Update app.py

app.py

@@ -6,12 +6,12 @@ def extrair_exames_formatado(pdf_file):
     if pdf_file is None:
         return "Nenhum arquivo enviado."
 
-    doc = fitz.open(pdf_file.name)
     texto = ""
-    for page in doc:
-        texto += page.get_text()
+    with fitz.open(pdf_file.name) as doc:
+        for page in doc:
+            texto += page.get_text()
 
-    def extrair_valor(padrao):
+    def buscar(padrao):
         match = re.search(padrao, texto, re.IGNORECASE)
         if match:
             return match.group(1).replace(",", ".").strip()
@@ -20,90 +20,89 @@ def extrair_exames_formatado(pdf_file):
     def k_format(v):
         try:
             n = float(v.replace(".", "").replace(",", "."))
-            if n >= 1000:
-                return f"{round(n / 1000, 1)}K"
-            return str(n)
+            return f"{round(n / 1000, 1)}K" if n >= 1000 else str(n)
         except:
             return v
 
-    eas_leuc = re.search(r"leucócitos[\s\n]*(\d+-\d+)", texto, re.IGNORECASE)
-    eas_hem = re.search(r"hem[áa]cias[\s\n]*(\d+-\d+)", texto, re.IGNORECASE)
-    eas_bact = re.search(r"bact[ée]rias[\s\n]*(\w+)", texto, re.IGNORECASE)
-    eas_final = ""
-    if eas_leuc and eas_hem and eas_bact:
-        eas_final = f"EAS: {eas_leuc.group(1)} LEUCÓCITOS + {eas_hem.group(1)} HEMÁCIAS + BACTÉRIAS {eas_bact.group(1).upper()} //"
+    # Coleta combinada de Leucócitos, Bastonetes e Segmentados
+    leuco = buscar(r"leuc[óo]citos\s+([\d.,]+)")
+    bastonetes = buscar(r"bastonetes\s+(\d+)\s*%")
+    segmentados = buscar(r"segmentados\s+(\d+)\s*%")
 
-    campos = [
-        ("LAC", r"lactato[\s\n]+([\d,\.]+)"),
-        ("AMIL", r"amilase[\s\n]+([\d,\.]+)"),
-        ("AC UR", r"[áa]cido[ \n]+[úu]rico[\s\n]+([\d,\.]+)"),
-        ("BT", r"bilirrubina total[\s\n]+([\d,\.]+)"),
-        ("BD", r"bilirrubina direta[\s\n]+([\d,\.]+)"),
-        ("BI", r"bilirrubina indireta[\s\n]+([\d,\.]+)"),
-        ("CAI", r"ioniz[áa]vel[\s\n]+([\d,\.]+)"),
-        ("CL-", r"cl[óo]ro[\s\n]+([\d,\.]+)"),
-        ("CR", r"creatinina[\s\n]+([\d,\.]+)"),
-        ("FAL", r"fosfatase alcalina[\s\n]+([\d,\.]+)"),
-        ("P", r"f[óo]sforo[\s\n]+([\d,\.]+)"),
-        ("GGT", r"gama.?gt[\s\n]+([\d,\.]+)"),
-        ("GLI", r"glicose[\s\n]+([\d,\.]+)"),
-        ("HB", r"hemoglobina[\s\n]+([\d,\.]+)"),
-        ("HT", r"hemat[óo]crito[\s\n]+([\d,\.]+)"),
-        ("LEUCO", r"leuc[óo]citos[\s\n]+([\d,\.]+)"),
-        ("B", r"bastonetes[\s\n]+([\d,\.]+)"),
-        ("SS", r"segmentados[\s\n]+([\d,\.]+)"),
-        ("PLT", r"plaquetas[\s\n]+([\d,\.]+)"),
-        ("LIP", r"lipase[\s\n]+([\d,\.]+)"),
-        ("MG++", r"magn[ée]sio[\s\n]+([\d,\.]+)"),
-        ("PCR", r"PCR[\s\n]+([\d,\.]+)"),
-        ("K+", r"pot[áa]ssio[\s\n]+([\d,\.]+)"),
-        ("PTN", r"prote[ií]na[s]? totais?[\s\n]+([\d,\.]+)"),
-        ("ALB", r"albumina[\s\n]+([\d,\.]+)"),
-        ("NA+", r"s[óo]dio[\s\n]+([\d,\.]+)"),
-        ("TGO", r"TGO[\s\n]+([\d,\.]+)"),
-        ("TGP", r"TGP[\s\n]+([\d,\.]+)"),
-        ("TAP", r"TAP.*?([\d,\.]+)"),
-        ("INR", r"INR.*?([\d,\.]+)"),
-        ("TTP", r"TTP.*?([\d,\.]+)"),
-        ("RELAÇÃO", r"relaç[aã]o.*?([\d,\.]+)"),
-        ("UR", r"ureia[\s\n]+([\d,\.]+)"),
-        ("CPK", r"CPK.*?([\d,\.]+)"),
-        ("CKMB", r"CKMB.*?([\d,\.]+)"),
-        ("TROPO", r"tropo.*?([<>=]?[\d,\.]+)")
-    ]
+    leuco_str = ""
+    if leuco:
+        leuco_str = f"LEUCO {k_format(leuco)}"
+        if bastonetes:
+            leuco_str += f" + {bastonetes}% B"
+        if segmentados:
+            leuco_str += f" + {segmentados}% SS"
 
-    output = [eas_final] if eas_final else []
-    temp_dict = {}
+    exames = {
+        "🟠 Renal / Eletrólitos": {
+            "UREIA": buscar(r"ureia\s+([\d.,]+)"),
+            "CR": buscar(r"creatinina\s+([\d.,]+)"),
+            "K+": buscar(r"pot[áa]ssio\s+([\d.,]+)"),
+            "NA+": buscar(r"s[óo]dio\s+([\d.,]+)"),
+            "CL-": buscar(r"cl[óo]ro\s+([\d.,]+)"),
+            "CAI": buscar(r"ioniz[áa]vel\s+([\d.,]+)"),
+            "CA TOTAL": buscar(r"c[áa]lcio total\s+([\d.,]+)"),
+            "MG++": buscar(r"magn[ée]sio\s+([\d.,]+)"),
+            "FÓS": buscar(r"f[óo]sforo\s+([\d.,]+)"),
+        },
+        "🟡 Hepático": {
+            "BT": buscar(r"bilirrubina total\s+([\d.,]+)"),
+            "BD": buscar(r"bilirrubina direta\s+([\d.,]+)"),
+            "BI": buscar(r"bilirrubina indireta\s+([\d.,]+)"),
+            "TGO": buscar(r"TGO\s+([\d.,]+)"),
+            "TGP": buscar(r"TGP\s+([\d.,]+)"),
+            "GGT": buscar(r"gama.?gt\s+([\d.,]+)"),
+            "FAL": buscar(r"fosfatase alcalina\s+([\d.,]+)"),
+            "ALB": buscar(r"albumina\s+([\d.,]+)"),
+            "PTN TOTAL": buscar(r"prote[ií]nas totais\s+([\d.,]+)"),
+            "GLOB": buscar(r"globulina\s+([\d.,]+)"),
+            "RELAÇÃO": buscar(r"relaç[aã]o A/G\s+([\d.,]+)")
+        },
+        "🔴 Hematológico": {
+            "HB": buscar(r"hemoglobina\s+([\d.,]+)"),
+            "HT": buscar(r"hemat[óo]crito\s+([\d.,]+)"),
+            "PLT": buscar(r"plaquetas\s+([\d.,]+)")
+        },
+        "🔵 Coagulação": {
+            "TAP": buscar(r"TEMPO\s+14,4|TAP.*?([\d.,]+)"),
+            "INR": buscar(r"INR\s*[:\-]?\s*([\d.,]+)"),
+            "TTP": buscar(r"TTPA.*?tempo\s+([\d.,]+)"),
+            "RELAÇÃO": buscar(r"relaç[aã]o.*?([\d.,]+)")
+        },
+        "🟢 Metabólico": {
+            "GLI": buscar(r"glicose\s+([\d.,]+)"),
+            "LIP": buscar(r"lipase\s+([\d.,]+)"),
+            "AMIL": buscar(r"amilase\s+([\d.,]+)"),
+            "AC UR": buscar(r"[áa]cido[ \n]+[úu]rico\s+([\d.,]+)"),
+            "LAC": buscar(r"lactato.*?([\d.,]+)"),
+            "PCR": buscar(r"PCR.*?([\d.,]+)")
+        }
+    }
 
-    for rotulo, regex in campos:
-        val = extrair_valor(regex)
-        if val:
-            temp_dict[rotulo] = val
+    output = []
+    if leuco_str:
+        output.append(leuco_str)
+    for sistema, dados in exames.items():
+        linha = []
+        for k, v in dados.items():
+            if v:
+                linha.append(f"{k}: {v}")
+        if linha:
+            output.append(f"{sistema} → " + " / ".join(linha))
 
-    if "LEUCO" in temp_dict:
-        l = k_format(temp_dict["LEUCO"])
-        b = f"{temp_dict['B']}% B" if "B" in temp_dict else ""
-        ss = f"{temp_dict['SS']}% SS" if "SS" in temp_dict else ""
-        combo = f"LEUCO {l}" + (f" + {b}" if b else "") + (f" + {ss}" if ss else "")
-        output.append(combo)
-        temp_dict.pop("LEUCO")
-        temp_dict.pop("B", None)
-        temp_dict.pop("SS", None)
-
-    ordem = [k for k, _ in campos if k not in ["LEUCO", "B", "SS"]]
-    for campo in ordem:
-        if campo in temp_dict:
-            output.append(f"{campo} {temp_dict[campo]}")
-
-    return " / ".join(output) if output else "Nenhum dado encontrado."
+    return "\n".join(output) if output else "Nenhum dado encontrado."
 
+# Gradio Interface
 with gr.Blocks() as demo:
-    gr.Markdown("## Extrator de Exames - PDF Clínico Compactado")
+    gr.Markdown("## 🧪 Extrator de Exames - PDF Clínico Compactado")
     with gr.Row():
-        pdf = gr.File(label="PDF de exames", file_types=[".pdf"])
-        btn = gr.Button("Extrair")
-
-    resultado = gr.Textbox(label="Exames extraídos", lines=4)
+        pdf = gr.File(label="📄 PDF de exames", file_types=[".pdf"])
+        btn = gr.Button("🔍 Extrair Exames")
+    resultado = gr.Textbox(label="📋 Resultados Extraídos", lines=20)
     btn.click(fn=extrair_exames_formatado, inputs=pdf, outputs=resultado)
 
 demo.launch()