more fields for gynandco

etalab_data/gynadco/scrap.py

@@ -20,6 +20,15 @@ def extraire_numero_telephone(line):
     else:
         return None
 
+def extraire_code_postal(line):
+    # Extraction du code postal dans la ligne
+    match = re.search(r'\b(\d{5})\b', line)
+    if match:
+        # Extraction des nombres correspondant au code postal
+        code_postal = match.group(1)
+        return code_postal
+    else:
+        return None
 # trouver si la ligne est une adresse en cherchant deux numéros distincts et une virgule
 def extraire_addr_line(line):
     # Extraction des nombres dans la ligne
@@ -69,22 +78,42 @@ for article in articles:
      # Recherche d'un numéro de téléphone dans l'article
     phone_number = None
     gender = 'unknown'
-    trans_friendly = 'unknown'
+    trans_friendly = ''
-    handles_violence = 'unknown'
+    handles_violence = ''
     visio_meeting = 'no'
-    pseudo_science = 'unknown'
+    pseudo_science = ''
-    premenstrual_syndrome = 'unknown'
+    premenstrual_syndrome = ''
-    accessible_cabinet = 'unknown'
+    accessible_cabinet = ''
-    tatoo = 'unknown'
+    tatoo = ''
-    toxico = 'unknown'
+    toxico = ''
-    sterilisation = 'unknown'
+    sterilisation = ''
-    abortion = 'unknown'
+    abortion = ''
-    endometriosis = 'unknown'
+    endometriosis = ''
-    bigbody = 'unknown'
+    bigbody = ''
-    poil = 'unknown'
+    poil = ''
-    bi = 'unknown'
+    bi = ''
+    lesbian = ''
+    pma = ''
+    ist = ''
+    pregnancy = ''
+    ivg = ''
+    generaliste = ''
+    gyneco = ''
+    sage_femme = ''
+    auto_prelev = ''
+    mycoses = ''
+    dyspareunie = ''
+    spoken = '' # langues parlées
+    diu = '' # dispositif intra utérin
+    puma= '' # PUMA (ex-CMU)
+    ame= '' # AME (Aide médicale d’état)
+    code_postal = ''
 
     for line in article.stripped_strings:
+        found = extraire_code_postal(line)
+        if found:
+            code_postal = found
+            address = line
         found = extraire_numero_telephone(line)
         if found:
             print(found)
@@ -111,6 +140,38 @@ for article in articles:
             bi = 'yes'
         if 'Stérilisation' in line:
             sterilisation = 'yes'
+        if 'DIU' in line:
+            diu = 'yes'
+        if 'Accompagnement grossesse' in line:
+            pregnancy = 'yes'
+        if 'suivi des IST' in line:
+            ist = 'yes'
+        if 'IVG' in line:
+            ivg = 'yes'
+        if 'Médecin généraliste' in line:
+            generaliste = 'yes'
+        if 'Sage-femme' in line:
+            sage_femme = 'yes'
+        if 'auto prélèvement' in line:
+            auto_prelev = 'yes'
+        if 'Conseils mycoses' in line:
+            mycoses = 'yes'
+        if 'Lesbiennes friendly' in line:
+            lesbian = 'yes'
+        if 'PMA' in line:
+            pma = 'yes'
+        if 'PUMA' in line:
+            puma = 'yes'
+        if 'AME' in line:
+            ame = 'yes'
+        if 'français, anglais' in line:
+            spoken = 'french;english'
+        if 'espagnol' in line:
+            spoken = (spoken+';spanish')
+        if 'Dyspareunie' in line:
+            dyspareunie = 'yes'
+        if 'Gynécologue' in line:
+            gyneco = 'yes'
         # chercher une adresse si on en a pas
         if not address:
             found = extraire_addr_line(line)
@@ -123,6 +184,7 @@ for article in articles:
     doctor = {
         "name": name,
         "address": address,
+        "address:code_postal": code_postal,
         "ref:FR:convention_secteur": sector,
         "contact:phone": phone_number,
         "contact:website": url,
@@ -131,8 +193,10 @@ for article in articles:
         "handles:gender:trans": trans_friendly,
         "handles:violence": handles_violence,
         "handles:premenstrual_syndrome": premenstrual_syndrome,
+        "handles:IST": ist,
         "accessible_cabinet": accessible_cabinet,
         "pseudo_science": pseudo_science,
+        "speaks": spoken,
         "handles:tatoo": tatoo,
         "handles:toxico": toxico,
         "handles:sterilisation": sterilisation,
@@ -142,6 +206,18 @@ for article in articles:
         "handles:hairy": poil,
         "handles:bigbody": bigbody,
         "handles:gender:bi": bi,
+        "handles:diu": diu,
+        "handles:mycoses": mycoses,
+        "handles:pregnancy": pregnancy,
+        "handles:abortion": ivg,
+        "handles:auto_prelevement": auto_prelev,
+        "handles:pma": pma,
+        "handles:dyspareunie": dyspareunie,
+        "handles:gender:lesbian": lesbian,
+        "healcare:generaliste": generaliste,
+        "healcare:sage_femme": sage_femme,
+        "healcare:gynecologist": gyneco,
+
     }
     doctors.append(doctor)