wip: feat: org-element specs

doerg/deps.edn

@@ -9,7 +9,8 @@
         com.rpl/specter {:mvn/version "1.1.6"}
         lambdaisland/deep-diff2 {:mvn/version "2.12.219"}
         mvxcvi/clj-cbor {:mvn/version "1.1.1"}
-        ch.qos.logback/logback-classic {:mvn/version "1.1.3"}}
+        ch.qos.logback/logback-classic {:mvn/version "1.1.3"}
+        org.clojure/test.check {:mvn/version "1.1.3"}}
  :paths ["src" "resources" "test"]
  :aliases
  {:test {:extra-deps {lambdaisland/kaocha {:mvn/version "1.91.1392"}}

doerg/doerg-parser/index.js

@@ -2,13 +2,17 @@
 
 const { parse } = require ("uniorg-parse/lib/parser.js");
 
+const opts = {
+    trackPosition: true
+}
+
 async function main () {
     const chunks = []
     for await (const chunk of process.stdin) {
 	chunks.push (chunk)
     }
     const orgText = Buffer.concat (chunks).toString ("utf8")
-    process.stdout.write (JSON.stringify (parse (orgText)))
+    process.stdout.write (JSON.stringify (parse (orgText, opts)))
 }
 
 main ()

doerg/resources/net/deertopia/doerg/preview-template.tex

@@ -48,6 +48,12 @@
 \newcommand{\optic}[3]{\opticname{#1}^\prime\;#2\;#3}
 \newcommand{\Optic}[5]{\opticname{#1}\;#2\;#3\;#4\;#5}
 
+% Default uses arrow glyphs from the active font, which are kinda ugly in the
+% case of Plex.
+\tikzcdset{
+  arrow style=tikz
+}
+
 \begin{document}
 \setlength\abovedisplayskip{0pt} % Remove padding before equation environments.
 %% \color[rgb]{0.000,0.000,0.004}\special{dvisvgm:currentcolor on}\setcounter{equation}{0}%

doerg/src/net/deertopia/doerg/element.clj

@@ -1,18 +1,26 @@
 (ns net.deertopia.doerg.element
-  (:require [babashka.process :as p]
+  (:refer-clojure :exclude [read-string type])
-            [net.deertopia.doerg.common :as common]
+  (:require
-            [clojure.string :as str]
+   [babashka.fs :as fs]
-            [clojure.zip]
+   [babashka.process :as p]
-            [babashka.fs :as fs]
+   [cheshire.core :as json]
-            [clojure.java.io :as io]
+   [clojure.core.match :refer [match]]
-            [cheshire.core :as json]
+   [clojure.java.io :as io]
-            [clojure.spec.alpha :as s]
+   [clojure.set :as set]
-            [spec-dict.main :refer [dict]]
+   [clojure.spec.alpha :as s]
-            [net.deertopia.doerg.config :as cfg]
+   [clojure.string :as str]
-            [com.rpl.specter :as sp]
+   [clojure.test.check.generators :as gen]
-            [clojure.tools.logging.readable :as lr])
+   [clojure.tools.logging.readable :as lr]
-  (:import (java.util UUID))
+   [clojure.zip :as z]
-  (:refer-clojure :exclude [read-string]))
+   [com.rpl.specter :as sp]
+   [com.rpl.specter.zipper :as sz]
+   [net.deertopia.doerg.common :as common]
+   [net.deertopia.doerg.config :as cfg]
+   [spec-dict.main :refer [dict]]
+   [clojure.tools.logging :as l])
+  (:import
+   (java.util UUID)))
+
 
 
 (def ^:dynamic *uniorg-timeout-duration*
@@ -34,14 +42,16 @@
     (if (zero? (:exit r))
       (-> r :out (json/parse-string (comp keyword camel->kebab))))))
 
-(declare gather-first-section)
+(declare gather-first-section gather-latex-paragraphs element-types)
 
-(defn read-string [s & {:keys [post-processors]
+(defn read-string
-                        :or {post-processors [gather-first-section]}}]
+  [s & {:keys [post-processors]
+        :or {post-processors [gather-first-section
+                              gather-latex-paragraphs]}}]
   (let [apply-post-processors (apply comp (reverse post-processors))]
     (with-in-str s
-     (-> (uniorg :in *in*)
+      (-> (uniorg :in *in*)
-         apply-post-processors))))
+          apply-post-processors))))
 
 
 
@@ -60,6 +70,9 @@
   (and (map? element)
        (contains? element :type)))
 
+(defn type [element]
+  (:type element))
+
 (defn of-type?
   "Return truthy if the Org node `element` is of type `type`. In the
   vararg case, return truthy if `element` is of any of the types
@@ -196,6 +209,22 @@
      :first-section-nodes of-first-section
      :rest remaining-nodes*}))
 
+(defn- element-bounds [& nodes]
+  (reduce (fn [acc {:keys [contents-begin contents-end]}]
+            (if (and (nat-int? contents-begin)
+                     (nat-int? contents-end))
+              (-> acc
+                  (update
+                   :contents-begin
+                   #(min (or % Integer/MAX_VALUE) contents-begin))
+                  (update
+                   :contents-end
+                   #(max (or % Integer/MIN_VALUE) contents-end)))
+              acc))
+          {:contents-begin nil
+           :contents-end nil}
+          nodes))
+
 (defn gather-first-section [node]
   (assert (of-type? node "org-data")
           "`gather-doerg-data` should be applied to the document root.")
@@ -203,8 +232,193 @@
         (split-sections (:children node))
         ;; TODO: Construct `:contents-begin` and `:contents-end` data
         ;; by spanning the children.
+        first-section (merge {:type "section"
+                              :children first-section-nodes}
+                             (apply element-bounds first-section-nodes))
         new-children (concat top-level-nodes
-                             (list {:type "section"
+                             (list first-section)
-                                    :children first-section-nodes})
                              rest)]
     (assoc node :children new-children)))
+
+(defn separated-by-explicit-paragraph-break?
+  "Returh truthy if each successive pair of elements is separated by
+  at least one explicit paragraph break; i.e. a blank line."
+  [& elements]
+  (match elements
+    [e₁ e₂ & es]
+    (and (< (-> e₁ :position :end :line)
+            (-> e₂ :position :start :line))
+         (recur es))
+    :else true))
+
+(defn swallow
+  ([predator prey]
+   (assert (greater-element? predator))
+   (-> predator
+       (update :children #(conj % prey))
+       (assoc-in [:position :end] (-> prey :position :end))))
+  ([predator prey & more-prey]
+   (reduce swallow predator (cons prey more-prey))))
+
+(defn gather-latex-paragraphs [node]
+  (->> node
+       (sp/transform
+        [postorder-walker (sp/must :children)]
+        (fn [children]
+          (loop [acc []
+                 cs (vec children)]
+            (match cs
+              ;; CASE: A paragraph followed by a LaTeX environment
+              ;; followed by a paragraph.  If there are no blank lines
+              ;; separating the three elements, absorb them into a
+              ;; single paragraph spanning the sum of their parts.
+              ([(para₁ :guard #(of-type? % "paragraph"))
+                (tex :guard #(of-type? % "latex-environment"))
+                (para₂ :guard #(of-type? % "paragraph"))
+                & rest]
+               :guard #(apply separated-by-explicit-paragraph-break? %))
+              (recur (conj acc
+                           ;; Swallow para₂'s /children/,
+                           ;; not para₂ itself.  Nested
+                           ;; paragraphs are not supported
+                           ;; by HTML.
+                           (apply swallow para₁ tex (:children para₂)))
+                     rest)
+              ;; CASE: A paragraph followed by a LaTeX environment.
+              ;; If there are no blank lines separating the paragraph
+              ;; from the LaTeX environment, the LaTeX environment
+              ;; shall become a child of the paragraph.
+              ([(para :guard #(of-type? % "paragraph"))
+                (tex :guard #(of-type? % "latex-environment"))
+                & rest]
+               :guard #(apply separated-by-explicit-paragraph-break? %))
+              (recur (conj acc (swallow para tex)) rest)
+              ;; CASE: Irrelevant or empty!
+              [c & rest]
+              (recur (conj acc c) rest)
+              [] acc))))))
+
+
+;;; Specs (top-level)
+
+;; Data taken from uniorg/index.d.ts
+
+(comment
+  (defn- typescript-enum->set [s]
+    (as-> s
+        it
+      (str/split it #" \| ")
+      (map camel->kebab it)
+      (into #{} it))))
+
+(def greater-element-types
+  #{"org-data" "section" "property-drawer" "drawer" "list" "list-item"
+    "quote-block" "verse-block" "center-block" "special-block"
+    "footnote-definition" "table"})
+
+(def element-types
+  #{"list-item-tag" "src-block" "comment-block" "latex-environment"
+    "keyword" "paragraph" "node-property" "example-block" "clock"
+    "planning" "diary-sexp" "fixed-width" "export-block"
+    "horizontal-rule" "comment" "table-row" "headline"})
+
+(def recursive-object-types
+  #{"citation" "footnote-reference" "superscript" "table-cell" "link"
+    "italic" "citation-common-prefix" "subscript" "citation-prefix"
+    "citation-common-suffix" "strike-through" "citation-reference"
+    "bold" "underline"})
+
+(def object-types
+  #{"line-break" "citation-suffix" "statistics-cookie" "timestamp"
+    "text" "verbatim" "citation-key" "export-snippet" "latex-fragment"
+    "entity" "code"})
+
+(s/def ::greater-element-type greater-element-types)
+(s/def ::element-type element-types)
+(s/def ::object-type object-types)
+(s/def ::recursive-object-type recursive-object-types)
+
+(s/def ::contents-begin nat-int?)
+(s/def ::contents-end nat-int?)
+
+(defmulti node-spec :type)
+
+(defn- unimplemented-spec [x]
+  (lr/warnf "unimplemented method for %s" (:type x))
+  any?)
+
+(defmethod node-spec :default [x] (unimplemented-spec x))
+
+(def ^:private nfe
+  "NFE — “no further expectations.” Used in sub-specs of `::element`
+  et al. for elements with no additional structure beyond that
+  provided by their parents."
+  (s/with-gen (constantly true)
+    (constantly (gen/return {}))))
+
+(s/def ::object
+  (dict {:type string?}))
+
+(s/def ::element
+  (dict ^:opt {:contents-begin ::contents-begin
+               :contents-end ::contents-end}
+        {:children (s/coll-of nfe :kind vector?)
+         :type string?}))
+
+(s/def ::node nil)
+
+(s/def ::greater-element
+  (dict {:contents-begin ::contents-begin
+         :contents-end ::contents-end
+         :children (s/coll-of ::node :kind vector?)
+         :type string?}))
+
+(s/def ::recursive-object
+  (dict ^:opt {:contents-begin ::contents-begin
+               :contents-end ::contents-end}
+        {:children (s/coll-of ::node :kind vector?)}))
+
+(s/def ::node (s/multi-spec node-spec :type))
+
+(comment
+  (use 'net.deertopia.doerg.repl)
+  (def doc (-> some-org-file slurp read-string))
+  (s/explain ::node doc))
+
+(s/def ::todo-keyword string?)
+(s/def ::priority string?)
+(s/def ::commented boolean?)
+(s/def ::level nat-int?)
+(s/def ::tags (s/coll-of string? :kind vector?))
+
+
+;;; Specs (specific objects)
+
+(def ^:private string-value (dict {:value string?}))
+
+(defmethod node-spec "text" [_] (s/merge ::object string-value))
+(defmethod node-spec "verbatim" [_] (s/merge ::object string-value))
+(defmethod node-spec "code" [_] (s/merge ::object string-value))
+(defmethod node-spec "bold" [_] ::recursive-object)
+(defmethod node-spec "italic" [_] ::recursive-object)
+
+
+;;; Specs (specific elements)
+
+(defmethod node-spec "headline" [_]
+  (s/merge ::element
+           (dict {:todo-keyword (s/nilable ::todo-keyword)
+                  :priority (s/nilable ::priority)
+                  :level ::level
+                  :commented ::commented
+                  :raw-value string?
+                  :tags ::tags})))
+
+
+;;; Specs (specific greater elements)
+
+(defmethod node-spec "org-data" [_]
+  ::greater-element)
+
+(defmethod node-spec "section" [_]
+  ::greater-element)

doerg/test/net/deertopia/doerg/element_test.clj

@@ -49,3 +49,47 @@
               first-paragraph-belongs-to-first-section?))
     (t/is (not (-> (parse-resource "first-paragraph-under-heading.org")
                    first-paragraph-belongs-to-first-section?)))))
+
+(defn walk-types [type & types]
+  [sut/postorder-walker #(apply sut/of-type? % type types)])
+
+(defn headline-matches? [re]
+  ())
+
+(defn- paragraph-ends-with-latex? [doc]
+  (let [type (-> (sp/select-first [(walk-types "paragraph")
+                                   (sp/must :children)
+                                   sp/LAST]
+                                  doc)
+                 sut/type)]
+    (t/is type "latex-environment")))
+
+(defn- paragraph-has-latex? [doc]
+  (t/is (sp/select-first [(walk-types "paragraph")
+                          (sp/must :children)
+                          sp/ALL
+                          #(sut/of-type? % "latex-environment")]
+                         doc)))
+
+(defn- paragraph-has-multiple-latex? [doc]
+  (let [paragraphs (sp/select (walk-types "paragraph") doc)]
+    (t/is (= 2 (count paragraphs)))
+    (let [[p₁ p₂] paragraphs]
+      (t/are [p ts] (= ts
+                       (sp/select [(sp/must :children)
+                                   sp/ALL (sp/view sut/type)] p))
+        p₁ ["text" "latex-environment"
+            "text" "latex-environment"]
+        p₂ ["text" "latex-environment"
+            "text" "latex-environment" "text"]))))
+
+(t/deftest paragraph-separation
+  (t/testing "paragraph ending with latex"
+    (-> (parse-resource "paragraph-ending-with-latex.org")
+        paragraph-ends-with-latex?))
+  (t/testing "paragraph surrounding latex"
+    (-> (parse-resource "paragraph-surrounding-latex.org")
+        paragraph-has-latex?))
+  (t/testing "paragraph with interleaved latex"
+    (-> (parse-resource "paragraph-with-multiple-latex.org")
+        paragraph-has-multiple-latex?)))

doerg/test/net/deertopia/doerg/element_test/paragraph-ending-with-latex.org

@@ -0,0 +1,7 @@
+#+title: paragraph ending with latex
+
+here is the paragraph,
+\begin{align*}
+\text{and here} &
+\\ & \text{is the \LaTeX}
+\end{align*}

doerg/test/net/deertopia/doerg/element_test/paragraph-surrounding-latex.org

@@ -0,0 +1,7 @@
+#+title: paragraph surrounding latex
+
+first part of paragraph
+\begin{equation*}
+\text{some \LaTeX \}:)}
+\end{equation*}
+last part of paragraph

doerg/test/net/deertopia/doerg/element_test/paragraph-with-multiple-latex.org

@@ -0,0 +1,24 @@
+#+title: paragraph with multiple latex environments
+
+* interleaved
+
+first part of paragraph
+\begin{equation*}
+\text{first \LaTeX\ environment}
+\end{equation*}
+second part of paragraph
+\begin{equation*}
+\text{second \LaTeX\ environment}
+\end{equation*}
+
+* fenceposted
+
+first fencepost
+\begin{equation*}
+\text{first fenceposted \LaTeX\ environment}
+\end{equation*}
+second fencepost
+\begin{equation*}
+\text{second fenceposted \LaTeX\ environment}
+\end{equation*}
+third fencepost