rough implementation of Python tool

breandan · Jan 28, 2025 · 2838676 · 2838676
1 parent a8a8a1a
commit 2838676
Show file tree

Hide file tree

Showing 3 changed files with 1,842 additions and 2 deletions.
diff --git a/src/commonMain/kotlin/ai/hypergraph/kaliningraph/parsing/CFG.kt b/src/commonMain/kotlin/ai/hypergraph/kaliningraph/parsing/CFG.kt
@@ -244,7 +244,12 @@ class Bindex<T>(
   val ntIndices: Map<T, Int> = indexedNTs.zip(indexedNTs.indices).toMap()
 ): List<T> by indexedNTs {
   constructor(map: Map<Int, T>) : this(map.values.toSet(), map.values.toList(), map.entries.associate { it.value to it.key })
-  operator fun get(s: T): Int = ntIndices[s] ?: 1.also { println("Unknown nonterminal: $s"); null!! }
+  operator fun get(s: T): Int = ntIndices[s] ?: 1.also {
+    println("Unknown nonterminal: $s");
+    try {
+      throw IllegalArgumentException("Unknown nonterminal: $s")
+    } catch (e: IllegalArgumentException) {e.printStackTrace()}
+    null!! }
   fun getUnsafe(s: T): Int? = ntIndices[s]
   override fun toString(): String = indexedNTs.mapIndexed { i, it -> "$i: $it" }.joinToString("\n", "Bindex:\n", "\n")
 }

diff --git a/src/commonMain/kotlin/ai/hypergraph/kaliningraph/parsing/Segmentation.kt b/src/commonMain/kotlin/ai/hypergraph/kaliningraph/parsing/Segmentation.kt
@@ -82,7 +82,7 @@ data class Segmentation(
       acc
     }
 
-  // Takes an IntRange of word indices and a String of words delimited by one or more whitespaces,
+// Takes an IntRange of word indices and a String of words delimited by one or more whitespaces,
 // and returns the corresponding IntRange of character indices in the original string.
 // For example, if the input is (1..2, "a__bb___ca d e f"), the output is 3..10
   fun IntRange.charIndicesOfWordsInString(str: String): IntRange {
@@ -125,6 +125,16 @@ fun preparseParseableLines(cfg: CFG, editorText: Σᐩ) {
     }
 }
 
+fun preparseParseableLines(cfg: CFG, editorText: Σᐩ, recognizer: (String) -> Boolean) {
+  editorText.lineSequence()
+    .filter { it.isNotBlank() && !it.containsHole() }
+    .forEach { line ->
+      segmentationCacheHTML.getOrPut(cfg.hashCode() + line.hashCode()) {
+        (if (recognizer(line)) line.also { println("Recognized $it") } else "<u>$line</u>".also { println("Unrecognized $it") })
+      }
+    }
+}
+
 fun getOrComputeSegmentations(cfg: CFG, editorText: Σᐩ): List<Segmentation> =
   editorText.split("---").last().lines() // Only preparse the section after the grammar
     .filter { it.isNotBlank() && !it.containsHole() }