added comments

feature_core.py

@@ -1,3 +1,4 @@
+# ONE OF THE CORE PROGRAMS OF THE PROJECT. REFERENCED BY feature_extractor and feature_extractor_web. 
 import re
 import subprocess
 import sys
@@ -50,7 +51,7 @@ def load_nlp_model(model_name: str = "tl_calamancy_md-0.2.0"):
 
     raise RuntimeError("Failed to load CalamanCy model. " + " | ".join(errors))
 
-
+# Merges sentences that contains dashes. Without this function, the model would split the sentence on every dash it encounters which is counterproductive.
 def merge_dash_sentences(doc) -> List:
     """Merge sentences split by dash tokens (from hyphenated words)."""
     dash_tokens = {"-"}
@@ -72,7 +73,7 @@ def merge_dash_sentences(doc) -> List:
             merged.append(sent)
     return merged
 
-
+# cleans the sentence, avoids misidentifying simple sentences as compound/complex
 def simple_clean(text: str) -> str:
     if not isinstance(text, str):
         return ""
@@ -81,7 +82,7 @@ def simple_clean(text: str) -> str:
     text = re.sub(r"[^\w\s\-.!?]", "", text)  # keep sentence-ending punctuation
     return text.strip()
 
-
+# gets the sentence, word, and token count
 def basic_counts(doc, original_text: str) -> Tuple[int, int, List]:
     tokens = [t for t in doc if not t.is_punct and not t.is_space]
     num_words = len(tokens)
@@ -106,29 +107,31 @@ def mean_lengths(tokens, num_words: int, num_sentences: int):
     mean_sentence_length = num_words / num_sentences if num_sentences else 0
     return round(mean_word_length, 4), round(mean_sentence_length, 4)
 
-
+# TTR. measures lexical diversity in a sample. Checks whether the vocabulary is rich or not. 
 def type_token_ratio(tokens, num_words: int):
     word_list = [t.text.lower() for t in tokens]
     return round(len(set(word_list)) / num_words if num_words else 0, 4)
 
-
 def count_filipino_syllables(word: str) -> int:
-    word = word.lower()
-    word = re.sub(r"[^a-z-]", "", word)
+    """Approximate Filipino syllable count by counting vowel nuclei."""
+    if not isinstance(word, str):
+        return 0
 
-    parts = word.split("-")
-    syllables = 0
-    for part in parts:
-        groups = re.findall(r"[aeiou]", part)
-        syllables += len(groups)
+    word = re.sub(r"[^a-z-]", "", word.lower())
+    if not word:
+        return 0
 
-    return syllables
+    syllables = 0
+    for part in filter(None, word.split("-")):
+        syllables += len(re.findall(r"[aeiou]", part))
 
+    return max(syllables, 1)
 
+# tags token that contains more than 3 syllables
 def polysyllabic_count(tokens) -> int:
     return sum(1 for t in tokens if count_filipino_syllables(t.text) >= 3)
 
-
+# Computes lexical density and part-of-speech ratios for the token list.
 def lexical_density_and_pos(tokens, num_words: int):
     content_pos = {"NOUN", "VERB", "ADJ", "ADV"}
     content_words = 0
@@ -153,7 +156,7 @@ def lexical_density_and_pos(tokens, num_words: int):
 
     return round(lexical_density, 4), pos_ratios
 
-
+# identifies foreign words by looking for letters foreign to the Filipino alphabet and computes its density.
 def foreign_word_density(tokens):
     english_ngrams = ["th", "ph", "sh", "ch", "wh", "ck", "qu"]
     foreign_letters = ["f", "v", "z", "x", "q", "j", "c"]
@@ -168,7 +171,7 @@ def foreign_word_density(tokens):
 
     return round(count / len(tokens) if tokens else 0, 4)
 
-
+# checks whether a sentence is a Subject-Verb-Object, or a Verb-Subject-Object
 def detect_svo_vso(doc):
     sentences = merge_dash_sentences(doc)
     if not sentences:
@@ -195,7 +198,7 @@ def detect_svo_vso(doc):
 
     return "Unknown"
 
-
+# detects keyword that identifies subordinate and coordinate clauses. Classifies the sentence based on whichever clause it has.
 def detect_sentence_type(doc):
     tokens = [t for t in doc if not t.is_punct and not t.is_space]
 
@@ -214,7 +217,7 @@ def detect_sentence_type(doc):
 
     return "Simple"
 
-
+# main func
 def extract_features(text: str, nlp) -> Dict[str, Any]:
     if not text or not isinstance(text, str):
         return {}

feature_extractor_web.py

@@ -1,3 +1,5 @@
+# USED BY THE WEB-APP FOR EXTRACTING FEATURES. 
+
 import logging
 from typing import Any, Dict
 

models/testing.py

@@ -1,3 +1,4 @@
+# THIS FILE IS NOT USED FOR THE CURRENT SYSTEM. ONLY USED FOR TRAINING EARLY VERSIONS OF THE SYSTEM.
 import pandas as pd
 
 df = pd.read_csv("corpus_clean.csv")

test_api.py

@@ -1,31 +0,0 @@
-import requests
-import json
-
-# Test health endpoint
-print("Testing health endpoint...")
-response = requests.get('http://localhost:5000/health')
-print(f"Health check: {response.json()}")
-print()
-
-# Test prediction
-print("Testing prediction...")
-test_texts = [
-    "Ang aso ay tumakbo sa parke.",
-    "Ang mga mag-aaral ay masigasig na nag-aaral para sa kanilang pagsusulit bukas.",
-    "Ang komprehensibong pagsusuri ng sosyo-ekonomikong kalagayan ay nagpapakita ng makabuluhang pagbabago."
-]
-
-for text in test_texts:
-    response = requests.post(
-        'http://localhost:5000/api/predict',
-        json={'text': text}
-    )
-    
-    if response.status_code == 200:
-        data = response.json()
-        print(f"\nText: {text[:50]}...")
-        print(f"Prediction: {data['prediction']['predicted_class']}")
-        print(f"Grade: {data['prediction']['grade_level']}")
-        print(f"Confidence: {max(data['prediction']['confidences'].values()):.3f}")
-    else:
-        print(f"Error: {response.json()}")