From 9f4151aa839c6f4752bf87dc1801514ab41e979a Mon Sep 17 00:00:00 2001
From: Yuanle Song <sylecn@gmail.com>
Date: Tue, 8 Sep 2015 11:49:07 +0800
Subject: [PATCH] v1.1.0 support ignoring lines that can't be parsed.
- small tweak on output guidelines
---
apache2-log-parser.cabal | 2 +-
operational | 16 ++++++++++++++++
src/Main.hs | 27 ++++++++++++++++++++++-----
3 files changed, 39 insertions(+), 6 deletions(-)
create mode 100644 operational
diff --git a/apache2-log-parser.cabal b/apache2-log-parser.cabal
index 64d0537..ef5c668 100644
--- a/apache2-log-parser.cabal
+++ b/apache2-log-parser.cabal
@@ -1,5 +1,5 @@
name: apache2-log-parser
-version: 1.0.1
+version: 1.1.0
synopsis: Simple project template from stack
description: Please see README.md
homepage: http://github.com/githubuser/apache2-log-parser#readme
diff --git a/operational b/operational
new file mode 100644
index 0000000..c64a9fe
--- /dev/null
+++ b/operational
@@ -0,0 +1,16 @@
+* COMMENT -*- mode: org -*-
+#+Date: 2015-09-08
+Time-stamp: <2015-09-08>
+#+STARTUP: content
+* later :entry:
+* current :entry:
+**
+** 2015-09-08 support reading input from stdin
+* done :entry:
+** DONE 2015-09-08 it doens't handle parse failures.
+Should warn about lines that failed to parse, ignore that line and continue
+parse the rest.
+
+Return value should be changed to failed lines [ByteString] and successfully
+parsed lines [AccessLog]. Parser should have a alternative failure case.
+
diff --git a/src/Main.hs b/src/Main.hs
index 902c1a7..e615313 100644
--- a/src/Main.hs
+++ b/src/Main.hs
@@ -8,10 +8,14 @@ import Data.Attoparsec.ByteString.Char8 hiding (option, take)
import Data.Word
import qualified Data.ByteString as B
import Data.ByteString (ByteString)
+import qualified Data.ByteString.Char8 as C8
import Options.Applicative hiding (Parser)
import qualified Options.Applicative as O
+import Data.Either (rights, lefts)
+import Control.Monad (when)
+
import Data.Time.Clock (getCurrentTime, diffUTCTime)
import Text.Printf (printf)
@@ -95,8 +99,13 @@ parseLine = do
,browser=browser
,responseTime=responseTime}
-logParser :: Parser [AccessLog]
-logParser = many $ parseLine <* endOfLine
+-- | if parse fail, return the whole line as Left. otherwise, return parsed
+-- AccessLog as Right.
+parseLineSafe :: Parser (Either ByteString AccessLog)
+parseLineSafe = (Right <$> (parseLine <* endOfLine)) <|> (Left <$> (tillChar '\n' <* endOfLine))
+
+logParser :: Parser [Either ByteString AccessLog]
+logParser = many $ parseLineSafe
----------------------------------
-- command line argument handling
@@ -174,7 +183,7 @@ countRequests logs = go Map.empty logs where
-- list m is ordered by view count DESC.
prettyPrintFrequentRequests :: Int -> [(RequestPair, Int)] -> IO ()
prettyPrintFrequentRequests totalRequestCount m = do
- putStrLn "== most frequently accessed URLs =="
+ printf "== most frequently accessed URLs (top %d) ==\n" (length m)
mapM_ ppLine m where
ppLine :: (RequestPair, Int) -> IO ()
ppLine ((verb, path), count) = do
@@ -199,6 +208,13 @@ parseFile args = do
else
printf "parsed %d lines in %d seconds. (%d line/s)\n"
linesParsed duration (linesParsed `div` duration)
+ -- warn about failed lines
+ let failures = lefts logs
+ when (not . null $ failures) (do
+ printf "== failed lines (%d) ==\n" (length failures)
+ mapM_ C8.putStrLn failures)
+ -- now show statistics on Right lines
+ let rlogs = rights logs
-- print slow requests if -g is used.
case slowRequestDuration args of
Nothing -> return ()
@@ -207,11 +223,12 @@ parseFile args = do
printf "== requests that take longer than %dus ==\n"
durationMicroseconds
mapM_ print
- (filter (\log -> responseTime log >= durationMicroseconds) logs)
+ (filter (\log -> responseTime log >= durationMicroseconds)
+ rlogs)
-- print top N most frequently viewed URL if -t is used.
case topNRequest args of
Nothing -> return ()
- Just n -> prettyPrintFrequentRequests linesParsed (take n (sortOn (Down . snd) (Map.toList (Map.filter (> minRequestCount) (countRequests logs)))))
+ Just n -> prettyPrintFrequentRequests linesParsed (take n (sortOn (Down . snd) (Map.toList (Map.filter (> minRequestCount) (countRequests rlogs)))))
main :: IO ()
main = parseArguments >>= parseFile
--
GitLab