diff options
| author | Alan Pearce | 2014-10-19 13:20:46 +0100 |
|---|---|---|
| committer | Alan Pearce | 2014-10-19 13:20:46 +0100 |
| commit | a4735ff01cdb3ac784820b0510bf07a33a561f1b (patch) | |
| tree | 79e382333d9192939bbd62d14745076ea208a784 | |
| parent | fe2e6c08a4f8d42317fe8e97a62e061209d4ed4d (diff) | |
| download | microformats-a4735ff01cdb3ac784820b0510bf07a33a561f1b.tar.lz microformats-a4735ff01cdb3ac784820b0510bf07a33a561f1b.tar.zst microformats-a4735ff01cdb3ac784820b0510bf07a33a561f1b.zip | |
Allow base-url to be overridden by parse argument
| -rw-r--r-- | src/microformats/parser.clj | 22 | ||||
| -rw-r--r-- | test/microformats/parser_expectations.clj | 6 |
2 files changed, 25 insertions, 3 deletions
diff --git a/src/microformats/parser.clj b/src/microformats/parser.clj index 21fcc26..9137eaa 100644 --- a/src/microformats/parser.clj +++ b/src/microformats/parser.clj @@ -387,8 +387,24 @@ [locs] (or (apply merge-with into (map parse-rel (select-rels locs))) {})) +(defprotocol node + (to-node [html])) + +(extend-protocol node + String + (to-node [html] + (some->> html str/trim html/html-snippet))) + +(extend-protocol node + clojure.lang.LazySeq + (to-node [html] + html)) + (defn parse "Parse a HTML string with microformats" - [html] - (let [document (some->> html str/trim html/html-snippet with-base-url (map z/xml-zip) first)] - {:items (some-> document walk) :rels (parse-rels document)})) + ([html] + (let [document (some->> html to-node with-base-url)] + (parse document (get-base-url document)))) + ([html base-url] + (let [document (some->> html to-node first (with-base-url base-url) z/xml-zip)] + {:items (some-> document walk) :rels (parse-rels document)}))) diff --git a/test/microformats/parser_expectations.clj b/test/microformats/parser_expectations.clj index f8c7bfb..dd7b335 100644 --- a/test/microformats/parser_expectations.clj +++ b/test/microformats/parser_expectations.clj @@ -305,6 +305,12 @@ <div class=\"h-card\"><img class=\"u-photo\" alt=\"Example User\" src=\"me.png\"></div> </body></html>")) +(expect-focused {:items '({:type ("h-card") :properties {:name ("Example User") + :photo ("http://not-example.com/me.png")}}) :rels {}} + (parse "<html><head><base href=\"http://example.com/\"></head><body> +<div class=\"h-card\"><img class=\"u-photo\" alt=\"Example User\" src=\"me.png\"></div> +</body></html>" "http://not-example.com/")) + (expect {:items '({:type ("h-card"), :properties {:photo ("http://blog.mozilla.org/press/files/2012/04/mitchell-baker.jpg"), :url ("http://blog.lizardwrangler.com/" "https://twitter.com/MitchellBaker"), |