ccomb · ccomb · Jun 24, 2026 · Jun 23, 2026 · Jun 23, 2026 · Jun 23, 2026
diff --git a/.gitignore b/.gitignore
@@ -1,6 +1,5 @@
 dist-newstyle
 dist/
-weeder.toml
 .hie
 *.o
 *\.bin

diff --git a/.hlint.yaml b/.hlint.yaml
@@ -0,0 +1,13 @@
+# HLint configuration for the VoLCA engine.
+#
+# Its presence activates the hlint step in CI, which otherwise skips when no config is
+# found. Run locally with: hlint src/
+#
+# The hygiene pass either *fixes* a hint or *baselines* it here with a one-line reason.
+# All default hints on src/ are fixed; the single baseline below is a deliberate
+# exception where the suggested rewrite conflicts with a stronger code-safety rule.
+
+# `cells !! 0` is guarded by `length cells > colIdx + 2`, so it never fails — but the
+# suggested `head cells` is a partial function we ban project-wide. Keep the guarded
+# indexing rather than reintroduce `head`.
+- ignore: {name: "Use head", within: Method.ParserCSV}
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,6 +3,15 @@
 ## [Unreleased]
 
 ### Added
+- A loaded database can be exported to any of the five supported formats —
+  SimaPro CSV, EcoSpold 1, EcoSpold 2, ILCD, and Brightway Excel — from both the
+  API and the CLI, so a database can be moved between tools or re-saved after edits.
+- A loaded database can be edited in place: copy it under a new name, delete a
+  filtered selection of activities, or relink it against a dependency through a
+  name-to-name alias map.
+- Activity records now carry separate `activity_name` and `product_name` fields,
+  instead of the old `name`/`product` that blurred an activity and its reference
+  product.
 - A partial EcoSpold2 import (a handful of `.spold` files cut from a full
   database) now becomes analyzable by loading its matching ecoinvent background
   as a dependency: each input is linked to the exact background activity it
@@ -19,6 +28,27 @@
   connect time to confirm they speak this engine's JSON format, so a version
   mismatch fails with a clear message instead of a confusing decode error.
 
+### Changed
+- Far broader EF 3.1 impact coverage: flows are matched by CAS number across
+  naming schemes, a substance registry bridges nomenclatures, sub-compartments
+  fall back sensibly, energy-carrier flows are characterized from their energy
+  content, land use is regionalized with per-country factors, and a large synonym
+  set links ecoinvent and Agribalyse flows to the method. Many products that
+  previously scored short are now characterized.
+- Large database uploads stream as raw bytes instead of base64-encoded JSON,
+  cutting memory use and time on big files.
+- A supplier substitution applies across every consumer of the substituted
+  product, not only the process you queried.
+
+### Fixed
+- ecoinvent waste-treatment activities import and score correctly: the reference
+  flow stays in the technosphere (these activities are no longer dropped) and
+  cross-database waste keeps the correct sign.
+- EcoSpold packages whose datasets live in a sub-directory now load.
+- Requesting a well-formed but non-existent process returns a clear "activity not
+  found" instead of a confusing error.
+- A missing reference CSV is reported as an error instead of crashing the load.
+
 ## [0.7.0] - 2026-05-29
 
 A month of engine work: a third flow kind for waste, regionalized impact

diff --git a/bench/Bench/Helpers.hs b/bench/Bench/Helpers.hs
@@ -14,7 +14,7 @@ import Data.Text (Text)
 
 import Database (buildDatabaseWithMatrices)
 import qualified Database.Loader as Loader
-import Types (Database, sdbActivities, sdbBioFlows, sdbTechFlows, sdbUnits)
+import Types (Database, sdbActivities, sdbBioFlows, sdbTechFlows, sdbUnits, sdbWasteFlows)
 import qualified UnitConversion as UC
 
 {- | Parse a fixture path and build the indexed 'Database' with matrices.
@@ -36,6 +36,7 @@ loadFullDatabase path = do
                     (sdbActivities sdb)
                     (sdbTechFlows sdb)
                     (sdbBioFlows sdb)
+                    (sdbWasteFlows sdb)
                     (sdbUnits sdb)
             case built of
                 Left err -> pure (Left ("buildDatabaseWithMatrices failed: " <> err))

diff --git a/bench/Bench/Lcia.hs b/bench/Bench/Lcia.hs
@@ -49,7 +49,6 @@ import Types (
     Indexes (..),
     ProductIndex (..),
     Unit (..),
-    emptyCrossDBLinkingStats,
  )
 import qualified UnitConversion as UC
 
@@ -145,6 +144,7 @@ emptyDatabase =
         , dbActivities = V.empty
         , dbTechFlows = M.empty
         , dbBioFlows = M.empty
+        , dbWasteFlows = M.empty
         , dbUnits = M.empty
         , dbIndexes = Indexes M.empty M.empty M.empty M.empty
         , dbTechnosphereTriples = U.empty
@@ -155,7 +155,7 @@ emptyDatabase =
         , dbBiosphereCount = 0
         , dbCrossDBLinks = []
         , dbDependsOn = []
-        , dbLinkingStats = emptyCrossDBLinkingStats
+        , dbLinkingStats = mempty
         , dbSynonymDB = Nothing
         , dbFlowsByName = M.empty
         , dbFlowsByCAS = M.empty
@@ -197,7 +197,7 @@ buildFixture =
             ]
 
         methods = [mkMethod m [] | m <- [1 .. nMethods]]
-        rawTables = [buildMethodTables M.empty (mappingsFor s) | s <- [1 .. nMethods]]
+        rawTables = [buildMethodTables M.empty M.empty (mappingsFor s) | s <- [1 .. nMethods]]
         filledTables = map (fillBroadcastVector unitCfg unitDB flowDB) rawTables
         setTables = buildMethodSetTables (zip methods filledTables)
 
@@ -338,7 +338,7 @@ registerReal = do
                             mappings <- mapMethodToFlows Builtin.defaultMappers db method
                             let unitDB = dbUnits db
                                 flowDB = dbBioFlows db
-                                tables0 = buildMethodTables M.empty mappings
+                                tables0 = buildMethodTables M.empty M.empty mappings
                                 !tables = fillBroadcastVector UC.defaultUnitConfig unitDB flowDB tables0
                                 !nCFs = length (methodFactors method)
                             putStrLn "[bench] lcia.real.score_method: computing inventory for first product..."

diff --git a/bench/Bench/Loader.hs b/bench/Bench/Loader.hs
@@ -27,7 +27,7 @@ import qualified Data.Text as T
 import qualified Database.CrossLinking as CL
 import qualified Database.Loader as Loader
 import qualified SynonymDB.Types as Syn
-import Types (cdlTotalInputs, sdbActivities)
+import Types (GeographyPolicy (..), cdlTotalInputs, sdbActivities)
 import qualified UnitConversion as UC
 
 import Bench.Json (BenchSpec (..), UnitOfWork (..))
@@ -132,6 +132,7 @@ registerCrossDbLinking = do
                             Syn.emptySynonymDB
                             UC.defaultUnitConfig
                             M.empty
+                            GeoGlobal
                             fgPath
                     case probeRes of
                         Left err -> do
@@ -179,6 +180,7 @@ crossLinkBench bgIndex fgPath = nfIO $ do
             Syn.emptySynonymDB
             UC.defaultUnitConfig
             M.empty
+            GeoGlobal
             fgPath
     case r of
         Left err -> evaluate (T.length err)

diff --git a/bench/Bench/Parsers.hs b/bench/Bench/Parsers.hs
@@ -177,7 +177,7 @@ registerSimaPro = do
         Nothing -> pure []
         Just path -> do
             -- Parse once to learn N_actual; the bench measurement re-parses fresh.
-            (acts, _, _, _) <- SP.parseSimaProCSV UC.defaultUnitConfig path
+            (acts, _, _, _, _) <- SP.parseSimaProCSV UC.defaultUnitConfig path
             let !n = length acts
             pure
                 [ BenchSpec
@@ -192,7 +192,7 @@ registerSimaPro = do
                     , bsMetric = "seconds"
                     , bsFixture = J.Fixture{J.fSource = F.fixtureSourceLabel F.Agribalyse, J.fSlice = "whole file"}
                     , bsAction = nfIO $ do
-                        (acts', _, _, _) <- SP.parseSimaProCSV UC.defaultUnitConfig path
+                        (acts', _, _, _, _) <- SP.parseSimaProCSV UC.defaultUnitConfig path
                         evaluate (length acts')
                     }
                 ]

diff --git a/src/API/Auth.hs b/src/API/Auth.hs
@@ -24,12 +24,9 @@ authMiddleware expectedPassword app req respond =
         isLoginEndpoint = requestMethod req == "POST" && path == "/api/v1/auth"
      in -- Only protect /api/ routes; static files and SPA index are public
         -- (so the browser can load the login page)
-        if not isApiRoute || isLoginEndpoint
+        if not isApiRoute || isLoginEndpoint || isAuthenticated expectedPassword req
             then app req respond
-            else
-                if isAuthenticated expectedPassword req
-                    then app req respond
-                    else respond unauthorized
+            else respond unauthorized
   where
     unauthorized =
         responseLBS

diff --git a/src/API/DatabaseHandlers.hs b/src/API/DatabaseHandlers.hs
@@ -2,7 +2,6 @@
 {-# LANGUAGE DataKinds #-}
 {-# LANGUAGE OverloadedStrings #-}
 {-# LANGUAGE ScopedTypeVariables #-}
-{-# LANGUAGE TypeOperators #-}
 
 {- |
 Module      : API.DatabaseHandlers

diff --git a/src/API/MCP.hs b/src/API/MCP.hs
@@ -52,7 +52,7 @@ import qualified Service
 import qualified Service.Aggregate as Agg
 import SharedSolver (SharedSolver, computeInventoryMatrixWithDepsCached, crossDBProcessContributions)
 import qualified SharedSolver
-import Types (Activity (..), BioFlowDB, BiosphereFlow (..), Database (..), Indexes (..), ProcessId, UUID, UnitDB, activityLocation, activityName, bfCompartmentName, bfCompartmentSub, exchangeIsInput, getUnitNameForBioFlow, isTechnosphereExchange, processIdToText, unresolvedCount)
+import Types (Activity (..), BiosphereFlow (..), Database (..), Indexes (..), ProcessId, UUID, UnitDB, activityLocation, activityName, bfCompartmentName, bfCompartmentSub, exchangeIsInput, getUnitNameForBioFlow, processIdToText, unresolvedCount)
 import UnitConversion (defaultUnitConfig)
 
 -- ---------------------------------------------------------------------------
@@ -1429,13 +1429,12 @@ mkMcpCrossDBEntry ::
     Text ->
     -- | method UUID text
     Text ->
-    BioFlowDB ->
     UnitDB ->
     -- | total score (for share %)
     Double ->
     ((Text, ProcessId), Double) ->
     IO Value
-mkMcpCrossDBEntry dbManager rootDbName mBaseUrl colName methodIdText flowDB unitDB score ((depDbName, pid), c) = do
+mkMcpCrossDBEntry dbManager rootDbName mBaseUrl colName methodIdText unitDB score ((depDbName, pid), c) = do
     mLd <- getDatabase dbManager depDbName
     let (actName, actLoc, prodName, pidText) = case mLd of
             Just ld ->
@@ -1708,7 +1707,7 @@ callGetContributingActivities dbManager mBaseUrl rid args =
             sorted = L.sortOn (\(_, c) -> negate (abs c)) (M.toList contributions)
             top = take lim sorted
             hasNeg = any (\(_, c) -> c < 0) top
-        rows <- liftIO $ mapM (mkMcpCrossDBEntry dbManager dbName mBaseUrl (lrCollection req) (lrMethodIdText req) mFlows mUnits score) top
+        rows <- liftIO $ mapM (mkMcpCrossDBEntry dbManager dbName mBaseUrl (lrCollection req) (lrMethodIdText req) mUnits score) top
         pure $
             toolSuccessJson rid $
                 object

diff --git a/src/API/OpenApi.hs b/src/API/OpenApi.hs
@@ -12,22 +12,22 @@ entry in 'API.Resources'. The actual spec derivation from 'LCAAPI'
 lives in 'API.Routes' to break an otherwise-circular dependency
 (API.Routes -> API.OpenApi -> API.Routes).
 -}
-module API.OpenApi (enrichWithResources) where
+module API.OpenApi (enrichWithResources, stampInfo) where
 
 import API.JsonOptions (strippedSchemaOptions)
 import API.Resources (Resource)
 import qualified API.Resources as R
 import API.Types
-import Control.Lens ((%~), (&), (?~), (^.))
-import Data.Aeson (Value, toJSON)
+import Control.Lens ((%~), (&), (.~), (?~), (^.))
+import Data.Aeson (Value)
 import qualified Data.HashMap.Strict.InsOrd as InsOrdHashMap
 import Data.OpenApi
 import qualified Data.OpenApi.Lens as OA
 import Data.Text (Text)
 import qualified Data.Text as T
-import Database.Manager (DatabaseSetupInfo, DependencyChoice, DependencyStatus, MissingSupplier)
+import Database.Manager (DatabaseSetupInfo)
 import Network.HTTP.Types.Method (StdMethod (..))
-import Types (LocationFallback, LocationKind, LocationUnresolved)
+import qualified Version
 
 {- | Orphan schema instance forward declaration for the login request body.
 The real type lives in "API.Routes"; this is defined there and re-imported
@@ -75,6 +75,17 @@ instance ToSchema BinaryContent where
             NamedSchema (Just "OctetStream") $
                 mempty & type_ ?~ OpenApiString & format ?~ "binary"
 
+{- | Stamp the spec's @info@ block so each generated @openapi.json@ self-identifies:
+@info.version@ = the engine version, @info.title@ = the engine name. servant-openapi3
+leaves both blank, which makes a published spec un-anchorable; stamping them lets a
+release's surface be diffed against the previous one.
+-}
+stampInfo :: OpenApi -> OpenApi
+stampInfo spec =
+    spec
+        & OA.info . OA.title .~ "VoLCA"
+        & OA.info . OA.version .~ T.pack Version.version
+
 {- | Walk the spec and stamp metadata from 'API.Resources' onto each
 resource-backed operation. Operations without a matching 'Resource'
 (e.g. infrastructure endpoints like @/auth@, @/logs@, @/version@) are

diff --git a/src/API/Routes.hs b/src/API/Routes.hs
@@ -382,7 +382,7 @@ Built in two steps:
      on @operationId@ (e.g. @"get_impacts"@).
 -}
 volcaOpenApi :: OpenApi
-volcaOpenApi = API.OpenApi.enrichWithResources (toOpenApi (Proxy :: Proxy LCAAPI))
+volcaOpenApi = API.OpenApi.stampInfo (API.OpenApi.enrichWithResources (toOpenApi (Proxy :: Proxy LCAAPI)))
 
 {- | Expand a named classification preset from config into the
 (system, value, exact) triples used by the Service/Aggregate layers.

diff --git a/src/CLI/Client.hs b/src/CLI/Client.hs
@@ -21,6 +21,7 @@ import qualified Data.ByteString.Base64 as B64
 import qualified Data.ByteString.Char8 as C8
 import qualified Data.ByteString.Lazy as BL
 import qualified Data.ByteString.Lazy.Char8 as BSL
+import Data.Char (isAsciiLower, isAsciiUpper, isDigit)
 import Data.List (intercalate, transpose)
 import Data.Maybe (catMaybes, fromMaybe, mapMaybe)
 import Data.Text (Text)
@@ -237,7 +238,7 @@ extractLoadedDbNames = fromMaybe [] . parseMaybe go
     go :: Value -> Parser [Text]
     go = withObject "resp" $ \obj -> do
         dbs <- obj .: "dlrDatabases"
-        fmap catMaybes $ mapM getName dbs
+        catMaybes <$> mapM getName dbs
     getName :: Value -> Parser (Maybe Text)
     getName = withObject "db" $ \db -> do
         status <- db .: "dsaStatus"
@@ -310,9 +311,9 @@ buildQuery params =
   where
     urlEncode = concatMap encodeChar
     encodeChar c
-        | c >= 'A' && c <= 'Z' = [c]
-        | c >= 'a' && c <= 'z' = [c]
-        | c >= '0' && c <= '9' = [c]
+        | isAsciiUpper c = [c]
+        | isAsciiLower c = [c]
+        | isDigit c = [c]
         | c `elem` ("-_.~" :: String) = [c]
         | otherwise = '%' : showHex2 (fromEnum c)
     showHex2 n = [hexDigit (n `div` 16), hexDigit (n `mod` 16)]
@@ -427,7 +428,7 @@ extractTable :: [Value] -> ([String], [[String]])
 extractTable [] = ([], [])
 extractTable rows@(Object first : _) =
     let keys = map fst (KM.toList first)
-        headers = map (Key.toString) keys
+        headers = map Key.toString keys
         dataRows = map (rowValues keys) rows
      in (headers, dataRows)
 extractTable rows = (["value"], map (\v -> [cellValue v]) rows)

diff --git a/src/CLI/Command.hs b/src/CLI/Command.hs
@@ -314,10 +314,10 @@ executeExportMatricesCommand registry database outputDir = do
             ehExport exporter ctx outputDir
         Nothing -> Service.exportUniversalMatrixFormat outputDir database
     reportProgress Info "Matrix export completed"
-    reportProgress Info $ "  - ie_index.csv (activity index)"
-    reportProgress Info $ "  - ee_index.csv (biosphere flow index)"
-    reportProgress Info $ "  - A_public.csv (technosphere matrix)"
-    reportProgress Info $ "  - B_public.csv (biosphere matrix)"
+    reportProgress Info "  - ie_index.csv (activity index)"
+    reportProgress Info "  - ee_index.csv (biosphere flow index)"
+    reportProgress Info "  - A_public.csv (technosphere matrix)"
+    reportProgress Info "  - B_public.csv (biosphere matrix)"
 
 -- | Execute plugin list command
 executePluginList :: PluginRegistry -> OutputFormat -> IO ()
@@ -338,14 +338,11 @@ executePluginList registry fmt = do
   where
     pluginEntry name typ backend mPriority =
         object $
-            concat
-                [
-                    [ "name" .= name
-                    , "type" .= (typ :: T.Text)
-                    , "backend" .= backendText backend
-                    ]
-                , maybe [] (\p -> ["priority" .= p]) mPriority
-                ]
+            [ "name" .= name
+            , "type" .= (typ :: T.Text)
+            , "backend" .= backendText backend
+            ]
+                ++ maybe [] (\p -> ["priority" .= p]) mPriority
     backendText Builtin = "builtin" :: T.Text
     backendText (External p) = "external:" <> T.pack p