
     int ref_ix;
     int def_count;
@@ -95,6 +109,26 @@ struct cmark_node {
   } as;
 };
 
+/**
+ * Syntax extensions can use this function to register a custom node
+ * flag. The flags are stored in the `flags` field of the `cmark_node`
+ * struct. The `flags` parameter should be the address of a global variable
+ * which will store the flag value.
+ */
+CMARK_GFM_EXPORT
+void cmark_register_node_flag(cmark_node_internal_flags *flags);
+
+/**
+ * DEPRECATED.
+ *
+ * This function was added in cmark-gfm version 0.29.0.gfm.7, and was
+ * required to be called at program start time, which caused
+ * backwards-compatibility issues in applications that use cmark-gfm as a
+ * library. It is now a no-op.
+ */
+CMARK_GFM_EXPORT
+void cmark_init_standard_node_flags(void);
+
 static CMARK_INLINE cmark_mem *cmark_node_mem(cmark_node *node) {
   return node->content.mem;
 }
@@ -118,6 +152,13 @@ static CMARK_INLINE bool CMARK_NODE_INLINE_P(cmark_node *node) {
 
 CMARK_GFM_EXPORT bool cmark_node_can_contain_type(cmark_node *node, cmark_node_type child_type);
 
+/**
+ * Enable (or disable) extra safety checks. These extra checks cause
+ * extra performance overhead (in some cases quadratic), so they are only
+ * intended to be used during testing.
+ */
+CMARK_GFM_EXPORT void cmark_enable_safety_checks(bool enable);
+
 #ifdef __cplusplus
 }
 #endif

data/ext/markly/parser.h

@@ -46,6 +46,7 @@ struct cmark_parser {
   /* Options set by the user, see the Options section in cmark.h */
   int options;
   bool last_buffer_ended_with_cr;
+  size_t total_size;
   cmark_llist *syntax_extensions;
   cmark_llist *inline_syntax_extensions;
   cmark_ispunct_func backslash_ispunct;

data/ext/markly/plaintext.c

@@ -16,23 +16,8 @@ static CMARK_INLINE void outc(cmark_renderer *renderer, cmark_node *node,
   cmark_render_code_point(renderer, c);
 }
 
-// if node is a block node, returns node.
-// otherwise returns first block-level node that is an ancestor of node.
-// if there is no block-level ancestor, returns NULL.
-static cmark_node *get_containing_block(cmark_node *node) {
-  while (node) {
-    if (CMARK_NODE_BLOCK_P(node)) {
-      return node;
-    } else {
-      node = node->parent;
-    }
-  }
-  return NULL;
-}
-
 static int S_render_node(cmark_renderer *renderer, cmark_node *node,
                          cmark_event_type ev_type, int options) {
-  cmark_node *tmp;
   int list_number;
   cmark_delim_type list_delim;
   int i;
@@ -46,14 +31,17 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
   // 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);
-    renderer->in_tight_list_item =
-        tmp && // tmp might be NULL if there is no containing block
-        ((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)));
+  if (entering) {
+    if (node->parent && node->parent->type == CMARK_NODE_ITEM) {
+      renderer->in_tight_list_item = node->parent->parent->as.list.tight;
+    }
+  } else {
+    if (node->type == CMARK_NODE_LIST) {
+      renderer->in_tight_list_item =
+        node->parent &&
+        node->parent->type == CMARK_NODE_ITEM &&
+        node->parent->parent->as.list.tight;
+    }
   }
 
   if (node->extension && node->extension->plaintext_render_func) {
@@ -79,13 +67,8 @@ static int S_render_node(cmark_renderer *renderer, cmark_node *node,
     if (cmark_node_get_list_type(node->parent) == CMARK_BULLET_LIST) {
       marker_width = 4;
     } else {
-      list_number = cmark_node_get_list_start(node->parent);
+      list_number = cmark_node_get_item_index(node);
       list_delim = cmark_node_get_list_delim(node->parent);
-      tmp = node;
-      while (tmp->prev) {
-        tmp = tmp->prev;
-        list_number += 1;
-      }
       // we ensure a width of at least 4 so
       // we get nice transition from single digits
       // to double

data/ext/markly/references.c

@@ -32,6 +32,7 @@ void cmark_reference_create(cmark_map *map, cmark_chunk *label,
   ref->title = cmark_clean_title(map->mem, title);
   ref->entry.age = map->size;
   ref->entry.next = map->refs;
+  ref->entry.size = ref->url.len + ref->title.len;
 
   map->refs = (cmark_map_entry *)ref;
   map->size++;

data/ext/markly/render.c

@@ -31,13 +31,7 @@ static void S_out(cmark_renderer *renderer, cmark_node *node,
   cmark_chunk remainder = cmark_chunk_literal("");
   int k = renderer->buffer->size - 1;
 
-  cmark_syntax_extension *ext = NULL;
-  cmark_node *n = node;
-  while (n && !ext) {
-    ext = n->extension;
-    if (!ext)
-      n = n->parent;
-  }
+  cmark_syntax_extension *ext = node->ancestor_extension;
   if (ext && !ext->commonmark_escape_func)
     ext = NULL;
 
@@ -182,6 +176,20 @@ char *cmark_render(cmark_mem *mem, cmark_node *root, int options, int width,
 
   while ((ev_type = cmark_iter_next(iter)) != CMARK_EVENT_DONE) {
     cur = cmark_iter_get_node(iter);
+    if (cur->extension) {
+      cur->ancestor_extension = cur->extension;
+    } else if (cur->parent) {
+      cur->ancestor_extension = cur->parent->ancestor_extension;
+    }
+    if (cur->type == CMARK_NODE_ITEM) {
+      // Calculate the list item's index, for the benefit of output formats
+      // like commonmark and plaintext.
+      if (cur->prev) {
+        cmark_node_set_item_index(cur, 1 + cmark_node_get_item_index(cur->prev));
+      } else {
+        cmark_node_set_item_index(cur, cmark_node_get_list_start(cur->parent));
+      }
+    }
     if (!render_node(&renderer, cur, ev_type, options)) {
       // a false value causes us to skip processing
       // the node's contents.  this is used for