Update app.py

app.py

@@ -291,50 +291,33 @@ def extract_text_from_pdf(uploaded_pdf) -> str:
                 text_parts.append(txt)
     return "\n".join(text_parts)
 
+def find_float_after_label(text: str, label: str, max_numbers: int = 2, window: int = 1200):
+    """
+    Cherche un label dans le texte extrait puis récupère les premiers nombres après ce label.
+    Version tolérante aux retours ligne / espaces / accents du PDF Zebris.
+    """
+    if not text or not label:
+        return []
+
+    text_norm = text.lower().replace("\xa0", " ")
+    label_norm = label.lower().replace("\xa0", " ")
 
-def find_float_after_label(text: str, label: str, max_numbers: int = 2):
-    idx = text.lower().find(label.lower())
+    idx = text_norm.find(label_norm)
     if idx == -1:
         return []
-    snippet = text[idx: idx + 500]
+
+    snippet = text[idx: idx + window]
+
     nums = re.findall(r"(\d+,\d+|\d+\.\d+|\d+)", snippet)
     out = []
     for n in nums[:max_numbers]:
-        out.append(float(n.replace(",", ".")))
+        try:
+            out.append(float(n.replace(",", ".")))
+        except Exception:
+            pass
     return out
 
 
-def extract_pdf_name(text: str):
-    m = re.search(r"Personne:\s*([A-Za-zÀ-ÿ\- ]+),\s*\d{2}/\d{2}/\d{4}", text)
-    if m:
-        return m.group(1).strip()
-    return None
-
-
-def estimate_attack_from_pdf(pdf_data: dict):
-    heel_peak_t = avg(pdf_data.get("heel_peak_time_pct_g"), pdf_data.get("heel_peak_time_pct_d"))
-    fore_peak_t = avg(pdf_data.get("fore_peak_time_pct_g"), pdf_data.get("fore_peak_time_pct_d"))
-    heel_force = avg(pdf_data.get("heel_force_g"), pdf_data.get("heel_force_d"))
-    fore_force = avg(pdf_data.get("fore_force_g"), pdf_data.get("fore_force_d"))
-    transition = avg(pdf_data.get("transition_g"), pdf_data.get("transition_d"))
-
-    if pd.notna(heel_peak_t) and pd.notna(fore_peak_t):
-        if heel_peak_t <= 12 and pd.notna(transition) and transition >= 0.055:
-            return "attaque talon"
-        if heel_peak_t > 15 and fore_peak_t < 50:
-            return "attaque avant-pied"
-
-    if pd.notna(heel_force) and pd.notna(fore_force) and fore_force != 0:
-        ratio = heel_force / fore_force
-        if ratio > 1.05 and pd.notna(transition) and transition >= 0.055:
-            return "attaque talon"
-        if ratio < 0.90 and pd.notna(transition) and transition <= 0.055:
-            return "attaque avant-pied"
-        return "attaque médio-pied"
-
-    return "indéterminée"
-
-
 def parse_zebris_pdf(uploaded_pdf):
     text = extract_text_from_pdf(uploaded_pdf)
     athlete_name = extract_pdf_name(text)
@@ -364,53 +347,195 @@ def parse_zebris_pdf(uploaded_pdf):
         "fore_peak_time_pct_d": np.nan,
     }
 
-    vals = find_float_after_label(text, "Instant du passage du talon vers\nl'avant-pied, s", 2)
+    # =========================
+    # TRANSITION talon -> avant-pied (s)
+    # =========================
+    vals = find_float_after_label(
+        text,
+        "Instant du passage du talon vers\nl'avant-pied, s",
+        max_numbers=2
+    )
     if len(vals) >= 2:
         data["transition_g"], data["transition_d"] = vals[0], vals[1]
+    else:
+        vals = find_float_after_label(
+            text,
+            "Instant du passage du talon vers l'avant-pied, s",
+            max_numbers=2
+        )
+        if len(vals) >= 2:
+            data["transition_g"], data["transition_d"] = vals[0], vals[1]
 
-    vals = find_float_after_label(text, "Force maximale, N\nForefoot (Three zones)", 2)
+    # =========================
+    # FORCE maximale, N
+    # =========================
+    vals = find_float_after_label(text, "Force maximale, N\nForefoot (Three zones)", max_numbers=2)
     if len(vals) >= 2:
         data["fore_force_g"], data["fore_force_d"] = vals[0], vals[1]
+    else:
+        vals = find_float_after_label(text, "Forefoot (Three zones)", max_numbers=2)
+        if len(vals) >= 2:
+            data["fore_force_g"], data["fore_force_d"] = vals[0], vals[1]
 
-    vals = find_float_after_label(text, "Midfoot (Three zones)", 2)
+    vals = find_float_after_label(text, "Midfoot (Three zones)", max_numbers=4)
     if len(vals) >= 2:
-        # attention : la première occurrence de Midfoot peut venir de la force ou d'une autre section
-        if pd.isna(data["mid_force_g"]):
-            data["mid_force_g"], data["mid_force_d"] = vals[0], vals[1]
+        # On prend les 2 premières valeurs trouvées après Midfoot
+        data["mid_force_g"], data["mid_force_d"] = vals[0], vals[1]
 
-    vals = find_float_after_label(text, "Heel (Three zones)", 2)
+    vals = find_float_after_label(text, "Heel (Three zones)", max_numbers=6)
     if len(vals) >= 2:
-        if pd.isna(data["heel_force_g"]):
-            data["heel_force_g"], data["heel_force_d"] = vals[0], vals[1]
-
-    vals = find_float_after_label(text, "Pression maximale, N/cm²\nForefoot (Three zones)", 2)
+        # Sur cette zone, le mot Heel revient plusieurs fois dans la page.
+        # Les 2 premières valeurs après le premier bloc correspondent bien ici aux forces.
+        data["heel_force_g"], data["heel_force_d"] = vals[0], vals[1]
+
+    # =========================
+    # PRESSION maximale, N/cm²
+    # =========================
+    vals = find_float_after_label(text, "Pression maximale, N/cm²\nForefoot (Three zones)", max_numbers=2)
     if len(vals) >= 2:
         data["fore_pressure_g"], data["fore_pressure_d"] = vals[0], vals[1]
-
-    vals = find_float_after_label(text, "Pression maximale, N/cm²\nMidfoot (Three zones)", 2)
+    else:
+        # fallback plus tolérant : on repart de "Pression maximale"
+        vals = find_float_after_label(text, "Pression maximale, N/cm²", max_numbers=8)
+        # ordre attendu dans ce PDF :
+        # fore G, fore D, mid G, mid D, heel G, heel D
+        if len(vals) >= 6:
+            data["fore_pressure_g"], data["fore_pressure_d"] = vals[0], vals[1]
+            data["mid_pressure_g"], data["mid_pressure_d"] = vals[2], vals[3]
+            data["heel_pressure_g"], data["heel_pressure_d"] = vals[4], vals[5]
+
+    if pd.isna(data["mid_pressure_g"]):
+        vals = find_float_after_label(text, "Pression maximale, N/cm²\nMidfoot (Three zones)", max_numbers=2)
+        if len(vals) >= 2:
+            data["mid_pressure_g"], data["mid_pressure_d"] = vals[0], vals[1]
+
+    if pd.isna(data["heel_pressure_g"]):
+        vals = find_float_after_label(text, "Pression maximale, N/cm²\nHeel (Three zones)", max_numbers=2)
+        if len(vals) >= 2:
+            data["heel_pressure_g"], data["heel_pressure_d"] = vals[0], vals[1]
+
+    # =========================
+    # TIMING pic de force, % phase d'appui
+    # =========================
+    vals = find_float_after_label(
+        text,
+        "Instant pic de force, % de phase d'appui\nForefoot (Three zones)",
+        max_numbers=2
+    )
     if len(vals) >= 2:
-        data["mid_pressure_g"], data["mid_pressure_d"] = vals[0], vals[1]
+        data["fore_peak_time_pct_g"], data["fore_peak_time_pct_d"] = vals[0], vals[1]
+    else:
+        vals = find_float_after_label(text, "Instant pic de force, % de phase d'appui", max_numbers=8)
+        # ordre attendu : fore G/D, mid G/D, heel G/D
+        if len(vals) >= 6:
+            data["fore_peak_time_pct_g"], data["fore_peak_time_pct_d"] = vals[0], vals[1]
+            data["mid_peak_time_pct_g"], data["mid_peak_time_pct_d"] = vals[2], vals[3]
+            data["heel_peak_time_pct_g"], data["heel_peak_time_pct_d"] = vals[4], vals[5]
+
+    if pd.isna(data["mid_peak_time_pct_g"]):
+        vals = find_float_after_label(
+            text,
+            "Instant pic de force, % de phase d'appui\nMidfoot (Three zones)",
+            max_numbers=2
+        )
+        if len(vals) >= 2:
+            data["mid_peak_time_pct_g"], data["mid_peak_time_pct_d"] = vals[0], vals[1]
+
+    if pd.isna(data["heel_peak_time_pct_g"]):
+        vals = find_float_after_label(
+            text,
+            "Instant pic de force, % de phase d'appui\nHeel (Three zones)",
+            max_numbers=2
+        )
+        if len(vals) >= 2:
+            data["heel_peak_time_pct_g"], data["heel_peak_time_pct_d"] = vals[0], vals[1]
+
+    # =========================
+    # FALLBACK spécial adapté à ton PDF page 8
+    # =========================
+    if (
+        pd.isna(data["transition_g"]) or
+        pd.isna(data["fore_force_g"]) or
+        pd.isna(data["fore_pressure_g"]) or
+        pd.isna(data["fore_peak_time_pct_g"])
+    ):
+        block_match = re.search(
+            r"Instant du passage du talon vers.*?Heel \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Force maximale, N.*?"
+            r"Forefoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Midfoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Heel \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Pression maximale, N/cm².*?"
+            r"Forefoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Midfoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Heel \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Instant pic de force, % de phase d'appui.*?"
+            r"Forefoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Midfoot \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±.*?"
+            r"Heel \(Three zones\)\s+Gauche\s+Droite\s+(\d+,\d+|\d+\.\d+)\±.*?(\d+,\d+|\d+\.\d+)\±",
+            text,
+            flags=re.S
+        )
 
-    vals = find_float_after_label(text, "Pression maximale, N/cm²\nHeel (Three zones)", 2)
-    if len(vals) >= 2:
-        data["heel_pressure_g"], data["heel_pressure_d"] = vals[0], vals[1]
+        if block_match:
+            vals = [float(x.replace(",", ".")) for x in block_match.groups()]
 
-    vals = find_float_after_label(text, "Instant pic de force, % de phase d'appui\nForefoot (Three zones)", 2)
-    if len(vals) >= 2:
-        data["fore_peak_time_pct_g"], data["fore_peak_time_pct_d"] = vals[0], vals[1]
+            data["transition_g"], data["transition_d"] = vals[0], vals[1]
 
-    vals = find_float_after_label(text, "Instant pic de force, % de phase d'appui\nMidfoot (Three zones)", 2)
-    if len(vals) >= 2:
-        data["mid_peak_time_pct_g"], data["mid_peak_time_pct_d"] = vals[0], vals[1]
+            data["fore_force_g"], data["fore_force_d"] = vals[2], vals[3]
+            data["mid_force_g"], data["mid_force_d"] = vals[4], vals[5]
+            data["heel_force_g"], data["heel_force_d"] = vals[6], vals[7]
 
-    vals = find_float_after_label(text, "Instant pic de force, % de phase d'appui\nHeel (Three zones)", 2)
-    if len(vals) >= 2:
-        data["heel_peak_time_pct_g"], data["heel_peak_time_pct_d"] = vals[0], vals[1]
+            data["fore_pressure_g"], data["fore_pressure_d"] = vals[8], vals[9]
+            data["mid_pressure_g"], data["mid_pressure_d"] = vals[10], vals[11]
+            data["heel_pressure_g"], data["heel_pressure_d"] = vals[12], vals[13]
+
+            data["fore_peak_time_pct_g"], data["fore_peak_time_pct_d"] = vals[14], vals[15]
+            data["mid_peak_time_pct_g"], data["mid_peak_time_pct_d"] = vals[16], vals[17]
+            data["heel_peak_time_pct_g"], data["heel_peak_time_pct_d"] = vals[18], vals[19]
 
     data["attaque_pdf"] = estimate_attack_from_pdf(data)
     return data
 
 
+def extract_pdf_name(text: str):
+    patterns = [
+        r"Personne:\s*([A-Za-zÀ-ÿ\- ]+),\s*\d{2}/\d{2}/\d{4}",
+        r"Personne:\s*([A-Za-zÀ-ÿ\- ]+)",
+    ]
+
+    for pattern in patterns:
+        m = re.search(pattern, text)
+        if m:
+            return m.group(1).strip()
+
+    return None
+
+
+def estimate_attack_from_pdf(pdf_data: dict):
+    heel_peak_t = avg(pdf_data.get("heel_peak_time_pct_g"), pdf_data.get("heel_peak_time_pct_d"))
+    fore_peak_t = avg(pdf_data.get("fore_peak_time_pct_g"), pdf_data.get("fore_peak_time_pct_d"))
+    heel_force = avg(pdf_data.get("heel_force_g"), pdf_data.get("heel_force_d"))
+    fore_force = avg(pdf_data.get("fore_force_g"), pdf_data.get("fore_force_d"))
+    transition = avg(pdf_data.get("transition_g"), pdf_data.get("transition_d"))
+
+    if pd.notna(heel_peak_t) and pd.notna(fore_peak_t):
+        if heel_peak_t <= 12 and pd.notna(transition) and transition >= 0.055:
+            return "attaque talon"
+        if heel_peak_t > 15 and fore_peak_t < 50:
+            return "attaque avant-pied"
+
+    if pd.notna(heel_force) and pd.notna(fore_force) and fore_force != 0:
+        ratio = heel_force / fore_force
+        if ratio > 1.05 and pd.notna(transition) and transition >= 0.055:
+            return "attaque talon"
+        if ratio < 0.90 and pd.notna(transition) and transition <= 0.055:
+            return "attaque avant-pied"
+        return "attaque médio-pied"
+
+    return "indéterminée"
+
+
 def match_pdf_to_athlete(pdfs_data, athlete_name):
     target = normalize_name(athlete_name)
     target_parts = set(target.split())