Create morphosyntax_interface_vOk-30-12-24.py
Browse files
modules/morphosyntax/morphosyntax_interface_vOk-30-12-24.py
ADDED
@@ -0,0 +1,247 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# modules/morphosyntax/morphosyntax_interface.py
|
2 |
+
|
3 |
+
import streamlit as st
|
4 |
+
import re
|
5 |
+
import logging
|
6 |
+
from spacy import displacy
|
7 |
+
|
8 |
+
# Se asume que la funci贸n perform_advanced_morphosyntactic_analysis
|
9 |
+
# y los m茅todos store_student_morphosyntax_base/iteration existen.
|
10 |
+
from ..morphosyntax.morphosyntax_process import perform_advanced_morphosyntactic_analysis
|
11 |
+
from ..database.morphosyntax_iterative_mongo_db import (
|
12 |
+
store_student_morphosyntax_base,
|
13 |
+
store_student_morphosyntax_iteration,
|
14 |
+
)
|
15 |
+
|
16 |
+
logger = logging.getLogger(__name__)
|
17 |
+
|
18 |
+
###########################################################################
|
19 |
+
def initialize_arc_analysis_state():
|
20 |
+
"""
|
21 |
+
Inicializa el estado de an谩lisis de arcos (base e iteraciones) si no existe.
|
22 |
+
"""
|
23 |
+
if "arc_analysis_state" not in st.session_state:
|
24 |
+
st.session_state.arc_analysis_state = {
|
25 |
+
"base_id": None,
|
26 |
+
"base_text": "",
|
27 |
+
"base_diagram": None,
|
28 |
+
"iteration_text": "",
|
29 |
+
"iteration_diagram": None,
|
30 |
+
}
|
31 |
+
logger.info("Estado de an谩lisis de arcos inicializado.")
|
32 |
+
|
33 |
+
###########################################################################
|
34 |
+
def reset_arc_analysis_state():
|
35 |
+
"""
|
36 |
+
Resetea completamente el estado de an谩lisis de arcos.
|
37 |
+
"""
|
38 |
+
st.session_state.arc_analysis_state = {
|
39 |
+
"base_id": None,
|
40 |
+
"base_text": "",
|
41 |
+
"base_diagram": None,
|
42 |
+
"iteration_text": "",
|
43 |
+
"iteration_diagram": None,
|
44 |
+
}
|
45 |
+
logger.info("Estado de arcos reseteado.")
|
46 |
+
|
47 |
+
###########################################################################
|
48 |
+
def display_arc_diagram(doc):
|
49 |
+
"""
|
50 |
+
Genera y retorna el HTML del diagrama de arco para un `Doc` de spaCy.
|
51 |
+
No imprime directamente en pantalla; regresa el HTML para
|
52 |
+
usar con `st.write(..., unsafe_allow_html=True)`.
|
53 |
+
"""
|
54 |
+
try:
|
55 |
+
diagram_html = ""
|
56 |
+
for sent in doc.sents:
|
57 |
+
svg_html = displacy.render(
|
58 |
+
sent,
|
59 |
+
style="dep",
|
60 |
+
options={
|
61 |
+
"distance": 100,
|
62 |
+
"arrow_spacing": 20,
|
63 |
+
"word_spacing": 30
|
64 |
+
}
|
65 |
+
)
|
66 |
+
# Ajustar tama帽os
|
67 |
+
svg_html = svg_html.replace('height="375"', 'height="200"')
|
68 |
+
svg_html = re.sub(
|
69 |
+
r'<svg[^>]*>',
|
70 |
+
lambda m: m.group(0).replace('height="450"', 'height="300"'),
|
71 |
+
svg_html
|
72 |
+
)
|
73 |
+
svg_html = re.sub(
|
74 |
+
r'<g [^>]*transform="translate\((\d+),(\d+)\)"',
|
75 |
+
lambda m: f'<g transform="translate({m.group(1)},50)"',
|
76 |
+
svg_html
|
77 |
+
)
|
78 |
+
# Envolver en contenedor
|
79 |
+
diagram_html += f'<div class="arc-diagram-container">{svg_html}</div>'
|
80 |
+
return diagram_html
|
81 |
+
|
82 |
+
except Exception as e:
|
83 |
+
logger.error(f"Error en display_arc_diagram: {str(e)}")
|
84 |
+
return "<p style='color:red;'>Error generando diagrama</p>"
|
85 |
+
|
86 |
+
###########################################################################
|
87 |
+
def display_morphosyntax_interface(lang_code, nlp_models, morpho_t):
|
88 |
+
"""
|
89 |
+
Interfaz principal para la visualizaci贸n de diagramas de arco
|
90 |
+
(Texto Base vs Iteraciones).
|
91 |
+
"""
|
92 |
+
# CSS para layout vertical y estable
|
93 |
+
st.markdown("""
|
94 |
+
<style>
|
95 |
+
.stTextArea textarea {
|
96 |
+
font-size: 1rem;
|
97 |
+
line-height: 1.5;
|
98 |
+
min-height: 100px !important;
|
99 |
+
height: 100px !important;
|
100 |
+
}
|
101 |
+
.arc-diagram-container {
|
102 |
+
width: 100%;
|
103 |
+
padding: 0.5rem;
|
104 |
+
margin: 0.5rem 0;
|
105 |
+
}
|
106 |
+
.divider {
|
107 |
+
height: 3px;
|
108 |
+
border: none;
|
109 |
+
background-color: #333;
|
110 |
+
margin: 2rem 0;
|
111 |
+
}
|
112 |
+
</style>
|
113 |
+
""", unsafe_allow_html=True)
|
114 |
+
|
115 |
+
# 1) Inicializar estados
|
116 |
+
initialize_arc_analysis_state()
|
117 |
+
arc_state = st.session_state.arc_analysis_state
|
118 |
+
|
119 |
+
# 2) Creamos pesta帽as: "Texto Base" y "Iteraciones"
|
120 |
+
tabs = st.tabs(["Texto Base", "Iteraciones"])
|
121 |
+
|
122 |
+
# =================== PESTA脩A 1: Texto Base ==========================
|
123 |
+
with tabs[0]:
|
124 |
+
st.subheader("An谩lisis de Texto Base")
|
125 |
+
|
126 |
+
# Bot贸n para iniciar nuevo an谩lisis
|
127 |
+
if st.button("Nuevo An谩lisis", key="btn_reset_base"):
|
128 |
+
# Solo limpiamos el estado; si requieres forzar reload,
|
129 |
+
# descomenta la siguiente l铆nea:
|
130 |
+
# st.experimental_rerun()
|
131 |
+
reset_arc_analysis_state()
|
132 |
+
|
133 |
+
# Textarea de texto base
|
134 |
+
arc_state["base_text"] = st.text_area(
|
135 |
+
"Ingrese su texto inicial",
|
136 |
+
value=arc_state["base_text"],
|
137 |
+
key="base_text_input",
|
138 |
+
height=150
|
139 |
+
)
|
140 |
+
|
141 |
+
# Bot贸n para analizar texto base
|
142 |
+
if st.button("Analizar Texto Base", key="btn_analyze_base"):
|
143 |
+
if not arc_state["base_text"].strip():
|
144 |
+
st.warning("Ingrese un texto para analizar.")
|
145 |
+
else:
|
146 |
+
try:
|
147 |
+
# Procesar con spaCy
|
148 |
+
doc = nlp_models[lang_code](arc_state["base_text"])
|
149 |
+
# Generar HTML del arco
|
150 |
+
arc_html = display_arc_diagram(doc)
|
151 |
+
arc_state["base_diagram"] = arc_html
|
152 |
+
|
153 |
+
# Guardar en Mongo
|
154 |
+
analysis = perform_advanced_morphosyntactic_analysis(
|
155 |
+
arc_state["base_text"],
|
156 |
+
nlp_models[lang_code]
|
157 |
+
)
|
158 |
+
base_id = store_student_morphosyntax_base(
|
159 |
+
username=st.session_state.username,
|
160 |
+
text=arc_state["base_text"],
|
161 |
+
arc_diagrams=analysis["arc_diagrams"]
|
162 |
+
)
|
163 |
+
if base_id:
|
164 |
+
arc_state["base_id"] = base_id
|
165 |
+
st.success(f"An谩lisis base guardado. ID: {base_id}")
|
166 |
+
|
167 |
+
except Exception as exc:
|
168 |
+
st.error("Error procesando texto base")
|
169 |
+
logger.error(f"Error en an谩lisis base: {str(exc)}")
|
170 |
+
|
171 |
+
# Mostrar diagrama base
|
172 |
+
if arc_state["base_diagram"]:
|
173 |
+
st.markdown("<hr class='divider'>", unsafe_allow_html=True)
|
174 |
+
st.markdown("#### Diagrama de Arco (Texto Base)")
|
175 |
+
st.write(arc_state["base_diagram"], unsafe_allow_html=True)
|
176 |
+
|
177 |
+
# ================== PESTA脩A 2: Iteraciones ==========================
|
178 |
+
with tabs[1]:
|
179 |
+
st.subheader("An谩lisis de Cambios / Iteraciones")
|
180 |
+
|
181 |
+
# Verificar que exista texto base analizado
|
182 |
+
if not arc_state["base_id"]:
|
183 |
+
st.info("Primero analiza un texto base en la pesta帽a anterior.")
|
184 |
+
return
|
185 |
+
|
186 |
+
# Mostrar texto base como referencia (solo lectura)
|
187 |
+
st.text_area(
|
188 |
+
"Texto Base (solo lectura)",
|
189 |
+
value=arc_state["base_text"],
|
190 |
+
height=80,
|
191 |
+
disabled=True
|
192 |
+
)
|
193 |
+
|
194 |
+
# Caja de texto para la iteraci贸n
|
195 |
+
arc_state["iteration_text"] = st.text_area(
|
196 |
+
"Texto de Iteraci贸n",
|
197 |
+
value=arc_state["iteration_text"],
|
198 |
+
height=150
|
199 |
+
)
|
200 |
+
|
201 |
+
# Bot贸n analizar iteraci贸n
|
202 |
+
if st.button("Analizar Cambios", key="btn_analyze_iteration"):
|
203 |
+
if not arc_state["iteration_text"].strip():
|
204 |
+
st.warning("Ingrese texto de iteraci贸n.")
|
205 |
+
else:
|
206 |
+
try:
|
207 |
+
# Procesar con spaCy
|
208 |
+
doc_iter = nlp_models[lang_code](arc_state["iteration_text"])
|
209 |
+
arc_html_iter = display_arc_diagram(doc_iter)
|
210 |
+
arc_state["iteration_diagram"] = arc_html_iter
|
211 |
+
|
212 |
+
# Guardar en Mongo
|
213 |
+
analysis_iter = perform_advanced_morphosyntactic_analysis(
|
214 |
+
arc_state["iteration_text"],
|
215 |
+
nlp_models[lang_code]
|
216 |
+
)
|
217 |
+
iteration_id = store_student_morphosyntax_iteration(
|
218 |
+
username=st.session_state.username,
|
219 |
+
base_id=arc_state["base_id"],
|
220 |
+
original_text=arc_state["base_text"],
|
221 |
+
iteration_text=arc_state["iteration_text"],
|
222 |
+
arc_diagrams=analysis_iter["arc_diagrams"]
|
223 |
+
)
|
224 |
+
if iteration_id:
|
225 |
+
st.success(f"Iteraci贸n guardada. ID: {iteration_id}")
|
226 |
+
|
227 |
+
except Exception as exc:
|
228 |
+
st.error("Error procesando iteraci贸n")
|
229 |
+
logger.error(f"Error en iteraci贸n: {str(exc)}")
|
230 |
+
|
231 |
+
# Mostrar diagrama de iteraci贸n
|
232 |
+
if arc_state["iteration_diagram"]:
|
233 |
+
st.markdown("<hr class='divider'>", unsafe_allow_html=True)
|
234 |
+
st.markdown("#### Diagrama de Arco (Iteraci贸n)")
|
235 |
+
st.write(arc_state["iteration_diagram"], unsafe_allow_html=True)
|
236 |
+
|
237 |
+
# Comparaci贸n vertical (uno abajo del otro)
|
238 |
+
if arc_state["base_diagram"] and arc_state["iteration_diagram"]:
|
239 |
+
st.markdown("<hr class='divider'>", unsafe_allow_html=True)
|
240 |
+
st.markdown("### Comparaci贸n Vertical: Base vs. Iteraci贸n")
|
241 |
+
|
242 |
+
st.markdown("**Diagrama Base**")
|
243 |
+
st.write(arc_state["base_diagram"], unsafe_allow_html=True)
|
244 |
+
|
245 |
+
st.markdown("---")
|
246 |
+
st.markdown("**Diagrama Iterado**")
|
247 |
+
st.write(arc_state["iteration_diagram"], unsafe_allow_html=True)
|