From 5977e1454d7738ddb086d37b20337e350e380790 Mon Sep 17 00:00:00 2001
From: Joris Guyonvarch
Date: Sat, 11 Apr 2015 11:50:48 +0200
Subject: Fetch first page ads of a given leboncoin url, fetch also the
 description page of each item.

---
 src/Parser/Detail.hs | 14 ++++++++++++++
 src/Parser/Resume.hs | 27 +++++++++++++++++++++++++++
 src/Parser/Utils.hs  | 30 ++++++++++++++++++++++++++++++
 3 files changed, 71 insertions(+)
 create mode 100644 src/Parser/Detail.hs
 create mode 100644 src/Parser/Resume.hs
 create mode 100644 src/Parser/Utils.hs

(limited to 'src/Parser')

diff --git a/src/Parser/Detail.hs b/src/Parser/Detail.hs
new file mode 100644
index 0000000..031d740
--- /dev/null
+++ b/src/Parser/Detail.hs
@@ -0,0 +1,14 @@
+module Parser.Detail
+  ( parseDetail
+  ) where
+
+import Text.HTML.TagSoup
+
+import Model.Detail
+
+import Parser.Utils
+
+parseDetail :: [Tag String] -> Detail
+parseDetail tags =
+  let description = getTagText "<div class=content>" tags
+  in  Detail { description = description }
diff --git a/src/Parser/Resume.hs b/src/Parser/Resume.hs
new file mode 100644
index 0000000..bd73912
--- /dev/null
+++ b/src/Parser/Resume.hs
@@ -0,0 +1,27 @@
+module Parser.Resume
+  ( parseResumes
+  ) where
+
+import Data.Maybe (catMaybes)
+
+import Text.HTML.TagSoup
+
+import Model.Resume
+
+import Parser.Utils
+
+parseResumes :: String -> [Resume]
+parseResumes page =
+  case sections (~== "<div class=list-lbc>") (parseTags page) of
+    [] ->
+      []
+    sectionTags : _ ->
+      let lbcTags = takeWhile (~/= "<div id=alertesCartouche>") sectionTags
+      in  catMaybes . fmap parseResume $ partitions (~== "<a>") lbcTags
+
+parseResume :: [Tag String] -> Maybe Resume
+parseResume item = do
+  name <- getTagText "<h2 class=title>" item
+  let price = getTagText "<div class=price>" item
+  url <- getTagAttribute "<a>" "href" item
+  return Resume { name = name, price = price, url = url }
diff --git a/src/Parser/Utils.hs b/src/Parser/Utils.hs
new file mode 100644
index 0000000..4864e00
--- /dev/null
+++ b/src/Parser/Utils.hs
@@ -0,0 +1,30 @@
+module Parser.Utils
+  ( getTagAttribute
+  , getTagText
+  ) where
+
+import Data.List (find, findIndex)
+import Data.Maybe (listToMaybe)
+
+import Text.HTML.TagSoup
+
+getTagAttribute :: String -> String -> [Tag String] -> Maybe String
+getTagAttribute selector attribute item =
+  find (~== selector) item >>= maybeTagAttribute attribute
+
+getTagText :: String -> [Tag String] -> Maybe String
+getTagText selector item =
+  case findIndex (~== selector) item of
+    Just index -> fmap trim $ safeGetAt (index + 1) item >>= maybeTagText
+    Nothing -> Nothing
+
+maybeTagAttribute :: String -> Tag String -> Maybe String
+maybeTagAttribute name (TagOpen _ xs) =
+  fmap snd . find (\(x, _) -> x == name) $ xs
+maybeTagAttribute attribute _ = Nothing
+
+trim :: String -> String
+trim = unwords . words
+
+safeGetAt :: Int -> [a] -> Maybe a
+safeGetAt index = listToMaybe . drop index
-- 
cgit v1.2.3