handle texts

bot.py

@@ -53,6 +53,7 @@ class MmBot(irc.IRCClient):
         self.nickname = self.botconf['nickname']
         self.codes = legifrance.legi('legifrance.conf')
         self.view = {}
+        self.view_content = {}
 
     def connectionMade(self):
         irc.IRCClient.connectionMade(self)
@@ -86,6 +87,10 @@ class MmBot(irc.IRCClient):
         rep+= u"   [Name Article], [Article du Name], ou [Article de la Name]\n"
         rep+= u"   alors le lien vers legifrance est affiché.\n"
         rep+= u"   Exemple: [CP 323-1], [323-1 du CP] ou [30 de la LCEN].\n"
+        rep+= u"De même quand il est rencontré un motif de la forme {Name Article},\n"
+        rep+= u"   {Article du Name} ou {Article de la Name}, alors le contenu\n"
+        rep+= u"   de l'article est affiché\n"
+        rep+= u"   Exemple: {CP 323-1}, {323-1 du CP} ou {30 de la LCEN}.\n"
         rep+= u"Commandes disponnibles en query:\n"
         rep+= u" - help\n"
         rep+= u" - list\n"
@@ -207,11 +212,11 @@ class MmBot(irc.IRCClient):
             # coeur de metier
             m = msg
             reg = r'(?:'
-            reg+= r'\[(?P<codename1>[A-Za-z0-9]+) (?P<article1>[^ \]]+)\]'
+            reg+= r'(?P<mark1a>(?:\[|\{))(?P<codename1>[A-Za-z0-9]+) (?P<article1>[^ \]]+)(?P<mark1b>(?:\]|\}))'
             reg+= r'|'
-            reg+= r'\[(?P<article2>[^ \]]+) du (?P<codename2>[A-Za-z0-9]+)\]'
+            reg+= r'(?P<mark2a>(?:\[|\{))(?P<article2>[^ \]]+) du (?P<codename2>[A-Za-z0-9]+)(?P<mark2b>(?:\]|\}))'
             reg+= r'|'
-            reg+= r'\[(?P<article3>[^ \]]+) de la (?P<codename3>[A-Za-z0-9]+)\]'
+            reg+= r'(?P<mark3a>(?:\[|\{))(?P<article3>[^ \]]+) de la (?P<codename3>[A-Za-z0-9]+)(?P<mark3b>(?:\]|\}))'
             reg+= r')'
             while True:
                 a = re.match('.*?%s'%reg, m)
@@ -222,25 +227,41 @@ class MmBot(irc.IRCClient):
                     codename = d['codename1']
                     article = d['article1']
                     form = '%s %s' % (codename, article)
+                    marks = (d['mark1a'],d['mark1b'])
                 if not d['codename2'] is None:
                     codename = d['codename2']
                     article = d['article2']
                     form = '%s du %s' % (article, codename)
+                    marks = (d['mark2a'],d['mark2b'])
                 if not d['codename3'] is None:
                     codename = d['codename3']
                     article = d['article3']
                     form = '%s de la %s' % (article, codename)
-
-                trop_jeune = False
-                if self.view.has_key( (channel,codename,article) ):
-                    if time.time()-self.view[(channel,codename,article)] < int(self.botconf['remain_time']):
-                        trop_jeune = True
-
-                if not trop_jeune:
-                    link = self.codes.get(codename,article)
-                    if not link is None:
-                        self.msg(channel,'[%s]: %s' % (form, link))
-                        self.view[(channel,codename,article)] = time.time()
+                    marks = (d['mark3a'],d['mark3b'])
+
+                if (marks[0] == '{' and marks[1]=='}') or (marks[0]=='[' and marks[1]=']') :
+
+                    trop_jeune = False
+
+                    if marks[0] == '{':
+                        used_view = self.view_content
+                    else:
+                        used_view = self.view
+
+                    if used_view.has_key( (channel,codename,article) ):
+                        if time.time()-used_view[(channel,codename,article)] < int(self.botconf['remain_time']):
+                            trop_jeune = True
+
+                    if not trop_jeune:
+                        if marks[0] == '[':
+                            link = self.codes.get(codename,article)
+                            if not link is None:
+                                self.msg(channel,'[%s]: %s' % (form, link))
+                        else:
+                            txt = self.codes.get_content(codename,article)
+                            if not txt is None:
+                                self.msg(channel,'{%s}:\n%s' (form, txt))
+                    used_view[(channel,codename,article)] = time.time()
                 m= re.sub(reg,'',m,1)
 
 

legifrance.py

@@ -29,6 +29,7 @@ import requests
 import re
 import time
 import configobj
+import html2text
 
 def get_articles_from_page(link):
     articles={}
@@ -50,6 +51,19 @@ def get_articles_from_page(link):
 
     return articles
 
+def get_article_content(link):
+    r = requests.get(link)
+
+    reg = '^.*<div class="corpsArt">(.*)</div>.*?Liens relatifs à cet article'
+    c = r.content.decode()
+    a = re.match(reg, c, re.DOTALL)
+    if a:
+        content = a.groups()[0]
+        h = html2text.HTML2Text()
+        h.ignore_links=True
+        return h.handle(content)
+    return None
+
 def get_code(codename,codeids):
 
     if not codename in codeids:
@@ -116,6 +130,17 @@ class legi:
                 return self.cache[name]['articles'][article]
         return None
 
+    def get_content(self,name,article,irc_format=False):
+        link = self.get(name,article)
+        if link is None:
+            return None
+        txt = get_article_content(link)
+        txt = re.sub('^\n+','',txt)
+        txt = re.sub('\n+$','',txt)
+        txt = re.sub('\n{3,}','\n\n',txt)
+        txt = re.sub('\n\n','\n \n',txt)
+        return txt
+
     def force_ct_reload(self,name):
         if name in self.codes:
             self.codes[name]['timestamp']=0