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

@@ -10,7 +10,10 @@
             [spec-dict.main :refer [dict]]
             [net.deertopia.doerg.config :as cfg]
             [com.rpl.specter :as sp]
-            [clojure.tools.logging.readable :as lr])
+            [clojure.tools.logging.readable :as lr]
+            [clojure.zip :as z]
+            [com.rpl.specter.zipper :as sz]
+            [clojure.core.match :refer [match]])
   (:import (java.util UUID))
   (:refer-clojure :exclude [read-string]))
 
@@ -34,14 +37,16 @@
     (if (zero? (:exit r))
       (-> r :out (json/parse-string (comp keyword camel->kebab))))))
 
-(declare gather-first-section)
+(declare gather-first-section gather-latex-paragraphs)
 
-(defn read-string [s & {:keys [post-processors]
-                        :or {post-processors [gather-first-section]}}]
+(defn read-string
+  [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*)
-         apply-post-processors))))
+      (-> (uniorg :in *in*)
+          apply-post-processors))))
 
 
 
@@ -209,12 +214,25 @@
                              rest)]
     (assoc node :children new-children)))
 
-(defn- neighbourly-mapcat [coll f]
-  (let [rest-coll (rest coll)]
-    (map f
-         coll
-         rest-coll
-         (concat (rest rest-coll) [nil]))))
+(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))))
 
 (comment
   (-> [1 2 3 4]
@@ -222,22 +240,43 @@
   (def doc (read-string (slurp some-org-file)))
 
   (let [r (atom [])
-        blah
-        (sp/transform
-         [postorder-walker
-          (sp/must :children)
-          (sp/collect-one sp/VAL)
-          sp/INDEXED-VALS
-          #_
-          #(of-type? (second %) "latex-environment")
-          #_
-          sp/ALL
-          #_
-          sp/INDEXED-VALS]
-         (fn [siblings x]
-           x)
-         doc)]
+        blah]
     @r))
 
 (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))))))

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

@@ -49,3 +49,32 @@
               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- paragraph-ends-with-latex? [doc]
+  (-> (sp/select-first [(walk-types "paragraph")
+                        (sp/must :children)
+                        sp/LAST]
+                       doc)
+      (sut/of-type? "latex-environment")))
+
+(defn- paragraph-has-latex? [doc]
+  (sp/select-first [(walk-types "paragraph")
+                    (sp/must :children)
+                    sp/ALL
+                    #(sut/of-type? % "latex-environment")]
+                   doc))
+
+(defn- paragraph-has-multiple-latex? [doc]
+  (sp/select-first [(walk-types "paragraph") (sp/must :children)]
+                   doc))
+
+(t/deftest paragraph-separation
+  (t/testing "paragraph ending with latex"
+    (t/is (->> (parse-resource "paragraph-ending-with-latex.org")
+               paragraph-ends-with-latex?)))
+  (t/testing "paragraph surrounding latex"
+    (t/is (->> (parse-resource "paragraph-surrounding-latex.org")
+               paragraph-has-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 +1,10 @@
+#+title: paragraph with multiple latex environments
+
+first part of paragraph
+\begin{equation*}
+\text{first \LaTeX\ environment}
+\end{equation*}
+second part of paragraph
+\begin{equation*}
+\text{second \LaTeX\ environment}
+\end{equation*}