word stats now collected more memory-efficient

src/word_stats.py

@@ -2,7 +2,9 @@ from collections import defaultdict, Counter
 
 class WordStats:
     def __init__(self, lemma_features):
-        self.all_words = []
+        self.raw_data = defaultdict(lambda: defaultdict(list))
+        self.all_words = 0
+
         self.rendered_words = {}
         self.frequent_words = {}
         self.num_words = {}
@@ -11,18 +13,16 @@ class WordStats:
         self.memoized_msd_merges = {}
 
     def add_words(self, words):
-        self.all_words.extend(words)
+        for w in words:
+            self.raw_data[w.lemma][w.msd].append(w.text)
+        self.all_words += len(words)
 
     def num_all_words(self):
-        return len(self.all_words)
+        return self.all_words
 
     def generate_renders(self):
         num_words = defaultdict(int)
-        data = defaultdict(lambda: defaultdict(list))
-        for w in self.all_words:
-            data[w.lemma][w.msd].append(w.text)
-
-        for lemma, ld in data.items():
+        for lemma, ld in self.raw_data.items():
             self.rendered_words[lemma] = {}
             freq_words = defaultdict(int)
             common_msd = "*" * 10
@@ -45,6 +45,9 @@ class WordStats:
             for (msd, txt), n in sorted(freq_words.items(), key=lambda x: -x[1]):
                 self.frequent_words[lemma].append((msd, txt, n))
 
+        self.num_words = dict(num_words)
+
+        #next, determine lemma's default msds
         lf = self.lemma_features
         for lemma in self.lemma_msd:
             cmsd = self.lemma_msd[lemma]
@@ -52,8 +55,6 @@ class WordStats:
                 self.lemma_msd[lemma] = "".join(
                     l1 if l1 != "-" else l2 for l1, l2 in zip(lf[cmsd[0]], cmsd)
                 )
-        
-        self.num_words = dict(num_words)
 
     def merge_msd(self, common_msd, new_msd):
         key = (common_msd, new_msd)