processPandocBiblio: Honour nocite directives

CHANGELOG.md

@@ -7,10 +7,12 @@ title: Releases
 ## Hakyll (unreleased)
 
 - GHC 9.12 compatibility: bump `template-haskell` upper bound to include 2.23
+- Add support for `nocite` metadata field to `processPandocBiblio` and
+    `processPandocBiblios` (#1058) (contribution by Tony Zorman)
 
 ## Hakyll 4.16.4.0 (2024-12-08)
 
-- Fixed an issue where compressing CSS with `clamp` expressions would 
+- Fixed an issue where compressing CSS with `clamp` expressions would
     result in invalid CSS (#1021) (contribution by Laurent P. René de Cotret)
 - Added `boolFieldM` (#1044) (contribution by 0xd34df00d)
 - Run HLint as part of GitHub Actions (#1045) (contribution by Yoo Chung)
@@ -118,7 +120,7 @@ title: Releases
     Alexander Batischev)
 - Allow `pandoc` 3.0. Note that the behavior of Hakyll's `readPandocBiblios` and
     `readPandocBiblio` is different whether pandoc 2 or 3 is installed
-    (contribution by Laurent P. René de Cotret) 
+    (contribution by Laurent P. René de Cotret)
 - Bump `mtl` upper bound to allow 2.3 (contribution by Alexander Batischev)
 - Bump `pandoc` upper bound to allow 3.1 (contribution by Laurent P. René de
     Cotret)

hakyll.cabal

@@ -252,7 +252,8 @@ Library
     Other-Modules:
       Hakyll.Web.Pandoc.Binary
     Build-Depends:
-      pandoc >= 2.11 && < 2.20 || >= 3.0 && < 3.7
+      pandoc >= 2.11 && < 2.20 || >= 3.0 && < 3.7,
+      pandoc-types >= 1.22 && < 1.24
     Cpp-options:
       -DUSE_PANDOC
 
@@ -317,7 +318,8 @@ Test-suite hakyll-tests
     Cpp-options:
       -DUSE_PANDOC
     Build-Depends:
-      pandoc >= 2.11 && < 2.20 || >= 3.0 && < 3.7
+      pandoc >= 2.11 && < 2.20 || >= 3.0 && < 3.7,
+      pandoc-types >= 1.22 && < 1.24
 
 
 Executable hakyll-init
@@ -351,4 +353,5 @@ Executable hakyll-website
     base      >= 4.12  && < 5,
     directory >= 1.0   && < 1.4,
     filepath  >= 1.0   && < 1.6,
-    pandoc    >= 2.11  && < 2.20 || >= 3.0 && < 3.7
+    pandoc    >= 2.11  && < 2.20 || >= 3.0 && < 3.7,
+    pandoc-types >= 1.22 && < 1.24

lib/Hakyll/Web/Pandoc/Biblio.hs

@@ -15,6 +15,7 @@
 {-# LANGUAGE Arrows                     #-}
 {-# LANGUAGE DeriveDataTypeable         #-}
 {-# LANGUAGE GeneralizedNewtypeDeriving #-}
+{-# LANGUAGE LambdaCase                 #-}
 {-# LANGUAGE OverloadedStrings          #-}
 module Hakyll.Web.Pandoc.Biblio
     ( CSL (..)
@@ -44,13 +45,16 @@ import           Hakyll.Core.Compiler.Internal
 import           Hakyll.Core.Identifier
 import           Hakyll.Core.Identifier.Pattern (fromGlob)
 import           Hakyll.Core.Item
+import           Hakyll.Core.Metadata          (getMetadataField)
 import           Hakyll.Core.Writable
 import           Hakyll.Web.Pandoc
 import           Text.Pandoc                   (Extension (..), Pandoc,
-                                                ReaderOptions (..),
+                                                PandocPure, ReaderOptions (..),
                                                 enableExtension)
 import qualified Text.Pandoc                   as Pandoc
+import           Text.Pandoc.Builder           (setMeta)
 import qualified Text.Pandoc.Citeproc          as Pandoc (processCitations)
+import           Text.Pandoc.Walk              (Walkable (query))
 import           System.FilePath               (addExtension, takeExtension)
 
 
@@ -116,7 +120,7 @@ processPandocBiblios :: Item CSL
                      -> [Item Biblio]
                      -> (Item Pandoc)
                      -> Compiler (Item Pandoc)
-processPandocBiblios csl biblios item = do
+processPandocBiblios csl biblios item' = do
     -- It's not straightforward to use the Pandoc API as of 2.11 to deal with
     -- citations, since it doesn't export many things in 'Text.Pandoc.Citeproc'.
     -- The 'citeproc' package is also hard to use.
@@ -126,6 +130,12 @@ processPandocBiblios csl biblios item = do
     --
     -- So we load the CSL and Biblio files and pass them to Pandoc using the
     -- ersatz filesystem.
+
+    -- Honour nocite metadata fields
+    item <- getUnderlying >>= (`getMetadataField` "nocite") >>= \case
+        Nothing -> pure item'
+        Just x  -> withItemBody (pure . setMeta "nocite" x) item'
+
     let Pandoc.Pandoc (Pandoc.Meta meta) blocks = itemBody item
         cslFile = Pandoc.FileInfo zeroTime . unCSL $ itemBody csl
         bibFiles = zipWith (\x y ->
@@ -148,19 +158,30 @@ processPandocBiblios csl biblios item = do
             Map.insert "bibliography"
               (Pandoc.MetaList $ map (Pandoc.MetaString . T.pack . fst) bibFiles) $
             meta
-        errOrPandoc = Pandoc.runPure $ do
-            Pandoc.modifyPureState addBiblioFiles
-            Pandoc.processCitations $ Pandoc.Pandoc biblioMeta blocks
-
-    pandoc <- case errOrPandoc of
-        Left  e -> compilerThrow ["Error during processCitations: " ++ show e]
-        Right x -> return x
 
+    pandoc <- do
+        let p = Pandoc.Pandoc biblioMeta blocks
+        p' <- case Pandoc.lookupMeta "nocite" biblioMeta of
+            Just (Pandoc.MetaString nocite) -> do
+                Pandoc.Pandoc _ b <- runPandoc $
+                    Pandoc.readMarkdown defaultHakyllReaderOptions nocite
+                let nocites = Pandoc.MetaInlines . flip query b $ \case
+                        c@Pandoc.Cite{} -> [c]
+                        _               -> []
+                return $ setMeta "nocite" nocites p
+            _ -> return p
+        runPandoc $ do
+            Pandoc.modifyPureState addBiblioFiles
+            Pandoc.processCitations p'
     return $ fmap (const pandoc) item
 
   where
     zeroTime = Time.UTCTime (toEnum 0) 0
 
+    runPandoc :: PandocPure a -> Compiler a
+    runPandoc with = case Pandoc.runPure with of
+        Left  e -> compilerThrow ["Error during processCitations: " ++ show e]
+        Right x -> pure x
 
 --------------------------------------------------------------------------------
 -- | Compiles a markdown file via Pandoc. Requires the .csl and .bib files to be known to the compiler via match statements.