strasbourg: add script to convert their current forms to wcs

strasbourg/reprise.py

@@ -0,0 +1,133 @@
+#! /usr/bin/env python
+# -*- coding: utf-8 -*-
+
+# fonctionnalités manquantes :
+#  - fieldset repeat
+#  - choicegroup (uniquement présent dans 104)
+#  - condition
+
+import os
+import sys
+import xml.etree.ElementTree as ET
+import csv
+
+import wcs
+from wcs.formdef import FormDef
+from wcs import fields
+
+def fix_encoding(x):
+    # their encoding process is real garbage, it looks like there's a
+    # broken post-processing step "beautifying" characters like quotes
+    # but it operates on latin1 bytes and fed utf8 bytes.
+    x = x.replace(u'\xc5\u2019', unicode('Œ', 'latin1'))
+    x = x.replace(u'\xc5\u201c', unicode('Œ', 'latin1'))
+    x = x.replace(u'\xc3\u201a', unicode('Â', 'latin1'))
+    x = x.replace(u'\xc3\u2030', unicode('É', 'latin1'))
+    x = x.replace(u'\xc3\u017d', unicode('Î', 'latin1'))
+    x = x.replace(u'\xc3\u20ac', unicode('À', 'latin1'))
+    x = x.replace(u'\xe2\u20ac\xa6', unicode('…', 'latin1'))
+    x = x.replace(u'\xe2\u20ac\u2122', "'")
+    try:
+        return x.encode('raw_unicode_escape').decode('utf-8').encode('utf-8')
+    except:
+        import pdb; pdb.set_trace()
+        raise
+
+tree = ET.parse(sys.argv[1])
+root = tree.getroot()
+
+formdef = FormDef()
+formdef.name = fix_encoding(root.attrib['name'])
+try:
+    formdef = [x for x in FormDef.select() if x.name == formdef.name][0]
+except IndexError:
+    pass
+print formdef.name
+formdef.fields = []
+
+field_id = 0
+
+def process_children(node):
+    global field_id
+    for field_node in node.getchildren():
+        field_id += 1
+        if field_node.tag in ('action', 'label'):
+            continue
+        if field_node.tag == 'title':
+            formdef.fields.append(fields.TitleField(id=str(field_id), type='title', label=fix_encoding(field_node.text)))
+            continue
+        if field_node.tag == 'htmlsection':
+            formdef.fields.append(fields.CommentField(id=str(field_id), type='comment', label='<div>%s</div>' % fix_encoding(field_node.text)))
+            continue
+        if field_node.tag == 'fieldset':
+            if 'repeat' in field_node.attrib:
+                print '!!! fieldset repeat'
+            try:
+                formdef.fields.append(fields.SubtitleField(id=str(field_id),
+                    type='subtitle',
+                    label=fix_encoding(field_node.find('label').attrib['plain'])))
+            except AttributeError:
+                pass
+            process_children(field_node)
+            continue
+
+        if field_node.tag != 'field':
+            raise Exception('unhandled tag type: %r' % field_node.tag)
+        field_type = field_node.attrib['type']
+        validation = field_node.find('validation')
+        if validation is not None and validation.attrib.get('type') == 'validate-email':
+            field_type = 'email'
+        if validation is not None and validation.attrib.get('type') == 'validate-datecal':
+            field_type = 'date'
+
+        if field_type == 'hidden':
+            continue
+
+        field = {
+            'textinput': fields.StringField,
+            'textarea': fields.TextField,
+            'email': fields.EmailField,
+            'checkbox': fields.ItemField,
+            'radio': fields.ItemField,
+            'select1': fields.ItemField,
+            'date': fields.DateField,
+            'fileupload': fields.FileField,
+        }[field_type](id=str(field_id),
+                label=fix_encoding(field_node.find('label').attrib['plain']))
+        formdef.fields.append(field)
+
+        if field_type == 'radio':
+            field.show_as_radio = True
+
+        if validation is not None and validation.attrib.get('required') == 'required':
+            field.required = True
+
+        if validation is not None and validation.attrib.get('type') == 'validate-custom':
+            if validation.text:
+                assert field_type == 'textinput'
+                field.validation = fix_encoding(validation.text[1:-1])
+
+        if field_node.attrib.get('size'):
+            size = field_node.attrib['size']
+            assert size.endswith('px')
+
+        if len(field_node.findall('hint')):
+            field.hint = fix_encoding(field_node.find('hint').text or '')
+
+        if field_type in ('checkbox', 'radio', 'select1'):
+            field.items = []
+            for choice in field_node.findall('choices/choice'):
+                field.items.append(fix_encoding(choice.find('label').attrib['plain']))
+
+process_children(root)
+
+for line in csv.reader(file(os.path.join(os.path.dirname(sys.argv[1]), 'formulaires_prod_reprise_données.csv')),
+        delimiter=','):
+    if line[0] + '.xml' == os.path.basename(sys.argv[1]):
+        formdef.fields.insert(0,
+                fields.CommentField(id=str(field_id+2), type='comment',
+                    label='<p>Origine: <a href="%s">%s</a></p>' % (line[7], line[7])))
+        formdef.workflow_options = {'email': line[8]}
+        break
+
+formdef.store()