From 1ab32d004caae32b8c04fe782af14909dd15e64c Mon Sep 17 00:00:00 2001
From: Fabien Freling <public@ffreling.com>
Date: Mon, 3 Mar 2014 23:33:21 +0100
Subject: [PATCH] Parse HTML document to retrieve title.

---
 ocaml/shell.ml      | 12 +++++++++---
 ocaml/simpleHttp.ml | 46 +++++++++++++++++++++++++++++++++++++++------
 2 files changed, 49 insertions(+), 9 deletions(-)

diff --git a/ocaml/shell.ml b/ocaml/shell.ml
index ecdbaa2..31dd7d3 100644
--- a/ocaml/shell.ml
+++ b/ocaml/shell.ml
@@ -15,9 +15,15 @@ let is_youtube_url url =
  * iter on all the items in the list *)
 let evaluate str =
   match str with
-  | str when is_youtube_url str -> SimpleHttp.get_http_title (SimpleHttp.get_body str); str
-  | str when is_url str -> SimpleHttp.get_http_title (SimpleHttp.get_body str); str
-  | _ -> str
+  | str when is_youtube_url str ->
+      (
+        let title = SimpleHttp.get_http_title (SimpleHttp.get_body str) in
+        match title with
+        | Some s -> s
+        | None -> ""
+      )
+  | str when is_url str -> str
+  | _ -> ""
 
 
 let () =
diff --git a/ocaml/simpleHttp.ml b/ocaml/simpleHttp.ml
index 36d9bdf..25e1f3a 100644
--- a/ocaml/simpleHttp.ml
+++ b/ocaml/simpleHttp.ml
@@ -14,20 +14,54 @@ Convenience.configure_pipeline
 
 let extract_string_value document =
   match document with
-  | Data(s) -> s
-  | _ -> ""
+  | Data(s) -> Some s
+  | _ -> None
 
 
-let rec get_title_element document =
+let find_string_value = function
+  |  Some s -> true
+  | None -> false
+
+
+let rec get_title_element_from_list doc_list =
+
+  try
+    List.find find_string_value (List.map get_title_element doc_list)
+  with
+    Not_found -> None
+
+and get_title_element document =
+
   match document with
-  | Element(e, args, sub) -> printf "%s: %s" "Element\n" e
-  | Data(s) -> printf "%s: %s" "Data\n" s
+  | Element("title", args, sub) ->
+      (
+        let title_candidates = List.map extract_string_value sub in
+        try
+          List.find find_string_value title_candidates
+        with
+          Not_found -> None
+      )
+  | Element(e, args, sub) -> get_title_element_from_list sub
+  | Data(s) -> None
+
+
+let rec print_document document =
+  match document with
+  | Element(e, args, sub) ->
+      printf "Element: %s\n" e;
+      List.iter print_document sub
+  | Data(s) -> printf "Data: %s\n" s
+
+
+let get_http_document body_str =
+  let ch = new Netchannels.input_string body_str in
+  Nethtml.parse ch
 
 
 let get_http_title body =
   let ch = new Netchannels.input_string body in
   let doc = Nethtml.parse ch in
-  get_title_element (List.hd doc)
+  get_title_element_from_list doc
 
 
 (* TODO: Log errors *)