Merge branch 'dev-phylo' into dev-merge

bin/gargantext-adaptative-phylo/Main.hs

+{-|
+Module      : Main.hs
+Description : Gargantext starter binary with Adaptative Phylo
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+
+Adaptative Phylo binaries
+ -}
+
+{-# LANGUAGE DataKinds          #-}
+{-# LANGUAGE DeriveGeneric      #-}
+{-# LANGUAGE FlexibleInstances  #-}
+{-# LANGUAGE NoImplicitPrelude  #-}
+{-# LANGUAGE StandaloneDeriving #-}
+{-# LANGUAGE TypeOperators      #-}
+{-# LANGUAGE OverloadedStrings  #-}
+{-# LANGUAGE Strict             #-}
+
+module Main where
+
+import Data.Aeson
+import Data.ByteString.Lazy (ByteString)
+import Data.Maybe (isJust, fromJust)
+import Data.List  (concat, nub, isSuffixOf, take)
+import Data.String (String)
+import Data.Text  (Text, unwords)
+
+import Gargantext.Prelude
+import Gargantext.Database.Types.Node (HyperdataDocument(..))
+import Gargantext.Text.Context (TermList)
+import Gargantext.Text.Corpus.Parsers.CSV (csv_title, csv_abstract, csv_publication_year)
+import Gargantext.Text.Corpus.Parsers (FileFormat(..),parseFile)
+import Gargantext.Text.List.CSV (csvGraphTermList)
+import Gargantext.Text.Terms.WithList (Patterns, buildPatterns, extractTermsWithList)
+import Gargantext.Viz.AdaptativePhylo
+import Gargantext.Viz.Phylo.PhyloMaker (toPhylo)
+
+import GHC.IO (FilePath) 
+import Prelude (Either(..))
+import System.Environment
+import System.Directory (listDirectory)
+import Control.Concurrent.Async (mapConcurrently)
+
+import qualified Data.ByteString.Lazy as Lazy
+import qualified Data.Vector as Vector
+import qualified Gargantext.Text.Corpus.Parsers.CSV as Csv
+
+
+---------------
+-- | Tools | --
+---------------
+
+
+-- | To print an important message as an IO()
+printIOMsg :: String -> IO ()
+printIOMsg msg = 
+    putStrLn ( "\n"
+            <> "------------" 
+            <> "\n"
+            <> "-- | " <> msg <> "\n" )
+
+
+-- | To print a comment as an IO()
+printIOComment :: String -> IO ()
+printIOComment cmt =
+    putStrLn ( "\n" <> cmt <> "\n" )
+
+
+-- | To get all the files in a directory or just a file
+getFilesFromPath :: FilePath -> IO([FilePath])
+getFilesFromPath path = do 
+  if (isSuffixOf "/" path) 
+    then (listDirectory path) 
+    else return [path]
+
+
+--------------
+-- | Json | --
+--------------
+
+
+-- | To read and decode a Json file
+readJson :: FilePath -> IO ByteString
+readJson path = Lazy.readFile path
+
+
+----------------
+-- | Parser | --
+----------------
+
+-- | To filter the Ngrams of a document based on the termList
+filterTerms :: Patterns -> (a, Text) -> (a, [Text])
+filterTerms patterns (y,d) = (y,termsInText patterns d)
+  where
+    --------------------------------------
+    termsInText :: Patterns -> Text -> [Text]
+    termsInText pats txt = nub $ concat $ map (map unwords) $ extractTermsWithList pats txt
+    --------------------------------------
+
+
+-- | To transform a Wos file (or [file]) into a readable corpus
+wosToCorpus :: Int -> FilePath -> IO ([(Int,Text)])
+wosToCorpus limit path = do 
+      files <- getFilesFromPath path
+      take limit
+        <$> map (\d -> let date' = fromJust $ _hyperdataDocument_publication_year d
+                           title = fromJust $ _hyperdataDocument_title d
+                           abstr = if (isJust $ _hyperdataDocument_abstract d)
+                                   then fromJust $ _hyperdataDocument_abstract d
+                                   else ""
+                        in (date', title <> " " <> abstr)) 
+        <$> concat 
+        <$> mapConcurrently (\file -> 
+              filter (\d -> (isJust $ _hyperdataDocument_publication_year d)
+                         && (isJust $ _hyperdataDocument_title d))
+                <$> parseFile WOS (path <> file) ) files
+
+
+-- | To transform a Csv file into a readable corpus
+csvToCorpus :: Int -> FilePath -> IO ([(Int,Text)])
+csvToCorpus limit path = Vector.toList
+    <$> Vector.take limit
+    <$> Vector.map (\row -> (csv_publication_year row, (csv_title row) <> " " <> (csv_abstract row)))
+    <$> snd <$> Csv.readFile path
+
+
+-- | To use the correct parser given a CorpusType
+fileToCorpus :: CorpusParser -> FilePath -> IO ([(Int,Text)])
+fileToCorpus parser path = case parser of 
+  Wos limit -> wosToCorpus limit path
+  Csv limit -> csvToCorpus limit path
+
+
+-- | To parse a file into a list of Document
+fileToDocs :: CorpusParser -> FilePath -> TermList -> IO [Document]
+fileToDocs parser path lst = do
+  corpus <- fileToCorpus parser path
+  let patterns = buildPatterns lst
+  pure $ map ( (\(y,t) -> Document y t) . filterTerms patterns) corpus
+
+
+--------------
+-- | Main | --
+--------------   
+
+
+main :: IO ()
+main = do
+
+    printIOMsg "Starting the reconstruction"
+
+    printIOMsg "Read the configuration file"
+    [args]   <- getArgs
+    jsonArgs <- (eitherDecode <$> readJson args) :: IO (Either String Config)
+
+    case jsonArgs of
+        Left err     -> putStrLn err
+        Right config -> do
+
+            printIOMsg "Parse the corpus"
+            mapList <- csvGraphTermList (listPath config)
+            corpus  <- fileToDocs (corpusParser config) (corpusPath config) mapList
+            printIOComment (show (length corpus) <> " parsed docs from the corpus")
+
+            printIOMsg "Reconstruct the Phylo"
+            let phylo = toPhylo corpus mapList config
+
+            printIOMsg "End of reconstruction"
\ No newline at end of file

bin/gargantext-phylo/Main.hs

@@ -216,7 +216,7 @@ main = do
                   (reBranchThr conf) (reBranchNth conf) (phyloLevel conf)
                   (RelatedComponents $ RCParams $ WeightedLogJaccard $ WLJParams (clusterTh conf) (clusterSens conf))
 
-      let queryView = PhyloQueryView (viewLevel conf) Merge False 1 [BranchAge,BranchBirth] [SizeBranch $ SBParams (minSizeBranch conf)] [GroupLabelIncDyn,BranchPeakInc] (Just (ByBranchBirth,Asc)) Json Flat True           
+      let queryView = PhyloQueryView (viewLevel conf) Merge False 1 [BranchAge,BranchBirth,BranchGroups] [SizeBranch $ SBParams (minSizeBranch conf)] [GroupLabelIncDyn,BranchPeakInc] (Just (ByBranchBirth,Asc)) Json Flat True           
 
       let phylo = toPhylo query corpus termList fis'
 

package.yaml

@@ -68,6 +68,8 @@ library:
   - Gargantext.Viz.Graph.Distances.Matrice
   - Gargantext.Viz.Graph.Index
   - Gargantext.Viz.Phylo
+  - Gargantext.Viz.AdaptativePhylo
+  - Gargantext.Viz.Phylo.PhyloMaker
   - Gargantext.Viz.Phylo.Tools
   - Gargantext.Viz.Phylo.Example
   - Gargantext.Viz.Phylo.LevelMaker
@@ -278,6 +280,31 @@ executables:
       - split
       - unordered-containers
 
+  gargantext-adaptative-phylo:
+    main: Main.hs
+    source-dirs: bin/gargantext-adaptative-phylo
+    ghc-options:
+    - -threaded
+    - -rtsopts
+    - -with-rtsopts=-N
+    - -O2
+    - -Wmissing-signatures
+    dependencies:
+      - aeson
+      - async
+      - base
+      - bytestring
+      - containers
+      - directory
+      - gargantext
+      - vector
+      - parallel
+      - cassava
+      - ini
+      - optparse-generic
+      - split
+      - unordered-containers      
+
   gargantext-import:
     main: Main.hs
     source-dirs: bin/gargantext-import

src/Gargantext/Viz/AdaptativePhylo.hs

+{-|
+Module      : Gargantext.Viz.AdaptativePhylo
+Description : Phylomemy definitions and types.
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+
+Specifications of Phylomemy export format.
+
+Phylomemy can be described as a Temporal Graph with different scale of
+granularity of group of ngrams (terms and multi-terms).
+
+The main type is Phylo which is synonym of Phylomemy (only difference is
+the number of chars).
+
+References:
+Chavalarias, D., Cointet, J.-P., 2013. Phylomemetic patterns
+in science evolution — the rise and fall of scientific fields. PloS
+one 8, e54847.
+-}
+
+{-# LANGUAGE DeriveGeneric, DeriveAnyClass #-}
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE TemplateHaskell   #-}
+{-# LANGUAGE MultiParamTypeClasses #-}
+
+module Gargantext.Viz.AdaptativePhylo where
+
+import Data.Aeson
+import Data.Aeson.TH (deriveJSON)
+import Data.Text   (Text, pack)
+import Data.Vector (Vector)
+import Data.Map (Map)
+import Data.Set (Set)
+
+import Gargantext.Core.Utils.Prefix (unPrefix)
+import Gargantext.Prelude
+import Gargantext.Text.Context (TermList)
+
+import GHC.Generics
+import GHC.IO (FilePath)
+import Control.DeepSeq (NFData)
+import Control.Lens (makeLenses)
+
+
+----------------
+-- | Config | --
+----------------  
+
+
+data CorpusParser = 
+      Wos {_wos_limit :: Int}
+    | Csv {_csv_limit :: Int}
+    deriving (Show,Generic,Eq) 
+
+
+data Proximity = 
+      WeightedLogJaccard 
+      { _wlj_sensibility   :: Double 
+      , _wlj_thresholdInit :: Double
+      , _wlj_thresholdStep :: Double }
+    | Hamming 
+    deriving (Show,Generic,Eq) 
+
+
+data TimeUnit = 
+      Year 
+      { _year_period :: Int
+      , _year_step   :: Int
+      , _year_matchingFrame :: Int }
+      deriving (Show,Generic,Eq) 
+
+
+data ContextualUnit = 
+      Fis 
+      { _fis_support :: Int
+      , _fis_size    :: Int }
+      deriving (Show,Generic,Eq)       
+
+
+data Config = 
+     Config { corpusPath     :: FilePath
+            , listPath       :: FilePath
+            , outputPath     :: FilePath
+            , corpusParser   :: CorpusParser
+            , phyloName      :: Text
+            , phyloLevel     :: Int
+            , phyloProximity :: Proximity
+            , timeUnit       :: TimeUnit
+            , contextualUnit :: ContextualUnit
+            , branchSize     :: Int  
+            } deriving (Show,Generic,Eq)
+
+
+defaultConfig :: Config
+defaultConfig = 
+     Config { corpusPath     = ""
+            , listPath       = ""
+            , outputPath     = ""
+            , corpusParser   = Csv 1000
+            , phyloName      = pack "Default Phylo"
+            , phyloLevel     = 2
+            , phyloProximity = WeightedLogJaccard 10 0 0.05
+            , timeUnit       = Year 3 1 5
+            , contextualUnit = Fis 2 4
+            , branchSize     = 3  
+            }
+
+instance FromJSON Config
+instance ToJSON Config
+instance FromJSON CorpusParser
+instance ToJSON CorpusParser
+instance FromJSON Proximity
+instance ToJSON Proximity
+instance FromJSON TimeUnit
+instance ToJSON TimeUnit
+instance FromJSON ContextualUnit
+instance ToJSON ContextualUnit
+
+
+-- | Software parameters
+data Software =
+     Software { _software_name    :: Text
+              , _software_version :: Text
+     } deriving (Generic, Show, Eq)
+
+defaultSoftware :: Software
+defaultSoftware = 
+      Software { _software_name    = pack "Gargantext"
+               , _software_version = pack "v4" }
+
+
+-- | Global parameters of a Phylo
+data PhyloParam =
+     PhyloParam { _phyloParam_version  :: Text
+                , _phyloParam_software :: Software
+                , _phyloParam_config   :: Config
+     } deriving (Generic, Show, Eq)
+
+defaultPhyloParam :: PhyloParam
+defaultPhyloParam =
+      PhyloParam { _phyloParam_version  = pack "v2.adaptative"
+                 , _phyloParam_software = defaultSoftware
+                 , _phyloParam_config   = defaultConfig }
+
+
+------------------
+-- | Document | --
+------------------
+
+
+-- | Date : a simple Integer
+type Date = Int
+
+-- | Ngrams : a contiguous sequence of n terms
+type Ngrams = Text
+
+-- | Document : a piece of Text linked to a Date
+data Document = Document
+      { date :: Date
+      , text :: [Ngrams]
+      } deriving (Eq,Show,Generic,NFData)  
+
+
+--------------------
+-- | Foundation | --
+--------------------
+
+
+-- | The Foundations of a Phylo created from a given TermList 
+data PhyloFoundations = PhyloFoundations
+      { _foundations_roots   :: !(Vector Ngrams)
+      , _foundations_mapList :: TermList
+      } deriving (Generic, Show, Eq)
+
+
+---------------------------
+-- | Coocurency Matrix | --
+---------------------------
+
+
+-- | Cooc : a coocurency matrix between two ngrams
+type Cooc =  Map (Int,Int) Double
+
+
+-------------------
+-- | Phylomemy | --
+-------------------
+
+
+-- | Phylo datatype of a phylomemy
+--  foundations : the foundations of the phylo
+--  timeCooc : a Map of coocurency by minimal unit of time (ex: by year)
+--  timeDocs : a Map with the numbers of docs by minimal unit of time (ex: by year)
+--  param : the parameters of the phylomemy (with the user's configuration)
+--  periods : the temporal steps of a phylomemy
+data Phylo =
+     Phylo { _phylo_foundations :: PhyloFoundations
+           , _phylo_timeCooc    :: !(Map Date Cooc)
+           , _phylo_timeDocs    :: !(Map Date Double)
+           , _phylo_param       :: PhyloParam
+           , _phylo_periods     :: Map PhyloPeriodId PhyloPeriod
+           }
+           deriving (Generic, Show, Eq)
+
+
+-- | PhyloPeriodId : the id of a given period
+type PhyloPeriodId = (Date,Date)
+
+-- | PhyloPeriod : steps of a phylomemy on a temporal axis
+--  id: tuple (start date, end date) of the temporal step of the phylomemy
+--  levels: levels of granularity
+data PhyloPeriod =
+     PhyloPeriod { _phylo_periodPeriod :: (Date,Date)
+                 , _phylo_periodLevels :: Map PhyloLevelId PhyloLevel
+                 } deriving (Generic, Show, Eq)   
+
+
+-- | Level : a level of clustering
+type Level = Int
+
+-- | PhyloLevelId : the id of a level of clustering in a given period 
+type PhyloLevelId  = (PhyloPeriodId,Level)
+
+-- | PhyloLevel : levels of phylomemy on a synchronic axis
+-- Levels description:
+-- Level 0: The foundations and the base of the phylo
+-- Level 1: First level of clustering (the Fis)
+-- Level [2..N]: Nth level of synchronic clustering (cluster of Fis)
+data PhyloLevel =
+     PhyloLevel { _phylo_levelPeriod :: (Date,Date)
+                , _phylo_levelLevel  :: Level 
+                , _phylo_levelGroups :: Map PhyloGroupId PhyloGroup
+                } 
+                deriving (Generic, Show, Eq)   
+
+
+type PhyloGroupId  = (PhyloLevelId, Int)
+
+-- | BranchId : (a level, a sequence of branch index)
+-- the sequence is a path of heritage from the most to the less specific branch
+type PhyloBranchId = (Level, [Int])
+
+-- | PhyloGroup : group of ngrams at each level and period
+data PhyloGroup = 
+      PhyloGroup { _phylo_groupPeriod   :: (Date,Date)
+                 , _phylo_groupLevel    :: Level
+                 , _phylo_groupIndex    :: Int
+                 , _phylo_groupSupport  :: Support
+                 , _phylo_groupNgrams   :: [Int]
+                 , _phylo_groupCooc     :: !(Cooc)
+                 , _phylo_groupBranchId :: PhyloBranchId
+                 , _phylo_groupLevelParents  :: [Pointer]
+                 , _phylo_groupLevelChilds   :: [Pointer]
+                 , _phylo_groupPeriodParents :: [Pointer]
+                 , _phylo_groupPeriodChilds  :: [Pointer]
+                 , _phylo_groupGhostPointers :: [Pointer]
+                 }
+                 deriving (Generic, Show, Eq)
+
+-- | Weight : A generic mesure that can be associated with an Id
+type Weight = Double
+
+-- | Pointer : A weighted pointer to a given PhyloGroup
+type Pointer = (PhyloGroupId, Weight)
+
+type Link = ((PhyloGroupId, PhyloGroupId), Weight)
+
+data Filiation = ToParents | ToChilds deriving (Generic, Show)    
+data PointerType = TemporalPointer | LevelPointer deriving (Generic, Show)                
+
+
+---------------------------
+-- | Frequent Item Set | --
+---------------------------
+
+-- | Clique : Set of ngrams cooccurring in the same Document
+type Clique   = Set Ngrams
+
+-- | Support : Number of Documents where a Clique occurs
+type Support  = Int
+
+-- | Fis : Frequent Items Set (ie: the association between a Clique and a Support)
+data PhyloFis = PhyloFis
+  { _phyloFis_clique  :: Clique
+  , _phyloFis_support :: Support
+  , _phyloFis_period  :: (Date,Date)
+  } deriving (Generic,NFData,Show,Eq)
+
+
+----------------
+-- | Lenses | --
+----------------
+
+makeLenses ''Config
+makeLenses ''Proximity
+makeLenses ''ContextualUnit
+makeLenses ''TimeUnit
+makeLenses ''PhyloFoundations
+makeLenses ''PhyloFis
+makeLenses ''Phylo
+makeLenses ''PhyloPeriod
+makeLenses ''PhyloLevel
+makeLenses ''PhyloGroup
+makeLenses ''PhyloParam
+
+------------------------
+-- | JSON instances | --
+------------------------
+
+
+$(deriveJSON (unPrefix "_foundations_"  ) ''PhyloFoundations)
\ No newline at end of file

src/Gargantext/Viz/Phylo.hs

@@ -80,8 +80,8 @@ data Phylo =
            , _phylo_foundations :: PhyloFoundations
            , _phylo_periods     :: [PhyloPeriod]
            , _phylo_docsByYears :: Map Date Double
-           , _phylo_cooc        :: Map Date (Map (Int,Int) Double)
-           , _phylo_fis         :: Map (Date,Date) [PhyloFis]
+           , _phylo_cooc        :: !(Map Date (Map (Int,Int) Double))
+           , _phylo_fis         :: !(Map (Date,Date) [PhyloFis])
            , _phylo_param       :: PhyloParam
            }
            deriving (Generic, Show, Eq)
@@ -156,7 +156,7 @@ data PhyloGroup =
                 , _phylo_groupNgramsMeta    :: Map Text [Double]
                 , _phylo_groupMeta          :: Map Text Double
                 , _phylo_groupBranchId      :: Maybe PhyloBranchId
-                , _phylo_groupCooc          :: Map (Int,Int) Double
+                , _phylo_groupCooc          :: !(Map (Int,Int) Double)
 
                 , _phylo_groupPeriodParents :: [Pointer]
                 , _phylo_groupPeriodChilds  :: [Pointer]
@@ -309,7 +309,7 @@ data SBParams = SBParams
 
 
 -- | Metric constructors
-data Metric = BranchAge | BranchBirth deriving (Generic, Show, Eq, Read)
+data Metric = BranchAge | BranchBirth | BranchGroups deriving (Generic, Show, Eq, Read)
 
 
 ----------------

src/Gargantext/Viz/Phylo/Example.hs

@@ -83,7 +83,7 @@ queryViewEx = "level=3"
 
 
 phyloQueryView :: PhyloQueryView
-phyloQueryView = PhyloQueryView 2 Merge False 2 [BranchAge,BranchBirth] [] [BranchPeakInc,GroupLabelIncDyn] (Just (ByBranchBirth,Asc)) Json Flat True
+phyloQueryView = PhyloQueryView 2 Merge False 2 [BranchAge,BranchBirth,BranchGroups] [] [BranchPeakInc,GroupLabelIncDyn] (Just (ByBranchBirth,Asc)) Json Flat True
 
 
 --------------------------------------------------

src/Gargantext/Viz/Phylo/LevelMaker.hs

@@ -22,9 +22,10 @@ module Gargantext.Viz.Phylo.LevelMaker
 import Control.Parallel.Strategies
 import Control.Lens                 hiding (both, Level)
 import Data.List                    ((++), sort, concat, nub, zip, last, null)
-import Data.Map                     (Map, (!), empty, singleton)
+import Data.Map                     (Map, (!), empty, singleton, size)
 import Data.Text (Text)
 import Data.Tuple.Extra
+import Data.Vector (Vector)
 import Gargantext.Prelude
 import Gargantext.Viz.Phylo
 import Gargantext.Viz.Phylo.Metrics
@@ -83,7 +84,7 @@ instance PhyloLevelMaker PhyloFis
     --------------------------------------
     -- | Level -> (Date,Date) -> [Fis] -> Map (Date,Date) [Fis] -> Phylo -> [PhyloGroup]
     toPhyloGroups lvl (d,d') l _ p =
-      let groups  = map (\(idx,fis) -> cliqueToGroup (d,d') lvl idx "" fis p) $ zip [1..] l
+      let groups  = map (\(idx,fis) -> cliqueToGroup (d,d') lvl idx "" fis (getPhyloCooc p) (getFoundationsRoots p)) $ zip [1..] l
           groups' = groups `using` parList rdeepseq
       in  groups' 
     --------------------------------------
@@ -111,7 +112,7 @@ addPhyloLevel' lvl m p = alterPhyloPeriods
                                     in  over (phylo_periodLevels)
                                         (\phyloLevels ->
                                             let groups = toPhyloGroups lvl pId (m ! pId) m p
-                                            in phyloLevels ++ [PhyloLevel (pId, lvl) groups]
+                                            in trace (show (length groups) <> " groups for " <> show (pId) ) $ phyloLevels ++ [PhyloLevel (pId, lvl) groups]
                                         ) period) p
 
 
@@ -121,8 +122,8 @@ addPhyloLevel' lvl m p = alterPhyloPeriods
 
 
 -- | To transform a Clique into a PhyloGroup
-cliqueToGroup :: PhyloPeriodId -> Level -> Int -> Text -> PhyloFis -> Phylo -> PhyloGroup
-cliqueToGroup prd lvl idx lbl fis p = PhyloGroup ((prd, lvl), idx) lbl ngrams 
+cliqueToGroup :: PhyloPeriodId -> Level -> Int -> Text -> PhyloFis -> Map Date (Map (Int,Int) Double) -> Vector Ngrams -> PhyloGroup
+cliqueToGroup prd lvl idx lbl fis cooc' root = PhyloGroup ((prd, lvl), idx) lbl ngrams 
     (getNgramsMeta cooc ngrams)
     -- empty
     (singleton "support" (fromIntegral $ getSupport fis)) 
@@ -132,10 +133,10 @@ cliqueToGroup prd lvl idx lbl fis p = PhyloGroup ((prd, lvl), idx) lbl ngrams
       where
         --------------------------------------
         cooc :: Map (Int, Int) Double
-        cooc = getMiniCooc (listToFullCombi ngrams) (periodsToYears [prd]) (getPhyloCooc p)
+        cooc = getMiniCooc (listToFullCombi ngrams) (periodsToYears [prd]) cooc'
         --------------------------------------
         ngrams :: [Int]
-        ngrams = sort $ map (\x -> getIdxInRoots x p)
+        ngrams = sort $ map (\x -> getIdxInRoots' x root)
                       $ Set.toList
                       $ getClique fis
         --------------------------------------
@@ -210,7 +211,11 @@ toNthLevel lvlMax prox clus p
   | otherwise     = toNthLevel lvlMax prox clus
                   $ traceBranches (lvl + 1)
                   $ setPhyloBranches (lvl + 1)
-                  $ transposePeriodLinks (lvl + 1)
+                  -- $ transposePeriodLinks (lvl + 1)
+                  $ traceTranspose (lvl + 1) Descendant
+                  $ transposeLinks (lvl + 1) Descendant
+                  $ traceTranspose (lvl + 1) Ascendant
+                  $ transposeLinks (lvl + 1) Ascendant
                   $ tracePhyloN (lvl + 1)
                   $ setLevelLinks (lvl, lvl + 1)
                   $ addPhyloLevel (lvl + 1)
@@ -238,7 +243,8 @@ toPhylo1 clus prox d p = case clus of
                        $ interTempoMatching Ascendant 1 prox
                        $ tracePhyloN 1
                        -- $ setLevelLinks (0,1)
-                       $ addPhyloLevel 1 (getPhyloFis phyloFis) phyloFis
+                       $ addPhyloLevel 1 (getPhyloFis phyloFis)
+                       $ trace (show (size $ getPhyloFis phyloFis) <> " Fis created") $ phyloFis
     where
       --------------------------------------
       phyloFis :: Phylo
@@ -279,6 +285,9 @@ tracePhyloN :: Level -> Phylo -> Phylo
 tracePhyloN lvl p = trace ("\n---------------\n--| Phylo " <> show (lvl) <> " |--\n---------------\n\n"
                       <> show (length $ getGroupsWithLevel lvl p) <> " groups created \n") p
 
+traceTranspose :: Level -> Filiation -> Phylo -> Phylo
+traceTranspose lvl fil p = trace ("----\n Transpose " <> show (fil) <> " links for " <> show (length $ getGroupsWithLevel lvl p) <> " groups\n") p
+
 
 tracePhyloBase :: Phylo -> Phylo
 tracePhyloBase p = trace ( "\n-------------\n--| Phylo |--\n-------------\n\n" 

src/Gargantext/Viz/Phylo/LinkMaker.hs

@@ -19,9 +19,9 @@ module Gargantext.Viz.Phylo.LinkMaker
 
 import Control.Parallel.Strategies
 import Control.Lens                 hiding (both, Level)
-import Data.List                    ((++), sortOn, null, tail, splitAt, elem, concat, delete, intersect, groupBy, union, inits, scanl, find)
+import Data.List                    ((++), sortOn, null, tail, splitAt, elem, concat, delete, intersect, elemIndex, groupBy, union, inits, scanl, find)
 import Data.Tuple.Extra
-import Data.Map                     (Map,(!),fromListWith,elems,restrictKeys,unionWith,member)
+import Data.Map                     (Map, (!), fromListWith, elems, restrictKeys, filterWithKey, keys, unionWith, unions, intersectionWith, member, fromList)
 import Gargantext.Prelude
 import Gargantext.Viz.Phylo
 import Gargantext.Viz.Phylo.Tools
@@ -222,6 +222,62 @@ interTempoMatching fil lvl _ p = updateGroups fil lvl (Map.fromList pointers) p
 ------------------------------------------------------------------------
 -- | Make links from Period to Period after level 1
 
+
+listToTuple :: (a -> b) -> [a] -> [(b,a)]
+listToTuple f l = map (\x -> (f x, x)) l
+
+
+groupsToMaps :: Ord b => (PhyloGroup -> b) -> [PhyloGroup] -> [Map PhyloGroupId PhyloGroup]
+groupsToMaps f gs = map (\gs' -> fromList $ listToTuple getGroupId gs')
+                  $ groupBy ((==) `on` f)
+                  $ sortOn f gs
+
+
+phyloToPeriodMaps :: Level -> Filiation -> Phylo -> [Map PhyloGroupId PhyloGroup]
+phyloToPeriodMaps lvl fil p =
+  let prdMap = groupsToMaps (fst . getGroupPeriod) (getGroupsWithLevel lvl p)
+  in case fil of  
+    Ascendant  -> reverse prdMap
+    Descendant -> prdMap
+    _          -> panic ("[ERR][Viz.Phylo.LinkMaker.phyloToPeriodMaps] Wrong type of filiation")
+
+
+trackPointersRec :: Filiation -> Map PhyloGroupId PhyloGroup -> [PhyloGroup] -> [PhyloGroup] -> [PhyloGroup]
+trackPointersRec fil m gs res =
+  if (null gs) then res
+  else if (Map.null m) then res ++ gs
+  else 
+    let g = head' "track" gs
+        pts  = Map.fromList $ getGroupPointers PeriodEdge fil g
+        pts' = Map.toList $ fromListWith (\w w' -> max w w') $ concat $ elems
+             $ intersectionWith (\w g' -> map (\(id,_w') -> (id, w))
+                                        $ getGroupPointers LevelEdge Ascendant g') pts m
+        res' = res ++ [case fil of 
+                        Ascendant  -> g & phylo_groupPeriodParents .~ pts' 
+                        Descendant -> g & phylo_groupPeriodChilds  .~ pts'
+                        _          -> panic ("[ERR][Viz.Phylo.LinkMaker.transposeLinks] Wrong type of filiation")]
+    in trackPointersRec fil (filterWithKey (\k _ -> not $ elem k (keys pts)) m) (tail' "track" gs) res'
+
+
+
+transposeLinks :: Level -> Filiation -> Phylo -> Phylo
+transposeLinks lvl fil p = 
+  let prdMap = zip (phyloToPeriodMaps (lvl - 1) fil p) (phyloToPeriodMaps lvl fil p)
+      transposed =  map (\(gs,gs') -> 
+                            let idx  = fromJust $ elemIndex (gs,gs') prdMap
+                                next = take (getPhyloMatchingFrame p) $ snd $ splitAt (idx + 1) prdMap
+                                groups = trackPointersRec fil (unions $ map fst next) (elems gs') []
+                            in  (getGroupPeriod $ head' "transpose" groups ,groups)
+                         ) prdMap
+      transposed' = Map.fromList $ (transposed `using` parList rdeepseq)
+  in alterPhyloGroups
+      (\gs -> if ((not . null) gs) && (lvl == (getGroupLevel $ head' "transpose" gs))
+              then transposed' ! (getGroupPeriod $ head' "transpose" gs)
+              else gs
+      ) p 
+
+
+
 -- | Transpose the parent/child pointers from one level to another
 transposePeriodLinks :: Level -> Phylo -> Phylo
 transposePeriodLinks lvl p = alterPhyloGroups

src/Gargantext/Viz/Phylo/PhyloExample.hs

+{-|
+Module      : Gargantext.Viz.Phylo.PhyloExample
+Description : Phylomemy example based on history of Cleopatre.
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+
+-- | Cesar et Cleôpatre
+-- | Exemple de phylomemie
+-- | French without accents
+-}
+
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE FlexibleContexts  #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE MultiParamTypeClasses #-}
+
+module Gargantext.Viz.Phylo.PhyloExample where
+
+import Data.List (sortOn, nub, sort)
+import Data.Map (Map)
+import Data.Text (Text, toLower)
+
+import Gargantext.Prelude
+import Gargantext.Text.Context (TermList)
+import Gargantext.Text.Terms.Mono (monoTexts)
+import Gargantext.Viz.AdaptativePhylo
+import Gargantext.Viz.Phylo.PhyloTools
+import Gargantext.Viz.Phylo.PhyloMaker
+
+import Control.Lens
+
+import qualified Data.Vector as Vector
+
+
+-----------------------------------------------
+-- | STEP 3 | -- Build the Level 1 of the Phylo
+-----------------------------------------------
+
+phylo1 :: Phylo
+phylo1 = appendGroups fisToGroup 1 phyloFis phyloBase
+
+
+---------------------------------------------
+-- | STEP 2 | -- Build the frequent items set
+---------------------------------------------
+
+
+phyloFis :: Map (Date,Date) [PhyloFis]
+phyloFis = toPhyloFis docsByPeriods (getFisSupport $ contextualUnit config) (getFisSize $ contextualUnit config) 
+
+
+docsByPeriods :: Map (Date,Date) [Document]
+docsByPeriods = groupDocsByPeriod date periods docs
+
+
+--------------------------------------------
+-- | STEP 1 | -- Init the Base of the Phylo
+--------------------------------------------
+
+
+phyloBase :: Phylo
+phyloBase = toPhyloBase docs mapList config
+
+
+phyloCooc :: Map Date Cooc
+phyloCooc = docsToTimeScaleCooc docs (foundations ^. foundations_roots)
+
+
+periods :: [(Date,Date)]
+periods = toPeriods (sort $ nub $ map date docs) (getTimePeriod $ timeUnit config) (getTimeStep $ timeUnit config)
+
+
+nbDocsByYear :: Map Date Double
+nbDocsByYear = docsToTimeScaleNb docs
+
+
+config :: Config
+config = 
+    defaultConfig { phyloName  = "Cesar et Cleopatre"
+                  , branchSize = 0
+                  , contextualUnit = Fis 0 0 }
+
+
+docs :: [Document]
+docs = map (\(d,t)
+    -> Document d ( filter (\n -> isRoots n (foundations ^. foundations_roots)) 
+                  $ monoTexts t)) corpus
+
+
+foundations :: PhyloFoundations
+foundations = PhyloFoundations (Vector.fromList $ map toLower actants) mapList 
+
+
+--------------------------------------------
+-- | STEP 0 | -- Let's start with an example
+--------------------------------------------
+
+
+mapList :: TermList
+mapList = map (\a -> ([toLower a],[])) actants
+
+
+actants :: [Ngrams]
+actants = [ "Cleopatre"   , "Ptolemee", "Ptolemee-XIII", "Ptolemee-XIV"
+          , "Marc-Antoine", "Cesar"   , "Antoine"      , "Octave"  , "Rome"
+          , "Alexandrie"  , "Auguste" , "Pompee"       , "Cassius" , "Brutus"]
+
+corpus :: [(Date, Text)]
+corpus = sortOn fst [ (-51,"Cleopatre règne sur l’egypte entre 51 et 30 av. J.-C. avec ses frères-epoux Ptolemee-XIII et Ptolemee-XIV, puis aux côtes du general romain Marc-Antoine. Elle est celèbre pour avoir ete la compagne de Jules Cesar puis d'Antoine, avec lesquels elle a eu plusieurs enfants. Partie prenante dans la guerre civile opposant Antoine à Octave, elle est vaincue à la bataille d'Actium en 31 av. J.-C. Sa defaite va permettre aux Romains de mener à bien la conquête de l’egypte, evenement qui marquera la fin de l'epoque hellenistique."), (-40,"Il existe relativement peu d'informations sur son sejour à Rome, au lendemain de l'assassinat de Cesar, ou sur la periode passee à Alexandrie durant l'absence d'Antoine, entre -40 et -37."), (-48,"L'historiographie antique lui est globalement defavorable car inspiree par son vainqueur, l'empereur Auguste, et par son entourage, dont l'interêt est de la noircir, afin d'en faire l'adversaire malfaisant de Rome et le mauvais genie d'Antoine. On observe par ailleurs que Cesar ne fait aucune mention de sa liaison avec elle dans les Commentaires sur la Guerre civile"), (-69,"Cleopatre est nee au cours de l'hiver -69/-686 probablement à Alexandrie."), (-48,"Pompee a en effet ete le protecteur de Ptolemee XII, le père de Cleopatre et de Ptolemee-XIII dont il se considère comme le tuteur."), (-48,"Ptolemee-XIII et Cleopatre auraient d'ailleurs aide Pompee par l'envoi d'une flotte de soixante navires."), (-48,"Mais le jeune roi Ptolemee-XIII et ses conseillers jugent sa cause perdue et pensent s'attirer les bonnes graces du vainqueur en le faisant assassiner à peine a-t-il pose le pied sur le sol egyptien, près de Peluse, le 30 juillet 48 av. J.-C., sous les yeux de son entourage."), (-48,"Cesar fait enterrer la tête de Pompee dans le bosquet de Nemesis en bordure du mur est de l'enceinte d'Alexandrie. Pour autant la mort de Pompee est une aubaine pour Cesar qui tente par ailleurs de profiter des querelles dynastiques pour annexer l’egypte."), (-48,"Il est difficile de se prononcer clairement sur les raisons qui ont pousse Cesar à s'attarder à Alexandrie. Il y a des raisons politiques, mais aussi des raisons plus sentimentales (Cleopatre ?). Il tente d'abord d'obtenir le remboursement de dettes que Ptolemee XII"), (-46,"Les deux souverains sont convoques par Cesar au palais royal d'Alexandrie. Ptolemee-XIII s'y rend après diverses tergiversations ainsi que Cleopatre."), (-47,"A Rome, Cleopatre epouse alors un autre de ses frères cadets, à Alexandrie, Ptolemee-XIV, sur l'injonction de Jules Cesar"), (-46,"Cesar a-t-il comme objectif de montrer ce qu'il en coûte de se revolter contre Rome en faisant figurer dans son triomphe la sœur de Cleopatre et de Ptolemee-XIV, Arsinoe, qui s'est fait reconnaître reine par les troupes de Ptolemee-XIII ?"), (-44,"Au debut de l'annee -44, Cesar est assassine par Brutus. Profitant de la situation confuse qui s'ensuit, Cleopatre quitte alors Rome à la mi-avril, faisant escale en Grèce. Elle parvient à Alexandrie en juillet -44."), (-44,"La guerre que se livrent les assassins de Cesar, Cassius et Brutus et ses heritiers, Octave et Marc-Antoine, oblige Cleopatre à des contorsions diplomatiques."), (-41,"Nous ignorons depuis quand Cleopatre, agee de 29 ans en -41, et Marc-Antoine, qui a une quarantaine d'annees, se connaissent. Marc-Antoine est l'un des officiers qui ont participe au retablissement de Ptolemee XII.  Il est plus vraisemblable qu'ils se soient frequentes lors du sejour à Rome de Cleopatre."), (-42,"Brutus tient la Grèce tandis que Cassius s'installe en Syrie. Le gouverneur de Cleopatre à Chypre, Serapion, vient en aide à Cassius."), (-42,"Cassius aurait envisage de s'emparer d'Alexandrie quand le 'debarquement' en Grèce d'Antoine et d'Octave l'oblige à renoncer à ses projets")]          
\ No newline at end of file

src/Gargantext/Viz/Phylo/PhyloExport.hs

+{-|
+Module      : Gargantext.Viz.Phylo.PhyloExport
+Description : Exportation module of a Phylo
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+-}
+
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE FlexibleContexts  #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE MultiParamTypeClasses #-}
+
+module Gargantext.Viz.Phylo.PhyloExport where
\ No newline at end of file

src/Gargantext/Viz/Phylo/PhyloMaker.hs

+{-|
+Module      : Gargantext.Viz.Phylo.PhyloMaker
+Description : Maker engine for rebuilding a Phylo
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+-}
+
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE FlexibleContexts  #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE MultiParamTypeClasses #-}
+
+module Gargantext.Viz.Phylo.PhyloMaker where
+
+import Data.List (concat, nub, partition, sort, (++))
+import Data.Map (Map, fromListWith, keys, unionWith, fromList, empty, toList, elems, (!), filterWithKey, restrictKeys)
+import Data.Set (size)
+import Data.Vector (Vector)
+
+import Gargantext.Prelude
+import Gargantext.Viz.AdaptativePhylo
+import Gargantext.Viz.Phylo.PhyloTools
+import Gargantext.Text.Context (TermList)
+import Gargantext.Text.Metrics.FrequentItemSet (fisWithSizePolyMap, Size(..))
+
+import Control.DeepSeq (NFData)
+import Control.Parallel.Strategies (parList, rdeepseq, using)
+import Debug.Trace (trace)
+import Control.Lens hiding (Level)
+
+import qualified Data.Vector as Vector
+import qualified Data.Set as Set
+
+
+------------------
+-- | To Phylo | --
+------------------
+
+
+toPhylo :: [Document] -> TermList -> Config -> Phylo
+toPhylo docs lst conf = phylo1
+    where
+        --------------------------------------
+        phylo1 :: Phylo
+        phylo1 = toPhylo1 docs phyloBase
+        --------------------------------------
+        phyloBase :: Phylo 
+        phyloBase = toPhyloBase docs lst conf
+        --------------------------------------
+
+
+
+--------------------
+-- | To Phylo 1 | --
+--------------------
+
+
+appendGroups :: (a -> PhyloPeriodId -> Level -> Int -> Vector Ngrams -> [Cooc] -> PhyloGroup) -> Level -> Map (Date,Date) [a] -> Phylo -> Phylo
+appendGroups f lvl m phylo =  trace ("\n" <> "-- | Append " <> show (length $ concat $ elems m) <> " groups to Level " <> show (lvl) <> "\n")
+    $ over ( phylo_periods
+           .  traverse
+           . phylo_periodLevels
+           .  traverse)
+           (\phyloLvl -> if lvl == (phyloLvl ^. phylo_levelLevel)
+                         then
+                            let pId = phyloLvl ^. phylo_levelPeriod
+                                phyloFis = m ! pId
+                            in  phyloLvl 
+                              & phylo_levelGroups .~ (fromList $ foldl (\groups obj ->
+                                    groups ++ [ (((pId,lvl),length groups)
+                                              , f obj pId lvl (length groups) (getRoots phylo) 
+                                                  (elems $ restrictKeys (phylo ^. phylo_timeCooc) $ periodsToYears [pId]))
+                                              ] ) [] phyloFis)
+                         else 
+                            phyloLvl )
+           phylo  
+
+
+fisToGroup :: PhyloFis -> PhyloPeriodId -> Level ->  Int -> Vector Ngrams -> [Cooc] -> PhyloGroup
+fisToGroup fis pId lvl idx fdt coocs =
+    let ngrams = ngramsToIdx (Set.toList $ fis ^. phyloFis_clique) fdt
+    in  PhyloGroup pId lvl idx
+                   (fis ^. phyloFis_support)
+                   ngrams
+                   (ngramsToCooc ngrams coocs)
+                   (1,[])
+                   [] [] [] [] []
+
+
+toPhylo1 :: [Document] -> Phylo -> Phylo
+toPhylo1 docs phyloBase = appendGroups fisToGroup 1 phyloFis phyloBase
+    where
+        --------------------------------------
+        phyloFis :: Map (Date,Date) [PhyloFis]
+        phyloFis =  toPhyloFis docs' (getFisSupport $ contextualUnit $ getConfig phyloBase) (getFisSize $ contextualUnit $ getConfig phyloBase)
+        --------------------------------------
+        docs' :: Map (Date,Date) [Document]
+        docs' =  groupDocsByPeriod date (getPeriodIds phyloBase) docs
+        --------------------------------------
+
+
+---------------------------
+-- | Frequent Item Set | --
+---------------------------
+
+
+-- | To apply a filter with the possibility of keeping some periods non empty (keep : True|False)
+filterFis :: Bool -> Int -> (Int -> [PhyloFis] -> [PhyloFis]) -> Map (Date, Date) [PhyloFis] -> Map (Date, Date) [PhyloFis]
+filterFis keep thr f m = case keep of
+  False -> map (\l -> f thr l) m
+  True  -> map (\l -> keepFilled (f) thr l) m
+
+
+-- | To filter Fis with small Support
+filterFisBySupport :: Int -> [PhyloFis] -> [PhyloFis]
+filterFisBySupport thr l = filter (\fis -> (fis ^. phyloFis_support) >= thr) l
+
+
+-- | To filter Fis with small Clique size
+filterFisByClique :: Int -> [PhyloFis] -> [PhyloFis]
+filterFisByClique thr l = filter (\fis -> (size $ fis ^. phyloFis_clique) >= thr) l
+
+
+-- | To filter nested Fis
+filterFisByNested :: Map (Date, Date) [PhyloFis] -> Map (Date, Date) [PhyloFis]
+filterFisByNested m = 
+  let fis  = map (\l -> 
+                foldl (\mem f -> if (any (\f' -> isNested (Set.toList $ f' ^. phyloFis_clique) (Set.toList $ f ^. phyloFis_clique)) mem)
+                                 then mem
+                                 else 
+                                    let fMax = filter (\f' -> not $ isNested (Set.toList $ f ^. phyloFis_clique) (Set.toList $ f' ^. phyloFis_clique)) mem
+                                    in  fMax ++ [f] ) [] l)
+           $ elems m 
+      fis' = fis `using` parList rdeepseq
+  in  fromList $ zip (keys m) fis' 
+
+
+-- | To transform a time map of docs innto a time map of Fis with some filters
+toPhyloFis :: Map (Date, Date) [Document] -> Int -> Int -> Map (Date,Date) [PhyloFis]
+toPhyloFis phyloDocs support clique = traceFis "Filtered Fis"
+                $ filterFisByNested 
+                $ traceFis "Filtered by clique size"
+                $ filterFis True clique (filterFisByClique)
+                $ traceFis "Filtered by support"
+                $ filterFis True support (filterFisBySupport)
+                $ traceFis "Unfiltered Fis" phyloFis
+    where
+        -------------------------------------- 
+        phyloFis :: Map (Date,Date) [PhyloFis]
+        phyloFis = 
+            let fis  = map (\(prd,docs) -> let lst = toList $ fisWithSizePolyMap (Segment 1 20) 1 (map text docs)
+                                           in (prd, map (\f -> PhyloFis (fst f) (snd f) prd) lst))
+                     $ toList phyloDocs
+                fis' = fis `using` parList rdeepseq
+            in fromList fis'
+        -------------------------------------- 
+
+
+--------------------
+-- | Coocurency | --
+--------------------
+
+
+-- | To build the local cooc matrix of each phylogroup
+ngramsToCooc :: [Int] -> [Cooc] -> Cooc
+ngramsToCooc ngrams coocs =
+    let cooc  = foldl (\acc cooc' -> sumCooc acc cooc') empty coocs
+        pairs = listToKeys ngrams
+    in  filterWithKey (\k _ -> elem k pairs) cooc
+
+
+-- | To transform the docs into a time map of coocurency matrix 
+docsToTimeScaleCooc :: [Document] -> Vector Ngrams -> Map Date Cooc
+docsToTimeScaleCooc docs fdt = 
+    let mCooc  = fromListWith sumCooc
+               $ map (\(_d,l) -> (_d, listToMatrix l))
+               $ map (\doc -> (date doc, sort $ ngramsToIdx (text doc) fdt)) docs
+        mCooc' = fromList
+               $ map (\t -> (t,empty))
+               $ toTimeScale (map date docs) 1
+    in   trace ("\n" <> "-- | Build the coocurency matrix for " <> show (length $ keys mCooc') <> " unit of time" <> "\n")
+       $ unionWith sumCooc mCooc mCooc'
+
+
+-----------------------
+-- | to Phylo Base | --
+-----------------------
+
+
+-- | To group a list of Documents by fixed periods
+groupDocsByPeriod :: (NFData doc, Ord date, Enum date) => (doc -> date) -> [(date,date)] -> [doc] -> Map (date, date) [doc]
+groupDocsByPeriod _ _   [] = panic "[ERR][Viz.Phylo.PhyloMaker] Empty [Documents] can not have any periods"
+groupDocsByPeriod f pds es = 
+  let periods  = map (inPeriode f es) pds
+      periods' = periods `using` parList rdeepseq
+
+  in  trace ("\n" <> "-- | Group " <> show(length es) <> " docs by " <> show(length pds) <> " periods" <> "\n") 
+    $ fromList $ zip pds periods'
+  where
+    --------------------------------------
+    inPeriode :: Ord b => (t -> b) -> [t] -> (b, b) -> [t]
+    inPeriode f' h (start,end) =
+      fst $ partition (\d -> f' d >= start && f' d <= end) h
+    --------------------------------------   
+
+
+-- | To count the number of docs by unit of time
+docsToTimeScaleNb :: [Document] -> Map Date Double
+docsToTimeScaleNb docs = 
+    let docs' = fromListWith (+) $ map (\d -> (date d,1)) docs
+        time  = fromList $ map (\t -> (t,0)) $ toTimeScale (keys docs') 1
+    in  trace ("\n" <> "-- | Group " <> show(length docs) <> " docs by " <> show(length time) <> " unit of time" <> "\n") 
+      $ unionWith (+) time docs'
+
+
+initPhyloLevels :: Int -> PhyloPeriodId -> Map PhyloLevelId PhyloLevel
+initPhyloLevels lvlMax pId = 
+    fromList $ map (\lvl -> ((pId,lvl),PhyloLevel pId lvl empty)) [1..lvlMax]
+
+
+-- | To init the basic elements of a Phylo
+toPhyloBase :: [Document] -> TermList -> Config -> Phylo
+toPhyloBase docs lst conf = 
+    let foundations  = PhyloFoundations (Vector.fromList $ nub $ concat $ map text docs) lst
+        params = defaultPhyloParam { _phyloParam_config = conf }
+        periods = toPeriods (sort $ nub $ map date docs) (getTimePeriod $ timeUnit conf) (getTimeStep $ timeUnit conf)
+    in trace ("\n" <> "-- | Create PhyloBase out of " <> show(length docs) <> " docs \n") 
+       $ Phylo foundations
+               (docsToTimeScaleCooc docs (foundations ^. foundations_roots))
+               (docsToTimeScaleNb docs)
+               params
+               (fromList $ map (\prd -> (prd, PhyloPeriod prd (initPhyloLevels (phyloLevel conf) prd))) periods)

src/Gargantext/Viz/Phylo/PhyloTools.hs

+{-|
+Module      : Gargantext.Viz.Phylo.PhyloTools
+Description : Module dedicated to all the tools needed for making a Phylo
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+-}
+
+{-# LANGUAGE FlexibleContexts  #-}
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE RankNTypes        #-}
+{-# LANGUAGE ViewPatterns      #-}
+
+module Gargantext.Viz.Phylo.PhyloTools where
+
+import Data.Vector (Vector, elemIndex)
+import Data.List (sort, concat, null, union, (++), tails, sortOn, nub)
+import Data.Set (Set, size)
+import Data.Map (Map, elems, fromList, unionWith, keys, member, (!), filterWithKey)
+import Data.String (String)
+
+import Gargantext.Prelude
+import Gargantext.Viz.AdaptativePhylo
+
+import Debug.Trace (trace)
+import Control.Lens hiding (Level)
+
+import qualified Data.Vector as Vector
+import qualified Data.List as List
+import qualified Data.Set as Set
+
+--------------
+-- | Misc | --
+--------------
+
+
+countSup :: Double -> [Double] -> Int
+countSup s l = length $ filter (>s) l
+
+
+elemIndex' :: Eq a => a -> [a] -> Int
+elemIndex' e l = case (List.elemIndex e l) of
+    Nothing -> panic ("[ERR][Viz.Phylo.PhyloTools] element not in list")
+    Just i  -> i
+
+
+---------------------
+-- | Foundations | --
+---------------------
+
+
+-- | Is this Ngrams a Foundations Root ?
+isRoots :: Ngrams -> Vector Ngrams -> Bool
+isRoots n ns = Vector.elem n ns
+
+-- | To transform a list of nrams into a list of foundation's index
+ngramsToIdx :: [Ngrams] -> Vector Ngrams -> [Int]
+ngramsToIdx ns fdt = map (\n -> fromJust $ elemIndex n fdt) ns
+
+
+--------------
+-- | Time | --
+--------------
+
+-- | To transform a list of periods into a set of Dates
+periodsToYears :: [(Date,Date)] -> Set Date
+periodsToYears periods = (Set.fromList . sort . concat)
+                       $ map (\(d,d') -> [d..d']) periods
+
+
+findBounds :: [Date] -> (Date,Date)
+findBounds dates = 
+    let dates' = sort dates
+    in  (head' "findBounds" dates', last' "findBounds" dates')
+
+
+toPeriods :: [Date] -> Int -> Int -> [(Date,Date)]
+toPeriods dates p s = 
+    let (start,end) = findBounds dates
+    in map (\dates' -> (head' "toPeriods" dates', last' "toPeriods" dates')) 
+     $ chunkAlong p s [start .. end]
+
+
+-- | Get a regular & ascendante timeScale from a given list of dates
+toTimeScale :: [Date] -> Int -> [Date]
+toTimeScale dates step = 
+    let (start,end) = findBounds dates
+    in  [start, (start + step) .. end]
+
+
+getTimeStep :: TimeUnit -> Int
+getTimeStep time = case time of 
+    Year _ s _ -> s
+
+getTimePeriod :: TimeUnit -> Int
+getTimePeriod time = case time of 
+    Year p _ _ -> p  
+
+getTimeFrame :: TimeUnit -> Int
+getTimeFrame time = case time of 
+    Year _ _ f -> f
+
+-------------
+-- | Fis | --
+-------------
+
+
+-- | To find if l' is nested in l
+isNested :: Eq a => [a] -> [a] -> Bool
+isNested l l'
+  | null l'               = True
+  | length l' > length l  = False
+  | (union  l l') == l    = True
+  | otherwise             = False 
+
+
+-- | To filter Fis with small Support but by keeping non empty Periods
+keepFilled :: (Int -> [a] -> [a]) -> Int -> [a] -> [a]
+keepFilled f thr l = if (null $ f thr l) && (not $ null l)
+                     then keepFilled f (thr - 1) l
+                     else f thr l
+
+
+traceClique :: Map (Date, Date) [PhyloFis] -> String
+traceClique mFis = foldl (\msg cpt -> msg <> show (countSup cpt cliques) <> " (>" <> show (cpt) <> ") "  ) "" [1..6]
+    where
+        --------------------------------------
+        cliques :: [Double]
+        cliques = sort $ map (fromIntegral . size . _phyloFis_clique) $ concat $ elems mFis
+        -------------------------------------- 
+
+
+traceSupport :: Map (Date, Date) [PhyloFis] -> String
+traceSupport mFis = foldl (\msg cpt -> msg <> show (countSup cpt supports) <> " (>" <> show (cpt) <> ") "  ) "" [1..6]
+    where
+        --------------------------------------
+        supports :: [Double]
+        supports = sort $ map (fromIntegral . _phyloFis_support) $ concat $ elems mFis
+        -------------------------------------- 
+
+
+traceFis :: [Char] -> Map (Date, Date) [PhyloFis] -> Map (Date, Date) [PhyloFis]
+traceFis msg mFis = trace ( "\n" <> "-- | " <> msg <> " : " <> show (sum $ map length $ elems mFis) <> "\n"
+                         <> "Support : " <> (traceSupport mFis) <> "\n"
+                         <> "Clique : "  <> (traceClique mFis)  <> "\n" ) mFis
+
+
+-------------------------
+-- | Contextual unit | --
+-------------------------
+
+
+getFisSupport :: ContextualUnit -> Int
+getFisSupport unit = case unit of 
+    Fis s _ -> s
+    -- _       -> panic ("[ERR][Viz.Phylo.PhyloTools.getFisSupport] Only Fis has a support")
+
+getFisSize :: ContextualUnit -> Int
+getFisSize unit = case unit of 
+    Fis _ s -> s
+    -- _       -> panic ("[ERR][Viz.Phylo.PhyloTools.getFisSupport] Only Fis has a clique size")  
+
+
+--------------
+-- | Cooc | --
+--------------
+
+
+listToCombi' :: [a] -> [(a,a)]
+listToCombi' l = [(x,y) | (x:rest) <- tails l,  y <- rest]
+
+listToEqual' :: Eq a => [a] -> [(a,a)]
+listToEqual' l = [(x,y) | x <- l, y <- l, x == y]
+
+listToKeys :: Eq a =>  [a] -> [(a,a)]
+listToKeys lst = (listToCombi' lst) ++ (listToEqual' lst)
+
+listToMatrix :: [Int] -> Map (Int,Int) Double
+listToMatrix lst = fromList $ map (\k -> (k,1)) $ listToKeys $ sort lst
+
+sumCooc :: Cooc -> Cooc -> Cooc
+sumCooc cooc cooc' = unionWith (+) cooc cooc'
+
+getTrace :: Cooc -> Double 
+getTrace cooc = sum $ elems $ filterWithKey (\(k,k') _ -> k == k') cooc
+
+--------------------
+-- | PhyloGroup | --
+--------------------
+
+getGroupId :: PhyloGroup -> PhyloGroupId 
+getGroupId group = ((group ^. phylo_groupPeriod, group ^. phylo_groupLevel), group ^. phylo_groupIndex)
+
+---------------
+-- | Phylo | --
+---------------
+
+addPointers :: PhyloGroup -> Filiation -> PointerType -> [Pointer] -> PhyloGroup
+addPointers group fil pty pointers = 
+    case pty of 
+        TemporalPointer -> case fil of 
+                                ToChilds  -> group & phylo_groupPeriodChilds  %~ (++ pointers)
+                                ToParents -> group & phylo_groupPeriodParents %~ (++ pointers)
+        LevelPointer    -> case fil of 
+                                ToChilds  -> group & phylo_groupLevelChilds %~ (++ pointers)
+                                ToParents -> group & phylo_groupLevelParents %~ (++ pointers)
+
+
+getPeriodIds :: Phylo -> [(Date,Date)]
+getPeriodIds phylo = sortOn fst
+                   $ keys
+                   $ phylo ^. phylo_periods
+
+
+getConfig :: Phylo -> Config
+getConfig phylo = (phylo ^. phylo_param) ^. phyloParam_config
+
+
+getRoots :: Phylo -> Vector Ngrams
+getRoots phylo = (phylo ^. phylo_foundations) ^. foundations_roots
+
+
+getGroupsFromLevel :: Level -> Phylo -> [PhyloGroup]
+getGroupsFromLevel lvl phylo = 
+    elems $ view ( phylo_periods
+                 .  traverse
+                 . phylo_periodLevels
+                 .  traverse
+                 .  filtered (\phyloLvl -> phyloLvl ^. phylo_levelLevel == lvl)
+                 . phylo_levelGroups ) phylo
+
+
+updatePhyloGroups :: Level -> Map PhyloGroupId PhyloGroup -> Phylo -> Phylo
+updatePhyloGroups lvl m phylo = 
+    over ( phylo_periods
+         .  traverse
+         . phylo_periodLevels
+         .  traverse
+         .  filtered (\phyloLvl -> phyloLvl ^. phylo_levelLevel == lvl)
+         . phylo_levelGroups
+         .  traverse 
+         ) (\group -> 
+                let id = getGroupId group
+                in 
+                    if member id m 
+                    then m ! id
+                    else group ) phylo
+
+
+------------------
+-- | Pointers | --
+------------------
+
+
+pointersToLinks :: PhyloGroupId -> [Pointer] -> [Link]
+pointersToLinks id pointers = map (\p -> ((id,fst p),snd p)) pointers
+
+
+-------------------
+-- | Proximity | --
+-------------------
+
+getSensibility :: Proximity -> Double
+getSensibility proxi = case proxi of 
+    WeightedLogJaccard s _ _ -> s
+    Hamming -> undefined
+
+getThresholdInit :: Proximity -> Double
+getThresholdInit proxi = case proxi of 
+    WeightedLogJaccard _ t _ -> t
+    Hamming -> undefined  
+
+getThresholdStep :: Proximity -> Double
+getThresholdStep proxi = case proxi of 
+    WeightedLogJaccard _ _ s -> s
+    Hamming -> undefined  
+
+
+----------------
+-- | Branch | --
+----------------
+
+ngramsInBranches :: [[PhyloGroup]] -> [Int]
+ngramsInBranches branches = nub $ foldl (\acc g -> acc ++ (g ^. phylo_groupNgrams)) [] $ concat branches
\ No newline at end of file

src/Gargantext/Viz/Phylo/SynchronicClustering.hs

+{-|
+Module      : Gargantext.Viz.Phylo.SynchronicClustering
+Description : Module dedicated to the adaptative synchronic clustering of a Phylo.
+Copyright   : (c) CNRS, 2017-Present
+License     : AGPL + CECILL v3
+Maintainer  : team@gargantext.org
+Stability   : experimental
+Portability : POSIX
+-}
+
+{-# LANGUAGE NoImplicitPrelude #-}
+{-# LANGUAGE FlexibleContexts  #-}
+{-# LANGUAGE OverloadedStrings #-}
+{-# LANGUAGE MultiParamTypeClasses #-}
+
+module Gargantext.Viz.Phylo.SynchronicClustering where
+
+import Gargantext.Prelude
+-- import Gargantext.Viz.AdaptativePhylo
+-- import Gargantext.Viz.Phylo.PhyloTools
+
+import Data.List (foldl', (++), null, intersect, (\\), union, nub, concat)
+
+--------------------
+-- | Clustering | --
+--------------------
+
+
+relatedComponents :: Eq a => [[a]] -> [[a]]
+relatedComponents graphs = foldl' (\mem groups -> 
+  if (null mem)
+  then mem ++ [groups]
+  else 
+    let related = filter (\groups' -> (not . null) $ intersect groups groups') mem
+    in if (null related)
+       then mem ++ [groups]
+       else (mem \\ related) ++ [union groups (nub $ concat related)] ) [] graphs 
\ No newline at end of file

src/Gargantext/Viz/Phylo/Tools.hs

@@ -241,6 +241,11 @@ getIdxInRoots n p = case (elemIndex n (getFoundationsRoots p)) of
     Nothing  -> panic $ "[ERR][Viz.Phylo.Tools.getIdxInRoots] Ngrams not in foundationsRoots: " <> cs n
     Just idx -> idx
 
+getIdxInRoots' :: Ngrams -> Vector Ngrams -> Int
+getIdxInRoots' n root = case (elemIndex n root) of
+    Nothing  -> panic $ "[ERR][Viz.Phylo.Tools.getIdxInRoots] Ngrams not in foundationsRoots: " <> cs n
+    Just idx -> idx    
+
 getIdxInVector :: Ngrams -> Vector Ngrams -> Int
 getIdxInVector n ns = case (elemIndex n ns) of
   Nothing  -> panic $ "[ERR][Viz.Phylo.Tools.getIdxInVector] Ngrams not in foundationsRoots: " <> cs n

src/Gargantext/Viz/Phylo/View/Export.hs

@@ -124,7 +124,10 @@ toDotLabel lbl = StrLabel $ fromStrict lbl
 setPeakDotNode :: PhyloBranch -> Dot DotId
 setPeakDotNode pb = node (toBranchDotId $ pb ^. pb_id) 
                       ([FillColor [toWColor CornSilk], FontName "Arial", FontSize 40, Shape Egg, Style [SItem Bold []], Label (toDotLabel $ pb ^. pb_peak)]
-                       <> (setAttrFromMetrics $ pb ^. pb_metrics))
+                       <> (setAttrFromMetrics $ pb ^. pb_metrics)
+                       <> [ setAttr "nodeType" "peak"
+                          , setAttr "branchId" ((pack $ show (fst $ getBranchId pb)) <> (pack $ show (snd $ getBranchId pb)))
+                          ])
 
 
 -- | To set a Peak Edge
@@ -134,9 +137,9 @@ setPeakDotEdge bId nId = edge bId nId [Width 3, Color [toWColor Black], ArrowHea
 
 colorFromDynamics :: Double -> H.Attribute
 colorFromDynamics d 
-  | d == 0    = H.BGColor (toColor PaleGreen) 
-  | d == 1    = H.BGColor (toColor SkyBlue)
-  | d == 2    = H.BGColor (toColor LightPink) 
+  | d == 0    = H.BGColor (toColor LightCoral)
+  | d == 1    = H.BGColor (toColor Khaki)
+  | d == 2    = H.BGColor (toColor SkyBlue)
   | otherwise = H.Color (toColor Black)
 
 
@@ -186,14 +189,19 @@ setHtmlTable pn = H.Table H.HTable
 -- | To set a Node
 setDotNode :: PhyloNode -> Dot DotId
 setDotNode pn = node (toNodeDotId $ pn ^. pn_id)
-                     ([FontName "Arial", Shape Square, toLabel (setHtmlTable pn)])
+                     ([FontName "Arial", Shape Square, toLabel (setHtmlTable pn)]
+                      <> [ setAttr "nodeType" "group"
+                         , setAttr "from" (pack $ show (fst $ getNodePeriod pn))
+                         , setAttr "to"   (pack $ show (fst $ getNodePeriod pn))
+                         , setAttr "branchId" ((pack $ show (fst $ getNodeBranchId pn)) <> (pack $ show (snd $ getNodeBranchId pn))) 
+                         ])
 
 
 -- | To set an Edge
 setDotEdge :: PhyloEdge -> Dot DotId
 setDotEdge pe 
   | pe ^. pe_weight == 100 = edge (toNodeDotId $ pe ^. pe_source) (toNodeDotId $ pe ^. pe_target)  [Width 2, Color [toWColor Red]]
-  | otherwise = edge (toNodeDotId $ pe ^. pe_source) (toNodeDotId $ pe ^. pe_target)  [Width 2, Color [toWColor Black]]
+  | otherwise = edge (toNodeDotId $ pe ^. pe_source) (toNodeDotId $ pe ^. pe_target)  [Width 2, Color [toWColor Black], Constraint True]
 
 
 -- | To set a Period Edge
@@ -213,8 +221,9 @@ viewToDot pv = digraph ((Str . fromStrict) $ pv ^. pv_title)
                           <> [setAttr "description" $ fromStrict $ pv ^. pv_description]
                           <> [setAttr "filiation"   $ (pack . show) $ pv ^. pv_filiation]
                           <> (setAttrFromMetrics $ pv ^. pv_metrics)
-                          <> [FontSize 30, LabelLoc VTop, Splines SplineEdges, Overlap ScaleOverlaps,
-                              Ratio AutoRatio, Style [SItem Filled []],Color [toWColor White]])
+                          <> [FontSize 30, LabelLoc VTop, NodeSep 1, RankSep [1], Rank SameRank, Splines SplineEdges, Overlap ScaleOverlaps
+                             , Ratio FillRatio
+                             , Style [SItem Filled []],Color [toWColor White]])
 
                 -- set the peaks
 
@@ -235,7 +244,10 @@ viewToDot pv = digraph ((Str . fromStrict) $ pv ^. pv_title)
 
                             -- set the period label
                             
-                            node (toPeriodDotId prd) [Shape Square, FontSize 50, Label (toPeriodDotLabel prd)]
+                            node (toPeriodDotId prd) ([Shape Square, FontSize 50, Label (toPeriodDotLabel prd)] 
+                                                   <> [setAttr "nodeType" "period", 
+                                                       setAttr "from" (fromStrict $ T.pack $ (show $ fst prd)),
+                                                       setAttr "to"   (fromStrict $ T.pack $ (show $ snd prd))])
 
                             mapM setDotNode $ filterNodesByPeriod prd $ filterNodesByLevel (pv ^. pv_level) (pv ^.pv_nodes)
                           

src/Gargantext/Viz/Phylo/View/Metrics.hs

@@ -36,6 +36,12 @@ addBranchMetrics id lbl val v = over (pv_branches
                                            else b) v
 
 
+branchGroups :: PhyloView -> PhyloView
+branchGroups v = foldl (\v' (bId,nb) -> addBranchMetrics bId "nbGroups" nb v') v
+               $ map (\(bId,ns) -> (bId,fromIntegral $ length ns)) 
+               $ getNodesByBranches v
+
+
 -- | To get the age (in year) of all the branches of a PhyloView
 branchAge :: PhyloView -> PhyloView
 branchAge v = foldl (\v' b -> let bId = (fst . (head' "branchAge")) b
@@ -63,6 +69,7 @@ processMetrics :: [Metric] -> Phylo -> PhyloView -> PhyloView
 processMetrics ms _p v = foldl (\v' m -> case m of
                                         BranchAge -> branchAge v'
                                         BranchBirth -> branchBirth v'
+                                        BranchGroups -> branchGroups v'
                                        -- _         -> panic "[ERR][Viz.Phylo.Example.processMetrics] metric not found"
                                         ) v ms