aktualizacja formatu originalDate

This commit is contained in:
Rafał Sobański 2021-05-01 05:27:55 +02:00
parent 11683b816c
commit 64ce5a8da7
9 changed files with 27 additions and 5 deletions

View File

@ -1,4 +1,4 @@
/home/arab/tajemnica-atari/shadow-library.cabal: /home/arab/tajemnica-atari/shadow-library.cabal:
hash: 714bd61ef22b868b423c5ab32edbbca76e241ecb2075dced142701878c34217a hash: 714bd61ef22b868b423c5ab32edbbca76e241ecb2075dced142701878c34217a
/home/arab/tajemnica-atari/app/tajemnicaatari.hs: /home/arab/tajemnica-atari/app/tajemnicaatari.hs:
hash: d6d30253647c843a66123cc3a966da1ec5db47b670cfbccc24a2b7289493980c hash: 2ae5ec6f34912c37ccd5a6788081b676990ad5acc12fe86d6d08ead9776b94c7

Binary file not shown.

View File

@ -13,6 +13,7 @@ Następnie należy wykonać następujące kroki:
``` ```
cd tajemnica-atari cd tajemnica-atari
stack install stack install
cd
./.local/bin/tajemnicaatari ./.local/bin/tajemnicaatari
``` ```

View File

@ -11,7 +11,7 @@ import Data.List.Utils (replace)
import Text.Regex.Posix import Text.Regex.Posix
import Text.Printf import Text.Printf
import Data.List.Utils import Data.List.Utils
extractRecords = extractLinksWithText "//a[@class='ramka']" -- pary adres-tytuł extractRecords = extractLinksWithText "//a[@class='ramka']" -- pary adres-tytuł
@ -22,7 +22,7 @@ extractRecords = extractLinksWithText "//a[@class='ramka']" -- pary adres-tytu
toShadowItem :: ((String, String), String) -> ShadowItem toShadowItem :: ((String, String), String) -> ShadowItem
toShadowItem ((url, articleTitle), yearlyTitle) = toShadowItem ((url, articleTitle), yearlyTitle) =
(defaultShadowItem url title) { (defaultShadowItem url title) {
originalDate = Just newFormatDate, originalDate = Just newDate,
itype = "periodical", itype = "periodical",
format = Just "zip", format = Just "zip",
finalUrl = url finalUrl = url
@ -30,13 +30,31 @@ toShadowItem ((url, articleTitle), yearlyTitle) =
where title = "Tajemnica Atari " ++ yearlyTitle ++ " " ++ (replace "\r\n" "" (replace "\r\n " "" articleTitle)) where title = "Tajemnica Atari " ++ yearlyTitle ++ " " ++ (replace "\r\n" "" (replace "\r\n " "" articleTitle))
date = getDate url date = getDate url
splitDate = split "_" date splitDate = split "_" date
newFormatDate = "19" ++ splitDate !! 1 ++ "-" ++splitDate !! 0 month = splitDate !! 0
year = splitDate !! 1
newDate = newFormatDate month year
getDate url = getDate url =
case url =~~ "/([1-9]_9[0-9]|1[0-2]_9[0-9]|[1-9]-[1-9]_9[0-9]|1[0-2]-1[0-2]_9[0-9])/" :: Maybe [[String]] of case url =~~ "/([1-9]_9[0-9]|1[0-2]_9[0-9]|[1-9]-[1-9]_9[0-9]|1[0-2]-1[0-2]_9[0-9])/" :: Maybe [[String]] of
Just [[_, year]] -> year Just [[_, year]] -> year
otherwise -> error $ "unexpected url: " ++ url otherwise -> error $ "unexpected url: " ++ url
newFormatDate :: String -> String -> String
newFormatDate month year =
if month =~ "^[1-9]$" then "19" ++ year ++ "-0" ++ month
else if month =~ "^1[0-2]$" then "19" ++ year ++ "-" ++ month
else splittedMonth month year
splittedMonth :: String -> String -> String
splittedMonth month year = do
if month =~ "^1*[0-9]-1*[0-9]$" then do
let firstMonth = split "-" month !! 0
let secondMonth = split "-" month !! 1
if firstMonth =~ "^[1-9]$" && secondMonth =~ "^[1-9]$" then "19" ++ year ++ "-0" ++ firstMonth ++ "-0" ++ secondMonth
else if firstMonth =~ "^[1-9]$" && secondMonth =~ "^1[0-9]$" then "19" ++ year ++ "-0" ++ firstMonth ++ "-" ++ secondMonth
else if firstMonth =~ "^1[1-9]$" && secondMonth =~ "^1[0-9]$" then "19" ++ year ++ "-" ++ firstMonth ++ "-" ++ secondMonth
else "wrong date"
else "wrong date"
main = do main = do
let start = "http://krap.pl/mirrorz/atari/horror.mirage.com.pl/pixel/" let start = "http://krap.pl/mirrorz/atari/horror.mirage.com.pl/pixel/"
@ -46,3 +64,6 @@ main = do
lLevel=0, lLevel=0,
webpage=start} webpage=start}
extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem) extractItemsStartingFromUrl shadowLibrary start (extractRecords >>> arr toShadowItem)