Use markdown parser based on commonmark-java (#668)

Upgrades `nextjournal/markdown` library to `{:mvn/version 0.6.157}` switching markdown parsing from `markdown-it` via GraalJS to `commonmark-java` (roughly 10 times faster). 

Makes Clerk is compatible with Java 22 or later (#642).

deps.edn

@@ -1,4 +1,5 @@
 {:paths ["src" "resources" "bb"]
+ :mvn/repos {"jitpack.io" {:url "https://jitpack.io"}}
  :deps {org.clojure/clojure {:mvn/version "1.10.3"}
         org.clojure/java.classpath {:mvn/version "1.0.0"}
         org.clojure/tools.analyzer {:mvn/version "1.1.0"}
@@ -8,8 +9,9 @@
         weavejester/dependency {:mvn/version "0.2.1"}
         com.nextjournal/beholder {:mvn/version "1.0.2"}
         org.flatland/ordered {:mvn/version "1.15.12"}
-
-        io.github.nextjournal/markdown {:mvn/version "0.5.148"}
+        io.github.nextjournal/markdown {:mvn/version "0.6.157"
+                                        ;; only used in cljs
+                                        :exclusions [applied-science/js-interop]}
         babashka/process {:mvn/version "0.4.16"}
         io.github.nextjournal/dejavu {:git/sha "4980e0cc18c9b09fb220874ace94ba6b57a749ca"}
         io.github.babashka/sci.nrepl {:mvn/version "0.0.2"}

notebooks/markdown.md

@@ -13,6 +13,7 @@ This notebook demoes feeding Clerk with markdown files. We currently make no ass
 Nextjournal Markdown library is able to ingest a markdown string
 
 ```clojure
+^{::clerk/visibility {:code :fold :result :hide}}
 (def markdown-syntax
   "Markdown: Syntax
 ================

src/nextjournal/clerk/parser.cljc

@@ -4,7 +4,6 @@
             [clojure.string :as str]
             [clojure.zip]
             [nextjournal.markdown :as markdown]
-            [nextjournal.markdown.parser :as markdown.parser]
             [nextjournal.markdown.transform :as markdown.transform]
             [rewrite-clj.node :as n]
             [rewrite-clj.parser :as p]
@@ -291,33 +290,17 @@
 #_(text-with-clerk-metadata-removed "(def ^::clerk/no-cache random-thing (rand-int 1000))" {'clerk 'nextjournal.clerk})
 #_(text-with-clerk-metadata-removed "^::clerk/bar [] ;; keep me" {'clerk 'nextjournal.clerk})
 
-(defn markdown-context []
-  (update markdown.parser/empty-doc
-          :text-tokenizers (partial map markdown.parser/normalize-tokenizer)))
-
-#_(markdown-context)
-
-(defn parse-markdown
-  "Like `n.markdown.parser/parse` but allows to reuse the same context in successive calls"
-  [ctx md]
-  (markdown.parser/apply-tokens ctx (markdown/tokenize md)))
-
 (defn update-markdown-blocks [{:as state :keys [md-context]} md]
-  (let [{::markdown.parser/keys [path]} md-context
-        doc (parse-markdown md-context md)
-        [_ index] path]
+  (let [doc (markdown/parse* (assoc md-context :content []) md)]
     (-> state
         (assoc :md-context doc)
         (update :blocks conj {:type :markdown
-                              :doc (-> doc
-                                       (select-keys [:type :content :footnotes])
-                                       ;; take only new nodes, keep context intact
-                                       (update :content subvec (inc index)))}))))
+                              :doc (select-keys doc [:type :content :footnotes])}))))
 
 (defn parse-clojure-string
   ([s] (parse-clojure-string {} s))
   ([{:as opts :keys [doc?]} s]
-   (let [doc (parse-clojure-string opts {:blocks [] :md-context (markdown-context)} s)]
+   (let [doc (parse-clojure-string opts {:blocks [] :md-context markdown/empty-doc} s)]
      (select-keys (cond-> doc doc? (merge (:md-context doc)))
                   [:blocks :title :toc :footnotes])))
   ([{:as _opts :keys [doc?]} initial-state s]
@@ -382,7 +365,7 @@
   (update doc :blocks #(filterv (some-fn :form (complement code?)) %)))
 
 (defn parse-markdown-string [{:as opts :keys [doc?]} s]
-  (let [{:as ctx :keys [content]} (parse-markdown (markdown-context) s)]
+  (let [{:as ctx :keys [content]} (markdown/parse* markdown/empty-doc s)]
     (loop [{:as state :keys [nodes] ::keys [md-slice]} {:blocks [] ::md-slice [] :nodes content :md-context ctx}]
       (if-some [node (first nodes)]
         (recur

src/nextjournal/clerk/viewer.cljc

@@ -21,7 +21,7 @@
                        [applied-science.js-interop :as j]])
             [nextjournal.clerk.parser :as parser]
             [nextjournal.markdown :as md]
-            [nextjournal.markdown.parser :as md.parser]
+            [nextjournal.markdown.utils :as md.utils]
             [nextjournal.markdown.transform :as md.transform])
   #?(:clj (:import (com.pngencoder PngEncoder)
                    (clojure.lang IDeref IAtom)
@@ -585,7 +585,7 @@
 
 (defn process-sidenotes [cell-doc {:keys [footnotes]}]
   (if (seq footnotes)
-    (md.parser/insert-sidenote-containers (assoc cell-doc :footnotes footnotes))
+    (md.utils/insert-sidenote-containers (assoc cell-doc :footnotes footnotes))
     cell-doc))
 
 (defn process-image-source [src {:as doc :keys [file package]}]
@@ -674,6 +674,14 @@
    :transform-fn (update-val transform-cell)
    :render-fn '(fn [xs opts] (into [:<>] (nextjournal.clerk.render/inspect-children opts) xs))})
 
+(defn lift-block-images
+  "Lift an image node to top-level when it is the only child of a paragraph."
+  [md-nodes]
+  (map (fn [{:as node :keys [type content]}]
+         (if (and (= :paragraph type) (= 1 (count content)) (= :image (:type (first content))))
+           (first content)
+           node)) md-nodes))
+
 (defn with-block-viewer [doc {:as cell :keys [type id]}]
   (case type
     :markdown (let [{:keys [content]} (:doc cell)
@@ -685,7 +693,8 @@
                                (process-sidenotes {:type :doc
                                                    :content (vec fragment)
                                                    ::doc doc} doc))]))
-                        (partition-by (comp #{:image} :type) content)))
+                        (partition-by (comp #{:image} :type)
+                                      (lift-block-images content))))
 
     :code (if (cell-visible? cell)
             [(with-viewer `cell-viewer (assoc cell ::doc doc))]