wip: fix: latex paragraphs again lol

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

@@ -232,23 +232,34 @@
         ;; TODO: Construct `:contents-begin` and `:contents-end` data
         ;; by spanning the children.
         first-section (merge {:type "section"
-                              :children first-section-nodes}
+                              :children (vec first-section-nodes)}
                              (apply element-bounds first-section-nodes))
-        new-children (concat top-level-nodes
+        new-children (vec (concat top-level-nodes
-                             (list first-section)
+                                  (list first-section)
-                             rest)]
+                                  rest))]
     (assoc node :children new-children)))
 
-(defn separated-by-explicit-paragraph-break?
+(defn- newline-final-paragraph?
-  "Returh truthy if each successive pair of elements is separated by
+  "Is `e` a paragraph, and does it end with a newline?"
-  at least one explicit paragraph break; i.e. a blank line."
+  [e]
+  (and (of-type? e "paragraph")
+       (some-> e :children last :value (str/ends-with? "\n"))))
+
+(defn consequtive-elements?
+  "Returh truthy if each successive pair of elements is NOT separated
+  by at least one explicit paragraph break; i.e. a blank line."
   [& elements]
   (match elements
-    [e₁ e₂ & es]
+    ([(e₁ :guard newline-final-paragraph?) e₂ & es] :seq)
-    (and (< (-> e₁ :position :end :line)
+    (and (= (-> e₁ :position :end :line)
             (-> e₂ :position :start :line))
          (recur es))
-    :else true))
+    ([e₁ e₂ & es] :seq)
+    (and (= (-> e₁ :position :end :line inc)
+            (-> e₂ :position :start :line))
+         (recur es))
+    ([_] :seq) true
+    ([] :seq) true))
 
 (defn swallow
   ([predator prey]
@@ -259,6 +270,22 @@
   ([predator prey & more-prey]
    (reduce swallow predator (cons prey more-prey))))
 
+(defn- paragraph-followed-by-tex? [children]
+  (match children
+    [(para :guard #(of-type? % "paragraph"))
+     (tex :guard #(of-type? % "latex-environment"))
+     & _]
+    (consequtive-elements? para tex)
+    :else false))
+
+(defn- paragraph-followed-by-paragraph? [children]
+  (match children
+    [(para₁ :guard #(of-type? % "paragraph"))
+     (para₂ :guard #(of-type? % "paragraph"))
+     & _]
+    (consequtive-elements? para₁ para₂)
+    :else false))
+
 (defn gather-latex-paragraphs [node]
   (->> node
        (sp/transform
@@ -271,16 +298,14 @@
               ;; 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"))
+              ([para tex & rest] :guard paragraph-followed-by-tex?)
-                (tex :guard #(of-type? % "latex-environment"))
-                & rest]
-               :guard #(apply separated-by-explicit-paragraph-break? %))
               (recur acc (vec (cons (swallow para tex) rest)))
-              ;; CASE:
+              ;; CASE: Similar to the paragraph-followed-by-tex case,
-              ([(para₁ :guard #(of-type? % "paragraph"))
+              ;; but instead of swallowing the entire second element,
-                (para₂ :guard #(of-type? % "paragraph"))
+              ;; we swallow the /children/ of the second element,
-                & rest]
+              ;; since paragraphs cannot be nested.
-               :guard #(apply separated-by-explicit-paragraph-break? %))
+              ([para₁ para₂ & rest]
+               :guard paragraph-followed-by-paragraph?)
               (recur acc (vec (cons (apply swallow para₁ (:children para₂))
                                     rest)))
               ;; CASE: Irrelevant or empty!

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

@@ -11,7 +11,8 @@
 
 (defn- parse-resource [path]
   (-> (str "net/deertopia/doerg/element_test/" path)
-      io/resource slurp sut/read-string))
+      io/resource slurp
+      (sut/read-string)))
 
 (t/deftest known-greater-elements
   (t/testing "known greater elements satisfy `greater-element?`"
@@ -44,41 +45,66 @@
       true)))
 
 (t/deftest first-paragraph-under-first-section
-  (t/testing "first paragraph should belong to a section"
+  (t/is (-> (parse-resource "first-paragraph-under-first-section.org")
-    (t/is (-> (parse-resource "first-paragraph-under-first-section.org")
+            first-paragraph-belongs-to-first-section?)))
-              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]
+(t/deftest first-paragraph-under-heading
+  (t/is (-> (parse-resource "first-paragraph-under-heading.org")
+            first-paragraph-belongs-to-first-section?
+            not)))
+
+(defn- walk-types [type & types]
   [sut/postorder-walker #(apply sut/of-type? % type types)])
 
-(defn- paragraph-ends-with-latex? [doc]
+(t/deftest paragraph-ending-with-latex
-  (let [type (-> (sp/select-first [(walk-types "paragraph")
+  (let [doc (parse-resource "paragraph-ending-with-latex.org")
+        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/deftest paragraph-surrounding-latex
-  (t/is (sp/select-first [(walk-types "paragraph")
+  (let [doc (parse-resource "paragraph-surrounding-latex.org")]
-                          (sp/must :children)
+    (t/is (sp/select-first [(walk-types "paragraph")
-                          sp/ALL
+                            (sp/must :children)
-                          #(sut/of-type? % "latex-environment")]
+                            sp/ALL
-                         doc)))
+                            #(sut/of-type? % "latex-environment")]
+                           doc))))
 
-(defn- paragraph-has-multiple-latex? [doc]
+(t/deftest paragraph-with-multiple-latex
-  (let [paragraphs (sp/select (walk-types "paragraph") doc)]
+  (let [doc (parse-resource "paragraph-with-multiple-latex.org")
+        paragraphs (sp/select (walk-types "paragraph") doc)]
     (t/is (= 2 (count paragraphs)))
     (let [[p₁ p₂] paragraphs]
       (doseq [[p ts] [[p₁ ["text" "latex-environment"
                            "text" "latex-environment"]]
                       [p₂ ["text" "latex-environment"
                            "text" "latex-environment" "text"]]]]
-        (t/is (= ts (sp/select [(sp/must :children)
+        (t/is (= (sp/select [(sp/must :children)
-                                sp/ALL (sp/view sut/type)] p)))))))
+                             sp/ALL (sp/view sut/type)] p)
+                 ts))))))
 
+(t/deftest paragraph-with-separate-latex
+  (let [doc (parse-resource "paragraph-with-separate-latex.org")
+        cs (sp/select [(walk-types "section")
+                       (sp/must :children)
+                       sp/ALL
+                       (sp/view sut/type)]
+                      doc)]
+    (t/is (= ["paragraph" "latex-environment"] cs))))
+
+(t/deftest paragraph-surrounding-separate-latex
+  (let [doc (parse-resource "paragraph-surrounding-separate-latex.org")
+        cs (sp/select [(walk-types "section")
+                       (sp/must :children)
+                       sp/ALL
+                       (sp/view sut/type)]
+                      doc)]
+    (t/is (= ["paragraph" "latex-environment" "paragraph"] cs))))
+
+#_
 (t/deftest paragraph-separation
   (t/testing "paragraph ending with latex"
     (-> (parse-resource "paragraph-ending-with-latex.org")
@@ -88,4 +114,10 @@
         paragraph-has-latex?))
   (t/testing "paragraph with interleaved latex"
     (-> (parse-resource "paragraph-with-multiple-latex.org")
-        paragraph-has-multiple-latex?)))
+        paragraph-has-multiple-latex?))
+  (t/testing "paragraph with separate latex"
+    (-> (parse-resource "paragraph-with-separate-latex.org")
+        paragraph-has-separate-latex?))
+  #_(t/testing "paragraphs surrounding separate latex"
+    (-> (parse-resource "paragraph-surrounding-separate-latex.org")
+        paragraph-surrounding-separate-latex?)))

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

@@ -0,0 +1,9 @@
+#+title: paragraphs surrounding separate latex
+
+a paragraph!
+
+\begin{gather*}
+\text{and now, an unrelated latex fragment}
+\end{gather*}
+
+more unrelated text

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

@@ -0,0 +1,7 @@
+#+title: paragraph with separate latex
+
+a paragraph!
+
+\begin{gather*}
+\text{and now, an unrelated latex fragment}
+\end{gather*}