update wiki registry. (#317)

Author: diasbruno

clib.json

@@ -9,7 +9,7 @@
     "which": "0.1.3",
     "stephenmathieson/str-flatten.c": "0.0.4",
     "commander": "1.3.2",
-    "stephenmathieson/wiki-registry.c": "0.0.4",
+    "clibs/wiki-registry.c": "0.1.2",
     "stephenmathieson/case.c": "0.1.3",
     "jwerle/fs.c": "0.2.0",
     "stephenmathieson/str-replace.c": "0.0.6",

deps/wiki-registry/package.json

@@ -1,7 +1,7 @@
 {
   "name": "wiki-registry",
-  "version": "0.0.4",
-  "repo": "stephenmathieson/wiki-registry.c",
+  "version": "0.1.2",
+  "repo": "clibs/wiki-registry.c",
   "description": "Turn a GitHub Wiki page into a package registry",
   "keywords": [ "registry", "github", "wiki" ],
   "license": "MIT",

deps/wiki-registry/wiki-registry.c

@@ -1,4 +1,3 @@
-
 //
 // wiki-registry.c
 //
@@ -96,31 +95,37 @@ wiki_registry_parse(const char *html) {
 
   GumboNode *body = gumbo_get_element_by_id("wiki-body", output->root);
   if (body) {
-    // grab all category `<h2 />`s
-    list_t *h2s = gumbo_get_elements_by_tag_name("h2", body);
-    list_node_t *heading_node;
-    list_iterator_t *heading_iterator = list_iterator_new(h2s, LIST_HEAD);
-    while ((heading_node = list_iterator_next(heading_iterator))) {
-      GumboNode *heading = (GumboNode *) heading_node->val;
-      char *category = gumbo_text_content(heading);
-      // die if we failed to parse a category, as it's
-      // almost certinaly a malloc error
-      if (!category) break;
-      trim(case_lower(category));
-      GumboVector *siblings = &heading->parent->v.element.children;
-      size_t pos = heading->index_within_parent;
-
-      // skip elements until the UL
-      // TODO: don't hardcode position here
-      // 2:
-      //   1 - whitespace
-      //   2 - actual node
-      GumboNode *ul = siblings->data[pos + 2];
-      if (GUMBO_TAG_UL != ul->v.element.tag) {
-        free(category);
+    GumboNode* markdown_body = ((GumboNode*)((GumboVector)body->v.element.children).data[1]);
+    GumboVector children = (GumboVector)markdown_body->v.element.children;
+
+    size_t count = children.length - 1;
+
+    for (size_t index = 0; index < count; index++) {
+      GumboNode *heading = (GumboNode *)children.data[index];
+      GumboNode *ul = NULL;
+
+      if (heading->v.element.tag != GUMBO_TAG_DIV) {
+        continue;
+      }
+
+      GumboAttribute *node_id = gumbo_get_attribute(&heading->v.element.attributes, "class");
+      if (node_id == NULL || strncmp(node_id->value, "markdown-heading", 16) != 0) {
         continue;
       }
 
+      for (; index < count; index++) {
+        ul = (GumboNode *)children.data[index];
+
+        if (ul->v.element.tag == GUMBO_TAG_UL) {
+          break;
+        }
+      }
+
+      list_t *h2 = gumbo_get_elements_by_tag_name("h2", heading);
+      char *category = gumbo_text_content(h2->head->val);
+      if (!category) break;
+      trim(case_lower(category));
+
       list_t *lis = gumbo_get_elements_by_tag_name("li", ul);
       list_iterator_t *li_iterator = list_iterator_new(lis, LIST_HEAD);
       list_node_t *li_node;
@@ -138,8 +143,6 @@ wiki_registry_parse(const char *html) {
       list_destroy(lis);
       free(category);
     }
-    list_iterator_destroy(heading_iterator);
-    list_destroy(h2s);
   }
 
   gumbo_destroy_output(&kGumboDefaultOptions, output);

test/search-basic.sh

@@ -3,14 +3,14 @@
 CACHE=$TMPDIR/clib-search.cache
 rm -f "$CACHE" 2> /dev/null
 
-N=$(clib search | wc -l)
+N=$(./clib-search | wc -l)
 # lame check for more than 100 lines of output
 [ "$N" -lt 100 ] && {
   echo >&2 "Expected \`clib search\` to return at least 100 results"
   exit 1
 }
 
-TRIM=$(clib search trim)
+TRIM=$(./clib-search trim)
 case "$TRIM" in
   *"stephenmathieson/trim.c"*)
     :