[hal] implement search with language

cabal.project

@@ -84,7 +84,7 @@ source-repository-package
 source-repository-package
     type: git
     location: https://gitlab.iscpif.fr/gargantext/crawlers/hal.git
-    tag: ac1bec2628cd2d6c8357b4af2f853f450c7b6123
+    tag: f7b928dab9eb14ff1a497d5e092c2224c133979e
 
 source-repository-package
     type: git

gargantext.cabal

 cabal-version: 1.12
 
+-- This file has been generated from package.yaml by hpack version 0.35.1.
+--
+-- see: https://github.com/sol/hpack
+
 name:           gargantext
-version: 0.0.6.9.9.7.5
+version:        0.0.6.9.9.7.5
 synopsis:       Search, map, share
 description:    Please see README.md
 category:       Data
@@ -47,7 +51,6 @@ library
       Gargantext.API.Node
       Gargantext.API.Node.Corpus.New
       Gargantext.API.Node.Corpus.Types
-      Gargantext.API.Node.Corpus.Update
       Gargantext.API.Node.File
       Gargantext.API.Node.Share
       Gargantext.API.Prelude
@@ -59,13 +62,12 @@ library
       Gargantext.Core.Text.Context
       Gargantext.Core.Text.Corpus.API
       Gargantext.Core.Text.Corpus.API.Arxiv
-      Gargantext.Core.Text.Corpus.API.OpenAlex
       Gargantext.Core.Text.Corpus.API.Pubmed
+      Gargantext.Core.Text.Corpus.API.OpenAlex
       Gargantext.Core.Text.Corpus.Query
       Gargantext.Core.Text.Corpus.Parsers
       Gargantext.Core.Text.Corpus.Parsers.CSV
       Gargantext.Core.Text.Corpus.Parsers.Date.Parsec
-      Gargantext.Core.Text.Corpus.Parsers.JSON
       Gargantext.Core.Text.List.Formats.CSV
       Gargantext.Core.Text.Metrics
       Gargantext.Core.Text.Metrics.CharByChar
@@ -111,7 +113,6 @@ library
       Gargantext.Database.Prelude
       Gargantext.Database.Query.Table.NgramsPostag
       Gargantext.Database.Query.Table.Node
-      Gargantext.Database.Query.Table.Node.Error
       Gargantext.Database.Query.Table.Node.UpdateOpaleye
       Gargantext.Database.Query.Table.User
       Gargantext.Database.Schema.Ngrams
@@ -160,6 +161,7 @@ library
       Gargantext.API.Node.Corpus.New.File
       Gargantext.API.Node.Corpus.New.Types
       Gargantext.API.Node.Corpus.Searx
+      Gargantext.API.Node.Corpus.Update
       Gargantext.API.Node.Document.Export
       Gargantext.API.Node.Document.Export.Types
       Gargantext.API.Node.DocumentsFromWriteNodes
@@ -205,9 +207,11 @@ library
       Gargantext.Core.Text.Corpus.Parsers.GrandDebat
       Gargantext.Core.Text.Corpus.Parsers.Iramuteq
       Gargantext.Core.Text.Corpus.Parsers.Isidore
+      Gargantext.Core.Text.Corpus.Parsers.JSON
       Gargantext.Core.Text.Corpus.Parsers.Json2Csv
       Gargantext.Core.Text.Corpus.Parsers.RIS
       Gargantext.Core.Text.Corpus.Parsers.RIS.Presse
+      Gargantext.Core.Text.Corpus.Parsers.Telegram
       Gargantext.Core.Text.Corpus.Parsers.Wikidata
       Gargantext.Core.Text.Corpus.Parsers.Wikidata.Crawler
       Gargantext.Core.Text.Corpus.Parsers.Wikimedia
@@ -313,6 +317,7 @@ library
       Gargantext.Database.Query.Table.Node.Contact
       Gargantext.Database.Query.Table.Node.Document.Add
       Gargantext.Database.Query.Table.Node.Document.Insert
+      Gargantext.Database.Query.Table.Node.Error
       Gargantext.Database.Query.Table.Node.Select
       Gargantext.Database.Query.Table.Node.Update
       Gargantext.Database.Query.Table.Node.User
@@ -356,6 +361,7 @@ library
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -440,6 +446,7 @@ library
     , ihaskell
     , ini
     , insert-ordered-containers
+    , iso639
     , jose
     , json-stream
     , lens
@@ -559,6 +566,7 @@ executable gargantext-admin
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -588,6 +596,7 @@ executable gargantext-cbor2json
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -621,6 +630,7 @@ executable gargantext-cli
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -661,6 +671,7 @@ executable gargantext-import
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -691,6 +702,7 @@ executable gargantext-init
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -721,6 +733,7 @@ executable gargantext-invitations
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -750,6 +763,7 @@ executable gargantext-phylo
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -793,6 +807,7 @@ executable gargantext-server
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -830,6 +845,7 @@ executable gargantext-upgrade
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -885,6 +901,7 @@ test-suite garg-test
       FlexibleInstances
       GADTs
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NamedFieldPuns
       NoImplicitPrelude
@@ -897,6 +914,7 @@ test-suite garg-test
       FlexibleContexts
       FlexibleInstances
       GeneralizedNewtypeDeriving
+      ImportQualifiedPost
       MultiParamTypeClasses
       NoImplicitPrelude
       OverloadedStrings
@@ -917,23 +935,15 @@ test-suite garg-test
     , gargantext
     , gargantext-prelude
     , hspec
-    , hspec-expectations >= 0.8.3
     , http-client
     , http-client-tls
-    , monad-control
     , mtl
-    , lens
     , parsec
     , patches-class
     , patches-map
-    , postgres-options
-    , postgresql-simple
     , quickcheck-instances
     , raw-strings-qq
-    , recover-rtti
-    , resource-pool
     , servant-job
-    , shelly
     , stm
     , tasty
     , tasty-hspec
@@ -941,7 +951,6 @@ test-suite garg-test
     , tasty-quickcheck
     , text
     , time
-    , tmp-postgres
     , unordered-containers
     , validity
   default-language: Haskell2010

package.yaml

-name: gargantext
-
-#         +------------ Layer 3
-#         | +--+------- Layer 2
-#         | | +-------- Layer 1  : New versions with API changes
-#         | | | +-----  Layers * : New versions with API additions
-#         | | | | +---  Layers * : New versions without API breaking changes
-#         | | | | |
-version: '0.0.6.9.9.7.5'
-synopsis: Search, map, share
-description: Please see README.md
-category: Data
-author: Gargantext Team
-maintainer: team@gargantext.org
-copyright:
-    - ! 'Copyright: (c) 2017-Present: see git logs and README'
-license: AGPL-3
-homepage: https://gargantext.org
-ghc-options:
-  - -Wall
-dependencies:
-  - extra
-  - text
-default-extensions:
-  - DataKinds
-  - DeriveGeneric
-  - FlexibleContexts
-  - FlexibleInstances
-  - GADTs
-  - GeneralizedNewtypeDeriving
-  - MultiParamTypeClasses
-  - NamedFieldPuns
-  - NoImplicitPrelude
-  - OverloadedStrings
-  - RankNTypes
-  - RecordWildCards
-  - StrictData
-data-files:
-  - ekg-assets/index.html
-  - ekg-assets/monitor.js
-  - ekg-assets/monitor.css
-  - ekg-assets/jquery.flot.min.js
-  - ekg-assets/jquery-1.6.4.min.js
-  - ekg-assets/bootstrap-1.4.0.min.css
-  - ekg-assets/chart_line_add.png
-  - ekg-assets/cross.png
-  - test-data/ngrams/GarganText_NgramsTerms-QuantumComputing.json
-  - test-data/phylo/bpa_phylo_test.json
-  - test-data/phylo/open_science.json
-  - .clippy.dhall
-library:
-  source-dirs: src
-  ghc-options:
-  - -Wincomplete-uni-patterns
-  - -Wincomplete-record-updates
-  - -Wmissing-signatures
-  - -Wunused-binds
-  - -Wunused-imports
-  - -Werror
-  - -freduction-depth=300
-  - -fplugin=Clippy
-  exposed-modules:
-  - Gargantext
-  - Gargantext.API
-  - Gargantext.API.Admin.Auth.Types
-  - Gargantext.API.Admin.EnvTypes
-  - Gargantext.API.Admin.Settings
-  - Gargantext.API.Admin.Orchestrator.Types
-  - Gargantext.API.Admin.Types
-  - Gargantext.API.Dev
-  - Gargantext.API.HashedResponse
-  - Gargantext.API.Ngrams
-  - Gargantext.API.Ngrams.Prelude
-  - Gargantext.API.Ngrams.Tools
-  - Gargantext.API.Ngrams.Types
-  - Gargantext.API.Node
-  - Gargantext.API.Node.Corpus.New
-  - Gargantext.API.Node.Corpus.Types
-  - Gargantext.API.Node.File
-  - Gargantext.API.Node.Share
-  - Gargantext.API.Prelude
-  - Gargantext.Core
-  - Gargantext.Core.NLP
-  - Gargantext.Core.Methods.Similarities
-  - Gargantext.Core.NodeStory
-  - Gargantext.Core.Text
-  - Gargantext.Core.Text.Context
-  - Gargantext.Core.Text.Corpus.API
-  - Gargantext.Core.Text.Corpus.API.Arxiv
-  - Gargantext.Core.Text.Corpus.API.Pubmed
-  - Gargantext.Core.Text.Corpus.API.OpenAlex
-  - Gargantext.Core.Text.Corpus.Query
-  - Gargantext.Core.Text.Corpus.Parsers
-  - Gargantext.Core.Text.Corpus.Parsers.CSV
-  - Gargantext.Core.Text.Corpus.Parsers.Date.Parsec
-  - Gargantext.Core.Text.List.Formats.CSV
-  - Gargantext.Core.Text.Metrics
-  - Gargantext.Core.Text.Metrics.CharByChar
-  - Gargantext.Core.Text.Metrics.Count
-  - Gargantext.Core.Text.Metrics.TFICF
-  - Gargantext.Core.Text.Prepare
-  - Gargantext.Core.Text.Search
-  - Gargantext.Core.Text.Terms
-  - Gargantext.Core.Text.Terms.Eleve
-  - Gargantext.Core.Text.Terms.Mono
-  - Gargantext.Core.Text.Terms.Multi
-  - Gargantext.Core.Text.Terms.Multi.Lang.En
-  - Gargantext.Core.Text.Terms.Multi.Lang.Fr
-  - Gargantext.Core.Text.Terms.Multi.RAKE
-  - Gargantext.Core.Text.Terms.WithList
-  - Gargantext.Core.Types
-  - Gargantext.Core.Types.Individu
-  - Gargantext.Core.Types.Main
-  - Gargantext.Core.Types.Query
-  - Gargantext.Core.Types.Phylo
-  - Gargantext.Core.Utils
-  - Gargantext.Core.Utils.Prefix
-  - Gargantext.Core.Viz.Graph
-  - Gargantext.Core.Viz.Graph.Index
-  - Gargantext.Core.Viz.Graph.Tools
-  - Gargantext.Core.Viz.Graph.Tools.IGraph
-  - Gargantext.Core.Viz.Graph.Types
-  - Gargantext.Core.Viz.Phylo
-  - Gargantext.Core.Viz.Phylo.API
-  - Gargantext.Core.Viz.Phylo.API.Tools
-  - Gargantext.Core.Viz.Phylo.PhyloExport
-  - Gargantext.Core.Viz.Phylo.PhyloMaker
-  - Gargantext.Core.Viz.Phylo.PhyloTools
-  - Gargantext.Core.Viz.Phylo.SynchronicClustering
-  - Gargantext.Core.Viz.Types
-  - Gargantext.Database.Action.Flow
-  - Gargantext.Database.Action.Flow.Types
-  - Gargantext.Database.Action.User.New
-  - Gargantext.Database.Admin.Config
-  - Gargantext.Database.Admin.Trigger.Init
-  - Gargantext.Database.Admin.Types.Hyperdata
-  - Gargantext.Database.Admin.Types.Node
-  - Gargantext.Database.Prelude
-  - Gargantext.Database.Query.Table.NgramsPostag
-  - Gargantext.Database.Query.Table.Node
-  - Gargantext.Database.Query.Table.Node.UpdateOpaleye
-  - Gargantext.Database.Query.Table.User
-  - Gargantext.Database.Schema.Ngrams
-  - Gargantext.Defaults
-  - Gargantext.Utils.Jobs
-  - Gargantext.Utils.Jobs.Internal
-  - Gargantext.Utils.Jobs.Map
-  - Gargantext.Utils.Jobs.Monad
-  - Gargantext.Utils.Jobs.Queue
-  - Gargantext.Utils.Jobs.Settings
-  - Gargantext.Utils.Jobs.State
-  - Gargantext.Utils.SpacyNLP
-  - Gargantext.Utils.Tuple
-  dependencies:
-  - HSvm
-  - KMP
-  - MissingH
-  - MonadRandom
-  - QuickCheck
-  - Unique
-  - accelerate
-  - accelerate-arithmetic
-  - accelerate-llvm-native
-  - accelerate-utility
-  - aeson
-  - aeson-pretty
-  - array
-  - async
-  - attoparsec
-  - auto-update
-  - base >=4.7 && <5
-  - base16-bytestring
-  - base64-bytestring
-  - blaze-html
-  - blaze-markup
-  - blaze-svg
-  - boolexpr
-  - bytestring
-  - case-insensitive
-  - cassava
-  - cereal     # (IGraph)
-  - cborg
-  - ghc-clippy-plugin
-  - conduit
-  - conduit-extra
-  - containers
-  - contravariant
-  - crawlerArxiv
-  - crawlerHAL
-  - crawlerISTEX
-  - crawlerIsidore
-  - crawlerPubMed
-  - cron
-  - cryptohash
-  - data-time-segment
-  - deepseq
-  - directory
-  - duckling
-  - ekg-core
-  - ekg-json
-  - exceptions
-  - fast-logger
-  - fclabels
-  - fgl
-  - filelock
-  - filepath
-  - formatting
-  - full-text-search
-  - fullstop
-  - gargantext-prelude
-  - gargantext-graph >= 0.1.0.0
-  - graphviz
-  - hashable
-  - haskell-igraph
-  - hlcm
-  - hsinfomap
-  - hsparql
-  - hstatistics
-  - http-api-data
-  - http-client
-  - http-client-tls
-  - http-conduit
-  - http-media
-  - http-types
-  - hxt
-  - ihaskell
-  - ini
-  - insert-ordered-containers
-  - jose
-  - json-stream
-  - lens
-  - lifted-base
-  - listsafe
-  - located-base
-  - logging-effect
-  - matrix
-  - monad-control
-  - monad-logger
-  - morpheus-graphql
-  - morpheus-graphql-app
-  - morpheus-graphql-core
-  - morpheus-graphql-subscriptions
-  - mtl
-  - natural-transformation
-  - network-uri
-  - opaleye
-  - openalex
-  - pandoc
-  - parallel
-  - parsec
-  - patches-class
-  - patches-map
-  - path
-  - path-io
-  - postgresql-simple
-  - pretty-simple
-  - probability
-  - process
-  - product-profunctors
-  - profunctors
-  - protolude
-  - pureMD5
-  - quickcheck-instances
-  - rake
-  - random
-  - rdf4h
-  - replace-attoparsec
-  - regex-compat
-  - regex-tdfa
-  - resource-pool
-  - resourcet
-  - safe
-  - scientific
-  - semigroups
-  - serialise
-  - servant
-  - servant-auth
-  - servant-auth-client
-  - servant-auth-server >= 0.4.4.0
-  - servant-auth-swagger
-  - servant-blaze
-  - servant-cassava
-  - servant-client
-  - servant-ekg
-  - servant-flatten
-  - servant-job
-  - servant-mock
-  - servant-multipart
-  - servant-server
-  - servant-swagger
-  - servant-swagger-ui
-  - servant-xml
-  - simple-reflect
-  - singletons # (IGraph)
-  - split
-  - stemmer
-  - stm
-  - swagger2
-  - taggy-lens
-  - tagsoup
-  - template-haskell
-  - temporary
-  - text-conversions
-  - text-metrics
-  - time
-  - time-locale-compat
-  - timezone-series
-  - transformers
-  - transformers-base
-  - tuple
-  - unordered-containers
-  - utf8-string
-  - uuid
-  - uri-encode
-  - validity
-  - vector
-  - wai
-  - wai-app-static
-  - wai-cors
-  - wai-extra
-  - wai-websockets
-  - warp
-  - wikiparsec
-  - websockets
-  - wreq
-  - xml-conduit
-  - xml-types
-  - xmlbf
-  - yaml
-  - zip
-  - zlib
-
-executables:
-  gargantext-server:
-    main: Main.hs
-    source-dirs: bin/gargantext-server
-    ghc-options:
-    - -O2
-    - -Wcompat
-    - -Wmissing-signatures
-    - -rtsopts
-    - -threaded
-    - -with-rtsopts=-N
-    - -with-rtsopts=-T
-    - -fprof-auto
-    dependencies:
-      - base
-      - cassava
-      - containers
-      - full-text-search
-      - gargantext
-      - gargantext-prelude
-      - ini
-      - optparse-generic
-      - postgresql-simple
-      - unordered-containers
-      - vector
-
-  gargantext-cli:
-    main: Main.hs
-    source-dirs: bin/gargantext-cli
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - aeson
-      - async
-      - base
-      - bytestring
-      - containers
-      - gargantext
-      - gargantext-prelude
-      - vector
-      - cassava
-      - ini
-      - optparse-generic
-      - split
-      - unordered-containers
-      - full-text-search
-
-  gargantext-phylo:
-    main: Main.hs
-    source-dirs: bin/gargantext-phylo
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - aeson
-      - async
-      - base
-      - bytestring
-      - containers
-      - directory
-      - gargantext
-      - gargantext-prelude
-      - vector
-      - parallel
-      - cassava
-      - ini
-      - optparse-generic
-      - split
-      - unordered-containers
-      - cryptohash
-      - time
-
-  gargantext-import:
-    main: Main.hs
-    source-dirs: bin/gargantext-import
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-      - servant-server
-
-  gargantext-init:
-    main: Main.hs
-    source-dirs: bin/gargantext-init
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-      - cron
-
-  gargantext-invitations:
-    main: Main.hs
-    source-dirs: bin/gargantext-invitations
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-
-
-  gargantext-upgrade:
-    main: Main.hs
-    source-dirs: bin/gargantext-upgrade
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-      - postgresql-simple
-      - cron
-
-  gargantext-admin:
-    main: Main.hs
-    source-dirs: bin/gargantext-admin
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-
-
-  gargantext-cbor2json:
-    main: Main.hs
-    source-dirs: bin/gargantext-cbor2json
-    ghc-options:
-    - -threaded
-    - -rtsopts
-    - -with-rtsopts=-N
-    - -O2
-    - -Wmissing-signatures
-    dependencies:
-      - gargantext
-      - gargantext-prelude
-      - base
-      - bytestring
-      - aeson
-      - serialise
-
-tests:
-   garg-test:
-     main: Main.hs
-     source-dirs: test
-     default-extensions:
-     - DataKinds
-     - DeriveGeneric
-     - FlexibleContexts
-     - FlexibleInstances
-     - GeneralizedNewtypeDeriving
-     - MultiParamTypeClasses
-     - NoImplicitPrelude
-     - OverloadedStrings
-     - RankNTypes
-     ghc-options:
-     - -threaded
-     - -rtsopts
-     - -with-rtsopts=-N
-     dependencies:
-     - QuickCheck
-     - aeson
-     - async
-     - base
-     - boolexpr
-     - bytestring
-     - conduit
-     - containers
-     - crawlerArxiv
-     - duckling
-     - gargantext
-     - gargantext-prelude
-     - hspec
-     - http-client
-     - http-client-tls
-     - mtl
-     - parsec
-     - patches-class
-     - patches-map
-     - duckling
-     - quickcheck-instances
-     - raw-strings-qq
-     - servant-job
-     - stm
-     - tasty
-     - tasty-hspec
-     - tasty-hunit
-     - tasty-quickcheck
-     - text
-     - time
-     - unordered-containers
-     - validity
-#  garg-doctest:
-#    main: Main.hs
-#    source-dirs: src-doctest
-#    ghc-options:
-#    - -O2
-#    - -Wcompat
-#    - -Wmissing-signatures
-#    - -rtsopts
-#    - -threaded
-#    - -with-rtsopts=-N
-#    dependencies:
-#    - doctest
-#    - Glob
-#    - QuickCheck
-#    - base
-#    - gargantext
-#    default-extensions:
-#    - DataKinds
-#    - DeriveGeneric
-#    - FlexibleContexts
-#    - FlexibleInstances
-#    - GeneralizedNewtypeDeriving
-#    - MultiParamTypeClasses
-#    - NoImplicitPrelude
-#    - OverloadedStrings
-#    - RankNTypes
-#

src/Gargantext/Core/Text/Corpus/API/Hal.hs

@@ -14,47 +14,67 @@ module Gargantext.Core.Text.Corpus.API.Hal
 
 import Conduit
 import Data.Either
+import Data.LanguageCodes qualified as ISO639
+import Data.Map.Strict qualified as Map
 import Data.Maybe
 import Data.Text (Text, pack, intercalate)
-import Servant.Client (ClientError)
-
 import Gargantext.Core (Lang(..))
+import Gargantext.Core.Text.Corpus.Parsers.Date qualified as Date
 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
-import qualified Gargantext.Defaults as Defaults
+import Gargantext.Defaults qualified as Defaults
 import Gargantext.Prelude
-import qualified Gargantext.Core.Text.Corpus.Parsers.Date as Date
-import qualified HAL            as HAL
-import qualified HAL.Client     as HAL
-import qualified HAL.Doc.Corpus as HAL
+import HAL qualified as HAL
+import HAL.Client qualified as HAL
+import HAL.Doc.Corpus qualified as HAL
+import Servant.Client (ClientError)
+
+toLang :: Lang -> Maybe ISO639.ISO639_1
+toLang DE = Just ISO639.DE
+toLang EL = Just ISO639.EL
+toLang EN = Just ISO639.EN
+toLang ES = Just ISO639.ES
+toLang FR = Just ISO639.FR
+toLang IT = Just ISO639.IT
+toLang PL = Just ISO639.PL
+toLang PT = Just ISO639.PT
+toLang RU = Just ISO639.RU
+toLang UK = Just ISO639.UK
+toLang ZH = Just ISO639.ZH
+toLang All = Nothing
 
 get :: Lang -> Text -> Maybe Int -> IO [HyperdataDocument]
 get la q ml = do
-  eDocs <- HAL.getMetadataWith q (Just 0) (fromIntegral <$> ml)
+  eDocs <- HAL.getMetadataWith [q] (Just 0) (fromIntegral <$> ml) (toLang la)
   either (panic . pack . show) (\d -> mapM (toDoc' la) $ HAL._docs d) eDocs
 
 getC :: Lang -> Text -> Maybe Int -> IO (Either ClientError (Maybe Integer, ConduitT () HyperdataDocument IO ()))
 getC la q ml = do
-  eRes <- HAL.getMetadataWithC q (Just 0) (fromIntegral <$> ml)
+  eRes <- HAL.getMetadataWithC [q] (Just 0) (fromIntegral <$> ml) (toLang la)
   pure $ (\(len, docsC) -> (len, docsC .| mapMC (toDoc' la))) <$> eRes
 --  case eRes of
 --    Left err -> panic $ pack $ show err
 --    Right (len, docsC) -> pure (len, docsC .| mapMC (toDoc' la))
 
 toDoc' :: Lang -> HAL.Corpus -> IO HyperdataDocument
-toDoc' la (HAL.Corpus i t ab d s aus affs struct_id) = do
+toDoc' la h@(HAL.Corpus { .. }) = do
+  -- printDebug "[toDoc corpus] h" h
   (utctime, (pub_year, pub_month, pub_day)) <-
-        Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) Just d)
+        Date.dateSplit la (maybe (Just $ pack $ show Defaults.year) Just _corpus_date)
+  let abstractDefault = intercalate " " _corpus_abstract
+  let abstract = case toLang la of
+        Nothing -> abstractDefault
+        Just l  -> fromMaybe abstractDefault (intercalate " " <$> Map.lookup l _corpus_abstract_lang_map)
   pure HyperdataDocument { _hd_bdd = Just "Hal"
-                         , _hd_doi = Just $ pack $ show i
+                         , _hd_doi = Just $ pack $ show _corpus_docid
                          , _hd_url = Nothing
                          , _hd_uniqId = Nothing
                          , _hd_uniqIdBdd = Nothing
                          , _hd_page = Nothing
-                         , _hd_title = Just $ intercalate " " t
-                         , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" aus
-                         , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" $ affs <> map (cs . show) struct_id
-                         , _hd_source = Just $ maybe "Nothing" identity s
-                         , _hd_abstract = Just $ intercalate " " ab
+                         , _hd_title = Just $ intercalate " " _corpus_title
+                         , _hd_authors = Just $ foldl (\x y -> x <> ", " <> y) "" _corpus_authors_names
+                         , _hd_institutes = Just $ foldl (\x y -> x <> ", " <> y) "" $ _corpus_authors_affiliations <> map (cs . show) _corpus_struct_id
+                         , _hd_source = Just $ maybe "Nothing" identity _corpus_source
+                         , _hd_abstract = Just abstract
                          , _hd_publication_date = fmap (pack . show) utctime
                          , _hd_publication_year = pub_year
                          , _hd_publication_month = pub_month

src/Gargantext/Core/Text/Terms/Multi.hs

@@ -59,6 +59,7 @@ tokenTag2terms (TokenTag ws t _ _) =  Terms ws t
 
 tokenTags :: NLPServerConfig -> Lang -> Text -> IO [[TokenTag]]
 tokenTags (NLPServerConfig { server = CoreNLP, url }) EN txt = tokenTagsWith EN txt $ corenlp url
+tokenTags (NLPServerConfig { server = CoreNLP, url }) FR txt = tokenTagsWith FR txt $ corenlp url
 tokenTags (NLPServerConfig { server = Spacy, url }) l txt = do
   -- printDebug "NLP Debug" txt
   tokenTagsWith l txt $ SpacyNLP.nlp url
@@ -95,4 +96,3 @@ cleanTextForNLP = unifySpaces . removeDigitsWith "-" . removeUrls
 
     removeUrls          = removeUrlsWith "http" . removeUrlsWith "www"
     removeUrlsWith w    = remove (DAT.string w *> many (DAT.notChar ' ') <* many DAT.space)
-