repair pointer events and add zoom and drag and drop in editor

chroma_viz/app.py

@@ -1,5 +1,7 @@
 from langchain.vectorstores import Chroma
+
 vectordb = Chroma.from_documents(data, embeddings, ids)
 
 from chromaviz import visualize_collection
-visualize_collection(vectordb._collection)
+
+visualize_collection(vectordb._collection)

integrator/Dockerfile

@@ -32,6 +32,9 @@ RUN apt install -y clang python3-dev build-essential uwsgi uwsgi-plugin-gevent-p
 #RUN pip install uwsgi==2.0.23 gevent
 #RUN pip install gevent-websocket
 #CMD ["pypy3", "server.py"]
+RUN pip install spacy
+RUN python -m spacy download en_core_web_sm
+
 ENTRYPOINT ["gunicorn", "-k", "geventwebsocket.gunicorn.workers.GeventWebSocketWorker", "-w", "1", "--threads", "100", "-b", "0.0.0.0:5000", "--timeout", "2000" , "--graceful-timeout", "200", "server:app"]
 
 #ENTRYPOINT uwsgi --plugins-list

integrator/cluster_hierarchy.py

@@ -1,8 +1,11 @@
 import numpy as np
 
+
 def single_linkage_clustering(data, n_clusters):
     # Calculate the Euclidean distance matrix
-    distance_matrix = np.sqrt(((data[:, np.newaxis] - data[np.newaxis, :]) ** 2).sum(axis=2))
+    distance_matrix = np.sqrt(
+        ((data[:, np.newaxis] - data[np.newaxis, :]) ** 2).sum(axis=2)
+    )
 
     # Initialize clusters as each individual point
     clusters = [{i} for i in range(len(data))]
@@ -11,8 +14,10 @@ def single_linkage_clustering(data, n_clusters):
         # Find the pair of clusters with minimum distance
         min_dist = np.inf
         for i in range(len(clusters)):
-            for j in range(i+1, len(clusters)):
-                dist = min(distance_matrix[x, y] for x in clusters[i] for y in clusters[j])
+            for j in range(i + 1, len(clusters)):
+                dist = min(
+                    distance_matrix[x, y] for x in clusters[i] for y in clusters[j]
+                )
                 if dist < min_dist:
                     min_dist = dist
                     closest_pair = (i, j)
@@ -23,6 +28,7 @@ def single_linkage_clustering(data, n_clusters):
 
     return clusters
 
+
 # Example usage
 data = np.array([[1, 2], [2, 3], [3, 4], [10, 10], [11, 11], [12, 12]])
 n_clusters = 2

integrator/combination_state.py

@@ -32,7 +32,7 @@ def __next__(self):
                         i_a, i_b = self.on_indices[a], self.on_indices[b]
                     else:
                         i_a, i_b = a, b
-                    if not self.on [i_a, i_b]:
+                    if not self.on[i_a, i_b]:
                         self._advance_generator()
                         self.yielded_count += 1
                         valid = False
@@ -68,10 +68,26 @@ def get_percentage(self):
         return round(self.yielded_count / math.comb(self.n, self.r), 2)
 
     def __getstate__(self):
-        return self.pool, self.r, self.indices, self.exhausted, self.yielded_count, self.on, self.on_indices
+        return (
+            self.pool,
+            self.r,
+            self.indices,
+            self.exhausted,
+            self.yielded_count,
+            self.on,
+            self.on_indices,
+        )
 
     def __setstate__(self, state):
-        self.pool, self.r, self.indices, self.exhausted, self.yielded_count, self.on, self.on_indices = state
+        (
+            self.pool,
+            self.r,
+            self.indices,
+            self.exhausted,
+            self.yielded_count,
+            self.on,
+            self.on_indices,
+        ) = state
         self.n = len(self.pool)
 
 

integrator/main.py

@@ -3,6 +3,7 @@
 import itertools
 import logging
 from pprint import pprint
+
 from classifier.result.predict import MODELS
 from integrator.serialize import serialize_graph_to_structure
 from integrator.states import states
@@ -37,14 +38,15 @@ def infer(model_name, t, valid_labels, tree=None, on=None, on_indices=None):
         )
     elif isinstance(t, list):
         if len(t) >= 1:
-
             if isinstance(t[0], tuple):
                 texts = []
                 keys = []
                 pre_computed_scores = []
                 for (n1, n2), score in t:
                     if n1 not in tree.index_text or n2 not in tree.index_text:
-                        logging.error(f"node {n1} or {n2} not in tree index: {tree.index_text}")
+                        logging.error(
+                            f"node {n1} or {n2} not in tree index: {tree.index_text}"
+                        )
                         continue
                     texts.append((tree.index_text[n1], tree.index_text[n2]))
                     keys.append((n1, n2))
@@ -63,11 +65,7 @@ def infer(model_name, t, valid_labels, tree=None, on=None, on_indices=None):
     ), list(score.view(-1, MODELS[model_name].config.n_samples).tolist())
 
     if pre_computed_scores:
-        score = [
-            [s, s]
-            for s in
-            pre_computed_scores
-        ]
+        score = [[s, s] for s in pre_computed_scores]
     lsk = [
         (l, s, k)
         for l, s, k in zip(labels, score, keys)
@@ -139,14 +137,18 @@ def score(model_name, t, relation, hash_id="default"):
     scores = [score for _, score in results]
     max_score = max(scores)
     min_score = min(scores)
-    results = [(key, (score - min_score) / (max_score - min_score)) for key, score in results]
+    results = [
+        (key, (score - min_score) / (max_score - min_score)) for key, score in results
+    ]
 
     return results
 
 
 # classify thesis, antithesis, synthesis
 def classifier(t, on=None, on_indices=None):
-    return infer("thesis_antithesis_synthesis", t, [1, 2, 3], on=on, on_indices=on_indices)
+    return infer(
+        "thesis_antithesis_synthesis", t, [1, 2, 3], on=on, on_indices=on_indices
+    )
 
 
 # classify hypernym, hyponym

integrator/reader/__init__.py

@@ -1,26 +1,32 @@
 from .reader_dialogue import parse_dialogue
 from .reader_enumerated import parse_enumerated
+from .reader_sentences import parse_sentences
 
 
-# Parse the text
 def parse_text(text):
 
     result = parse_enumerated(text)
+
+    if not result:
+        result = parse_sentences(text)
     if not result:
         result = parse_dialogue(text)
+
     return result
+
+
 def get_inputs(filename):
     with open(filename) as f:
         text = f.read()
     return parse_dialogue(text)
+
+
 if __name__ == "__main__":
 
     from pprint import pprint
 
-
-    if __name__ == "__main__":
-        # Test
-        text = """
+    # Test
+    text = """
     6.1265 Logic can always be conceived to be such that every proposition is its own proof.
     
     
@@ -35,8 +41,8 @@ def get_inputs(filename):
     6.13 Logic is not a theory but a reflexion of the world.
         """
 
-        parsed_data = parse_text(text)
-        pprint(parsed_data)
+    parsed_data = parse_text(text)
+    pprint(parsed_data)
 
-        pprint(get_inputs("texts/euclid.txt"))
-        pprint(get_inputs("texts/phaedrus.1b.txt"))
+    pprint(get_inputs("texts/euclid.txt"))
+    pprint(get_inputs("texts/phaedrus.1b.txt"))

integrator/reader/reader_sentences.py

@@ -0,0 +1,35 @@
+from pprint import pprint
+
+import spacy
+
+# Load the spaCy model
+nlp = spacy.load("en_core_web_sm")
+
+def parse_sentences(text):
+    # Use spaCy's pipeline for sentence segmentation
+    doc = nlp(text)
+    sentences = [sent.text.strip() for sent in doc.sents]
+    return sentences
+
+def get_inputs(filename):
+    with open(filename, 'r') as f:
+        text = f.read()
+    return parse_sentences(text)
+
+if __name__ == "__main__":
+    dialogue_text = """
+    My dear Phaedrus, whence come you, and whither are you going?
+    I come from Lysias the son of Cephalus, and I am going to
+    take a walk outside the wall, 
+    for I have been sitting with him the
+    whole morning; and our common friend Acumenus tells me that it is
+    much more refreshing to walk in the open air than to be shut up in
+    a cloister.
+    There he is right. Lysias then, I suppose, was in the town?
+    """
+
+    parsed_dialogue = parse_sentences(dialogue_text)
+    pprint(parsed_dialogue)
+
+    # Example of reading from a file
+    # pprint(get_inputs("texts/phaedrus.1b.txt"))

integrator/tree.py

@@ -194,15 +194,17 @@ def draw_graph(self, graph=None, root=None, path="graph.png", text_relation=True
         except Exception as e:
             logging.error(f"Error in drawing graph {e}", exc_info=True)
 
-    def pull(self, n_samples, relations, on=None, on_indices =None):
+    def pull(self, n_samples, relations, on=None, on_indices=None):
         if not tuple(relations) in self.iterators:
             self.iterators[tuple(relations)] = CustomCombinations(
                 list(self.node_index.keys()), n_samples, on=on, on_indices=on_indices
             )
         for k in self.iterators[tuple(relations)]:
             yield [(i, self.index_text[i]) for i in k]
 
-    def pull_batch(self, batch_size, n_samples, relations=None, on=None ,on_indices =None):
+    def pull_batch(
+        self, batch_size, n_samples, relations=None, on=None, on_indices=None
+    ):
         """
         :param batch_size:
         :param n_samples:
@@ -215,7 +217,10 @@ def pull_batch(self, batch_size, n_samples, relations=None, on=None ,on_indices
                     *[
                         ([i for i, _ in samples], [s for _, s in samples])
                         for samples in maxislice(
-                            self.pull(n_samples, relations, on=on, on_indices=on_indices), batch_size
+                            self.pull(
+                                n_samples, relations, on=on, on_indices=on_indices
+                            ),
+                            batch_size,
                         )
                     ]
                 )

web/package.json

@@ -20,6 +20,7 @@
     "react": "^18.2.0",
     "react-dom": "^18.2.0",
     "react-markdown": "^8.0.7",
+    "react-pan-zoom": "^1.0.2",
     "react-query": "^3.39.3",
     "react-scripts": "5.0.1",
     "react-zoom-pan-pinch": "^3.0.8",

web/src/lib/position.js

@@ -46,3 +46,21 @@ export function postProcessTitle(title) {
   if (!title || typeof title == 'object') return null
   return title?.replace('.md', '')?.replace('_.', '')
 }
+
+export const pointInTriangle = (px, py, ax, ay, bx, by, cx, cy) => {
+  const v0 = [cx - ax, cy - ay]
+  const v1 = [bx - ax, by - ay]
+  const v2 = [px - ax, py - ay]
+
+  const dot00 = v0[0] * v0[0] + v0[1] * v0[1]
+  const dot01 = v0[0] * v1[0] + v0[1] * v1[1]
+  const dot02 = v0[0] * v2[0] + v0[1] * v2[1]
+  const dot11 = v1[0] * v1[0] + v1[1] * v1[1]
+  const dot12 = v1[0] * v2[0] + v1[1] * v2[1]
+
+  const invDenom = 1 / (dot00 * dot11 - dot01 * dot01)
+  const u = (dot11 * dot02 - dot01 * dot12) * invDenom
+  const v = (dot00 * dot12 - dot01 * dot02) * invDenom
+
+  return u >= 0 && v >= 0 && u + v < 1
+}

web/src/lib/useMuuriGrid.js

@@ -1,6 +1,7 @@
 import { useEffect } from 'react'
 import Muuri from 'muuri'
 import { idToSize, idToZIndex, positionToId } from '../ui/editor/Puzzle'
+import { pointInTriangle } from './position'
 
 const useMuuriGrid = (gridRef, options, size, props) => {
   useEffect(() => {
@@ -9,7 +10,29 @@ const useMuuriGrid = (gridRef, options, size, props) => {
 
       const grid = new Muuri(gridRef.current, options)
       grid.size = size
+      grid.on('dragStart', (item, event) => {
+        console.log('DRAG START', item, event)
+        const element = item.getElement()
+        const rect = element.getBoundingClientRect()
+
+        // Triangle vertices based on the rectangle
+        const ax = rect.left
+        const ay = rect.bottom
+        const bx = rect.right
+        const by = rect.bottom
+        const cx = rect.left + rect.width / 2
+        const cy = rect.top
+
+        // Get the mouse position
+        const mouseX = event.clientX
+        const mouseY = event.clientY
 
+        // Check if the click is inside the triangle
+        if (!pointInTriangle(mouseX, mouseY, ax, ay, bx, by, cx, cy)) {
+          event.srcEvent.preventDefault()
+          return
+        }
+      })
       grid.on('dragEnd', (item, event) => {
         console.log('DRAG RELEASE END', item, event)
 
@@ -30,6 +53,18 @@ const useMuuriGrid = (gridRef, options, size, props) => {
           props.action(newId)
           return
         }
+
+        // Find the triangle-text sub-divs
+        const elementTriangleText = element.querySelector('.triangle-text')
+        const targetTriangleText = targetElement.querySelector('.triangle-text')
+
+        // Swap their font sizes
+        if (elementTriangleText && targetTriangleText) {
+          const tempFontSize = elementTriangleText.style.fontSize
+          elementTriangleText.style.fontSize = targetTriangleText.style.fontSize
+          targetTriangleText.style.fontSize = tempFontSize
+        }
+
         targetElement.setAttribute('id', oldId)
         element.setAttribute('id', newId)
 
@@ -38,8 +73,10 @@ const useMuuriGrid = (gridRef, options, size, props) => {
         element.style.width = idToSize(newId)
         element.style.height = idToSize(newId)
 
-        targetElement.style.zIndex = idToZIndex(oldId)
-        element.style.zIndex = idToZIndex(newId)
+        const oldZIndex = targetElement.style.zIndex
+
+        targetElement.style.zIndex = element.style.zIndex
+        element.style.zIndex = oldZIndex
 
         props.socket.timeout(3000).emit(
           'save_params',
@@ -67,6 +104,9 @@ const useMuuriGrid = (gridRef, options, size, props) => {
       return () => {
         // Remove event listener for dragReleaseEnd event
         grid.off('dragReleaseEnd')
+
+        grid.off('dragStart')
+        grid.off('dragEnd')
         try {
           grid.destroy()
         } catch (e) {